The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Federal Reserve Bank of New York Summer-Fall 1994 Volume 19 Quarterly Review 1 The Credit R ating Industry Richard Cantor and Frank Packer 27 The Price R isk of Options Positions Arturo Estrella, Darryll Hendricks, John Kambhu, Soo Shin, and Stefan Walter 44 M aking Sense of the Profits of Foreign Firms in the United States D avidS. Las ter and Robert N. McCauley 76 The Baby Boom Generation and A ggregate Savings Richard Cantor and Andrew Yuengert 92 M ortgage Security H edging and the Yield Curve J u lia D. Fernald, Frank Keane, and Patricia C. Mosser 101 Has Excess Capacity Abroad Reduced U.S. Inflationary Pressures? James A. Orr 107 Recent Trends in the Profitability of Credit Card Banks Andrea Meyercord 112 Regional Employment Trends in the Second District Charles Steindel and Lois Banks Number 2 Federal Reserve Bank of New "York Quarterly Review Summer-Fall 1994 Table o f Contents A Volume 19 Number 2 r t ic l e s 1 T he C r e d it R a t in g I n d u s t r y Richard Cantor and Frank Packer Investors and regulators have been increasing their reliance on the opinions of the credit rating agencies. This article shows that although the ratings provide accurate rank-orderings of default risk, the meaning of specific letter grades varies over time and across agencies. N oting that current regulations do not explicitly adjust for agency differences, the authors argue that a reassessment of the use of ratings and the adequacy of public oversight is overdue. 27 T he P r ic e R is k of O p t io n s P o sit io n s Arturo Estrella, Darryll Hendricks, John Kambhu, Soo Shin, and Stefan Walter This article evaluates supervisory approaches to the measurement and capital treatment of the price risk of options positions. The authors find that approximate value-at-risk rules tend to provide better estimates of potential losses than simple strategy-based rules. The value-at-risk rules are particularly effective when they adjust for nonlinear changes in options prices. The authors also consider the reporting burdens posed by the different approaches and the consistency of the rules with existing and proposed supervisory frameworks. 44 M a k in g Se n se of th e P r o f it s of F o r e i g n F ir m s in t h e U n it e d S ta tes DavidS. Las ter and Robert N . McCauley The scant profit of foreign firms operating in the United States has prom pted many observers to suggest that these firms understate their income to reduce their U .S. tax liability. This article offers an alternative expla nation: a surge in foreign acquisitions of U .S. firms in the 1980s drove down the average profitability of foreign-owned firms in this country. The authors also find that high leverage and transfer pricing have con tributed to the low U .S. profits of foreign firms. 76 T h e B a b y B o o m G e n e r a t io n a n d A g g r e g a t e S a v in g s Richard Cantor and Andrew Yuengert Many analysts predict a resurgence in national savings as baby boomers approach retirement. This analysis of demographic trends and survey measures of savings and income suggests that such expectations may be ill founded. Although baby boomer saving rates will likely rise over the next twenty years, aggregate saving may not increase because other, low-saving age groups will be claim ing an increasing share of the population. Moreover, despite a reputation for free spending, many baby boomers have accumulated substantial savings already and may not raise their saving rates aggressively in their later years. I n B r ie fs 92 M ortgage S e c u r it y H Y e d g in g a n d t h e ie l d C urve Ju lia D. Fernald, Frank Keane, and 'Patricia C. Alosser The authors find that the use of Treasury securities to hedge mortgage-backed security extension risk may have magnified increases in long-term interest rates after the tightening of monetary policy in early 1994. Substantial increases in the duration of m ortgage securities appear to have caused realignments of hedges and portfolios that, in turn, had a significant im pact on the short-run movements of the Treasury market, partic ularly for ten-year securities. This phenomenon may have altered the short-run dynamics of the yield curve and thus changed the transmission of monetary policy. 101 H as E x c e s s C a p a c it y A b r o a d R educed U.S. I n f l a t i o n a r y P r e s s u r e s ? James A. Orr This article examines whether the sizable amount of excess capacity abroad in recent years has eased U.S. inflationary pressures by keeping im port prices from rising as fast as the prices of U.S.-produced goods. The analysis finds that overall import price growth has roughly kept pace with U .S. inflation because the effects of lower inflation abroad have been offset by exchange rate changes. In the case of Japan, dollar depreciation has rendered excess capacity basically ineffective against U.S. inflationary pressures. 107 R ecent T r e n d s in t h e P r o f it a b il it y of C r e d it C a r d B a n k s Andrea Meyercord Com petition among credit card issuers has increased sharply over the past few years. Despite this trend, the profitability of credit card banks not only remains high relative to the rest of the banking industry but con tinues to grow. This article examines recent trends in credit card bank profitability and, by looking beyond the aggregate data, uncovers some important differences between credit card banks owned by bank holding companies and those owned by nonbank firms. 112 R e g io n a l E m plo ym en t T r e n d s in t h e Second D is t r ic t Charles Steindel and Lois Banks It is well known that job growth in the Second D istrict as a whole has not kept pace with national trends over the last few years. This article offers a different perspective by assessing job trends in specific regions within the District. The authors conclude that employment growth has resumed in most of the D istrict and that in areas such as Northern New Jersey and Albany, the gaps with the national data are either small or narrowing. The areas in the District that have been lagging are mainly those feeling the effects of corporate restructuring and defense cuts— ills that should abate over time. Note to Q u a r t e r l y R e v ie w With this issue o f the Readers: Q u a r t e r l y R e v ie w , w e introduce a new cover design a n d page form at. Our intention is to make the publication visually more invitin g a n d the information in the articles easier to absorb. The new cover preserves the sim plicity o f our earlier design but is brighter a n d more arresting. In recognition o f the contributions o f in dividual authors to this publication, we have also chosen to include the authors’ names along w ith the titles o f their articles on the cover o f the Q u a r t e r l y R e v ie w . We have opened up the page layouts a n d chosen a new typeface that is at once elegant an d h igh ly readable. Pullout quotations in each article give emphasis to the author’s findings, a n d color highlights the principal headings. We have stream lined the graphics in our articles, using the w hite o f the page as a backdrop fo r charts a n d tables. In addition, the charts are now more compatible in design w ith the tables. C oinciding w ith the change in the Q u a r te rly ^ ph ysical appearance is a change in our method o f documentation. We have replacedfootnotes with author-date citations in the text that are keyed to a list o f references at the end o f each article. Explanatory notes w ill appear as endnotes. We hope that you w ill be pleased w ith the changes in our publication. The Credit Rating Industry Richard Cantor and Frank Packer A s financial market complexity and borrower are used as the basis o f most investor guidelines and govern diversity have grown over time, investors and ment regulations, the variations in meaning could have seri regulators have increased their reliance on the ous im p licatio n s. M oreover, as the num ber o f agencies opinions of the credit rating agencies. At the increases, differences in ratings may encourage borrowers to same time, the number of rating agencies operating in the “shop” for the most favorable ratings. In light of the possi United States and abroad has risen sharply. Together, these b ilitie s for ratin g s m isu se, the current reevaluation o f trends have prompted market participants and policymakers ratings-dependent regulations and the adequacy of public to reassess the perform ance o f the agencies and the adequacy oversight seems well justified. of public oversight of the ratings industry. This article pro ratings-dependent regulations, and the reliability and com T h e E v o l u t io n a n d E c o n o m ic s of th e R a tin g s I n d u st r y R a t in g A g e n c y O r ig in s , O w n e r s , a n d S ym bols parability o f the agencies’ ratings. We examine the corre The precursors of bond rating agencies were the mercantile spondence of ratings with default rates and report differences credit agencies, which rated merchants’ ability to pay their among major agencies in their ratings for junk bonds, inter financial obligations. In 1 8 4 1 , in the wake of the financial cri national banks, and mortgage-backed securities. sis o f 1 8 3 7 , Louis Tappan established the first mercantile vides background for such a reassessment by investigating the evolution and economics of the industry, the growth of Our findings raise several questions about the cur credit agency in N ew York. R ob ert D un su b seq u en tly rent uses o f ratings. W hile the agencies provide accurate acquired the agency and published its first ratings guide in rank-orderings of default risk, the meanings o f specific rat 1859- A ings vary over time and across agencies. Since these ratings 1849 sim ilar m ercantile rating agency was formed in by John Bradstreet, who published a ratings book in F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 1 1857. In 1933, the two agencies were consolidated into Dun contradict the agencies’ ratings. To the extent that the analy and B r a d s tr e e t, w hich becam e the ow ner o f M o o d y ’s ses underlying these recommendations are made public, they Investors Service in 1962. provide alternative perspectives to the ju dgm en ts o f the The expansion of the ratings business to securities rating agencies. ratings began in 1909 when John Moody started to rate U.S. As capital flows in international financial markets railroad bonds. A year later, Moody extended his ratings have shifted from the banking sector to capital markets, cred activity to utility and industrial bonds. Poor’s Publishing it ratings have also begun to make a mark overseas. Credit Company issued its first ratings in 1916, Standard Statistics ratings are in use in the financial markets of most developed Company in 1922, and the Fitch Publishing Company in economies and several em erging market countries as well 1924. The number of bond rating agencies in the U.S. revert (Dale and Thom as 1991)- W ith demand rising in foreign ed to three when Standard Statistics and Poor’s Publishing countries, the number of foreign-based rating agencies has Com pany m erged to form Standard and Poor’s (S& P) in increased. Along with the four largest U.S. raters, one other 1941. The most significant new entry in the United States U .S., one British, two Canadian, and three Japanese firms are since that time has been the Chicago-based D uff and Phelps, listed among the world’s “most influential” rating agencies which began to provide bond ratings for a wide range of com by the Financial Times in its publication Credit Ratings Inter panies in 1982, although it had researched public utility national. The principal characteristics of all eleven agencies companies since 1932. Another m ajor ratings provider— are reported in Table 1. McCarthy, Crisanti, and Maffei— was founded in 1975 and The ownership structures of the U.S. rating agencies acquired by Xerox Financial Services before its fixed income do not generally present serious conflict of interest problem s.1 rating and research service was merged into D uff and Phelps The major agencies are all either independent or owned by in 1991. nonfinancial companies, though two had until recently been The four major rating agencies face additional com owned by financial companies. M oody’s is a subsidiary of Dun petition from more specialized agencies. For example, Thom and Bradstreet, which dominates the market for commercial son Bankwatch and IBCA in the United States exclusively credit ratings. Standard and Poor’s is a subsidiary of McGraw- rate financial institutions, and A.M . Best rates insurance H ill, a m ajor publishing company with a strong business companies’ claims-paying abilities. More generally, the ana information focus. Fitch, initially a publishing company, was lysts employed by many financial institutions regularly make bought by an independent investors group in 1989- D uff and recommendations to buy or sell that im plicitly confirm or Phelps Credit R atings is a subsidiary of D u ff and Phelps, Table 1 S e l e c t e d B o n d R a t i n g A g e n c ie s ■■... ■...v . . . . Year Ratings First Published Credit Rating Agency Home Country Year o f SEC Designation Employees Ownership Principal Ratings Areas 19 0 9 M oody’s Investors Service (“M oody’s”) U.S. 1975 674 Dun and Bradstreet Full service 1922 Fitch Investors Service (“Fitch”) U.S. 1975 200 + Independent Full service 19 2 3 19 7 2 Standard and Poor’s Corporation (“S&P") U.S. 700+ M cG raw-H ill Full service Canadian Bond Rating Service (“CBRS") Canada 1975 N .A. 26 Independent Full service (Canada) 19 7 4 Thomson BankW atch (“Thom ”) U.S. 40 Thomson Company Financial institutions 1975 Japanese Bond Rating Institute (“JB R I”) Japan 19 9 1 N.A. 91 Japan Economic Journal (Nikkei) Full service (Japan) 19 7 7 D ominion Bond Rating Service (“D BRS”) Canada N .A. 20 Independent Full service (Canada) 19 7 8 IBCA, Ltd. (“IB C A”) U .K . 19 9 0 50 Independent Financial institutions 19 8 0 D uff and Phelps Credit Rating Co. (“D u ff’) U.S. 1982 Full service Japanese Credit Rating Agency (“J C R A ”) Japan N .A. 160 61 D uff and Phelps Corp. 1985 Financial Institutions Full service (Japan) 1985 Nippon Investor Service Inc. (“NIS”) Japan N.A. 70 Financial Institutions Full service (Japan) 1975 McCarthy, Crisanti, and Maffei (“M CM ”) (no longer in operation) U.S. 19 8 3 N .A. Acquired by D uff and Phelps in 19 9 1 Full service (U.S.) 2 FRBN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Inc., whose affiliates offer investment management, financial the performance risk o f m ortgage servicers, and the price consulting, and investment research services. By late 1994, volatility of mutual funds and mortgage-backed securities. however, D u ff and Phelps C redit R atin gs is expected to Increased foreign demand has also led to a dramatic become an independent company as its shares are spun off to overseas expansion of the established U .S. rating agencies. the shareholders of D uff and Phelps, Inc., itself a closely held Over the past ten years, M oody’s has opened offices in Tokyo, company. Thomson Bankwatch was a subsidiary of Keefe, London, Paris, Sydney, Frankfurt, and Madrid, and now rates Bruyette, and Woods, a brokerage firm, until March 1989, the securities of approximately 1,200 non-U.S. issuers (out of when it was sold to the Thomson Corporation, a large private more than 4 ,500 total). Standard and Poor’s has set up offices international publishing conglomerate. Most of the non-U.S. in Tokyo, London, Paris, M elbourne, Toronto, Frankfurt, firms are also independent. The London-based rating agency, Stockholm, and Mexico City, and has established affiliations IBCA, is independently owned, as are the two Canadian rat or acquired local rating agencies in Sweden, Australia, Spain, ing agencies. Two of the rating agencies from Japan, however, and Mexico. D u ff and Phelps has formed joint ventures in are owned by consortia of financial institutions, including M exico and several other Latin Am erican countries. The some for which credit ratings are issued. established U .S. agencies appear to have a competitive advan Over time, the agencies have expanded the depth and frequency of their coverage. The four leading U.S. credit tage over their foreign counterparts in the business of provid ing independent, credible securities ratings. rating agencies rate not only the long-term bonds issued by The bond ratings assigned by all the rating agencies U .S. corporations, but also a w ide variety o f other debt are meant to indicate the likelihood of default or delayed pay instruments: municipal bonds, asset-backed securities, pre ment of the security. Most of the rating agencies have long ferred stocks, medium -term note programs, shelf registra had their own system of symbols— some using letters, others tions, private placements, commercial paper programs, and using numbers, many both— for ranking the risk of default bank certificates of deposit. More recently, ratings have been from extremely safe to highly speculative. Gradually, howev applied to other types of risks, including the counterparty er, a rough correspondence among the major agencies’ ratings risk posed by derivative products companies and other insti has emerged (Table 2).2 To provide finer rating gradations to tutions, the claim s-paying ability of insurance companies, help investors distin gu ish more carefully am ong issuers, Table 2 L o n g -T erm S enior D e bt R a ting S ymbols Investment Grade Ratings Speculative Grade Ratings S&P and others M oody’s Interpretation AAA Aaa Highest quality S&P and others M oody’s Interpretation BB + Bal Likely to fulfill BB Ba2 obligations; ongoing BB- Ba3 uncertainty AA + AA A al B+ B1 High risk Aa2 B B2 obligations AA- Aa3 B- B3 A+ A1 Strong payment A A2 capacity A- A3 High quality CCC+ ccc Current vulnerability Caa to default, or in CCC- default (Moody’s) BBB + Baal Adequate payment c Ca In bankruptcy or BBB Baa2 capacity D D default, or other BBB- Baa3 marked shortcoming Notes: The other agencies listed in Table 1 use the rating symbols o f the first column, with the exception o f DBRS (H and L symbols in place o f + and —) and CBRS (H and L symbols in place o f + and —, and + symbols that correspond to second and third letters). The agencies follow a variety o f policies w ith respect to the number o f ratings symbols given below B—. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 3 Fitch in 1973, Standard and Poor’s in 1974, and Moody’s in Agencies charge fees that vary with the size and type 1982 started attaching plus and minus symbols to their rat of issue, but a representative fee on a new long-term corporate ings. Other modifications of the grading schemes — includ bond issue ranges from 2 to 3 basis points of the principal for ing the addition of a “credit watch” category to denote that a each year the rating is maintained. Normally, the charge for rating is under review — have also become standard. In the any one bond issue has both a floor and a ceiling, and negoti remainder of this article, the symbols currently employed by ated rates are available for frequent issuers. For issuers o f com Standard and Poor’s, Fitch, D uff and Phelps, and others are mercial paper, M oody’s and Standard and Poor’s m aintain used to refer to the ratings of all agencies. quarterly charges based on amounts outstanding (up to 7 basis points) plus an annual fee. T h e T r a n s it io n t o C h a r g in g I ssuers a n d t h e R ole o f R epu ta tio n encourage agencies to assign higher ratings to satisfy issuers, Agencies initially provided public ratings of an issuer free of the agencies have an overriding incentive to maintain a repu charge, and financed their operations solely through the sale tation for high-quality, accurate ratings. If investors were to of publications and related materials. However, the publica lose confidence in an agency’s ratings, issuers would no longer tions, which were easily copied once published, did not yield believe they could lower their funding costs by obtaining its sufficient returns to justify intensive coverage. As the de ratings. As one industry observer has put it, “every time a rat mand on rating agencies for faster and more comprehensive ing is assigned, the agency’s name, integrity, and credibility service increased, the agencies began to charge issuers for rat are on the line and subject to inspection by the whole invest- W hile the current payment structure may appear to ings. They then used these revenues to expand services and products and to compete with private sector analysts at other financial institutions. The default of Penn Central on $82 million of com mercial paper in 1970 was a catalyst in the transition to W bile the cu rren t p a ym en t stru ctu re m ay appear to encourage agen cies to a ssign h igh er ra tin gs to charging issuers. The commercial paper market had grown very rapidly in the 1960s with little regard for credit quality. sa tisfy issuers, the agen cies h a ve an o verrid in g Investors tended to assume that any firm with a household in cen tive to m ain ta in a reputation f o r h ig h - name was an acceptable credit risk. W hen Penn Central defaulted during the 1970 recession, investors began to ques q u ality, a ccu ra te ratings. tion the financial condition of many companies and refused to roll over their commercial paper. Facing a liquidity crisis, many of these companies also defaulted. To reassure nervous ment comm unity” (Wilson 1994). Over the years, the disci investors, issuers actively sought credit ratings, and it be pline provided by reputational considerations appears to have came established market practice that new debt issues com been effective, with no major scandals in the ratings industry ing to m arket have at least one credit ratin g. W ith the of which we are aware.1 demand for rating services rising, the agencies found they In addition to putting an agency’s reputation at risk, were able to impose charges on issuers. Fitch and M oody’s inaccurate ratings m ight expose the agency to costly legal started to charge corporate issuers for ratings in 1970, and dam ages. However, the threat of legal liability for rating Standard and Poor’s followed suit a few years later. (Standard agencies has not yet materialized. Class action suits have been and Poor’s started to charge municipal bond issuers for rat brought against rating agencies following major failures — ings in 1968.) Now, according to one estimate, roughly four- such as the Washington Public Power Supply System default fifths of Standard and Poor’s revenue comes from issuer fees in 1983 and the Executive Life bankruptcy in 1991— but the (Ederington and Yawitz 1987). cases were dropped before verdicts were reached. Digitized 4 for FRASER FRBN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 T he R a t in g s P r o c e s s and U n so l ic it e d R a t in g s ings in these areas are often substantially lower than the The process of obtaining a rating can be lengthy, requiring solicited ratings and can affect the yield paid at issuance. Pro significant time and effort on the part of the debt-issuer and ponents claim that unsolicited ratings provide a powerful its underwriter as well as the agency. The agencies base their check against rating shopping, the practice of hiring only ratings on both quantitative and qualitative assessments of those agencies that offer favorable ratings. Critics complain the borrowing company’s condition and the special provi that unsolicited ratings are based on incomplete information, sions of the particular security at hand. A staff committee at because communication with the issuer is limited. Although the agency usually votes on a recommendation by a senior an agency assigning unsolicited ratings may appear to have analyst after presentation and debate. The rating assigned, an incentive to be unduly conservative so as to reward those often accompanied by explanatory analysis, is first communi firms that do pay for its ratings, this incentive may be offset cated to the issuer and underwriter, and then to the public at by the need to maintain a reputation for analytical credibility large. The issuer frequently has the opportunity to appeal a (Monro-Davis 1994). rating if it is not satisfied, but in general the ratings process is structured to hear the best case the issuers have to present T he U se of R a t in g s in R e g u l a t io n s before the rating is assigned. (More discussion of the informa- Introduced as guides for unsophisticated investors, credit rat tion-gathering and decision process can be found in Wilson ings have acquired several new uses. Many mutual funds and 1994 and Ederington and Yawitz 1987.) pension funds place limits on the amount of a portfolio that The agencies maintain very different policies about can be invested in non-investment-grade securities. Debt assigning ratings not requested by the issuer. Some agencies issuers and investors frequently introduce ratings explicitly will issue ratings only upon request; other agencies will issue into the covenants of their financial contracts and seek guid unsolicited ratings. Standard and Poor’s rates all taxable secu ance from the agencies on the structuring of their financial rities in the U.S. domestic market registered by the Securities transactions. and Exchange Commission (SEC), regardless of whether the As ratings have gained greater acceptance in the rating was requested and paid for by the issuer. Standard and marketplace, regulators of financial markets and institutions Poor’s will not, however, assign unsolicited ratings for struc have increasingly used ratings to simplify the task of pruden tured securities and bonds issued by foreign companies tial oversight. The reliance on ratings extends to virtually all because it views the nonpublic information provided by the financial regulators, including the public authorities that issuer to be essential for analyzing these securities. Moody’s oversee banks, thrifts, insurance companies, securities firms, shares Standard and Poor’s policy of rating all SEC-registered, capital markets, mutual funds, and private pensions. The U.S. co rp o rate se cu ritie s, but Moody’s freq u en tly issu es u n so early re gu lato ry u ses of ra tin g s drew on ly on the agen cy dis licited ratings on structured securities and foreign bonds as tinctions between investment grade securities, or those rated well. In contrast, both Fitch and Duff and Phelps refrain from BBB and above, and speculative securities, those rated BB assigning unsolicited ratings to any security. Moreover, Duff and below. Regulations required that extra capital be held and Phelps will only make a rating public upon the request of against speculative securities or prohibited such investments its client (Ederington and Yawitz 1987). altogether. Although the distinction between investment Moody’s and Standard and Poor’s usually receive fees grade and speculative securities remains an important one, for ratings they would have issued anyway because companies over time, regulatory capital requirements, disclosure re want the opportunity provided by the formal rating process quirements, and investment prohibitions have increasingly to put their best case before the agencies. Moody’s unsolicited been tied to other letter grades as well. The history of selected ratings of issuers of structured securities and foreign bonds uses of ratings by regulators is summarized in Table 3.4 are more controversial because such assessments are not part Since the regulators adopted ratings-dependent of an overall policy to rate all such securities. Unsolicited rat rules, they have had to specify which agencies would qualify FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 5 for consideration under their regulations. The SEC currently to be charged against capital. Similar rules were adopted by designates six agencies as “nationally recognized statistical many state banking departments. rating organizations” (NRSROs), and the other regulators In 1936, the Office of the Comptroller and the Feder generally rely on the SEC’s designations. Given the large al Reserve went further, prohibiting banks altogether from number of designated agencies (and at least as many agencies holding bonds not rated BBB or above by at least two agencies. have applications pending), regulations must include meth The new rules had far-reaching consequences because 891 of ods for dealing with rating disagreements among the agen 1,975 bonds listed on the New York Stock Exchange were cies. Most regulations simply accept either the highest rating rated below BBB in 1936. Still in force for banks today, these or the second highest rating, but the insurance regulators restrictions on investments were extended to thrifts in 1989- conduct independent analyses to resolve disagreements As of the early 1930s, regulators of insurance com among the agencies. The first approach is arbitrary and per panies were relying on ratings to help determine the capital haps inflationary, while the second approach incurs the cost of to be put aside for securities held. In 1951, the National establishing in-house analytical capacity. Association of Insurance Commissioners (NAIC) established a system of internal quality categories in which the top-qual- T r a d i t i o n a l U se o f R a tin g s : D is tin g u is h in g ity classification corresponded to ratings of BBB and above, I n v e s tm e n t G r a d e F ro m S p e c u la tiv e S e c u r itie s effectively establishing uniformity in the definition of On the heels of a sharp decline in credit quality in 1931, the “investment grade” across bank and insturance regulators Office of the Comptroller of the Currency ruled that bank (West 1973).5 holdings of publicly rated bonds had to be rated BBB or bet Regulatory rules based on the distinction between ter by at least one rating agency if they were to be carried at investment grade and speculative securities have since book value; otherwise the bonds were to be written down to expanded. The SEC has required dealers to hold extra capital market value and 50 percent of the resulting book losses were against their inventories of speculative or “junk” bonds since Table 3 S e l e c t e d U se s o f R a t i n g s in R e g u l a t i o n Year Adopted Ratings-Dependent Regulation Minimum Rating How Many Ratings? Regulator /Regulation 19 3 1 Required banks to m ark-to-m arket lower rated bonds BBB 2 OCC and Federal Reserve examination rules 19 3 6 Prohibited banks from purchasing “speculative securities” BBB Unspecified OCC, FDIC, and Federal Reserve joint statement 19 5 1 Imposed higher capital requirements on insurers’ lower rated bonds Various N.A. N AIC mandatory reserve requirements 1975 Imposed higher capital haircuts on broker/dealers’ below-investment-grade bonds BBB 2 SEC amendment to Rule 1 5 c 3 -l: the uniform net capital rule 19 8 2 Eased disclosure requirements for investment grade bonds BBB 1 SEC adoption o f Integrated Disclosure System (Release # 6383) 19 8 4 Eased issuance o f nonagency mortgage-backed securities (MBSs) AA 1 Congressional prom ulgation o f the Secondary Mortgage M arket Enhancement A ct o f 1 9 8 4 19 8 7 Permitted margin lending against MBSs and (later) foreign bonds AA 1 Federal Reserve Regulation T 19 8 9 Allow ed pension funds to invest in high-rated asset-backed securities A 1 Departm ent o f Labor relaxation o f ERISA Restriction (PTE 89-8 8 ) 1989 Prohibited S&Ls from investing in below-investment-grade bonds BBB 1 Congressional prom ulgation o f the Financial Institutions Recovery and Reform Act of 1989 19 9 1 Required money market mutual funds to lim it holdings o f low-rated paper A l* It SEC amendment to Rule 2a-7 under the Investment Company A ct o f 1 9 4 0 19 9 2 Exempted issuers o f certain asset-backed securities from registration as a m utual fund BBB 1 SEC adoption o f Rule 3a-7 under the Investment Company A ct o f 1 9 4 0 19 9 4 Proposal W ou ld impose varying capital charges on banks’ and S&Ls’ holdings o f different tranches o f asset-backed securities AAA & BBB 1 Federal Reserve, OCC, FDIC, OTS Proposed Rule on Recourse and Direct Credit Substitutes * Highest ratings on short-term debt, generally implying an A —long-term debt rating or better. | If issue is rated by only one N RSRO , its rating is adequate; otherwise, two ratings are required. 6 FR BN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 1975. In 1989, Congress passed legislation prohibiting to essentially creating the nonagency mortgage-backed secu thrifts from investing in junk bonds. In 1993, the Basle rities market, SMMEA established a new regulatory cutoff Committee on Bank Supervision proposed in its market risk rating. The higher AA rating was chosen because mortgage- guidelines that internationally active commercial banks deal backed securities with full or partial government backing— ing in securities should hold extra capital against their non- the reference securities to which the new securities were com investment-grade bond inventories as well. (This passage in pared— were virtually all rated AAA or AA at the time. the proposal mirrors a similar statement in the European A few years later, the Federal Reserve Board, which Community’s Capital Adequacy Directive governing the had previously refrained from expanding its use of ratings activities of security dealers domiciled in the Community.) beyond the basic investment grade requirement for bank The achievement of an investment grade rating eases portfolio investments, also began to incorporate an AA cutoff the burden of disclosure for the issuer of the securities. In in certain of its prudential rules affecting bank supervision. 1982, the SEC started to require less detailed disclosure at In recognition of the expanded role given to ratings by the issuance for investment grade securities. In 1993, the SEC Congress, the Board began to use AA as a cutoff in rules for adopted Rule 3a-7, which made the investment grade rating determining the eligibility of mortgage-related securities a criterion for easing the public issuance of certain asset- (1987) and foreign bonds (1989) as collateral for margin backed securities (Cantor and Demsetz 1993). lending.6 Embedding the investment grade distinction in reg The single A rating has also served as a cutoff. The ulations has simplified prudential oversight of financial insti Labor Department, in its role as overseer of the private pen tutions. Some of these regulations have, as a by-product, sion industry, adopted a regulation in 1988 permitting pen adversely affected the availability and cost of funds to below- sion fund investments in asset-backed securities rated single- investment-grade borrowers. West (1973) and Carey et al. A or better (Baron and Murch 1993). The A rating gained (1993) show that spreads rose for borrowers rated BB follow further regulatory importance in 1990 when the N A IC ing the adoption of regulations affecting bank and insurance adopted new capital rules that applied the least burdensome company investments in below-investment-grade securities. capital charge to bonds with the NAIC quality designation corresponding to a public rating of A or above. T he E m e r g e n c e of N ew C u t o f f R a t in g s Short-term ratings too have been important tools of Regulators are increasingly using ratings other than BBB as recent regulation. In 1991, the SEC adopted amendments to thresholds in their rules. Each new regulatory use appears to Rule 2a-7 of the Investment Company Act of 1940 that have encouraged other regulators to expand their reliance on imposed ratings-based restrictions on money market mutual ratings. Some of these new rules have greatly influenced the fund investments.7 Following the adoption of these amend development of capital markets. ments, mutual fund holdings of lower quality paper fell to In 1984, to promote the development of a mortgage-backed securities market without the support of gov- zero, and the total amount of lower quality paper outstanding declined sharply (Crabbe and Post 1992). ernment-related agencies (Government National Mortgage Some regulations have gone beyond specific cutoff Association, Federal National Mortgage Association, and levels by incorporating schedules of multiple rating levels Federal Flome Loan Mortgage Corporation), Congress passed and corresponding restrictions and charges. As part of its the Secondary Mortgage Market Enhancement Act (SMMEA). 1990 reform of rating procedures, the NAIC increased the This act eased issuance and enhanced the marketability of number of its quality categories from four to six and applied mortgage-backed securities rated AAA or AA. In particular, different regulatory restrictions to each category. Four years it allowed these securities to be marketed up to six months in later, the Federal Financial Institutions Examination Council advance of the delivery of their underlying collateral and (1994) joined bank and thrift regulators, including the Fed exempted them from most states’ blue sky laws. In addition eral Reserve, in a proposal to adjust capital charges on deposi F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 7 tory institutions’ holdings of structured securities on the basis of credit ratings. Nonetheless, the informality of the process and the opaqueness of the acceptance criteria raise serious problems. The requirement that an agency be widely used by major T h e D e s ig n a tio n o f N R S R O s investors before it can be designated as an N RSRO clearly Under most current ratings-dependent regulations in the favors incum bents. G iven the grow ing im portance of United States, ratings matter only if they are issued by an NRSRO status, new entrants in the ratings business who lack NRSRO . The SEC first applied the NRSRO designation to agencies in 1975 in referring to agencies whose credit ratings could be used to determine net capital requirements for bro Regulations generally refer directly to NRSRO ker-dealers. Subsequently, the term was taken up by regula rating levels without allowances fo r differences tors other than the SEC and even by the private investment across agencies. community. When the phrase N RSRO was first used, the SEC was referring to the three agencies that had a national pres this status may find it increasingly difficult to attract a wide ence at that time, Moody’s, Standard and Poor’s, and Fitch. following in the investment community. These concerns may But as the public bond market and rating industry grew over become more acute as the SEC considers applications from time, other agencies have sought NRSRO designation from foreign rating agencies. the SEC. In 1982, Duff and Phelps received designation, fol At present, the SEC does not require N RSR O s to lowed by IBCA and Thomson Bank Watch in 1991 and 1992, have uniform rating standards. In particular, the Commission respectively. The designation of the latter two has been limit has no explicit rule that “equivalent” letter grades must cor ed to their ratings for banks and financial institutions only. In respond to similar expected default rates. Nonetheless, regu 1983, the SEC granted NRSRO status to McCarthy, Crisanti, lations generally refer directly to N RSRO rating levels with and Maffei; however, this company’s credit rating franchise out allowances for differences across agencies.8 Unless the was acquired by Duff and Phelps in 1991- At least six foreign way in which regulations use ratings is changed, all N RSRO rating agencies currently have applications outstanding with ratings of a certain level ought to correspond to the same level the SEC for designation as NRSROs. of credit risk. To achieve such consistency, the SEC may have At present, the SEC’s procedures and conditions for to develop additional acceptance criteria and ongoing moni designating agencies as NRSRO s are not very explicit. If a toring capacity. In recognition of these concerns, the SEC has rating agency requests N R SR O status from the SEC, the published a “concept release” that invites rating agencies, SEC’s staff will undertake an investigation, analyzing data corporations, and investors to comment on “the role of rat supplied by the rating agency about its history, ownership, ings in federal securities laws and the need to establish formal employees, financial resources, policies, and internal proce procedures for designating and monitoring the activities of dures. Nevertheless, the principal test applied by the SEC to N RSRO s” (SEC 1994a). any agency seeking N R SR O status is that the agency be “nationally recognized by the predominant users of ratings in R e s o l v in g D is a g r e e m e n t s the United States as an issuer of credible and reliable ratings” A (SEC 1994a). In effect, the SEC requires that the market Most ratings-dependent regulations only require that a bond already place substantial weight on the judgment of a rating issue carry a single N R SR O ’s rating. However, issuers in the agency. Market acceptance is determined by polling on an United States commonly obtain at least two ratings on pub informal basis. By giving the market a role in selecting licly issued securities. Since both Moody’s and Standard and NRSRO s, the SEC intends to weed out agencies that have not Poor’s rate virtually all public corporate bond issues, a dual already established a reputation for accurate ratings. rating is fairly automatic. As a consequence, differences of FR BN Y 8 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 a m o n g the R a t in g g e n c ie s opinion across the rating agencies inevitably arise. Regula the corresponding public credit rating. In practice, the SVO tors have had to find a way to resolve these differences because concentrates its resources on (1) determining a quality cate most of their rules key off specific letter grades. Their gory for unrated private placement securities and (2) resolv approaches to the problem take two forms— explicit rules ing differences of opinion among the agencies, where the and independent analysis. SVO may choose either the higher or lower rating (NAIC The most common approach is to adopt an explicit 1994). At the cost of establishing the capacity to undertake rule, recognizing either the highest or the second highest rat independent analysis, the N A IC has developed a discre ing, regardless of the number or level of the other ratings. tionary use of ratings that calls for judgment in the interpre The second-highest rating rule attempts to strike a balance tation of split ratings and permits certain ratings to be dis between a conservative policy (eliminating the highest rat counted if they are viewed as too high. ing) and a liberal policy (not necessarily using the lowest rat ing). When the ratings industry was dominated by Moody’s and Standard and Poor’s, this rule was effectively conservative T he R e l ia b il it y of R a t in g s In this section, we review the rating agencies’ historical records in measuring relative and absolute risks of corporate bond defaults. Many of the current uses of ratings presume Agency ratings have been a less reliable guide... to absolute credit risks: default probabilities associated with specific letter ratings have drifted over time. accuracy on both counts. To be meaningful, ratings must, at a minimum, provide a reasonable rank-ordering of relative credit risks. In addition, however, ratings ought to provide a reliable guide to absolute credit risk. In other words, the rat ings levels corresponding to regulatory cutoffs should have a fairly stable relationship to default probabilities over time. Our review of the corporate bond defaults data assembled by since the lower of two ratings was also the lowest rating. As Moody’s and Standard and Poor’s suggests that the agencies the number of NRSRO s has increased and issuers have begun do a reasonable job of assessing relative credit risks: lower to obtain three, four, or more ratings, the policy is potentially rated bonds do in fact tend to default more frequently than more liberal. Although regulators could conceivably adopt a higher rated bonds. Agency ratings have been a less reliable more conservative rule (such as the lowest rating), in areas guide, however, to absolute credit risks: default probabilities such as structured finance where Moody’s and Standard and associated with specific letter ratings have drifted over time. Poor’s do not attempt to rate every issue, issuers could re sp o n d by d r o p p in g a g en cies th a t a ssig n e d the low er ra tin g s. Our review is limited to Moody’s and Standard and Poor’s ratings because only these agencies have a long history The second approach, used by the NAIC, resolves of rating a large number of corporate issues. We present data differences of opinion among the rating agencies through primarily from Moody’s because it has published more histor independent analysis. The N A IC ’s Securities Valuation ical data than Standard and Poor’s. By and large, however, we Office (SVO) assigns each bond held by an insurance compa believe that the patterns observed in Moody’s ratings are also ny to one of six quality categories, and each category has a dif present in Standard and Poor’s ratings, and we provide some ferent implication for mandatory reserves. The six quality support for this view in the text. In addition, the analysis is categories are meant to correspond to different N RSRO pub limited to corporate bond ratings and excludes commercial lic ratings. (Category 1 corresponds to AAA,AA, and A; 2 to paper ratings, m unicipal bond ratings, or asset-backed BBB; 3 to BB; 4 to B; and 5 or 6 to CCC,C, or D ratings, bonds. In these other markets, a study of rating reliability is depending on the rating agency.) However, the SVO staff is not possible either because defaults have been too rare, the free to assign a rating that differs from the bond’s public cred data are too hard to obtain, or the history of the market is too it rating as long as their judgment implies a downgrade from short. FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 9 M e a su r in g tion not already embedded in market yields. Even if ratings R e l a t iv e C r e d it R is k s Some very simple tests suggest that the rating industry mea do not contain independent information about credit risk, sures relative credit risks with reasonable accuracy. The capi the use of ratings by investors and regulators may make sense tal markets seem to validate the agencies’ judgments by pric if ratings offer an efficient summary of this information. ing lower rated bonds at higher average yields. Moreover, Measuring ratings performance by contemporane both average short-term and long-term default rates are cor ous market yields, however, does not control for waves of related in a sensible way with credit ratings. This evidence market optimism or pessimism. The accumulation of ex post implies that ratings provide a useful rank ordering of credit evidence on bond performance provides a more precise score risks. card on ratings. Moody’s and Standard and Poor’s have made For U.S. corporate bonds, market yields are general such evidence available in their corporate bond default stud ly closely related to their credit ratings. Table 4 reports the ies, which calculate historical default rates among classes of average yield spreads between corporate bonds and U.S. Trea rated issuers. suries by rating category for issues rated by Standard and These studies indicate that lower corporate bond Poor’s between 1973 and 1987. Each letter grade decline cor ratings have indeed been associated with a higher probability responds to a distinct increase in average yield spreads. The o f default. The results of the M oody’s study (M oody’s pattern of increasing yields as the ratings category is lowered Investors Service 1994) are summarized in Chart 1, which is extremely robust and holds without exception across all reviews the default rates among rated issuers between 1970 years of the sample (Altman 1989)- While this correlation and 1993. The upper left panel in Chart 1 presents the one- may seem unsurprising and perhaps a weak test of ratings year default rate for the entire sample of rated bonds. Mea reliability, Artus, Garrigues, and Sassenou (1993) put forth sured to 1/10 of a percentage point, the one-year default rates evidence that, for the French bond market, a direct relation are zero for all bonds rated A and above. The one-year default ship between yield and the ratings of the largest French bond rate rises to 2/10 of a percentage point for BBB issuers, and rating agency is either weak or nonexistent. 1.8 and 8.3 percent for BB and B rated issuers, respectively. This simple association of yields and ratings in the The other three panels of Chart 1 show how the U.S. bond market need not indicate the presence of a causal default probabilities across Moody’s rating categories change relationship. Rather, it may simply mean that the capital as the time horizon is lengthened to five, ten, and fifteen markets and the rating agencies basically agree on the factors years.9 While the default probability increases with the time that measure credit risk. Although the literature is volumi horizon for each rating category, the negative relation nous (see Ederington and Yawitz 1987), the evidence is between default probability and ratings remains intact. A mixed on whether credit ratings contain additional informa- similar historical default study (Brand, K itto, and Bahar 1994) covering bonds rated by Standard and Poor’s between Table 4 1981 and 1993 basically confirms the conclusions drawn S preads betw een C o rporate B o n d s a n d U.S. T reasuries 1973—87 Averages from the longer term study by Moody’s. Rating AAA Consistent with the traditional importance of the Basis Points investment grade/non-investment-grade distinction, the 43 AA 73 A 99 BBB 1 66 BB B 299 404 CCC 724 probability of default rises most dramatically once the invest ment grade barrier is breached. In the Moody’s study, over a five-year time horizon, the default probability is six times higher for bonds rated BB than for those rated BBB. In con Source: Altm an (198 9 ). Note: Based on equally weighted averages o f monthly spreads per rating cate trast, the comparable ratio of default probabilities for B-rated versus BB-rated issues is much lower at 2.2, as is the ratio for gory. Spreads for BB and B represent data for 1 9 7 9 —87 only; spreads for CCC, BBB-rated versus A-rated issues at 3.2. The same ratios for data for 1 9 8 2 —87 only. 10 FR BN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 the Standard and Poor’s study were 4.8 (BB versus BB B), 3.0 default probabilities at the different ratings levels should (BBB versus A), and 1.9 (B versus BB), respectively. exhibit relative stability over frequencies longer than the business cycle. In fact, legislators and financial regulators are M ea su r in g A bsolute C r e d it R isks presuming such a stability when they embed specific credit The agencies do not intend their ratings to imply precisely rating thresholds into law and regulation. the same default probabilities at every point in time. In par The reliability o f ratings as predictors of absolute ticular, they are reluctant to make ratings changes based sim credit risks can be evaluated by examining the default rates ply on cyclical considerations even though the frequency of associated with different ratings over time, particularly if the defaults within rating categories clearly rises in recessions.10 time horizon is long enough to incorporate both ends of the But even if cyclical variability in short-term default rates is business cycle. Using M oody’s data between 1970 and 1994, an inevitable result of a longer term perspective, long-term Chart 2 reviews the progress of five-year cumulative default Chart 1 A v e r a g e D e f a u l t R a t e s b y C r e d it R a t in g P ercen t 10 One-Year Default Rates: 1 9 7 0 -9 3 AAA AA A BBB BB B AAA P ercent P ercent 50 50 Ten-Year Default Rates: 1 9 7 0 -8 4 AAA AA A Fifteen-Year Default Rates: 1 9 7 0 -7 9 BBB AAA AA A BBB BB B Source: Moody's Investors Service 1994. FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 11 rates for investment grade and non-investment-grade bonds. In retrospect, the rise in default rates is unsurprising The initial spike in 1970 for non-investment-grade bonds given the general deterioration in credit ratios within rating stems from the default that year of Penn Central and twenty- classes that began in the m id-1980s. Chart 3 shows that the six other railroad companies; default rates decreased dramati median fixed-charge coverage and leverage ratios o f indus cally the next year. For cohorts established since January trial firms with B B B , BB , and B credit ratings from Standard 1971, however, the cumulative default rate within all rating and Poor’s generally worsened between 1985 and 1991. classes B B B and below has increased roughly threefold. The These data suggest that a relaxation of credit standards may 1971 to 1989 increase is from 0.4 percent to 0.8 percent for have occurred,11 perhaps as a result of the view commonly A-rated bonds, 1.1 percent to 3.2 percent for BB B-rated held in the late 1980s that even healthy corporations should bonds, 5.1 percent to 19-7 percent for BB-rated bonds, and increase leverage. In sum, the experience since 1970 indicates 11.1 percent to 34.3 percent for B-rated bonds. Five-year that the correspondence of ratings to default probabilities is default rates now lie well above the highs of 1970. subject to substantial change over time. T h ough five-year d efau lt rates rose d u rin g the R a tin g s D ifferen ces across A g en cies growth of the junk bond market in the 1980s, deterioration in performance was common to both investment grade and Differences among the agencies over specific ratings are com non-investment-grade samples. The increase in default rates mon, unavoidable, and even desirable to the extent that dis actually began with the 1 976, 1977, and 1978 cohorts, agreem ents prom ote better understanding. N onetheless, whose five-year defaults rates incorporated defaults that these differences can be highly problematic for ratings-based occurred through the end of 1980, 1981, and 1982, respec regulations in which the ratings o f any two N R SR O s are sub tively. The rising trend in default rates, therefore, was initial stitutable. Some of the observed differences can be attributed ly related to the early 1980s recession but continued on to alternative rating methodologies; others are the results of through the decade. the judgmental element in the ratings process. Many of the Chart 2 T r e n d s in F iv e -Y e a r D e f a u lt R a te s b y C r e d it R a tin g Percent Percent 35 “ 1 BBB Irivestment-Grade Is suers B elow-I nvestment-(jrad e Issuers A 1 b 1 1 25 \ #aa»• ♦ ♦ A ♦■if• ♦ • • i i 1970 72 * ♦ ♦ ♦ * # • • 76 78 1 15 1 V I1 \ ¥ K *T ' BB y\ < K r r __ i . ____ 1 1 1 1' * 10 * i — l/ . - ? 74 •• • • • * .. 1 •• • • • • \ 20 y ___ A’ .___ . I 1 1 #> • * r / / 30 .V p i./\ i i 80 82 84 « 86 * i i 88 89 0 L 1970 . . .. i....i .. 72 74 t 76 i l 78 1_L I I _ l 80 82 84 1 86 Source: Moody's Investors Service 1994. Note: The five-year default rate indicates the share of issuers with a given rating at the beginning of the year that defaulted within the following five years. FRBN Y 12 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 i i .... 88 89 differences, however, may reflect system atic differences ships among the scales are imprecise, the implicit presump among agencies in the acceptable level of risk in any ratings tion of the ratings-dependent regulations is that the corre category. In this section, we review some of the basic differ sponding ratings levels o f the different N R SR O s represent ences in agency m ethodologies, average ratings, and rank equivalent levels of credit risk and are interchangeable. orderings of credit risks. We examine some of these differ Periodically, the rating agencies articulate unique ences in the context of three important areas of competition ratings philosophies. For example, although M oody’s and w ithin the in dustry— ratin gs for new-issue junk bonds, Standard and Poor’s are primarily concerned with the likeli banks, and asset-backed securities. hood of default on interest or principal, Moody’s is prepared to give a higher rating to an asset-backed security that is like R a t in g D isa g r eem en ts St e m m in g F r o m A lternative M e t h o d o l o g ie s addition, in the area of rating sovereign credit risks, Moody’s Although each agency publishes formal definitions of its var is more reluctant to assign a higher rating to a country’s ious letter ratings, these definitions provide very little insight domestic currency obligations relative to its foreign currency into the source of agency rating differences. The definitions obligations than is Standard and Poor’s (Purcell, Brown, imply that a different likelihood of default is associated with Chang, and Dam rau 1 9 9 3 ) . The other agencies also differ each letter grade, but do not quantify these differences.12 In from their counterparts in certain particulars. For example, addition, rating agencies do not explicitly compare their rat unlike other agencies, D u ff and Phelps som etim es gives ings with those of other agencies. As a practical matter, how higher ratings for the medium-term notes than for the longer ever, it appears that m arket participants have historically term securities of the same issuers. And IBCA assigns higher viewed the Moody’s and Standard and Poor’s scales as roughly ratings to certain non-U.S. banks than do the U.S. agencies equivalent and that the other agencies have attem pted to because it attaches more weight to a foreign government’s align their scales against those two. But while the relation- implicit support of the banking system. Individual agencies R a t io R a t io o f e a r n in g s t o f ix e d c h a r g e s Coverage o f d e b t t o a sse ts Leverage B rated firms **» > „ B B B rated firms % • • • ........... ■ . V X . - " ' X \ - .** ^ BB rated firms ----------- ^ \ \ _____ ♦* V i i 84 ____ _______ m 86 i s..... i s i i ...... 88 90 I I 92 / I 1982 I I 84 —I 86 BBB rated firms ......................... *•« * __ _______ i / / .+ ' - X B rated firms 1982 / j i i % \ i i 1 B B rated firms ✓ \/ M e d ia n C o v e r a g e a n d L e v e r a g e R a t io s o f I n d u s t r ia l Fir m s b y C r e d it R a t in g 1 Chart 3 ■..........................................................................................................................................................-. t- ly to recover most of its principal in the event of default.13 In I ..... ! 88 I ....... 1 90 1 92 Source: Standard and Poor's. Note: Data are three-year moving averages. F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 13 often describe the bases for their positions in their docu ingly, M oody’s and Standard and Poor’s ratings were very ments, but how their methodologies differ from other agen highly correlated at 0 .9 7 , revealing a general consensus cies’ generally m ust be inferred. regarding rank ordering of relative risks.15 This rough equivalence in the rating standards of B r o a d D ifferen ces O bserved in R a tin g s Moody’s and Standard and Poor’s does not seem to extend to Beattie and Searle (1992a) summarize the ratings differences other rating agencies. Table 5 compares the ratings given by observed in a large sam ple o f lon g-term cred it ratin gs nine agencies with those given by Moody’s to the same bor assigned in 1990 by twelve of the leading international rat rowers. (Moody’s ratings are used as the basis of comparison ing agencies and recorded by the Financial Times in its quar simply because this agency has the most ratings in the data terly publication Credit Ratings International. A m ong the set.) Three measures of ratings differences are presented: the 5,284 rating pairs examined for 1,853 rated borrowers, 44 frequency of agreement, the correlation coefficients, and the percent agreed precisely, 35 percent differed by one rating average ratings differences. Standard and Poor’s agrees most notch, 14 percent differed by two notches, and 6 percent dif closely with M oody’s (64 percent), while the percentage fered by three or more notches. (A “rating notch” is, for exam agreement varies among the rest to a low o f 11 percent for the ple, the gap between an A and A + rating.) The differences Japan Credit R ating Agency. Compared with Standard and across agencies as measured by the frequency of agreement Poor’s ratings, the ratings of the other agencies exhibit larger com pound two potential sources of disagreem ent— mean average absolute ratings differences and less correlation with rating scales and rank orderings. Moody’s ratings. The two largest N R S R O s, M oody’s and Standard These differences in ratings reflect not only differ and Poor’s, assign very sim ilar average ratings and rank order ences in rank orderings of credit risks, but, to a large extent, ings o f credit risks. O f the 1,398 cases in 1990 in which differences in rating scales. Chart 4 shows that most of the senior debt ratings were assigned by both companies, 64 per cent were assigned the same rating, 16 percent were rated higher by M oody’s, and 20 percent were rated higher by Stan dard and Poor’s. The average (mean) difference in their rat Chart 4 A v e r a g e D i f f e r e n c e in R a t in g s b e t w e e n M o o d y 's a n d O t h e r A g e n c ie s in 1990 ings (including all those cases where their ratings were the same) was only five-one-hundredths of a notch.14 N ot surpris- R a t in g " n o t c h e s ” 2.0 Table 5 S e n i o r D e b t R a t i n g s o f N in e R a t i n g A g e n c ie s C o m p a r e d w it h M o o d y ’s R a t i n g s in 1 9 9 0 Name o f Agency Number o f Join tly Rated Companies Percentage o f Ratings That Are Equal Correlation between Ratings Scales* Average Ratings D ifferences! (“ +”=higher; ” =lower) 0 .7 8 CBRS 37 38 DBRS 28 0.8 3 0 .7 2 D uff 51 524 50 0 .9 2 -0 .2 5 0 .3 8 Fitch 295 47 0.9 0 0 .2 9 IBCA 13 4 28 0.83 0 .0 5 JBR I 65 11 0.6 7 1.75 3 43 26 0.90 -1 .0 4 33 13 9 8 33 64 0.6 3 0 .9 7 1.0 9 0.05 MCM NIS S&P Lower ratings than Moody's -1.5 I_______ I_______ I_______ 1_______ I_______ I_______ 1_______ I_______ 1_______ f MCM DBRS IBCA S&P Fitch D&P CBRS NIS JBRI Source: Beattie and Searle 1992. Source: Beattie and Searle 1 992a. * The Pearson product-mom ent correlation. t Differences are measured in rating “notches.” For example, the gap between A+ and A - is two ratings notches. FR BN Y 14 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Note: Rating notches are the gaps between ratings. For example, the gap between A+ and A- is two notches. Average differences are calculated using only the ratings of issuers that were rated by both Moody's and the other agency. agencies rate higher than M oody’s, although M cCarthy, (18.2 percent) in 1993 (Fridson 1994). O f the nearly 700 new Crisanti, and Maffei (since merged with D u ff and Phelps) U .S. junk issues between 1989 and 1993 listed in First rates on average a whole rating notch lower. The two largest Boston’s annual High-Yield Handbook, 96 percent are rated American agencies after M oody’s and Standard and Poor’s, by both Moody’s and Standard and Poor’s. Ju n k bond issuers, Fitch and D uff and Phelps, each rate about a third of a notch however, are increasingly seeking third and fourth ratings as higher than Moody’s .16 The Canadian Bond Rating Service, well. D uff and Phelps and Fitch, which respectively rated 16 N ip p on Investors Service, and the Ja p a n C redit R atin g and 4 percent of all issues in the 1989-93 period, have signif Agency rated on average 0.78, 1.09, and 1.75 notches higher icantly increased their rating activity in the past few years than Moody’s, respectively. (see Table 6). The junk bond sample reveals more striking differ R a tin g s fo r N ew -I ssue J u n k B o n d s ences in agency measurements of absolute and relative credit From the point of view of regulatory practice, a rise in the risks than does the broad sample. Standard and Poor’s and number of rating agencies increases the likelihood that mar M oody’s are much more often at odds in the ratings they gin al borrowers w ill m eet m in im um ratin gs thresholds assign junk bond issuers: if we compare the junk bond ratings in Table 7 with the ratings for the broad sample in Table 5, we find smaller frequencies of agreement and smaller correla A rise in the num ber o f ra tin g agen cies increases tion coefficients. The providers of third (and fourth) opinions the lik elih ood th a t m a rgin a l borrow ers w ill meet in this sector, D uff and Phelps and Fitch, also appear to dis agree with Moody’s with greater regularity and on a greater m inim um ra tin gs thresholds. scale in the junk bond sample. because (1) natural variation in opinion increases the proba bility o f receiving at least one satisfactory rating, and (2) some rating agencies may have higher average rating scales Table 7 C r e d it R a t in g s A s s ig n e d t o J u n k B o n d I ssu e r s in 1 9 8 9 - 9 3 : C o m p a r in g t h e R a t in g s o f S& P, D u ff & P h e l p s , a n d F it c h w it h M o o d y ’s R a t in g s enabling more borrowers to meet regulatory cutoffs. We can definitions of investment grade securities by docum enting Name of Agency Number o fjo in tly Rated Companies agency disagreements in the junk bond market. S&P 67 2 41 0.83 D uff 11 3 28 33 14 0 .7 9 -0 .0 0 3 0 .9 65 0 .6 9 1.3 9 3 observe the impact of m ultiple rating agencies on regulatory As generally defined, any issue is considered “junk” that has at least one rating below the BB B - level from either Moody’s or Standard and Poor’s. After falling off in the early 1990s, junk bond issues reached a new high of $57 billion Fitch Percentage o f Ratings That Are Equal Correlation between Ratings Scales* Average Ratings Differences! (“ +”=higher; lower) Sources: First Boston 19 90—94; Federal Reserve Bank of New York staff estimates. * The Pearson product-mom ent correlation. f Differences are measured in rating “notches.” For example, the gap between A+ and A —is two ratings notches. Table 6 M a r k et S hares of N f.w -I ssu e U .S . J unk B o n d R a t in g s : 1 9 8 9 - 9 3 Percent o f New Issues Year M oody’s S&P D uff 19 8 9 19 9 0 100 80 99 80 40 19 9 1 19 9 2 100 100 94 97 19 9 3 97 98 Fitch Memo: Total Number o f Issues 116 24 0 0 12 24 5 13 4 233 301 7 5 42 Percent o f D ollar Volum e Memo: Total Volume (Billions o f Dollars) Year M oody’s S&P D uff 19 8 9 19 9 0 100 100 21 99 99 72 0 0 19 9 1 19 9 2 100 100 21 24 9 .9 99 99 28 5 19 9 3 99 99 19 7 3 8 .9 54.1 Fitch 2.9 0.5 Sources: First Boston 1 9 9 0 —94; Federal Reserve Bank o f New York staff estimates. F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 15 For the newer rating agencies, many of the observed grade rating from the major two agencies obtained a third differences may be related to a difference in their absolute opinion. O f these thirty-four firms, twenty-nine obtained a scales in rating credit risks. W hile M oody’s and Standard and second investment grade rating. Am ong issuers that received Poor’s rate about the same on average for jointly rated issues, m arginally below -investm ent-grade ratings (B B -ratin gs) D uff and Phelps and Fitch ratings are between 1 and 1.5 rat from both M oody’s and Standard and Poor’s, 26 percent ing notches higher than M oody’s or Standard and Poor’s. obtained a third rating. O f these thirty-four firms, sixteen These differences greatly exceed those reported in Table 5 for obtained an investment grade rating. In sum, the demand for the aggregate sam ple o f bond issues. Thus, differences of third ratings increases with the issuer’s proximity to invest opinion between the two largest agencies and the smaller ment grade, and the opportunity to seek third and fourth rat agencies appear to be greater for junk bonds than for invest ings has enabled a number of firms to achieve investment ment grade securities. grade status under certain regulations. Given the possibilities for split ratings, the decision to employ a third rating agency is not random. Chart 5 relates I n t e r n a t io n a l B a n k R a tin g s the frequency with which issuers seek a third rating to the As capital markets have become increasingly global, interna ratin gs received from M oody’s and Standard and Poor’s. tional considerations have taken on greater importance in the Issuers are more likely to obtain a third rating if they receive ratings industry. U.S. rating agencies have been expanding near-investment-grade or mixed (speculative grade/invest their presence overseas, and non-U.S. rating agencies have ment grade) ratings from M oody’s and Standard and Poor’s. been proliferating. In this section, we review international In particular, 46 percent of the firms with one investment ratings differences in the senior debt ratings of banks. Credit ratings are particularly important to banks (through counterparty exposure lim its, letters of credit, and Chart 5 nondeposit sources of funds),17 and a large number of ratings T h e D e c is io n t o O b t a in a T h ir d R a t in g Percent in the industry are cross-border ratings. In addition, the potential designation of certain foreign agencies as N R SR O s o f is s u e s 60 may have considerable im pact on the activities of foreign banks in the United States. Dominant in many other industry sectors, M oody’s and Standard and Poor’s are the leading agencies in the rat ing of banks. O f the 1018 banks worldwide for which long term bond ratings were available in January 1994, M oody’s and Standard and Poor’s rated 64 and 55 percent, respective ly (Financial Times 1994).18 When the sample is lim ited to just those 580 rated banks dom iciled outside the U nited States, Moody’s and Standard and Poor’s still rated 57 and 46 percent, while IBC A was in third position at 31 percent. BBB/BB BB/BB C o m b in a t io n BB/B o f r a t in g s a s s ig n e d b y B/B M o o d y 's a n d B/CCC or CCC/CCC S&P their home countries, most of the other agencies tend to spe cialize in the ratings of banks of their own nationality (see Source: First Boston 1990-94. Notes: The sample consists of 671 junk bond issues brought to market between 1989 and 1993. The issues received the following combinations of ratings from Moody's and Standard and Poor's: 74 rated BBB/BB, 132 rated BB/BB, 79 rated BB/B, 359 rated B/B, and 27 rated B/CCC or CCC/CCC. Each bar in the chan indicates the fraction of issues that were given a third rating from another agency. Digitized 16 for FRASER FRBN Y W hile these three leading agencies rate many banks outside Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Table 8). Agencies appear to disagree more in their measure ment of credit risks for banks than in their risk measurement for other industries. In Table 9, the bank ratings of nine lead ing rating agencies generally show lower frequencies o f Japanese agencies, the wider disagreem ent reflects higher agreement and higher absolute ratings differences relative to average rating differentials. Moody’s than does the broader ratings sample described in National differences in methodology and approach Table 5 .19 The differences are greater for the agencies of some may also help explain the variation in international bank rat countries than for the agencies of others: in particular, the ings. For example, the accounting for nonperforming loans ratings of Japanese agencies differ much more from those of and reserves is not standardized by country, and opinions vary M oody’s than do the ratings of other agencies. W hat accounts widely regarding the extent to which particular governments for the wide disagreement? For the U.S. and Canadian agen lend im plicit support to specific banks in the banking sys cies, agreement concerning relative risk declines as we move tem. Indeed, judgm ents regarding controversial issues are from the broad ratings sample to the bank sample, as evi related to some degree to the nationality of rating agencies. denced by lower correlation coefficients. By contrast, for the When raters are from the same country, agreement about the relative ranking of issuers, as measured by the coefficient of correlation, tends to be higher than when they are not.20 Table 8 Are observed bank ratings consistent with earlier P e r c e n t a g e M a r k e t S h a r e s o f In t e r n a t io n a l B a n k R a t i n g s in 1994 research concluding that agencies judge issuers from their Agency Home Country A ll Banks U.S. Banks Non-U.S. Banks CBRS DBRS JC R A JB R I NIS Fitch D uff M oody’s S&P Thom IBCA Canada Canada Japan Japan Japan U.S. U.S. U.S. U.S. U.S. U .K . 2.1 0.2 4 .0 5.5 4 .0 3.1 5.0 8 .4 17 .6 6 9 .6 50.3 9.7 3 0 .0 0.4 10.0 0.0 0.0 0.0 7.6 5.8 9.4 17.1 3 6 .8 7 5 .8 6 6 .3 16 .9 23.5 0.8 0.8 64 .3 3 5 .9 3.4 35.7 Memo: Home Country Ratings as a Percentage o f Total Each i 9 5 .2 9 6 .1 7 8 .9 7 9 .3 7 0 .0 9 4 .9 9 7 .6 50 .8 6 1 .9 8 1.3 9.2 Sources: Financial Times 1994; Federal Reserve Bank of New York staff estimates. own country more leniently (Beattie and Searle, 1992b)? When the ratings of all banks evaluated by both home-coun try and foreign agencies are aggregated, the average home rating exceeds the average foreign rating by one-half of a rat- Chart 6 A v e r a g e S e n io r D e b t R a t in g s A s s ig n e d t o B a n k s b y H o m e a n d F o r e ig n A g e n c ie s Ratings Table 9 I n t e r n a t io n a l B a n k R a t in g s o f N i n e R a t in g A g e n c ie s C o m p a r e d w i t h M o o d y ’s R a t i n g s i n Name of Agency Number o f Jointly Rated Companies Percentage o f Ratings That Are Equal 1994 Correlation between Ratings Scales* Average Ratings Differences'^ (“+”=higher; ” =lower) CBRS 11 9 0 .5 2 0 .3 6 DBRS 17 0 .6 1 0 .8 4 - 0 .5 3 0 .1 7 D uff 13 9 29 42 Fitch 68 44 0 .7 7 0 .3 8 IBCA 206 38 0.88 0 .5 1 JC R A 19 0.8 2 JBR I 19 0.7 3 0 .8 1 2 .6 3 2 .4 2 0.7 7 NIS 35 11 0 6 S&P 351 37 Canadian banks Japanese banks U.K. banks 2 .4 0 - 0 .1 5 Sources: Financial Times 1994; Federal Reserve Bank o f New York staff estimates. * The Pearson product-mom ent correlation. f Differences are measured in rating “notches.” For example, the gap between A+ and A —is two ratings notches. U.S. banks A ll U.S., Canadian, Japanese, and U.K. banks Sources: Financial Times 1994; Federal Reserve Bank of New York staff estimates. Notes: Observations are limited to banks with both home country and foreign country senior debt ratings. Sample sizes for the five bank groups (from left to right) are 100, 17, 35, 18, and 170. FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 17 ing notch. However, the results differ greatly depending on International ratings differences are o f particular the nationality of the bank (Chart 6). While U .S. and Canadi importance at the present time because the SEC is reviewing an banks receive lower home ratings than foreign ratings, numerous applications for N R SR O designation from agen Japanese and U .K . banks receive higher home ratings. At cies of foreign countries. Differences among the agencies of least in this sample, observed differences between home and different countries and the tendency of many agencies to foreign ratings reflect the relative toughness of each country’s focus on the rating of banks from their home countries imply agencies rather than a more general hom e-country bias. that if N R SR O status were to be granted to the two Canadian W hether the rated bank was from the same country or not and three Japanese rating agencies, the number of Canadian bore little relationship to the differences between the ratings and Japanese banks reaching regulatory cutoff ratings would of non-U.S. agencies and M oody’s. increase considerably. As Chart 7 shows, o f the fifty-three Chart 7 Im p l i c a t i o n s o f E x p a n d i n g t h e N u m b e r o f N a t i o n a l l y R e c o g n i z e d S t a t i s t i c a l R a t i n g O r g a n i z a t i o n s ( N R S R O s ) Japanese Banks Canadian Banks Top N RSRO Rating Top N RSRO Rating Top O verall Rating Top Overall Rating Source: Financial Tim s 1994. Notes: The Japanese sample and the Canadian sample consist of seventy-two and fifty-three banks, respectively. Top NRSRO rating is the highest long-term rating of those assigned by Duff, Fitch, IBCA, Moody's, S&P, and Thomson. Top overall rating is the highest long-term rating of those assigned by the NRSROs and the following non-NRSROs: CBRS, DBRS, JBRI, JCRA, and NIS. FRBN Y 18 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Canadian banks with senior debt ratings listed in the Finan Standard and Poor’s was the undisputed leader in M BS and cial Times Credit Ratings International (1994), the share receiv A BS ratings. In the late 1980s, Moody’s caught up consider ing an N R SR O credit rating of at least AA- would rise from ably and D u ff and Phelps made significant inroads in the 23 percent to 55 percent. Sim ilarly, o f the seventy-four A BS market. Since then, Fitch has made great strides in mar Japanese banks with senior debt ratings listed in the same ket share, actually leading the m arket for M BSs in 1994, publication, the share receiving an N R SR O rating of at least while the other agencies’ shares have fluctuated. Charts 9 and AA- would rise from 20 percent to 48 percent. 10 summarize the available data on these two markets since 1989- (Since more than one agency can rate each security, the R a tin g s fo r M o r t g a g e - a n d A sset -B a ck ed S ecu rities sum of the shares exceeds 100 percent.) U nlike the corporate bond m arket, the M BS and C o m p etition am on g the ratin g agencies is p articu larly A BS m arkets are lim ited alm ost entirely to highly rated marked in the rating of mortgage-backed and asset-backed issues, typically either AA or AAA for M BSs, and A or AAA securities (MBSs and ABSs).21 Issuers often seek ratings from for ABSs. The need for high ratings appears to arise from the just one or two companies, and as we see below, Fitch and D uff advantages regulations confer on highly rated (particularly and Phelps have increased market share. Banks and securities M BS) issues and from investors’ concerns about the quality of firms generally consult directly with the rating agencies to the collateral as well as their unfamiliarity with the compli find out how M BSs and ABSs can be structured to obtain high cated structures of the securities. The relatively small share of credit ratings. The agencies analyze the asset pools to be secu M BSs and ABSs that are rated less than A consist largely of ritized to determine the adequacy of the credit support under “B ” tranches that are subordinate to much larger, highly lying each tranche of structured transactions. Agency dis rated senior tranches. The subordinated tranches tend to be agreements normally center on the criteria that establish the amount of credit enhancement required for a specific rating. These differences of opinion are not normally evident in the ratings per se because issuers structure their securities to obtain the desired ratin gs from the agen cies they hire. (Moody’s occasionally assigns unsolicited ratings that indicate its disagreement with the higher ratings assigned by other agencies.) Market observers have expressed concern that com Chart 8 Is s u a n c e o f A s s e t - b a c k e d a n d N o n a g e n c y M o r t g a g e - b a c k e d S e c u r it ie s Billions of dollars 120 -------------- petitive pressures have led agencies to compete on ratings cri teria, potentially underm ining the reliability of the ratin gs.22 Industry analysts normally distinguish between two broad categories of M BSs, those backed by government agen cies such as the Federal National M ortgage Association and private label issues that securitize jumbo m ortgages, com mercial m ortgages, and various other so-called nonconform ing first m ortgages that the governm ent agencies do not securitize. The ABS market securitizes shorter duration asset pools such as credit card receivables, auto loans, and home equity loans. The M BS and ABS markets have grown very rapidly since 1989 (Chart 8). Rating agency market shares for M BS and ABS rat ings have shifted considerably over time. In the m id-1980s, 1989 90 91 92 93 94 Sources: Asset Sales Report; Inside Mortgage Finance; Federal Reserve Bank of New York staff estimates. Note: Mortgage-backed and asset-backed volumes for 1994 are annualized using data through April and June, respectively. FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 19 privately placed and are often rated by just a single agency or carry no rating at all. emergence of market competition have been particularly dra matic in the case of private-label mortgage-backed securities. M BS and A BS structures typically contain credit U n til the m id -1 9 8 0 s, Standard and Poor’s was the only protection so that the securities are less risky than the underly agency ra tin g these se cu ritie s, and its requ ired cred it ing asset pools. The forms of the credit enhancements vary enhancements for reaching target ratings represented the widely and include bank letters of credit, bond insurance com industry standard. In 1986, Moody’s entered the market with pany gu aran tees, subordin ated interests, cash collateral criteria that were slightly different. W hile its standards were accounts, and reinvestment of the excess cash flows generated stricter than those o f Standard and Poor’s in some areas, by the asset pools themselves. Since all enhancements are cost M oody’s set lower enhancement requirem ents for certain ly, issuers prefer structures that achieve a given rating with the types of mortgage pools (shorter term, negative amortization, smallest enhancements and choose rating agencies with the and convertible adjustable rate mortgages) and subsequently most lenient credit enhancement requirements, provided the gain ed m arket share in those areas. In 1987 and 19 8 8 , agencies’ ratings carry sufficient weight in the capital market. Moody’s issued some unsolicited ratings (in areas where its In principle, securities with lower credit enhancements can be standards were stricter than those of Standard and Poor’s) and discounted by the market. However, in practice, the market caused yields to rise on these securities. In response, some has trusted agencies to be prudent in the determination of issuers changed their M BS structures and hired Moody’s. By credit support requirem ents and has not required higher 1989, M oody’s share of the M BS business exceeded that of yields from issuers that have switched to agencies with lower Standard and Poor’s. enhancement requirements (Bruskin 1994). In 1990, Fitch began rating mortgage-backed secu The evolution o f credit rating standards and the rities using a model o f required credit enhancement that Chart 10 Chart 9 R a t in g A g e n c y M a r k e t S h a r e s o f A sse t - b a c k e d R a t in g A g e n c y M a r k e t S h a r e s o f M o r t g a g e - b a c k e d S e c u r it ie s Iss u a n c e S e c u r it ie s Is s u a n c e Percentage P ercentage o f to t a l dollar vo lu m e o f to tal dollar volum e l00 ------------------------------ / / 20 ------------------------------------------- / ------------------------------------------------f Duff B,« / 0 I___ — —-1 —— ■J_____________ I___ I___________I_______ I 1988 91 90 91 92 93 94 Sources: Internal agency data; Asset Sales Report', Federal Reserve Bank of New York staff estimates. Sources: Internal agency data; Asset Sales Report; Federal Reserve Bank of New York staff estimates. Notes: Market shares for 19 9 4 are based on data through June. The sum o f the market shares exceeds 100 percent because many issues receive multiple ratings. Notes: Market shares for 1994 are based on data through April. The sum of the market shares exceeds 100 percent because many issues receive multiple ratings. FRBN Y 20 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 based its worst-case scenarios on the Texas recession of the letter ratings as thresholds for determining capital charges 1980s. This approach resulted in required enhancements and defining investment prohibitions. Although the agencies below those of Standard and Poor’s (lower by as much as 50 make no such assurances, the current use of ratings in regula percent for certain balloon payment m ortgages), whose tion assumes a stable relationship between ratings and model extrapolated from the mortgage default experience of default probabilities. The historical record suggests other the Great Depression. Duff and Phelps followed suit with a wise: although ratings usefully order credit risks at any point framework similar to Fitch’s in 1992. Standard and Poor’s, in time, specific letter ratings corresponded to higher default which once had a monopoly, saw its market share slide to 5 5 risks in the 1980s than in the 1970s. percent in 1993 as many issuers who had at one time The increasing number of agencies also poses prob employed only Standard and Poor’s, then later used Standard lems for the existing structure of ratings-based regulations. and Poor’s together with Moody’s, had switched to a pairing Some agencies appear to have different absolute scales, rating of Fitch and Moody’s. bonds higher or lower on average than other agencies. How Most recently, in December 1993, Standard and ever, even normal variations in opinion across agencies with Poor’s came out with revised criteria for credit enhancements the same basic scales confounds the application of existing that implied a 30 percent reduction on average across a vari regulations. These problems multiply as the number of agen ety of mortgage pool types. Explanations for the changes fol cies and the differences of opinion among them increase. lowed the next month (Standard and Poor’s 1994). Some of The impact of multiple rating agencies and ratings the revisions were in those areas in which the agency had been differences is apparent in the case studies of junk bond, bank losing market share, including shorter term mortgages. Rival debt, and m ortgage-backed securities ratings. For junk rating agencies claimed that the move represented a competi bonds, the availability of third opinions enables many bor tive attem pt to win back market share. In the first four rowers to climb out of the speculative grade zone into invest months of 1994, Standard and Poor’s has regained market ment grade territory. In the area of bank debt ratings, differ share largely at Moody’s expense. It is difficult to tell at this ences of opinion are particularly great between agencies of point whether the shift reflects issuers moving from one different countries and imply that the designation of more agency to another or merely a growth in issuance by firms foreign agencies as NRSRO s will allow more foreign banks to that use Standard and Poor’s ratings (see Schultz 1994 and achieve higher ratings. Regarding private-label mortgage- Inside Mortgage Securities 1994a, 1994b). backed securities, intensifying competition among the four Clearly, M BS credit enhancem ent levels have declined over the history of the market. Analysts and agen major agencies has been associated with downward revisions of required enhancement levels. cies note that this in part reflects a progression along the The Securities and Exchange Commission (1994a, learning curve: more information has become available over 1994b) is currently reconsidering its procedures for designat time about the performance of such securities, reducing the ing nationally recognized agencies (NRSROs), the role of rat degree of uncertainty. Skeptics remark that competitive pres ings in regulations, and the degree of public oversight and sures can lead to increased pressures to review standards. mandatory ratings disclosure. Questions for which comments Whether or not agencies compete on criteria, it does appear have been solicited include: that the incentive to innovate in structured finance ratings tends to favor lower enhancement levels. • What are the proper objective criteria to consider when determining N RSRO status? C o n c l u s io n • Is it appropriate for NRSROs to charge issuers for Regulators, like investors, value the cost savings achieved ratings, and in particular, to vary the charge with through the use of ratings in the credit evaluation process. As the size of the transaction? a result, they have have come to employ a variety of specific • Would further regulatory oversight of NRSRO s F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 21 be appropriate and what type of oversight would agencies. O f course, any changes in policy would entail com that be? plex tradeoffs; specific proposals and their implications will • Should issuers be required to disclose activities surely be explored in future research. such as rating shopping— soliciting preliminary The SEC has also invited comment on whether it indications from numerous rating agencies in should continue to employ an N RSRO concept. Although order to identify the agency that will provide the dropping the designation of N RSRO s would be a radical highest rating? measure, it might encourage regulators to revise their current The SEC’s questions all raise the possibility of addi use of ratings and to adjust for ratings differences across time tional oversight or disclosure of NRSRO activity and the rat and agency. Ratings can and do play an important and valu ings process. Such measures could conceivably address some able role in the functioning and oversight of financial mar of the issues raised in this paper by improving the intertem kets. But at a m inim um , regulators and investors alike poral stability of default rates within ratings category and should be critical users and should regularly review their reducing differences am ong the officially design ated application of ratings to the decisions they make. FRBN Y 22 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Endnotes 1. Government policy has helped to avert conflicts of interest. The Federal Reserve Board discouraged a proposed acquisition of Duff and Phelps by Security Pacific Bank in 1984. The Board ruled that if the merger were to take place, Duff and Phelps would be prohibited from issuing public rat ings because Security Pacific would effectively be rating its own borrowers (Ederington and Yawitz 1987). 2. See the description of the agencies’ ratings methodologies in the Financial Times 1994, pp. 25-79. Comparisons across the agencies’ rating scales are more difficult in the lowest part of the range near default, where the agencies carry different numbers of ratings (Dale and Thomas 1991). 3. For a more critical view, see Stein 1992. Note 8 continued time to time as they deem appropriate. Further, when determining ratings, rating agencies neither use uniform criteria nor weigh the same criteria equally” (Grafton 1992). 9. The Moody’s study calculates the default rate formally as a weightedaverage cumulative default rate, which is the complement of the product of weighted-average marginal survival rates. For details concerning rate cal culations, see the Appendix in Moody’s Investors Service 1994. 10. As reported by Fons (1991), most of the cyclical variation in the aggre gate default rate on corporate bonds cannot be explained by cyclical varia tions in Moody’s ratings on bond outstandings. Moreover, since yield spreads between high- and low-rated bonds tend to rise during recessions, 4. Dale and Thomas (1991) and Baron and Murch (1993) provide compre hensive discussions of the current use of ratings by regulators in the United States and abroad. Harold (1938) provides a detailed account of the earliest uses of ratings in the United States. market pricing is consistent with a perceived rise in the default probabili ties of lower rated issues relative to those of higher rated issues during recessions. Alternatively, the rise in spreads in recessions may merely reflect a concurrent rise in the market’s aversion to default risk or other supply and demand factors. 5. Under this system, insurance companies were allowed to request that specific BB rated bonds be treated the same as those more highly rated bonds in the top quality category. This practice, which became common over time, was halted by reforms adopted by the NAIC in 1990. 11. Wigmore (1990) documents a much more severe difference between the 1986-88 average credit ratios and the 1983-85 ratios than is suggested by the Standard and Poor’s data presented in Charts 7 and 8 for bonds rated BB or B. According to Wigmore, the Standard and Poor’s data understate 6. In analyzing the margin rules for mortgage-related securities, Federal the decline in credit quality because a greater proportion of the junk bonds Reserve Board staff reasoned thus: “The question of using bond ratings by a recognized service as a criterion for margin eligibility was discussed when the initial definition o f ‘OTC margin bond’ was under consideration. The leveraging events (mergers, acquisitions, and leveraged buyouts); there fore, their current credit ratios were much weaker than the historical credit National Association of Securities Dealers proposed a rating standard at that time and most securities dealers endorsed its use for non-listed bonds in comment letters, but the Board declined to adopt such a requirement. issued in the later period were issued by companies in conjunction with ratios included in the Standard and Poor’s data. In contrast, Fridson (1991) argues that much of the apparent deterioration in credit ratios and increas es in default rates for B-rated issues in the late 1980s can be explained by an Since that time, however, the SEC has used these evaluations of third par ties as a means of categorizing some debt securities; regulatory examiners increase in the proportion of issuers rated “B-” as opposed to “B” or “B + use them to determine investment grade; and the United States Congress has mandated their use in the statutory definition under consideration. Staff believes that developments subsequent to the 1978 decision warrant a departure from the Board’s earlier decision” (Board of Governors 1987). 12. Moody’s and Standard and Poor’s now provide ex post analyses of cor porate bond defaults by rating categories. Variations in default probabili ties for Moody’s and Standard and Poor’s can, therefore, be inferred from 7. These rules key off the agencies’ short-term ratings, limiting money funds from holding more than 5 percent of their assets in paper rated A2 by 13. Standard and Poor’s is generally less willing to base ratings on expect ed recoveries even though it has always made such distinctions, as have all Standard and Poor’s (P2 by Moody’s) or more than 1 percent in any paper of a single A2/P2 issuer. Issuers with these weaker short-term ratings typical the other agencies, for different classes of debt issued by the same firm. Whenever a firm defaults on its subordinated debts, its senior debt is these studies. ly have long-term bond ratings that are still rated well above the invest almost always drawn into default as well. Nevertheless, agencies regularly ment grade cutoff. award higher ratings to the senior debt because its expected recovery rate is higher. 8. When voicing its concerns over the 1991 amendments to Rule 2a-7, the Securities Industry Association noted that “rating categories were not 14. Other authors with other data sets also note a rough equivalence designed as regulatory tools and NRSROs may change their criteria from between Moody’s and Standard and Poor’s ratings (Perry 1985, Ederington N o tes FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 23 E n d n o te s (Continued) Note 14 continued Note 17 continued 1986, Ederington and Yawitz 1987). Moreover, Billingsley, Lamy, Marr, and Thompson (1985) show that the market views Moody’s and Standard and Poor’s ratings as equivalent because the yields on bond issues with split ratings do not depend on which agency assigned the higher rating. banks were also put on CreditWatch. 15. The Pearson product-moment correlation coefficient, which can range from -1.00 to a maximum value of 1.00, measures the extent to which rank orderings agree while removing any confounding effects of differences in average rating scores and differences in units of measurement. 18. The ratings of the French rating agency S&P-ADEF, a joint venture founded in 1990 by Standard and Poor’s and Agence d e v alu ation Financiere, are counted as Standard and Poor’s ratings for the purposes of calculating global market share. 19- The one consistent exception is IBCA, which shows more agreement with Moody’s on these measures for banks than does the wider sample. This finding may reflect IBCA’s initial specialization in the rating of financial 16. These agencies acknowledge that their ratings are higher than Moody’s and Standard and Poor’s on average; however, they attribute some of the ratings difference to sample selection bias. They argue that ratings from Fitch or Duff and Phelps are only sought when there is a strong expec tation of improving upon Moody’s and Standard and Poor’s ratings. When Fitch or Duff and Phelps might, in fact, rate lower, their ratings are not institutions and the limitation of its NRSRO designation to that area. 20. The mean of the Pearson product-moment correlation for ratings of agencies from the same country is 0.858, compared with a mean of 0.775 for the correlation coefficients for the ratings of agencies from different countries. The standard errors of measurement for the two coefficients are purchased. 0.018 and 0.054, respectively. 17. For example, U.S. issuers of commercial paper and long-term securi 21. In this article, we follow industry practice in using “asset-backed secu rities” (ABSs) in the more narrow sense that excludes mortgage-backed ties often obtain bank letters of credit in order to achieve targeted credit ratings, but the attractiveness of such backing depends greatly on the cred it rating of the bank issuing the letter of credit. When Standard and Poor’s put three Japanese banks on its CreditWatch list (with negative implica securities (MBSs). tions) in March 1994, the bond issues of 144 U.S. bond issuers and 46 U.S. commercial paper issues that were backed by letters of credit from these ties (1994a, 1994b). This section’s discussion of the evolution of ratings cri teria for MBSs draws heavily from these sources. 22. See Bruskin (1988, 1994), Schultz (1994), and Inside Mortgage Securi R eferences Altman, Edward I. 1989. “Measuring Corporate Bond Mortality and Per formance.” J o u r n a l OF F in a n c e (September): 909-22. Baron, Neil, and Leah Murch. 1993- “Statutory and Regulatory Uses of Rat ings in the United States and Other Jurisdictions.” Fitch Investors Ser vices, January 14. Ang,J., andK. Patel. 1975. “Bond Rating Methods: Comparison and Vali dation.” J o u r n a l of F in a n c e (May): 631-40. Beattie, Vivien, and Susan Searle. 1992a. “Bond Ratings and Inter-Rater Agreem ent.” J Artus, Patrick, Jean Garrigues, and Mohamed Sassenou. 1993- “Interest Rate Costs and Issuer Ratings: The Case of French CP and Bonds.” J o u r n a l of I n t e r n a t io n a l S e c u r it ie s M a r k e t s 24 FRBNY Q (S u m m er): o u rn a l of I n t e r n a t io n a l S e c u r it ie s M a r k e t s 167-72. (Autumn): 211-18. u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 N otes R eferen ces {Continued) _____ . 1992b. “Credit-Rating Agencies: The Relationship between Rater Federal Financial Institutions Examination Council. 1994. “Risk Based Capi In t e r n a tal Requirements— Recourse and Direct Credit Substitutes; Proposed Rule.” F e d e r a l R e g is t e r , vol. 59, no. 100, pp. 27116-41. Billingsley, Randall, Robert Lamy, M. Wayne Man, and G. Rodney Thompson. Financial Times. 1994. F in a n c ia l T im es C r e d it R a t in g s I n t e r n a t io n a l Agreement and Issuer/Rater Characteristics.” J o u r n a l t io n a l S e c u r it ie s M a r k e t s of (Winter): 371-5. 1985. “Split Ratings and Bond Reoffering Yields.” F in a n c ia l M a n a g e m ent (First Quarter). (Spring): 59-65. First Boston. 1990-94. H ig h - Y ie l d H a n d b o o k , annual issues. Board of Governors of the Federal Reserve System. 1987. “Marginability of Mortgage-Related Securities.” Office Correspondence from Division of Banking Supervision and Regulation to the Board of Governors, Fons, Jerome S. 1991. “An Approach to Forecasting Default Rates.” M o o d y ’s S p e c ia l R e p o r t . April 15. Brand, Leo, Thomas C. Kitto, and Reza Bahar. 1994. “ 1993 Corporate Fridson, MartinS. 1991. “Are Bond Ratings Consistent Over Time?” J u n k B o n d R e p o r t e r , February 25. Default, Rating Transition Study Results.” STANDARD AND P o o r ’s _____ . 1994. “The State of the High Yield Bond Market.” J o u r n a l C r e d it W e e k I n t e r n a t io n a l , June 6. A p p l ie d C o r p o r a t e F in a n c e Bruskin, Eric. 1988. “The Role of the Rating Agencies.” CONVENTIONAL P a s s - T h r o u g h Q u a r t e r l y . Mortgage Securities Research, Goldman of (Spring):85-97. Grafton, K. Susan. 1992. “The Role of Ratings in the Federal Securities Laws.” I n s ig h t s (August): 22-27. Sachs, July. _____ . 1994. “The Rating Game: You Pays Your Money and You Take Harold, Gilbert. 1938. B o n d R a t in g s as a n I n v e s t m e n t G u id e . Y o u r C h o ic e .” M o r t g a g e M a r k e t C o m m e n t , M o r t g a g e S e c u r itie s R esearch , G o ld m a n Sach s, Ja n u a r y 7. I n s id e M o r t g a g e S e c u r it i e s . 1994a. “ In M B S R a t in g S e rv ic e B a t t le , H isto r ic a l L ead er S & P Q u ie tly D ro p s in to T h ird P lace d u r in g Cantor, Richard, and R. Demsetz. 1993- “Securitization, Loan Sales, and the Credit Slowdown.” F e d e r a l R e ser v e B a n k R e v ie w (Summer): 27-38. of 1993.” M arch 18. N ew Y o r k Q u a r t e r l y _____ 1994b. “Four-Way Battle for Private-Label Rating Business Heats Up As Moody’s Loses Market Share.” June 17. Carey, Mark, Stephen Prowse, John Rea, and Gregory Udell. 1993. FEDERAL R e ser v e B u l l e t in (February): 77-92. Lore, Kenneth. 1991. M o r t g a g e - B a c k e d S e c u r it i e s , 1990-91 Edition. New York: Clark, Boardman Company. Crabbe, Leland, and Mitchell Post. 1992. “The Effect of SEC Amendments to Rule 2a-7 on the Commercial Paper Market.” Board of Governors of the Federal Reserve System Working Paper no. 199- 1970-1993. ” January. Dale, RichardS., and Stephen H. Thomas. 1991. “T h e R e g u la to r y U s e o f C r e d it R a t in g s in In te rn a tio n al F in a n c ia l M a r k e ts.” J o u r n a l o f I n t e r n a t io n a l S e c u r it ie s M a r k e t s (S p rin g ): Moody’s Investors Service. 1994. “Corporate Bond Defaults and Default Rates: Monro-Davis, Robin. 1994. “Unsolicited Ratings Are for Investors.” T h e T r e a s u r e r (February): 30-33- 9-18. National Association of Insurance Commissioners, Securities Valuation Office. Ederington, Louis. 1986. “Why Split Ratings Occur.” F in a n c ia l M a n a g e m ent 1994. V a l u a t io n of S e c u r it ie s M a n u a l , vols. 1-2. (Spring): 37-47. O’Neill, Leo C. 1993. Speech before the IOSCO Annual Conference, Mexico Ederington, Louis, and Jess Yawitz. 1987. “The Bond Rating Process.” In Edward Altman, ed., H a n d b o o k John Wiley and Sons. N o tes of F in a n c ia l M a r k e t s . City, October 24-26. New York: Perry, Larry G. 1985. “The Effect of Bond Rating Agencies on Bond Rat ing Models.” J o u r n a l FRBNY Q of F in a n c ia l R e s e a r c h u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 (Winter): 307-15. 25 R efe re n c e s (Continued) Furcell, John, Ernest Brown, Joyce Chang, and Dirk Damrau. 1 9 9 3 . “ F erm en t o v e r D e v e lo p in g C o u n tr y R a t in g s : T h e C a se o f M e x ic o .” E m e r g i n g Standard and Poor’s. 1994. “Improving Outlook Spurs MBS Revisions.” S t a n d a r d a n d P o o r ’s C r e d it w e e k , January 10. M a r k e t s R e s e a r c h , S a lo m o n B ro th e rs, F eb ru ary 2. Stein, Benjamin. 1992. A L i c e n s e Schultz, Abby. 1994. “S&P Faces Criticism in Mortgage Area.” W a l l S t r e e t J o u r n a l , January 4. Securities and Exchange Commission. 1994a. “Nationally Recognized Statisti cal Rating Organizations.” Release Nos. 33*7085, August 31. to S t e a l : T he U n t o ld St o r y of M ic h a e l M i l k e n a n d t h e C o n s p i r a c y t o B i l k t h e N a t i o n . N e w Y ork: S im o n an d Sch uster. West, Richard R. 1973. “Bond Ratings, Bond Yields, and Financial Reg ulation: Some Findings.” J o u r n a l 159-68. of L a w a n d E c o n o m ic s (April): Securities and Exchange Commission. 1994b. “Disclosure of Security Ratings.” Release No. 33-7086, August 31. FR BN Y 26 Q u a r t e r l y R e v ie w /S u m m e r - F a ll Wigmore, Barrie. 1990. “The Decline in Credit Quality of New-Issue Junk Bonds.” F in a n c ia l A n a l y s t s J o u r n a l (September-October): 53-62. 1994 N otes The Price Risk of Options Positions: Measurement and Capital Requirements Arturo Estrella, Darryll Hendricks, John Kambhu, SooShin, and Stefan Walter G lobal m arkets for option products, both of the different methods in providing capital coverage for exchange-traded and over-the-counter, have potential losses on a series of option portfolios. expanded rapidly in recent years. The Bank We find that the simple strategy methods provide for International Settlements (1994) reports only a rough measure of potential losses. Moreover, for mar that outstanding options now exist on notional principal ket participants with large option positions, the simple strat amounts totaling at least $3-3 trillion. In the last four years, egy approach could lead to an excessive reporting burden. the outstanding interest rate, commodity, and equity-related The value-at-risk methods, which are based on option pric options of U.S. commercial banks have grown more than 40 ing models, tend to provide better estimates of the market percent annually and the banks’ foreign exchange options risk inherent in a position. The accuracy of the value-at-risk more than 16 percent per year. approach is also found to be significantly enhanced by adjust The expansion of options markets underscores the ments for gamma risk, the risk that an options price changes need for supervisors to develop sound methods of monitoring in a nonlinear fashion as a result of large movements in the the risks associated with these markets. This article assesses price of the underlying instrument.1 different supervisory approaches to the measurement and capital treatment of the market risk of options— the risk that Background an d M eth odology an options contract will decline in value with changes in mar T he U n iq u e R i s k s ket prices or rates. The methods for measuring the market Like most other instruments, options contracts entail both risk of options positions examined in the article fall into two price (or market) risk and credit risk. Market risk arises when broad categories: simple strategy methods and value-at-risk the value of the interm ediary’s portfolio is sensitive to (or price sensitivity) methods. We compare the performance changes in market prices or rates. On some occasions, inter of O p t io n s F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 27 mediaries will attempt to eliminate such risk by engaging in the charge. The most efficient charges would, for example, offsetting transactions; that is, they attempt to “hedge” the require substantial data about the composition of the options market risk away. On other occasions, however, intermedi portfolio and the risk factors affecting the portfolio’s value, as aries may attempt to earn a risk premium for bearing the well as estimates of the sensitivities of the portfolio’s value to market risk. Credit risk arises because a financial asset, such movements in the risk factors. To process such information in as a purchased option or a business loan, could become worth a timely fashion, an institution must be willing to commit less if the counterparty to the asset does not make good on its significant resources. obligations. This article focuses on market risk. The complexity of the capital calculation itself is The form that market risk takes in options markets also an important consideration. Complex supervisory charges can be quite different from its form in other markets. This is may be difficult to implement uniformly. Moreover, the more true in part because the values of options contracts can change specific the rule, the greater the opportunities for finding extremely rapidly— often far more rapidly as a percentage of exceptions or exclusions that were not intended. Neverthe their value than do the assets that underlie options contracts. less, complex calculations could lead to more accurate and In addition, the price sensitivity and volatility of a position efficient charges, thereby lessening the regulatory burden on can themselves change quickly, further complicating the risk the options market. management of options positions. An intermediary must A final consideration that is particularly important in constantly track the changes in the volatility of its portfolio setting supervisory capital for options is the use of option pric that result from market movements. This task can be particu ing models in the calculation. The development of these mod larly difficult in periods of great market stress and lowered els has been critical to the growth of the options market, and liquidity, such as that experienced in the market for European currency options in September 1992. Alert to these difficulties, supervisors must consider carefully how supervisory capital can best reduce the detri mental impact of options risks in the financial markets. There is a definite trade-off between the efficiency o f the capital charge and the resources required to compute the charge. C o n s id e r a t io n s in S e t t in g S u p e r v is o r y C a p it a l R e q u ir e m e n t s Determining appropriate supervisory capital standards for options positions involves several choices. A sufficient level of their use in the market is pervasive. However, the markets do prudence could probably be achieved by simply setting very not always conform to the assumptions of the models. In par high standards without regard to how the risks change in ticular, options pricing models typically assume continuous response to changing market conditions. This approach price changes and liquid markets. The risk of systemic prob could, however, require far more capital than is actually lems, however, is very likely to go hand-in-hand with a loss of needed. The costs of this excessive safety would then translate market liquidity and discontinuous price movements. There into a slowdown in potentially beneficial options trading or fore, in setting supervisory capital requirements, regulators perhaps a relocation of this trading to jurisdictions not may not wish to rely solely on pricing models. imposing such onerous standards. More accurate measures of the risks of an options C urrent an d P r o p o s e d S u p e r v is o r y A ppro ach es portfolio require more information about the composition of At the international level, there are a number of different the portfolio and more calculations. As the analysis below existing and proposed supervisory capital treatments for the makes clear, there is a definite trade-off between the efficiency market risk of options. The European Community’s Capital of the capital charge and the resources required to compute Adequacy Directive (CAD), which takes effect in January FRBNY Q 28 u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 1996, includes a simple capital requirement for options that C a p it a l R u l e s for O p t io n s applies to all commercial banks and securities firms licensed Capital rules for the price risk of options fall into two cate in the Community. The Basle Committee on Banking Super gories: value-at-risk (or price sensitivity) rules and strategy- vision’s April 1993 market risk proposal (BIS proposal) based rules. The value-at-risk approach uses option pricing includes capital requirements for options similar to those of models to estimate a portfolio’s potential losses. The strategy- the CAD. Unlike the CAD, however, the BIS proposal also based rules, by contrast, apply various formulas to the market considers allowing banks to apply a more sophisticated sce price of an option and its underlying asset. nario-based methodology to calculate their capital require ment for options positions. While strategy-based rules estimate potential losses very roughly, the option pricing models of the value-at-risk The rules of both the United Kingdom’s Securities approach in principle provide a much more accurate measure and Futures Authority (SFA) and the U.S. Securities and of potential changes in the value of options. Strategy-based Exchange Commission (SEC) already include capital require rules have the additional drawback of requiring a complex ments for the market risk of options. The SFA’s requirements parsing of a portfolio to rearrange its components into the range from simple rules for small options players to a sce particular set of strategies recognized for capital purposes. nario-based approach for more sophisticated institutions. The value-at-risk approach uses a simple aggrega The SEC ’s capital requirements are based on a series of tion of all positions relating to a given underlying asset (for options trading strategies that are commonly employed by example, a particular equity or exchange rate). This aggrega financial institutions. tion of positions of the value-at-risk approach more reliably accounts for offsetting within a portfolio than does the strate M eth odology of E v a l u a t io n gy-based approach, which recognizes offsetting in only a Conceptually, measuring the market risk of options for capi piecemeal fashion— and only to the extent that positions fit tal purposes consists of two basic steps. The first entails mak into the trading strategies recognized by supervisors. Note, ing assumptions about the potential movement of risk factors however, that this conclusion applies only to aggregation that could affect the value of an options portfolio over a across instruments sensitive to a given underlying asset. The defined holding period and at a certain level of confidence. appropriate approach to aggregation of positions across mul This step may also account for correlations among the differ tiple underlyings is a separate problem. ent risk factors. The second step consists of measuring the sensitivity of options positions to the assumed movements in V a l u e - at- R is k A the underlying risk factors to arrive at an estimate of the port A portfolio’s value at risk is a measure of its potential losses, folio’s potential gain or loss. Each step can be carried out with where the losses are expressed in terms of some confidence various degrees o f sophistication. This article focuses level (for example, a loss of such magnitude that it is likely to primarily on the second step, examining the performance of a occur in only one month out of a hundred). This risk mea number of capital rules in approximating potential portfolio surement approach lends itself quite readily to the construc losses resulting from given movements in the underlying risk tion of a capital requirement. Specifically, the potential losses factors. a portfolio might suffer are estimated and then used to deter The following section describes various methods of setting regulatory capital and the rationale that underlies ppro ach es mine an amount of capital sufficient to cover these losses with the desired level of confidence. each. Although in several cases the methods overlap with those of specific regulatory entities, our intent is not to assess S c e n a r io - b a s e d a n d S im u l a t io n M e t h o d s w it h F u ll the particular rules of different regulatory bodies, but to R e v a lu a t io n evaluate the generic approaches that might be taken towards The most precise method for estimating value at risk entails regulatory capital.2 calculating a portfolio’s gains and losses by using option pric FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 29 ing models to explicitly revalue the portfolio over a set of pos the principal approach of the European Com m unity’s CA D , tulated price changes. The resulting capital charge is simply is based only on the option’s delta value, while other rules use the largest loss over the set of postulated price changes, mea both the delta and gam m a values. Chart 1 illustrates the sured at a certain level of confidence. results of using first- and second-order approximations in the The postulated price changes can be obtained using case of a written call option. either a scenario approach or a simulation method. The sce N ote that the second-order approxim ation is not nario approach revalues the portfolio at several distinct values necessarily more accurate than the first-order approximation, of the underlying asset within a given interval— for example, particularly if the changes in the underlying value are large. within the interval defined by the current underlying price Thus, for applications that rely on accurate measures of port plus or minus three standard deviations of monthly moves. folio changes resulting from large market movements (for Alternatively, changes in the price of the underlying asset example, “stress testing”), models involving approximations may be sim ulated, using either historical price changes or should be used with caution. Monte Carlo methods. In either type of simulation, the entire An advantage of the price sensitivity approach is its portfolio is revalued at each point generated by the simula simplicity. It requires only the most basic options-related tion. Either the largest loss or some conservative percentile of data used by standard risk management systems: the stan the losses, depending on the desired degree of confidence, can dard deviation (volatility) of the underlying asset, the portfo then be selec te d as the valu e at risk . T h e sim u la tio n lio’s delta value, and the portfolio’s gam m a value. Addition approaches allow for sam pling o f portfolio value changes over ally, a convenient feature o f deltas and gam m as is their a more continuous range of price changes in the underlying additivity across instruments written on the same underlying asset than does the scenario approach, which focuses on a asset, allowing offsetting exposures within a portfolio to be more limited number of specific price movements. properly netted and reflected in the capital charge. The scenario approach is included in the SFA capital rules as a preferred alternative. It is also similar to the SPAN system of margin requirements used by a number of deriva tives exchanges. Chart 1 M e t h o d s B a s e d o n P r ic e S e n s it iv it ie s — F ir s t - a n d W r it t e n C a l l O p t io n F ir s t - a n d S e c o n d -O r d e r A p p r o x im a t io n s o f a S e c o n d - O r d e r A p p r o x im a t io n s O p t io n v a l u e Another class o f value-at-risk rules relies on a simple approxi Initial underlying value = strike price = 100 Volatility = 15 percent Interest rate = 7 percent mation of an option’s price sensitivity to changes in the price of the underlying asset (see Appendix I). All of these rules use various combinations o f an option’s basic price sensitivity measures— the option’s delta and gam m a values. An option’s delta value is the change in the price of an option resulting from a small change in the price of the underlying asset. An option’s gam m a value measures the change in delta with respect to movements in the price of the underlying asset. The gam m a value can be used not only to improve upon the approximation obtained from the delta value but also to eval uate the quality of hedges— a delta-hedged options position with a large negative gam m a is vulnerable to large changes in the price of the underlying asset. The delta-equivalent rule, Digitized FRBNY Q 30 for FRASER u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 -150 I I I 1 I I I____I____I____I____I____I____1____I____I____I____1____I 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 U n d e r ly in g v a lu e Nevertheless, price sensitivity approximations, even applies to securities firms holding options for proprietary those that incorporate gamma values, may underestimate trading, and Paragraph (c)(2)(x), which applies to market- potential losses because the price sensitivity at current prices making firms. Both rules apply a lower capital charge for cer may not be representative of an option’s behavior at other tain offsetting options strategies, but Paragraph (c)(2)(x) rec prices. Option prices do not move in a linear fashion. Some ognizes more offsetting strategies than does Appendix A.3 options portfolios can have a very small gamma at the current The SEC has released for public comment (Federal price of the underlying asset but possess a much more nega Register 1994) a proposal that would permit broker-dealers tive gamma at different prices (for example, a written out-of- to calculate their capital requirements for listed options the money put). In this case, price sensitivity approaches can using a scenario-based approach similar to the one described underestimate the fall in the portfolio’s value resulting from a earlier in this section.4 According to the SEC’s proposal, large decline in the price of the underlying asset. “haircuts for options and related positions, when computed Both the scenario and simulation approaches can using this model, would more accurately reflect the risk also be combined with price sensitivity approximations. In inherent in broker-dealers’ option positions.” SEC require this case, the portfolio is not explicitly revalued at each simu ments for over-the-counter options would continue to be cal lated price change, but instead the changes in portfolio value culated using the strategy-based rules outlined in this paper. are approximated for each simulated price change. Although The strategy-based capital rules do not closely paral the approximations are based on option pricing models, their lel most firms’ risk control and/or trading systems because use will inevitably lead to a loss of accuracy relative to the the capital charges may not reflect portfolio-wide risks. For explicit use of the models themselves. instance, the risk of a delta-hedged short call strategy is equal to the risk of a delta-hedged covered call strategy with the I n c o r p o r a t io n o f V o l a t il it y R is k same set of parameters. However, the capital charges could be The discussion to this point has addressed only potential different depending on the strategy chosen to compute them. losses resulting from changes in the price of the underlying asset. Unlike other instruments, however, options are also Q u a n t it a t iv e A n a l y s is of P rice R is k R ules exposed to changes in price volatility. A capital adjustment This section assesses the performance of the two classes of for volatility risk can be incorporated in the price sensitivity rules discussed in the previous section. Under the value-at- approximation approach and the simulation or scenario- risk approach, we consider three different methods for con based methods. structing a capital charge: (1) the delta-equivalent rule, which bases the capital charge on the delta-equivalent S t r a t e g y - b a s e d R u les amount (that is, delta multiplied by the value of the underly The capital requirement under strategy-based rules is derived ing position); (2) the Taylor series rule, which supplements from a series of defined options trading strategies commonly the delta-equivalent rule with an adjustment for gamma employed by financial institutions. The strategy-based (positive or negative); and (3) the gamma rule, which supple approach may recognize offsetting for certain types of trades, ments the delta-equivalent rule with an adjustment for nega but in general, it does not easily accommodate the netting of tive gamma. Under the strategy-based approach, we consider opposite positions. Since strategy-based rules are collections the SEC’s Appendix A rule and C2X rule. of formulas that apply to various specific positions, it is We evaluate the accuracy with which these rules impossible to define a generic form for these rules. Thus, to measure the price risk of a variety of options positions and the test this type of rule, it is convenient to use its implementa degree to which they provide adequate capital levels for the tion by a particular regulatory authority, the SEC. portfolios’ potential losses. Broadly, the analysis performed in The SEC’s Net Capital Rule (15(c)3-l) for options this exercise reveals that the approximate value-at-risk rules contains two strategy-based rules: Appendix A, which tend to estimate portfolio risk more accurately and thus pro F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 31 duce more efficient levels of capital than the simple strategy- short position, where gamma is negative. based rules. Within the value-at-risk category, the gamma Each sample set contains thirty-five portfolios con and Taylor series rules provide a more accurate measure of structed from different combinations of common options potential losses than does the delta-equivalent rule, high positions. Among the thirty-five portfolios are several that lighting the importance of an adjustment for gamma risk. correspond to strategies recognized directly by the SEC rules. Other portfolios are then added by building on the SEC M strategies. Portfolios include naked positions, simple strate eth odology The riskiness of an options portfolio is measured as the maxi gies, and various combinations of SEC-recognized strategies. mum potential loss that would occur over a range of changes Delta-hedged portfolios are also included to generate results in the underlying asset price. Using a scenario-based method that more closely parallel gains and losses from realistic trad ology, we calculate gains and losses by revaluing each portfo ing practices. The composition of each portfolio and the para lio over a set of postulated price and volatility changes to find meters used to compute options prices and risks are listed in the largest loss (price and volatility assumptions are listed in Table 1.6 Portfolio sizes are normalized to ensure that differ Table 1). Specifically, the underlying price is varied within a ences in potential losses across portfolios result from the port folios’ inherent riskiness rather than their size.7 Within the value-at-risk category, the gamma and Taylor series rules provide a more accurate measure o f potential losses than does the deltaequivalent rule, highlighting the importance o f C o m p a ris o n o f C a p i t a l R u le s E f f ic ie n c y o f C a pita l R u l es To evaluate the efficiency of individual rules, we perform a linear regression of the capital charge produced by a given rule on the largest loss as measured by the scenario approach discussed in the previous section. There are thus thirty-five an adjustment fo r gamma risk. data points in each regression, corresponding to the thirtyfive portfolios considered. The R 2 coefficient, which is one range of ± 3 standard deviations of one-month price changes measure of correlation between two variables, reflects the in discrete increments of 5 percent of the initial price. The strength of the linear relationship between portfolio losses volatility of the underlying price is varied within a range of and capital requirements. Rules with higher R 2 coefficients ± 5 percentage points of the initial volatility in increments of tend to assign capital levels that are more closely correlated to one percentage point.5 The parameters used in this analysis potential losses. R 2 coefficients for the rules examined are provide a consistent benchmark for comparing the different summarized in Table 2. capital rules. Although the parameters are intended to be The Taylor series and gamma charge rules display a realistic, they are not intended to represent a prescribed stronger relationship between capital charges and losses than absolute level of confidence. does the delta-equivalent rule. The reason is that the capital We measure potential losses for a set of options port charge of the delta-equivalent rule is based only on a portfo folios with 30 days to maturity and for otherwise equivalent lio’s delta value at a given point in time and therefore does not portfolios with 180 days to maturity. Analyzing the effect of capture the nonlinearity of an option’s price changes. In con an option’s time to maturity on the accuracy of a rule is trast, both the Taylor series and gamma charge rules approxi important because of its influence on an option’s time value mate the option’s nonlinear price behavior by reflecting the and gamma. Gamma is substantially larger for options near gamma values in the capital computation. ing expiration— especially around the money— than for Overall, both strategy-based rules— Appendix A options with longer time to maturity. The effect of gamma and C2X — exhibit significantly lower R 2 coefficients than risk is generally greatest for portfolios consisting of a net the value-at-risk rules, suggesting that capital charges de- Digitized 32 for FRASER FRBN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Table 1 O p t io n s P o r t f o l io s * * 1 2 * 3 4 Portfolio Position Strike Price Short call-in the money Short 1 call 90 Short call-o u t o f the money Short call—at the money Short 1 call Short 1 call 110 100 Delta-hedged/short call-in the money Portfolio #1, delta-hedged 5 Delta-hedged/short call-ou t o f the money Portfolio #2, delta-hedged 6 Delta-hedged/short call—at the money Portfolio #3, delta-hedged * 7 Covered short call-in the money Long 1 underlying Short 1 call 90 * 8 Covered short call-out o f the money Long 1 underlying Short 1 call 110 * 9 Covered short call-at the money Long 1 underlying Short 1 call 10 Synthetic long futures w ith split strike Strikes o f put and call are different. Long 1 call Short 1 put * 11 Bull call spread Strike o f short is higher than strike o f long. Long 1 call Short 1 call 12 Delta-hedged/bull call spread Portfolio # 11, delta-hedged * 13 Bear call spread Strike o f short is lower than strike o f long. Long 1 call Short 1 call 100 110 90 95 115 115 95 14 Delta-hedged/bear call spread Portfolio # 13, delta-hedged * 15 Bear call spread/symmetric to # 15 Sym m etrically in the money and out o f the money, as in #15 Long 1 call Short 1 call 16 Delta-hedged/bear call/symmetric Portfolio # 15, delta-hedged * 17 Short straddle Short 1 call 18 Delta-hedged/short straddle Portfolio # 17, delta-hedged * 19 Protected call-in the money (synthetic put) Short 1 Underlying Long 1 call 90 * 20 Protected call-ou t o f the money (synthetic put) Short 1 Underlying Long 1 call 110 * 21 Protected call—at the money (synthetic put) Short 1 Underlying Long 1 call 100 22 Delta-hedged/protected call-in the money Portfolio # 19, delta-hedged 23 24 Delta-hedged/protected call-o u t o f the money Portfolio #20, delta-hedged Delta-hedged/protected call—at the money Portfolio # 21, delta-hedged 105 85 100 110 * 25 Short put—in the money Short 1 put * 26 Short p u t-ou t o f the money Short 1 put 90 * 27 Short put—at the money Short 1 put * 28 Long put—in the money Long 1 put 100 110 * 29 Long p u t-o u t o f the money Long 1 put 90 * 30 Long p u t-at the money Long 1 put 31 Ratio call spread 2:1 ratio Long a call with a lower strike and short 2 otherwise identical calls w ith a higher strike Long 1 call Short 2 calls 100 100 110 32 Delta-hedged/ ratio call Portfolio #31 delta-hedged 33 Ratio call backspread 2:1 ratio Short a call w ith a lower strike and long 2 otherwise identical calls with a higher strike Long 2 calls Short 1 call 34 Delta-hedged/ratio call backspread V ertical box Combination o f a bull spread and a bear spread Long call at higher strike and short call at lower strike Long put at higher strike and short put at lower strike 35 100 90 Portfolio #33 delta-hedged Long 1 call Short 1 call Long 1 put Short 1 put 115 95 115 95 Note: The parameters used in the portfolio simulation program are as follows: U nderlying price = 10 0 Annual dividend rate = 0 percent Annual volatility = 3 0 percent Interest rate = 3.5 percent Time to expiration = 1 8 0 days (set 1) 3 0 days (set 2 ) *Option strategies that are officially recognized under the SEC’s Appendix A and C 2X capital rules. F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 33 rived under the simple strategy-based approach are not closely Appendix A slope coefficient is significantly above one for the correlated with the potential losses of options portfolios. The 30-day options, and below one (although not significantly so) low R 2 could be attributed to the fact that strategy-based for 180-day options. The C 2 X coefficients are significantly rules provide only a rough measure of the portfolios’ risk, lower than those of the other rules and are closer to zero than largely because they do not account for internal risk manage to one. ment practices or changing market conditions. The capital coverage of the various rules is captured One notable exception to this overall finding occurs graphically in Charts 2 through 6. The “45 degree” lines in when options are nearer to maturity (30-day series). In this the charts depict capital-to-loss ratios o f 1:1. The points case the correlation between capital charges and potential above the 45 degree line in charts 2 through 6 represent port losses is stronger under Appendix A than under the delta- folios for which the capital requirements exceed portfolio equivalent rule. The consistently high capital charge of the losses. The points falling below the 45 degree line represent Appendix A rule appears to mesh well with the high gam m a portfolios for which the capital requirements are insufficient risk often associated with short-dated options.8 In contrast, to cover portfolio losses. From the charts, it is clear that the the delta-equivalent rule performs less efficiently in this case gam m a and Taylor measures produce the patterns most sim i because it does not capture gam m a risk. lar to the 45 degree line itself, while the patterns for strategybased rules are the least similar. Another approach to the evaluation of the methods C o v e r a g e o f C a p it a l R u le s C apital rules should provide not only a high correlation is to exam ine the “errors” produced by each m ethod. In between capital requirements and losses but also adequate Charts 2 through 6, capital deficits and surpluses can be mea coverage of portfolio losses. One way to evaluate coverage is sured by the vertical distance, expressed in absolute dollar to compare the slope coefficients of regression lines produced amounts, of a portfolio from the 45 degree line. To compare by each rule. A slope coefficient gives the marginal amount of the aggregate m agnitude of the capital shortages and excesses capital that corresponds to a $ 1 difference in the risk exposure resulting from each rule, we summed the portfolio deficits of a portfolio. For example, a slope coefficient of 1 indicates and surpluses separately (Table 2). that an incremental loss of $1 is associated with $1 more of From a supervisory perspective, the gam m a charge capital. The slope coefficients are also summarized in Table 2. rule appears slightly more attractive than the Taylor series The differences between slope coefficients for the rule. The total deficit is higher under the Taylor series rule Taylor series and gam m a charge rules are slight. In contrast, than under the gam m a charge rule. As Appendix I explains in the delta-equivalent rule produces m aterially lower slope greater detail, the Taylor series rule gives capital credit when coefficients than do the gam m a and Taylor series rules. The ever gam m a is positive, while the gam m a rule does not pro- Table 2 C om pa rison of V alue - a t -R isk a n d Stra teg y -B ased R ules Statistic 30 -D ay O ptions R 2 coefficient Slope coefficient Total capital deficit Total capital surplus 180-D ay O ption s R 2 coefficient Slope coefficient Total capital deficit Total capital surplus Taylor Gamma Delta Appendix A C 2X 0.842 0.96* 188 173 0.8 28 0 .9 1* 158 274 0.381 0.52 497 204 0.555 1.54 131 1176 0 .0 26 0.1 9 634 580 0 .9 74 1.02* 60 16 0.927 1.02* 33 74 0.667 0.6 9 187 62 0.1 12 0.6 7* 112 695 0.0 10 0.12 253 325 *The hypothesis that the slope coefficient equals 1 cannot be rejected at the 5 percent level. 34 FRBN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Chart 2 T a y l o r S e r ie s R u l e C a p it a l C a p it a l ch arge La r g e s t ch arge L lo ss a r g e s t lo ss Note: To maintain comparability of scale across charts, the coordinate pair (161, 205) is not shown in the left panel. Chart 3 G a m m a C h a r g e R ule C a p it a l C a p it a l ch arge ch arg e J 200 La r g e st La r g e s t lo ss lo ss Note: To maintain comparability of scale across charts, the coordinate pair (16 1, 205) is not shown in the left panel. F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 35 Chart 4 D e l t a Eq u iv a l e n t R ule C a p it a l ch arge C a p it a l La r g e s t ch arge lo ss La r g e s t loss Chart 5 SEC A p p e n d ix A R u le C a p it a l C a p it a l ch arge La r g e s t La r g e s t lo ss Note: To maintain comparability of scale across charts, the following coordinate the right panel. 36 FRBN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll ch arge 1994 lo ss are not shown: (100,264) (149,408) in the left panel and (2,99) (53,149) in vide for such credit. Under the delta-equivalent rule, the gy-based rules may be partly attributable to their intended m agnitude of uncovered losses is much greater than under coverage of nonprice risks. When the results of the strategy- either o f the foregoing m ethods. In Chart 4, the capital based rules are examined on a portfolio-by-portfolio basis, charges computed for several portfolios under the delta rule however, a systematic pattern appears. For portfolios repre diverge significantly from the 45 degree line. senting SEC-recognized strategies, the strategy-based rules If we turn from the value-at-risk approaches to the tend to underestim ate the potential loss, generating very strategy-based rules, we find that Appendix A and C 2 X tend sm all capital requirem ents. For p ortfolio s representing either to underestimate risk significantly, yielding large capi strategies not recognized by the SEC, the rules tend to overes tal deficits, or to overcompensate for risk, producing large timate the potential loss, generating correspondingly large capital surpluses. The C 2 X rule, which is designed for mar capital requirements.9 ket makers on the trading floor, could result in especially Finally, all of the rules provide a better estimate of large uncovered losses. It produces the highest deficits among potential losses for the set o f portfolios with 180 days to all of the capital rules considered. m aturity than for the set with 30-day options. As noted The Appendix A rule appears to perform well when options with 30 days to maturity are considered. However, above, this result can be a ttrib u ted to the high level of gam m a risk associated with short-dated options. Table 2 shows that Appendix A also leads to significant capi In summary, our evaluation o f the efficiency and tal surpluses. By incorporating only simple trading strate coverage of the capital rules indicates that the approximate gies, Appendix A does not entirely allow for hedging or off value-at-risk rules tend to perform better than the simple setting portfolio effects, causing substantial excess capital strategy-based rules. W ithin the category o f value-at-risk burdens for a number of portfolios. rules, the gam m a and Taylor series rules provide relatively The large capital surpluses produced by the strate- sim ilar results. Both outperform the delta-equivalent rule, Chart 6 SEC C 2 X R u le C a p it a l ch arge C a p it a l La r g est lo ss ch arge La r g e st lo ss F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 37 and the gamma rule tends to be a bit more conservative than Consequently, these methods do not create excessive report the Taylor series rule. ing burdens for banks that rely on only a few simple options strategies, primarily for hedging purposes. But while the V o l a t il it y R is k A d d - o n s capital requirement of simple strategy-based methods may To investigate the potential benefits of a volatility risk add be relatively easy to calculate, these methods can lead to sub on, we supplement the Taylor series rule with a charge equal stantial reporting burdens for more sophisticated institutions to the absolute value of the vega of each position multiplied that carry out a wide range of trading strategies, since each by a 5 percentage point change in volatility.10 Since vega rises type of position would require a different capital requirement. with the tenor of the option, the volatility add-ons are sub In addition, capital requirements that rely exclusively on the stantially larger for the options with remaining maturity of strategy-based method may compel regulators to revise their 180 days than for those with remaining maturity of 30 days. capital requirements repeatedly as financial institutions Across all of the positions, the add-on specification increases develop new options instruments and trading strategies. total required capital by 4.2 percent for options with remain Com pared w ith the sim ple strategy m ethods, ing maturity of 30 days and 14.4 percent for options with methods based on price sensitivity require a higher level of remaining maturity of 180 days. sophistication on the part of both banks and regulators. Nev For the 30-day positions the vega add-on has little ertheless, price sensitivity methods rely on standard risk tangible effect on measures of capital efficiency. The R 2 management techniques already employed by most major increases from 0.842 to 0.844, while the slope increases from financial institutions to manage the risk of their options port 0.96 to 0.97. With respect to coverage, the total capital folios. Most larger banks use internal risk management sys deficit decreases from 188 to 168, while the surplus increases tems that track on an ongoing basis the delta, gamma, and from 173 to 189- The effect of the volatility risk add-on is volatility risks of their options portfolios. As a result, in the larger for the 180-day option positions. The R2 of the regres sion of capital on largest loss rises from 0.974 to 0.978, while the slope rises from 1.02 to 1.06. The total capital deficit declines from 60 to 21, while the surplus rises from 16 to 31. In summary, a volatility risk add-on certainly does not reduce the efficiency of the capital charge, but neither does it markedly increase it relative to the Taylor series rule. The volatility risk add-on does increase coverage of the largest losses as measured by the total capital deficit. an d methods based on price sensitivity require a higher level o f sophistication on the part o f both banks and regulators. Nevertheless, price sensitivity methods rely on standard risk management techniques already employed by O t h e r C o n s id e r a t io n s S im p l ic it y Compared with the simple strategy methods, most major financial institutions to manage R e p o r t in g B u r d e n In general, the introduction of specific supervisory capital requirements for the market risks arising from banks’ options the risk o f their options portfolios. positions requires a certain level of sophistication on the part of both banks and supervisors. Nevertheless, the proposals case of larger options players, the price sensitivity approach examined in this paper vary significantly in their complexity should impose lower reporting burdens than the strategy- and therefore in their potential reporting burden. based approach because the methodology reflects banks’ Strategy-based methods can be easily applied by existing hedging practices and can be applied uniformly to even the least sophisticated options players because they do all options positions. In addition, the price sensitivity not require the use of complicated options pricing models. approach can simplify the task of regulators: it is flexible FRBN Y 38 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 enough to incorporate market innovations in options instru St ra te g y- base d M ments and trading strategies without requiring additional The principal advantage of strategy-based capital require capital standards. ments is their simplicity. Strategy-based methods do not eth ods Scenario methods also tend to be more complex than require the use of option pricing models and are therefore simple strategy-based methods. However, like the price sensi most appropriate for banks that carry out only a limited tivity approach, these methods rely on variables that most amount of options business, primarily for the purpose of larger banks already monitor through their standard risk man hedging. However, simple strategy-based methods can only agement systems. Simulation methods are the most complex roughly estimate the potential losses of an options portfolio. to implement and may lead to excessive reporting burdens for all but the most sophisticated banking institutions. A further drawback of the strategy-based method is that it may pose reporting problems for sophisticated options players. Each trading strategy is subject to a different capital C o n f o r m it y w it h E x is t i n g an d P r o po se d R ules requirement, and the approach is generally unrelated to the The methods examined in this article also vary in the degree internal risk management systems employed by banks. of their conformity with existing capital measurement Moreover, the method lacks the flexibility to incorporate approaches. By combining credit risk and market risk in one future market developments in options instruments and capital charge, the simple strategy method conflicts with the trading strategies. As a result, regulators must continuously SFA, CAD, and proposed BIS market risk standards, which upgrade and expand their capital requirements. apply separate capital charges for the two types of risk. In addition, while the SFA, CAD, and BIS proposals allow for a V a l u e -at- R is k M limited number of simple strategy-based trades, none of The value-at-risk methods examined above differ from each these supervisory approaches permits an across-the-board other primarily in the treatment of gamma risk. One advan application of strategy-based methods. In contrast, the price tage of these methods is that they can be incorporated in the sensitivity approach can easily be incorporated into the capi framework of the SFA, the CAD, and the proposed BIS and tal framework of the SFA, the CAD, and the BIS and SEC SEC market risk capital requirements. proposals, each of which allows for a price-sensitivity-based D e l t a - E q u i v a l e n t M e t h o d . The delta-equivalent rule method. provides banks with a relatively simple capital rule. Since eth ods The SFA allows a scenario method, while the BIS most banks that are active in options activities generally mea market risk proposal considers permitting scenario or simula sure delta risk as part of their overall risk management strategy, tion methods as alternatives to the price sensitivity approach. the delta-equivalent method would not present an excessive However, because the scenario approach does not strictly reporting burden for these banks. Further advantages of the employ the building block methodology as defined by the delta-equivalent method are that it can be applied to all BIS market risk proposal— which distinguishes specific from options positions uniformly, it may allow for portfolio effects, general market risk— it might have to be “carved out” of the and it is flexible enough to incorporate market innovations. building block methodology of the BIS framework. However, the delta-equivalent rule can seriously underestimate potential losses in the case of large price move Sum m ary of F in d in g s ments because it does not incorporate gamma risk. In addi We have compared various supervisory approaches to the cap tion, it cannot by itself account for changes in the volatility of ital treatment of the market risk of options. Our comparison the underlying instrument. rested on three criteria: capital coverage of potential losses, D e l t a - E q u iv a l e n t M e t h o d w it h G a m m a A d j u s t m e n t . simplicity and reporting burden, and conformity with exist The delta-equivalent method with gamma adjustment is ing and proposed supervisory approaches. The main findings similar to the delta-equivalent method except that it also of the analysis are summarized below. modifies the capital requirement for the risk that delta will FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 39 change as the price of the underlying moves. Our calculations sophistication but generally rely on options pricing models show that adjusting for gam m a risk (with either the Taylor or to measure the gains or losses on each option position in gam m a rules) typically leads to a significantly im proved response to defined movements in the price or volatility of the approximation o f an options portfolio’s potential losses. underlying instruments. Scenario methods with full revalua D elta -E q u iv a l e n t M e t h o d w it h G a m m a a n d V o l a t il it y tion implicitly account for delta and gam m a risk, and they A d ju s t m e n t s . In addition to measuring delta and gam m a can also explicitly account for volatility risk. As a result, these risk, this approach also takes into account volatility risk, the methods may provide a more exact measure of potential losses risk that the variability of the underlying instrument’s price than the approxim ation-based price sensitivity m ethods. will increase or decrease, affecting the price of the option. A Sim ulation m ethods, using either real historical data or a volatility adjustment could play an important role, depend Monte Carlo methodology, may facilitate the incorporation ing on the specific portfolios involved.11 of other variables, such as an option’s time value or its sensi S c e n a r io - b a se d and S im u l a t io n M e t h o d s w it h tivity to changes in interest rates. F u l l R e v a l u a t i o n . Scenario m ethods can vary in their A p p e n d ix I: D etailed D esc r ipt io n o f V alue - at -R isk R ules B asic F r a m e w o r k replaced by a measure o f the asset’s price volatility. For exam The change in the value (V) o f an options portfolio associated ple, with a m ultiple m of the standard deviation of the price with a given change in the price (u ) o f the underlying asset is change, equation 2 becomes: given by: <3) (1) AV = V ( « + A # ) - V («). AV = ^ (mCT) + 2 (mCT>2' The m ultiple of the standard deviation determines the confi T h is change may be approxim ated using a second-order dence level of the capital rule.13 The m agnitude o f value at risk typically depends on Taylor series expansion: both the portfolio’s sensitivity to changes in the price of the /o\ (2) <3V a 1 d2V a 2 underlying asset and the price volatility o f the underlying AV ' T u A * + 2 ~Su1 Aa ’ asset. In term s o f equation 3, if the p ortfolio were well where AV and A u are changes in the values o f the portfolio hedged, then the price sensitivity terms dNIdu and d2V/d&2 and the underlying asset, and the partial derivatives dVIdu would both be small because of the offsetting positions in the and d2Y ldu2 are the portfolio’s delta and gam m a values.12 portfolio, thereby leadin g to a low capital requirem ent. To determine the size o f the capital requirement, the Given the price sensitivity of the portfolio, however, the capi measure o f price sensitivity is combined with a measure o f the tal requirement would also be smaller if the underlying asset volatility of the underlying asset’s price. In terms of equation had a small price volatility: a smaller likelihood o f large price 2, the change in the price o f the underlying asset, A u, is changes implies that potential losses will be smaller. FRBN Y 40 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 A p p e n d ix I: D e t a i l e d D e s c r ip t io n o f V a lu e - a t - R is k R u le s M eth o d s B ased o n P rice S en sitiv ities — F irst - a n d S e c o n d -O r d e r A pp r o x im a t io n s Delta-Equivalent Rule. The delta-equivalent rule uses only the first term in equation 2, the delta value of the portfolio: (Continued) price increase as well as a price decrease (of magnitude mcr) is applied to the Taylor series equation, and the capital charge is the largest resulting loss: /e\ v- i • r/dV „ 1 d2V, o w dV, \ 1 d2V , n1, (5) K = lm in C (¥ m a + 2 ^ ( m a ) 2) , ( ^ ( - m a ) + 2 du2 (ma)2),0]L (4) K -I^ m al, For a book of written options (when options risk is where K is the capital charge and m is a m ultiple of the stan most extreme), the Taylor series rule will always produce a dard deviation that determines the degree of confidence in larger capital requirement than the delta-equivalent. How the capital rule. The delta-equivalent is the basic component ever, for a book of purchased options, the Taylor series rule of the C A D ’s rule and the proposed BIS rule. will require less capital than the delta-equivalent because the latter overestimates the price risk o f purchased options. Example. Consider a portfolio consisting of a $ 1 0 0 Gamma Charge Rule. A more conservative application o f long position in the underlying asset and a written gam m a values is to use the sum of the absolute values o f delta call option on $ 1 0 0 of the asset. If the delta of the and gam m a (when gam m a is negative): option equals .2 5 , then the delta of this portfolio is (6) K = l^(ma)l + lmin(^-^(maf,0)l. equal to .75 (the written call has a negative delta of —.25 and the underlying has a delta o f 1, where both This gam m a charge capital requirement is one of the alterna are weighted equally since the underlying amount is tive rules in the SFA capital requirements. the same for each position). If we assume that a three The capital charge in this rule is at least as large as standard deviation confidence interval for the capital the Taylor series rule (equation 5) because of the way the delta rule implies a $20 change in the price of the under and gam m a interact in the Taylor series when they have dif lying asset, then the capital charge will be $15 ferent signs. For a portfolio o f w ritten options (negative ($ 1 5 gam m a), the Taylor series rule and the gam m a charge rule = ( .7 5 ) x ($ 2 0 ) ) . produce identical capital charges. In a portfolio of purchased The delta-equivalent can often underestimate risk options, however, the delta term will overestimate potential because it is a linear approximation to the price sensitivity of losses, and in the Taylor series rule, the positive gam m a (of options, which are inherently nonlinear. This approximation the long options) will reduce the capital charge by moderat error can be reduced by incorporating an option’s gam m a ing the effects o f the delta term in equation 5. The gam m a value, a measure o f the nonlinearity o f an option’s price sensi charge rule (equation 6), however, does not provide such tivity. The augmented delta-equivalent rules include such an “credit” for the delta’s overestimate o f risk. In this case (a adjustment. book of purchased options), the delta-equivalent and the Taylor Series Rule. One use of the gam m a value to augment gam m a charge rule produce the same capital charge. the delta-equivalent rule is the direct application o f the Tay lor series in equations 2 and 3. An option’s price sensitivity Example. Recall the earlier example o f the delta- can be different for increases and decreases in the price of the equivalent rule, and suppose that the option had a underlying security. Hence, to obtain the capital charge, a gam m a value o f -0 .1 . The gam m a charge, in this F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 41 A p p e n d i x I: D e t a i l e d D e s c r i p t i o n o f V a l u e - a t - R i s k R u l e s case, is $20 =1 (—0.1) ($20)2/2 I, and the total capital (Continued') or, if an approximation is used, requirement is $35 = $15 + $20. (The $20 term that is squ ared in the ca lc u latio n is derived from a v o la tility o f 20 p ercen t w eigh ted by the $ 1 0 0 underlying asset; the $35 term is the sum o f the where Acr denotes the change in volatility. In equation 8, the d e lta ch arge from the earlier exam p le and the options’ volatility sensitivity (vega or lambda) is weighted by gam m a charge.) the change in volatility, while in equation 7 the options are revalued at different volatilities. The SFA capital require I n c o r p o r a t io n o f V o la tility R isk ments incorporate such a volatility risk add-on, using a one C ap ital coverage for v o latility risk can be incorporated percentage point change in volatility. Although equation 7 is through an additional capital charge. This charge would be a more exact estim ate, an option’s sensitivity to volatility based on changes tends to be a linear relationship, and the use o f a volatility sensitivity measure as in equation 8 is a reasonable (7) estimate of volatility risk. V (a + Act) - V(cr) A p p e n d ix II: E xam ples o f S trategy - based R ules The following two examples of strategy-based rules are taken Strategy 2: Long call option and short the underlying stock from the SE C ’s Appendix A rule, which applies to securities firms holding options for proprietary trading. Capital charge: If the call is out-of-the-money, the lesser of a) 30 percent of the underlying, or S t r a t e g y 1 : Long call option b) the out-of-the-money amount. If the call is in-the-money, no charge is Capital charge: 50 p ercen t o f the m arket value o f the option. 42 FRBN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 required. Endnotes 1. A more detailed discussion of options risk and supervisory capital rules 8. Appendix A appears to overcompensate substantially for losses of a few is found in a related paper by the same authors: “Options Positions: Risk portfolios, resulting in points with extreme capital levels: capital-to-loss Measurement and Capital Requirements,” Federal Reserve Bank of New ratios under the Appendix A rule are as high as 69 :1, in contrast to 6:1 under the delta equivalent rule. Charts 4 and 5 illustrate the differences in capital levels. York Research Paper. 2. The paper cited in the preceding note provides a detailed account of specific methodologies actually used or proposed by regulatory authorities. 9 . Portfolios that have extreme capital-to-loss ratios are synthetic long 3. For examples of specific strategy-based rules, see Appendix II. futures, delta-hedged protected call, delta-hedged bull call, delta-hedged bear call, ratio call spread, protected call, delta-hedged ratio call spread, 4. In the interim, the SEC’s Division of Market Regulation issued a “no action letter” on March 15, 1994, that effectively allows broker-dealers to use the proposed method. 5. The annual volatility of underlying price changes is assumed to be 30 percent. Historical volatility for the U.S. S&P 500 index has typically ranged from 10 to 20 percent, but individual stocks can display much higher volatilities. The average volatility for individual stocks in the Dow delta-hedged ratio call back spread, and box spread. The portfolios that represent SEC-recognized strategies are so indicated in Table 1. 10. Vega is defined as the change in the price of an option in response to a unit change in the volatility of the underlying asset. 11. Volatility risk is largest for one-sided portfolios (all written or all pur chased), for options with longer maturities, and for options that are close to the money. Jones Industrial Average is currently around 30 percent. 12. These delta and gamma values represent net portfolio values that are 6. Capital requirements under the SEC rules could differ depending on the sequence in which recognized strategies are pulled out from a portfolio. the arithmetic sum (taking account of the signs) of the deltas and gammas of all instruments and transactions in the portfolio. Recognizing that a firm is likely to use the lowest cost interpretation of the capital rule, we chose the sequence that yielded the lowest level of capital requirement. 13. Although the degree of confidence provided by a specific multiple of the standard deviation depends on the precise underlying probability dis tribution, a rough idea of the degree of protection provided by, say, two or 7. The normalization method used is a delta-based normalization; that is, the larger of the sum of gross negative delta-equivalent values or the sum of three standard deviations may be obtained. Given a normal probability dis tribution, at three standard deviations, losses will exceed capital with a gross positive delta-equivalent values of the options in a portfolio, excluding positions in the underlying asset, is normalized to $ 100.00. Further discussion likelihood of 1 in 700, whereas at 2.33 standard deviations, losses will exceed capital with a likelihood of one in a hundred. of approaches to normalization is contained in the paper cited in note 1. R eferen ces Bank, for International Settlements. 1994. A n n u a l R e p o r t . F e d e r a l R e g is te r . 1994. Vol. 59, No. 54, March 21. N otes FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 43 Making Sense of the Profits of Foreign Firms in the United States David S. Laster and Robert N. McCauley T he scant profit of foreign firms operating in the industry (Appendix I). United States has emerged as one of the biggest The pretax income reported by foreign firms has puzzles in international finance. That 4.7 mil remained low, dropping to $4 billion in 1990, according to lion workers using $1.8 trillion in assets to gen the most recently published Internal Revenue Service data erate sales of $1.2 trillion could fail to turn a profit strikes (Chart 1, bottom panel). Had foreign firms earned the same many as unbelievable. Could foreign companies have paid return on sales as U.S. firms, they would have made an addi $316 billion in the past decade for firms earning $10.7 bil tional $32.1 billion in profits. Some claim that such addi lion in the year before acquisition only to lose money overall tional profits would yield the U.S. Treasury substantially on their holdings in 1992— a year in which U.S.-owned firms larger tax revenues each year (U.S. Congress, House Commit earned record profits?1 tee on Ways and Means 1990, pp. 186, 250, 288, 300). Although foreign firms have earned lower U.S. prof This article attributes the depressed earnings of for its than their domestic counterparts since World War II, the eign firms to the firms’ rapid buildup of U.S. operations in gap has widened substantially in the last two decades. In the late 1970s and the 1980s. These companies paid top dol manufacturing, the gap in return on equity averaged 3.4 per lar for underperforming U.S. firms, borrowed heavily, and cent in 1951-75, then doubled to 6.8 percent in 1976-80, then spent freely on investment and marketing. As the share and reached 8.8 percent in 1981-91 (Chart 1, top panel). of recently acquired foreign firms in the United States rose in Returns worsened in petroleum, wholesale and retail trade, the 1980s, aggregate returns deteriorated. and finance and insurance as well (Chart 1, middle panel). The article also investigates two other explanations Realized returns also deteriorated in real estate, and mark-to- often advanced for the low returns of foreign firms: 1) a weak market losses wiped out much of the foreign stake in this dollar has depressed the firms’ profits, and 2) foreign firms are 44 FRBN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Chart 1 P r o f i t a b i l i t y o f F o r e i g n F i r m s in t h e U n it e d S t a t e s P ercen t 20 Return on Position in Manufacturing The longstanding gap between the returns of foreign-owned manufacturing firms in the United States and U.S. manufacturing firms widened in the late 1970s and again in the 1980 s. 15 10 5 0 1951 60 70 80 90 93 P ercen t 10 Return on Sales by Sector of Foreign Firms in the United States 6 Foreign firms' returns in other sectors deteriorated in the 1980 s as well. , 4 2 0 -2 B illio n s o f d o lla r s 40 30 As a result, the taxable income of foreign firms remained low. 10 0 -10 1983 84 85 86 87 88 89 90 Sources: U.S. Department of Commerce, Bureau of the Census and Bureau of Economic Analysis, balance o f payments data (top panel) and operations data (middle panel); Internal Revenue Service (bottom panel). Notes: In the top panel, petroleum products are included in U.S. manufacturing but excluded from foreign manufacturing; before 1974, U.S. manufacturing included affiliates abroad. Top panel plots both post-tax profit and interest earned by foreign firms as a percentage of their equity and debt claims on affiliates. Position is average of beginning-of-year and end-of-year data on a historic-cost basis; position so measured was only 14 percent lower than the position on a current-cost basis in 1993. Middle panel plots only post-tax profit as a percentage of sales. Bottom panel plots pretax profits. FRBNY Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 45 understating their earnings to avoid paying U.S. taxes. We acquisitions market. Because foreign firms generally denomi find no clear evidence for the first claim and some support for nate their U.S. affiliates’ debt in dollars, any cheap foreign the second. Firms with the most incentive and opportunity to currency debt confers little advantage. The cost of foreign shift profits out of the country report lower profits than other equity matters far more. When the stock exchange in Tokyo firms. Nevertheless, the rapid rate at which foreign firms or London places a higher value on a given stream of earnings divest their U.S. subsidiaries suggests that many investments than does the N ew York Stock Exchange, a Jap an ese or really have performed poorly. British firm can outbid a U.S. firm and still satisfy its share The last section of the article considers the im plica holders. In the late 1970s, foreign companies took advantage tions of our findings. Ju st as the rush of foreign acquisitions of low U.S. equity prices in the first postwar wave of foreign in the 1980s depressed returns, so the subdued pace of such acquisitions. The more sizable surge of foreign acquisitions in acquisitions in the 1990s points to higher returns in the near the 1980s drew strength from the high valuations in foreign future. Improved profits of foreign firms should narrow the equity markets, especially the Japanese market. R egression analysis o f acquisitions by companies internal or fiscal deficit but widen the external, current from the seven m ajor source countries o f foreign direct account deficit. investment shows the importance of both foreign and U.S. A c q u is it io n s : C auses a n d C o n seq u en c es equity prices in the tim ing of purchases over the 1980-92 The U .S. Com m erce D epartm en t defines foreign direct period. We set out to relate the variation over time in acquisi investment as a U.S. company or partnership in which a for tion activity (measured against home-country G D P) to equi eign entity holds a voting share of more than 10 percent. The ty prices, real interest rates, overall economic growth, the term “foreign direct investment” may conjure up images of exchange rate, unit labor costs, and other plausible variables construction workers building car factories in the Midwest. Yet such “greenfield” entry represents a small share of the increase in foreign holdings of U.S. corporate assets: for every dollar foreign investors spend to establish a new business, they spend five dollars to acquire existing ones.2 Chart 2 O u t l a y s b y F o r e ig n In v e s t o r s f o r t h e A c q u is it io n of U.S. B B il l io n s C auses o f F o r e ig n A c q u is it io n A c tiv ity in t h e 1 980s u s in e s s e s of 19 8 7 d ollars 70 60 Before exam ining how strong acquisition activity drove down the aggregate returns o f foreign com panies in the 50 U n ited States, let us consider the reasons for grow th in 40 acquisitions (Chart 2). As mergers and acquisitions acceler ated in the United States in the mid- to late 1980s, foreign firms won more and more bidding contests. After a wave of 30 20 activity in 1978-81 that carried the foreign share of U .S. acquisitions outlays to a fifth or a quarter of total U.S. m erg 10 ers, foreign acquisitions subsided only to surge to a third of 0 total activity in 1987-90 (Merrill Lynch Business Services 1992, pp. 7, 50). Foreign firms’ cost of equity advantage (McCauley and Zimm er 1994, 1989) permitted them to outbid domes tic firms for corporations “in play” in the U.S. mergers and FRBNY 46 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 1975 80 85 90 94 Sources: U.S. Department of Commerce, Bureau of Economic Analysis; Merrill Lynch Business Brokerage and Valuation, Mergerstat Review. Notes: Reported dollar amounts were converted from current to 1987 dollars using the GDP implicit price deflator. The 1994 datum is through June, annualized. (Table 1). Equity prices exert a substantial influence: a 10 per sitions: poor selection, high acquisition prices, heavy debt, cent rise in foreign prices relative to U.S. prices raises foreign and a rise in post-acquisition operating expenses. Profitabili ty improves only slowly over time as investment and consoli acquisitions by 18 to 32 percent.3 dation expenditures pay off, as new managers learn through C o n seq u en c es o f A c q u isit io n s experience, and as foreign paren ts sell o ff un successful The wave of foreign acquisitions in the 1980s raised the share acquisitions. of foreign-owned assets that were recently acquired. This The profitability of new acquisitions traces a J curve share can be measured as the ratio of outlays by foreigners in (Chart 4), dropping after acquisition and then recovering acquiring and establishing U .S. firms to the total foreign over tim e.5 Consider the factors behind this profile. ownership stake in U .S.-based firm s (Chart 3).4 By 1990 about half of foreign holdings in the United States had come P o o r S e l e c t io n . Foreign firms buying U.S. firms do not get into foreign hands by acquisition in the previous five years. the p ic k o f the litte r (C h art 5). In the 1 9 8 0 s, fo reign Another 10 percent o f foreign holdings had been newly multinationals bought U .S. manufacturers that were only a established in that same period. qu arter as p ro fita b le as a broad U .S . norm (Landefeld, The preponderance o f new acquisition s helps to Lawson, and Weinberg 1992, p. 83). explain the weak aggregate profitability o f foreign firm s because recently acquired firms tend to have low returns. We review several factors that depress the profits of recent acqui- Chart 3 I m m a t u r i t y o f F o r e i g n F ir m s in t h e U n i t e d S t a t e s Acquisition and Establishment Outlays over the Preceding Five Years in Relation to Foreign Direct Investment Position Table 1 R e g r e s s io n A n a l y sis o f F o r e ig n A c q u is it io n s in t h e U n ite d S ta te s, 1980-92 Percent Dependent variable: Acquisition outlays in the United States as a fraction of 70 Acquisitions and establishments (Bureau of Econoinic Analysis) source-country GDP for the United Kingdom, Japan, Canada, Germany, France, the Netherlands, and Switzerland 60 1 8 2 * ** 0 .4 2 Real U.S. equity price —3 .2 1 * * 1.42 - 0 .1 2 *** 0 .0 4 Foreign unit labor costs relative to U.S. 0 .7 4 0.60 Real foreign GD P growth 0 .0 8 0.05 Real foreign GD P growth, lagged 0 .0 8 * Real U.S. GD P growth Real U.S. GD P growth, lagged 0 .0 8 * 0 .0 7 * 0.05 0 .0 4 0 .0 4 Foreign currency per dollar Real foreign bond yield Real U.S. bond yield Exchange rate volatility Time trend 0.01 - 0 .20 ** -0 .1 7 0.05 0 .0 4 \ $ A / y \ # / i / i / i* /I i / # / 50 !!* !l!l!| I l® i 1 Real foreign equity price Independent Variable A $ / 40 *% Coefficient Standard Error Acqu isitions 30 — (Bureau o f Eco V - ' ' f 0 .0 9 2.20 0.10 t % \ V »% \ \\ *% \ I \ * \ * \\ t% % \ % 1 % V % \ \ \ % » \ \ » Acquisitions (Mergerstat) ** \\ % % % V 20 * N = 91 Adjusted R 2 = 0.55 Sources: U.S. Department o f Commerce, Bureau o f Economic Analysis; national sources. 10 i 1977 i i i . . . i i i i 80 -J. ..ill__ 1 85 ...... L.........J 90 i......1 94 Notes: Observations are weighted by 19 9 2 position. Dependent variables are de-meaned by country. A ll variables are expressed in logarithmic form except for GDP growth rates, bond yields, and the time trend. Estimated intercept is not reported. Real equity price is stock index deflated by GD P deflator. Real bond yield is nominal bond yield less contemporaneous inflation as measured Sources: U.S. Department of Commerce, Bureau of Economic Analysis; Merrill Lynch Business Brokerage and Valuation, Mergerstat Review. Notes: Foreign direct investment position is valued at current cost. The 1994 datum is through June, annualized. by GDP deflator. ♦Significant at 10 percent level. ♦♦Significant at 5 percent level. ♦♦♦Significant at 1 percent level. F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 47 O f course, foreign companies did not buy money- as the percent excess of the price per share at acquisition over losers exclusively. The m odal target made a profit in the price per share pre-bid, have not consistently shown that for broad range of the U.S. norm, and some were very profitable eign firms pay higher premia than domestic firms (Cakici, indeed (Chart 6). Nonetheless, year after year, 40 percent of Hessel, and Tandon 1991; Cebenoyan, Papaioannou, and the foreign firms’ targets lost money, and many lost substan Travlos 1992; Dewenter 1992; Swenson 1993; Harris and tial sum s.6 Ravenscraft 1991; K ogut and Chang 1991). Foreign firms could be paying premia similar to those paid by U.S. firms, H ig h A c q u is it io n P r ic e s . Foreigners pay full price for sub- say 30 to 50 percent, for a set of firms that contains more par performers. A s noted at the outset, in 1980-90, foreign money-losers than do the targets of U.S. firms. In that case, firm s p a id $ 3 1 6 b illio n for firm s th at had earn ed an aggregate of $10.7 billion in the year before acquisition. This price-earnings ratio of nearly thirty-to-one well exceeded the F oreign firm s b u yin g U.S. firm s do not g et the price-earnings ratio for the Standard and Poor’s 500, which pick o f the litter. ranged from the high teens to the low twenties in the late 1980s, when most of the acquisitions took place.7 The observation that foreign firms in aggregate pay prices well above market norms for a dollar of earnings need foreign firms would not overpay by the test of these studies not imply that they pay more for their U.S. targets than do but would nonetheless pay a hefty aggregate price-earnings U .S. acquirers in m erger contests. Takeovers, foreign or ratio. Similarly, if foreign firms concentrated their purchases dom estic, require a prem ium to be paid over usual share in periods with higher acquisition premia (that is, the late prices. Thus, studies measuring acquisitions premia, defined 1980s) or in industries with high premia, then they could pay a full price without paying more than U.S. acquirers in Chart 4 Chart 5 P r o f i t a b i l i t y o f F o r e i g n A c q u i s i t i o n s in t h e U n it e d St a t e s Foreign-owned Manufacturing Firms P r o f it a b il i t y o f Fo r e ig n A c q u is it io n s in Y e a r b e f o r e A c q u is it io n C o m p a r e d w it h P r o f it a b il it y o f A ll U .S . M a n u f a c t u r i n g F i r m s R e t u r n o n s a l e s in p e r c e n t 5 --------------------------------------------------------------------------------------------------Average U.S. manufacturing returns in 1990=3-8 0 e t u r n o n sa l e s in p e r c e n t p Returns of newly acquired U.S. manufacturers, 1.6 1987-90 \ R 2.2 .♦*'3.0 / » / / 12 1 1 -----------------------------\ \ / Returns reported in 1990 tax returns, by year after incorporation • \ -2.6 / U.S. manufacturing 1irms t\V I - * * A._ i i i i ...... ...1...........1...........1............1.......... 1......... i............1 -1 0 1 2 3 Y 4 5 6 7 8 9 i 10+ 1978 e a r s a f t e r a c q u i s i t i o n /i n c o r p o r a t i o n i i 80 i *% \ /\ Acc uisitions i n \ X marlufacturing -■5 i \ % % % All acquisitions \ i i .... ...J _ 85 I----- J_... x ... . ... ....• .....J 90 Sources: U.S. Department of Commerce; Internal Revenue Service, Corporate Statistics Branch. Sources: U.S. Department of Commerce, Bureau of Economic Analysis and Bureau of the Census. Notes: 10+ includes corporations that did not report year of incorporation. Returns o f newly acquired U.S. firms represent an average for year before acquisition. Notes: Manufacturing excludes petroleum products. Values plotted for acquisitions in each year show profitability of firms acquired in subsequent year. FRBNY Q 48 u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 92 those periods or in those industries. N ote also that the stud age of foreign firms in the United States, just as such acquisi ies of acquisition premia cited above take the firm as the unit tions pushed up U.S. firms’ leverage from 1984 through the of analysis, whereas the price-earnings ratio of thirty gives end of the decade. more weight to large transactions and so could reflect a few big money-losers. The m ore leveraged foreign in d u stries show ed weaker profits. Across manufacturing, the greater the excess of foreign firms’ debt ratio over that of their domestic coun H ea v y D e b t : Servicing the debt that often finances a high- terparts, the weaker the foreign firms’ relative performance as priced acquisition eats up profits. R JR -N abisco— to offer a measured by return on assets (Chart 8). purely domestic example— changed from a money-spinner How much did higher leverage reduce affiliate prof to a m oney-loser after it was taken private in a h igh ly its? Foreign-owned manufacturing firms financed their $429 leveraged buyout in 1989- billion in assets with 10.9 percentage points more debt in In manufacturing, foreign companies in the United 1990 than did their U .S. counterparts. The resulting $47 bil States have generally operated with more debt than have U.S. lion in excess debt, at an interest rate of 8 percent, lowered firms. Including debt to parent group, foreign firms’ lever taxable profit by $3-8 billion. This figure is just shy o f an esti age— the ratio of debt to assets— ranged from 5 to 15 per mate of $4.0 billion derived by a parallel calculation using centage points higher than that of domestic manufacturers Internal Revenue Service data on net interest paid in relation (Chart 7). The increase in foreign firms’ leverage here in the to sales by “foreign-controlled” and “other dom estic” manu late 1980s contrasted with the general deleveraging trend of facturers (Table 2). The latter calculation, extended to all companies in major countries abroad (Remolona 1990). We industries, shows the total shortfall of taxable income owing hypothesize that debt-financed acquisitions in the late 1970s to comparatively high leverage to be $14.5 billion, almost and again in the late 1980s (Chart 2) ratcheted up the lever half of the profit gap (Table 2; Chart 1, bottom panel). Chart 7 Chart 6 Leverage of P r o f it a b ility D is trib u tio n o f U.S. F irm s A c q u i r e d b y F o r e i g n F ir m s 1 9 8 7 - 8 9 U.S. M a n u f a c t u r in g F ir m s O w n e d b y Fo r e ig n In v e s t o r s R a t io o f d e b t t o a s se t s in p e r c e n t 65 Percent — o f a f f il ia t e s in in d ic a t e d r a n g e Ratio for foreign firms including intercompany debt 60 50 S 55 40 50 *» U.S. firms 30 45 „ "" 9•* —‘ 40 20 35 -......i 1977 J _J___ 80 l I Ratio for foreign firms excluding intercompany debt 1.......1 l 1 85 1 1 i 1 90 92 10 Sources: U.S. Department of Commerce, Bureau of Economic Analysis and Bureau of the Census. 0 -20.0 or less -10 .0 to -19-9 Percent 0.0 to -9.9 More than 0.0 to 9 9 10.0 to 19-9 20.0 or more r e t u r n o n sa l e s in y e a r b e f o r e a c q u is it io n Source: Fahim-Nader 1994, p.58. Notes: Debt denotes current liabilities plus long-term debt. Payables to foreign parents have been subtracted from debt, and receivables from foreign parents have been subtracted from assets, in the numerator and denominator, respectively, of the ratio for foreign firms excluding intercompany debt. Petroleum products are included for U.S. companies but excluded for foreign affiliates. F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 49 P o s t -A c q u i s i t i o n C o s t s : Profits deteriorate further after therefore explain just $1.7 billion of the $4 billion decline in acquisition. In manufacturing, acquisition targets in 1987- profits. 90 managed after-tax profits o f $2.0 billion in the year before The sharp decline in profitability of newly acquired acq u isitio n (C hart 5 ).8 Yet 1 9 9 0 tax returns show that firms thus seems to derive largely from a deterioration in foreign manufacturers incorporated in 1987-90 collectively operating margins. Stepped-up investment, increased mar lost $2.1 billion. keting expenditures, and consolidation expenses all raise This $4 billion decline in profits derived from two operating costs.10 Anecdotes support this inference. An exec sources: finance and acco u n tin g charges and op eratin g utive from Sony of America described the high initial invest results. If foreign manufacturers incorporated in 1987-90 ments required after the acquisition of Colum bia Pictures financed 10.9 percent more of their assets with debt than did and Tristar as follows: “The operation spent a lot of money the domestic manufacturers (as indicated above), their interest first year....We had to make more movies. They had to spend expense would have been $1 billion higher.9 A lso, newly more. They renovated the Culver City studio. They installed acquired firms sometimes increase their depreciation expense a lot of new technology from Sony.” 11 To cite another exam by revaluing tangible assets or increase their amortization by ple, Thomson of France bought RCA for $1 billion in cash revaluing intangible assets. Under the generous assumption and assets five years ago, quadrupled capital spending in two that recently incorporated firms accounted for the entire dif years, and only began to post operating profits, not including ference between “foreign-controlled” and “other dom estic” interest payments, in 1992 (Browning 1993, p. A7). manufacturers’ propensity to deduct depreciation and amor In this matter, foreign acquisitions perform differ tization expenses, the extra expense would have amounted to ently from U .S. acquisitions in the period. In contrast to only $0.7 billion (Policy Econom ics Group, K P M G Peat recently incorporated foreign-owned firms (Chart 9), U.S.- M arwick, 1994, pp. IV-2 through IV-4). Taken together, owned manufacturers incorporated in the three preceding higher debt, depreciation, and am ortization expenses can years turned a profit in 1990. U.S. manufacturing industries with heavy (unrelated) acquisition activity in the 1980s show declines in nonproduction workers, a development that is Chart 8 con sisten t w ith im proved operatin g resu lts (Caves and R e tu r n o n A s s e t s a n d L e v e r a g e o f F o re ig n - o w n e d Krepps 1993, pp. 251-54). U.S. leveraged buyouts, hardly F i r m s 1 9 7 7 -9 2 Manufacturing Industry Averages F o r e ig n a f f il ia t e r e t u r n s m in u s Table 2 d o m e s t ic r e t u r n s in p e r c e n t Q--------------------------------------------------------------------Primary metals L e v e r a g e a n d P r o f i t o f F o r e i g n -o w n e d F i r m s i n t h e U n it e d S t a tes Return gap = -2.56 - 0.17 * Debt ratio gap Ratio o f Net Interest to Sales (Percent) ForeignOther controlled Domestic Corporations Corporations Difference (3) ( 1) ( 2) 2 .7 1 1.82 .89 446 4.0 Wholesale and retail trade 1.28 1.00 0.2 8 416 1.2 Finance,insurance and real estate —9.00 - 1 7 .2 0 8.20 113 9.3 2.97 - 0.02 86 0.0 10 6 0 14.5 2.95 Total Fo r e ig n 5 10 15 20 25 30 a f f il ia t e d e b t r a t io m in u s d o m e s t ic d e b t r a t io in p e r c e n t Sources: U.S. Department of Commerce, Bureau of Economic Analysis and Bureau of the Census; Federal Reserve Bank of New York staff estimates. Notes: Debt ratio denotes liabilities as a proportion of assets. The t-statistic is in parentheses. FRBN Y 50 Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Additional Profit (3 )x (4 ) (Billions of Dollars) Manufacturing Others 0 Sales (Billions of Dollars) (4) Source: Hobbs 19 9 3 . Notes: Foreign-controlled corporations and other domestic corporations w ith in finance, insurance, and real estate are said to differ in that most foreign banks operate as branches, which are excluded from these data. The negative net expense is said to reflect financial firms’ earning interest on their unbor rowed funds. See Policy Economics Group 19 9 4 . typical acquisitions but the only ones where the target firm phenomenon both here and abroad ascribe it to payoffs from continues to report independent results, generally show investment in plant, equipment, and market share, the slack improvements in operating results, usually because o f cut ening o f consolidation expenses, m anagerial learning, and backs in investment and working capital (Long and Raven- the divestment of unsuccessful enterprises. A study of 1966 scraft 1993; Kaplan and Stein 1993). benchmark data on U.S. direct investment abroad found that A d iffere n ce in p o s t- a c q u is itio n e x p e n d itu re the return on equity earned by U.S.-owned manufacturers between U.S. and foreign acquirers is entirely consistent with rose by 1 percent per year since incorporation.12 More recent our explanation of the surge in foreign acquisitions. If foreign Internal Revenue Service data on foreign operations of U.S. buyers enjoyed a cost of equity advantage, then they were multinational corporations confirm this strong positive rela able to put more money in a U.S. firm than its previous man tio n sh ip betw een p r o fita b ility and experien ce am o n g agement could justify. By contrast, domestic acquisitions in m an u factu rin g affiliates in 19 8 8 ; this relation sh ip also the 1980s have been interpreted as a device for disciplining applies to affiliates in the finance and trade sectors (Lutzy and managers to accept the consequences of their (high U.S.) cost M iller 1992, p. 86). of capital (Blair and Litan 1990). The relation holds even more strongly for foreign firms’ operations in the U nited States. Cross-sectional data S u b s e q u e n t R e c o v e r y o f P r o f i t s : D irect in vestm ent for 1990, provided at our request by the Corporation Statis profits improve with age. Researchers who have observed this tics Branch of the Internal Revenue Service, show that return Chart 9 P r o f i t a b i l i t y o f F o r e i g n F irm s in t h e U n i t e d S t a t e s i n 1 9 9 0 , b y N u m b e r o f Y e a r s s i n c e I n c o r p o r a t i o n R eturn R o n e q u i t y in p e r c e n t 12 ---------------------------------------- e t u r n o n e q u i t y in p e r c e n t Manufacturing Finance, Insurance, and Real Estate Wholesale and Retail Trade All Industries -8 L 0-3 years 4-6 years 7-9 years 10 or more years 0-3 years 4-6 years 7-9 years 10 or more years Source: Internal Revenue Service, Corporation Statistics Branch. Note: Affiliates whose age was not reported are grouped together with affiliates of ten or more years of age. F R B N Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 51 on equity improves at a rate of about 1.2 percent per year has a median value of 4 percent per year (Chart 10). Thus, sub (Chart 9). The results are equally strong in a fixed sample over stantial new investment and expansion of existing businesses the years 1980-87.13 m ust occur each year just for the existing stock o f foreign direct investment to remain stable. When new foreign invest S u b s e q u e n t D i v e s t m e n t : R eturns im prove over tim e as ment slackened in 1992, the number of Americans working foreign firm s sell off their U .S . operations at a su rp ris for foreign firms declined for the first time in at least fifteen in gly rapid rate. B ecause d raw in g inferences from the years (Zeile 1994). The attrition rate reached as high as 10 p o s t- m e r g e r p e rfo rm an c e o f firm s p o se s d iffic u lt ie s percent in 1986 (Chart 10). At the latter figure, the half-life of (M agenheim and M ueller 1988; Franks, H arris and Tit- foreign holdings in the United States is seven years— a year man 1991; Healy, Palepu, and R uback 1992), researchers longer than Renault’s investment in American Motors and have m easured the success o f dom estic m ergers by their three years longer than Volkswagen’s venture producing cars survival over tim e. T h u s, analysts have interpreted the in Pennsylvania (Hood and Young 1986, pp. 175-78). sale by th e m i d - 1 9 8 0 s o f 4 7 p e rc e n t o f 6 0 0 0 f i r ms U .S. D ir e c t I n v e s t m e n t A b r o a d : A natural test for any explanation o f low profits o f foreign firm s in the U nited States is whether it can account for the respectable perfor R eturns im prove over tim e as fo r eig n firm s sell mance of U .S. direct investment abroad, which earned 8.3 o f f th eir U.S. operations a t a su rp risin gly percent on position valued at current cost in 1993 (Weinberg ra p id rate. 1994, p. 114; Scholl 1994, p. 63; see also Kapler 1994). Our explanation, which relates profitability to the recency o f acquisitions, passes this test. Internal Revenue Service data acquired between 1950-76 (Ravenscraft and Scherer 1987) and the sale by 1989 of 44 percent of 271 large firms acquired between 1971 and 1982 (K aplan and W eisbach 1992) as demonstrating that hoped-for results remained elusive. Sim i larly, major corporations had divested half of the 2000 acqui Chart 10 A t t r i t i o n o f F o r e i g n F i r m s in t h e U n i t e d S t a t e s Proportion of Foreign Firms' Employees Whose Firm Was Sold, Liquidated, or Cut Back sitions that they had made between 1950 and 1986 by the Percent latter date (Porter 1987). 14 The Bureau of Economic Analysis has measured the sales and liquidations of whole divisions by foreign firms in the United States for selected years. Liquidation refers to the fate of the foreign owners’ interest rather than the fate of the corporate assets. Thus, the 1992 data (Zeile 1994, p. 157) Sales and liquidations 12 10 8 apparently include the Allied-Federated store chain, which passed from its Canadian owner’s hands when the firm exited 6 Chapter 11, and perhaps the Santa Fe railroad, whose Canadi 4 an m inority owners, the real estate developers the Reichmanns, sold off their share. The unit for this count is employ 2 ees: the number of U.S. workers at firms or divisions sold or liquidated by foreign-owned firms. We chart these numbers as H Cutbacks 0 i U n 11i n■ i ii ■■■■■■■ 1984 85 86 89 90 91 92 a fraction of all Americans working for foreign-owned firms at the end of the previous year.14 The attrition rate so measured Digitized 52 for FRASER FRBN Y Q u a r t e r l y R e v ie w /S u m m e r - F a ll 1994 Sources: Howenstine 1987, p.38; Herr 1988, p.6 1; Bezirganian 1993, p.92; and Zeile 1994, p.156. for 1988 show that the median year of incorporation (mea wedge between the profitability of foreign and domestic sured by sales) for U.S.-owned firms abroad was between I960 and 1964 overall and between 1955 and 1959 for man ufacturers (Lutzy and Miller 1992, p. 86). By contrast, Inter nal Revenue Service data for 1990 indicate that the median foreign-owned firm in the United States dated only to the late firms. in the United States in the 1980s. As foreign firms’ recent acquisitions bulked larger in their overall holdings here, the 1970s, a generation later. Despite the shortcomings of years low returns associated with these acquisitions dragged down since incorporation as a measure of the maturity of direct the aggregate profitability of foreign firms. In summary, the after-effects of a wave of acquisi tions help explain the declining profitability of foreign firms investment stocks, U.S. direct investment abroad clearly has Before proceeding to consider the role of the weak stood the test of time and shows it in its returns. Moreover, dollar, let us estimate how the characteristics that we have U.S. manufacturers abroad showed weak profits in the 1950s ascribed to foreign firms in the United States—higher lever and 1960s, when their holdings were growing rapidly. age and recency of acquisition (immaturity)—may have hurt profits. As noted above, if foreign firms paid the same net Foreign acquisitions of interest in relation to receipts as U.S. firms, the profits of the U.S. firms have to some extent yielded poor returns for the former would have been higher by $14.5 billion in 1990. If same reasons as have domestic acquisitions. Foreigners pay all foreign firms earned as much as the ones that had been acquisition premia between 30 and 50 percent, not much here for ten years, then they would have earned $11.9 billion more than U.S. acquirers. Rapid rates of divestiture suggest more income in 1990 (Table 3). On this showing, leverage that foreign and U.S. managers alike experience disappoint and recency account for much of the 1990 shortfall of profits ment with their acquisitions. Foreign firms may differ from of $32.1 billion (Chart 1, bottom panel). F o r e ig n v e r s u s U.S. A c q u isit io n s: domestic acquirers, however, in their selection of targets. W hile domestic firms purchase targets whose profits are at or T h e D o l l a r ’s V a l u e near industry standards (Ravenscraft and Scherer 1987, pp. Observers have proposed another explanation for the low prof 56-74), foreign firms buy a large proportion of money-losers. its of foreign-owned companies in the United States: the sharp Another major difference lies in money spent after acquisi depreciation of the dollar since 1985. Because U.S. affiliates of tion: new foreign owners open their wallets wider than new foreign firms import nearly twice as much as they export, a U.S. owners. Even absent any differences between foreign decline in the dollar could easily raise their dollar-denominat and domestic acquisitions, the fact that recently acquired ed input costs more than it raises their export revenues. firms constitute a substantially larger share of foreign firms Certain high-profile industries provide support for than of domestic firms allows acquisition activity to drive a this hypothesis. Affiliates engaged in automotive wholesale Table 3 M a t u r i t y a n d P r o f it , 1 9 9 0 Return on Sales (Percent) A ll Affiliates M anufacturing W holesale and retail trade Incorporated before 19 8 1 Profit (Billions o f dollars) Incorporated after 19 8 1 Sales (Billions o f Dollars) A t p re - 1 9 8 1 Return on Sales Actual Difference 1.8 3.0 0.3 446 13 .4 8.1 5.3 - 0.6 - 0 .5 - 1.6 416 - 2.1 - 2.6 0.5 Finance, insurance, and real estate —0.6 0.5 - 2.1 113 0.6 - 0 .7 1.3 Other - 1.1 4 .7 - 4 .7 85 4 .0 -0 .9 4.0 4 .9 Total 10 6 0 15 .9 * 1 1 .9 Source: Internal Revenue Service, Corporation Statistics Branch. *Datum is sum o f four hypothetical profits above. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 53 trade have suffered a sharp decline in profitability owing to a weak dollar (Chart 11). In particular, each 10 percent depre centage points, which at 1992 levels amounts to $850 mil man 1987; Froot and Klemperer 1989; Hooper and Mann 1989; Okuro 1989; Marston 1990; Knetter 1994; Rangan and Lawrence 1993).16 Chart 11 suggests that foreign auto companies have not only maintained dollar prices on their lion. During the mid-1980s, when the dollar was strongest, exports but have also squeezed wholesaling margins in the return on assets peaked at over 9 percent. In 1990 and 1991, United States—a strategy that serves the same goal of keep with the dollar 35 percent lower, the affiliates suffered losses.15 ing prices to U.S. consumers competitive. In short, the pric The compression of profits in automotive wholesale ing strategies of multinational firms, at least in autos, have trade suggests that the conventional analysis of how exporters offset the weakening of the dollar by much more than inter to the United States have responded to the weak dollar is national trade prices can demonstrate. ciation of the dollar reduces their return on assets by 1.5 per incomplete. Economists have measured the varying extent to W hile the evidence on automobile trade is com which foreign exporters respond to a strengthening of their pelling, a comprehensive industry-by-industry review finds currencies by permitting higher dollar costs to pass through that the impact of the dollar’s value on foreign firms’ prof to higher dollar export prices or, alternatively, by slashing itability is small and ambiguous. The dollar should have lit profit margins to price to the U.S. market. These studies have tle impact on affiliate profits in services, finance, insurance, found that foreign companies have responded to a weak dollar real estate, construction, transportation, or food stores, which by maintaining dollar prices on exports to the United States together constitute 22 percent of the gross product of direct to some extent at the expense of their profit margins (Krug- investment in the United States, because imports and exports combined amount to less than 5 percent of sales in these seven industries (Zeile 1994, pp. 173,177). For each of the remain ing twenty-eight industries (see endnote 41), we regress Chart 11 return on assets on the logarithm of the trade-weighted U.S. T h e D o l l a r E x c h a n g e R a te a n d P r o f it a b ili t y In M o t o r exchange rate as computed by the Federal Reserve Board for V e h ic le W h o le s a le T ra d e 19 7 7 -9 2 R the years 1977-92. e t u r n o n a s s e t s in p e r c e n t 10 ^ 1985 n ^ Return on Assets = 2.68 + 0.15* (Exchange rate-100) _ (7.83) 19840/ ^ 8 ---------------R2 = 0.81 N = 1 6 ------------------------- X -------------- 1979Q 1988 □ 1980 q 1986 1981 □ 1977 U „/ 1983 12). In four industries, profit showed a statistically signifi cant (at the .01 level) contemporary or lagged response to the 1982 □ dollar (Table 4). In contrast to automotive wholesale affili ates, whose profits suffer from a weak dollar, foreign firms in t—11989 several export-intensive industries—mining, agricultural 1 1991 1990 L__ 80 their profit to the dollar’s exchange rate. The regression results suggest that a weak dollar reduces profits in eleven industries but boosts profits in seventeen industries (Chart .✓ '1987 1 9 9 3 ^ 1978 The twenty-eight industries vary in the response of wholesale trade, and agriculture—show higher profits when 1 90 T ...i..... I 1 1 1 100 11 0 12 0 13 0 140 1 150 r a d e -w e i g h t e d d o l l a r e x c h a n g e r a t e Sources: U.S. Department of Commerce, Bureau of Economic Analysis; Board of Governors of the Federal Reserve System; Federal Reserve Bank of New York staff estimates. Notes: The exchange rate used is the Federal Reserve Board's multilateral trade-weighted dollar index (March 1973 =100). The profitability data also include manufacturing results for some automotive firms whose primary activity was in wholesale trade. The t-statistic is in parentheses. The regression of return on sales on the exchange rate yields an estimated equation: return on sales =0.85+0.05* (exchange rate-100). The t-statistic is 8.09; the R-squared is 0.82. Digitized F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 54 for FRASER the dollar is w eak.17 W hether exchange rate effects are summed across just the industries that show a significant relation or across all twenty-eight industries (Table 4), the net impact of a 10 percent dollar depreciation is well under $1 billion. Explanations for the $32.1 billion shortfall in profits must be found elsewhere. Given that affiliates import substantially more than they export, the finding that a weak dollar does little to depress affiliate profits seems surprising. Foreign-owned wholesalers account for most of the excess of imports over exports but, as we have seen, their parents share their profit Chart 12 Ef f e c t o f D o l l a r Ex c h a n g e R a t e o n F o r e ig n F ir m s ’ P r o f it s Distribution across Industries N compression when the dollar weakens. In U.S. manufactur u m b e r o f in d u s t r i e s ing, both exporting firms and import-competing firms tend to benefit from dollar weakness (Hung 1992-93).18 More over, compared with the wholesalers, foreign manufacturers .i 1 ■ in are larger and their profits stronger to begin with, so their — Less -0.15 -0.12 -0.09 -0.06 -0.03 0 than to to to to to to -0.15 -0.12 -0.09 -0.06 -0.03 0 0.03 E s t im a t e d 0.03 to 0.06 0.06 to 0.09 0.09 to 0.12 more weight. ■ i r improvement in response to the dollar’s weakness carries 0.12 Greater to than 0.15 0.15 P r o f it S h if t in g Public discussion of the low profitability of foreign firms has centered on assertions that they understate their profits to c o e f f ic ie n t Sources: U.S. Department of Commerce, Bureau of Economic Analysis; Board of Governors of the Federal Reserve System; Federal Reserve Bank of New York staff estimates. Notes: Return on assets is regressed on a constant and the natural logarithm of the trade-weighted dollar. The regression was run for each of twenty-eight industries for the years 1977-92. The exchange rate used is the Federal Reserve Board’s multilateral trade-weighted dollar index (March 1973 =100). A positive coefficient implies that profits rise when the dollar appreciates; a negative coefficient, that profits fall when the dollar appreciates. avoid U.S. taxes. A multinational firm can shift profits from one tax jurisdiction to another in at least two different ways. A firm’s treasurer can load debt onto its operations in one country and thereby shelter the income earned there with tax deductions for interest paid. Or a firm can arrange for its operations in one tax jurisdiction to pay those in other juris dictions high prices for goods and services, or outsize royalty or interest payments. Table 4 T h e D o l l a r E x c h a n g e R a t e a n d t h e P r o f i t a b i l i t y o f F o r e i g n F i r m s in t h e U n i t e d S t a t e s Annual Data 19 7 7 —92 Forecasted Impacts Regression Results Industry Contemporaneous Exchange Rate Beta R-squared Lagged Exchange Rate Beta 1 9 9 1 Industry Assets (Billions R -squared o f Dollars) Estimated Impact o f a 10 Percent Depreciation o f the D ollar (M illions o f Dollars) Subsequent Year Same Year Selected industries M otor vehicles and equipment wholesale trade 0 . 16 0 ** 0 .8 0 0 .0 1 3 3 * * (0.035) 0 .5 1 58.2 -9 3 1 -7 7 4 0 .6 4 -0 .1 6 5 * * (0.034) 0.62 22.0 35 6 363 -0 .0 5 3 * * (0 .0 1 2 ) 0 .5 8 -0 .0 5 6 * * 0 .6 0 9-1 48 51 -0 .0 8 7 * * 0 .5 4 0.36 4.2 37 31 Total o f selected industries -490 Memo: Sum o f estimated impacts for all 28 industries in sample -6 7 8 -3 2 9 482 (0 .022 ) M ining -0 .1 6 2 * * (0.032) Farm-product raw materials wholesale trade Agriculture, forestry, and fishing (0 .02 2 ) (0 .0 1 2 ) - 0 .0 7 4 * (0.027) Sources: U.S. Department o f Commerce, Bureau o f Economic Analysis; Board o f Governors o f the Federal Reserve System. Notes: Return on assets is regressed on a constant and the natural logarithm o f the trade-weighted exchange rate. The indicated regression is run for each o f twenty-eight industries for the years 1 9 7 7 —92 . Selected industries are those whose contemporaneous or lagged regressions are significant at the 1 percent level. Standard errors are in parentheses. Estimated intercepts are not reported. *Significant at 5 percent level. **Significant at 1 percent level. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 55 The U.S. tax code tries to prevent both forms o f prof burden. A scan o f corporate tax rates in major source coun it shifting. Foreign firm s lose the full tax deductibility o f tries suggests at first that foreign companies do not have interest payments if the fraction o f their global consolidated much to gain by shifting income out o f the United States. debt in the United States is higher than the fraction o f their That is, companies based in most foreign countries would pay assets in the United States .19 Similarly, multinational firms as much in home country taxes on an extra dollar o f income as lose deductions for internal, or transfer, prices if they differ they would pay in combined U.S. and home country taxes if from prices that would be set in arm’s-length transactions. they reported the income in the U nited States (Landefeld, In the remainder o f this section we consider whether Lawson, and W einberg 19 9 2 , p. 84).21 the low profit reported by foreign-owned firms in the United Such a comparison does not prove that foreign firms States stems from profit shifting strategies that allow these face no immediate tax incentive to shift profits out o f the firms to avoid U.S. taxes. United States, however, for at least two reasons.22 First, statu tory tax rates at home may overstate the effective tax rate .23 E a r n in g s S t r ip p in g t h r o u g h E x cessive Second, foreign firm s need not shift profit to th eir home L e v e r ag e country; they can shift profit to low-tax third countries as As established above, the U.S. affiliates of foreign firms oper well. ate at higher leverage in the U nited States than do U.S. firms. But the tax code specifies a different standard o f comparison: R does the interest-bearing debt o f a U.S. affiliate (including Shifting income to low-tax countries pays off best for firms debt to foreign affiliates) in relation to U.S. assets exceed the from countries, mostly in Continental Europe, that apply the foreign firm ’s consolidated worldwide ratio o f debt to assets principle o f territorial taxation to their home firms. Firms (U.S. Congress, Join t Committee on Taxation 19 8 9 , p. 34)? headquartered in these countries are exempt from local taxes That leverage o f foreign firms’ affiliates in the United States on their foreign profits. W h en these firms transfer income rose in the 1980 s while leverage was falling in most major out o f the United States into a low-tax third country, they source countries increases the likelihood that some compa pay only the taxes o f that third country. By contrast, in a nies attempted to reduce their taxes here by burdening their country that applies the principle o f w orldw ide taxation, U.S. affiliates w ith excessive debt. Lacking the company-by- such as Japan or Britain, a firm receives a credit for taxes paid company data necessary to test for excessive debt, we simply abroad but m ust pay ad dition al taxes to its home fiscal note that the evidence does not perm it the inference that for authorities when it brings home income from a tax haven. eign firms burden their U.S. operations with more than their T hu s, firm s fro m c o u n trie s u n iv e rs a lly a p p ly in g th e fair share o f debt. territorial principle o f taxation stand to benefit most from eturns of Fo r e ig n F ir m s by H ome Ta x R e g im e : transferring income out o f the United States .24 T r a n sfe r P r ic in g How do firms from the two kinds o f countries com The second method o f shifting profits, transfer pricing, may pare in their U.S. profitability? Broadly following previous help explain w hy foreign firm s report lower profits than analysts (Slemrod 19 9 0 ; Auerbach and Hasset 19 93), we look domestic firms. But it cannot readily explain why the prof for differences in behavior across firms from the two types of itability o f foreign firms declined in the 1980 s, a trend better countries. The U.S. operations o f firms from countries that explained by acquisition activity. Transfer pricing could only tax strictly on the territorial principle report lower profitabil help to account for developments in the 1980s if the U.S. tax ity (Chart 13 ) than do the U .S. operations o f firm s from authorities failed to step up the enforcement effort as foreign countries that tax w orldw ide profits (U nited K ingd om , acquisitions accelerated during this period .20 Japan) or tax profits from nontreaty countries (Canada, Ger Foreign firms m ight want to shift profits out o f the many ).25 This finding is no more than suggestive, however, in U nited States if they could thereby lower their global tax view o f the different m ix and vintage o f industries across Digitized 56for FRASER FRBN Y Q u a r t e r l y R e v i e w /S u m m e r -F a l l 1994 countries. We now consider another approach to the transfer An analysis of nineteen major manufacturing indus pricing question that focuses not on country-specific incen tries for the period 1977-92 shows that industries in which affiliates import a higher fraction of their sales from their tives to shift income, but on industry-specific opportunities to shift income. parent group report lower returns on sales (Chart 14).26 Reported profits vary inversely with the opportunity to R eturns of F o r e ig n F ir m s an d Extent of transfer income. Trade between the U.S. operations of This finding is only suggestive. Imports from par foreign firms and the firms’ affiliated companies abroad has ents may simply proxy for overall imports. In that case, the exceeded $100 billion per year since the mid-1980s and in observed relationship could mean only that a weak dollar 1992 reached about $200 billion. Two-thirds of such trade crimped the profits of manufacturing industries heavily takes the form of imports from the foreign parent group. dependent on imports. This relationship between low profits In t r a f ir m T r a d e : Such extensive intrafirm international trade sug and high imports, however, gains further credibility when it gests that modest deviations from arm’s-length pricing could survives in a regression analysis controlling for overall import succeed in shifting substantial income abroad. Indeed, mer propensities and exchange rate effects (Appendix II).27 chandise trade issues account for 75 percent of the adjust How big might profit shifting through the pricing ments that the Internal Revenue Service recommends (U.S. of imports from related parties be? The relationship between Department of Treasury and Internal Revenue Service, 1988, profitability and imports from parents across manufacturing Appendix B, p. 1). Accordingly, we investigate whether a suggests that for every additional dollar of imports from par larger opportunity to manipulate prices is associated with ents, returns are 24 cents lower. Since manufacturing affili weaker reported profits in the United States. ates of foreign firms imported $33.2 billion from their par ents in 1990, transfer pricing could reduce profits by as much as $8.0 billion. By this calculation, transfer pricing in Chart 13 Chart 14 P r o f i t a b i l i t y o f F o r e i g n M a n u f a c t u r e r s in t h e U n it e d S t a t e s b y H o m e T a x R e g im e R P r o f it a b il i t y a n d Im p o r t s f r o m F o r e ig n P a r e n t s 1 9 7 7 - 9 2 Manufacturing Industry Averages e t u r n o n s a l e s in p e r c e n t 6 ----------—---------------------------------------- R e t u r n o n s a le s in p e r c e n t 6 ------------------ Credit: Affiliates ultim ately owned by firms in countries that tax all or some foreign income and give a credit for taxes paid: Japan, United Kingdom, Canada, Germany Exempt: * -4 L. 1977 78 J__ L 79 80 ^ J___L 81 82 83 J___L 84 85 86 87 89 90 91 92 Sources: U.S. Department of Commerce, Bureau o f Economic Analysis; Organization for Economic Cooperation and Development 1991. Note: Canada and Germany exempt income from those countries with which they have tax treaties but tax income from other countries. , (-3 .62 ) Soaps and cleaners r, Affiliates ultim ately owned by firms in countries that completely exempt foreign income from income taxes: Netherlands, France, Switzerland Return on sales = 2.89 - 0.24 * imports from parents as a percentage of sales □->.—□ O R = 0.44 □ l~1 Printingand*****’* publishing \ □ Stone clay, and glass Im p o r t s Electrical machinery □ Nonelectrical m ach in eryv\ , s^ ■4 1------------------- 1........ ... 5 N = 19 Transportation equipment _ i -------------------1 1 10 15 20 i 25 fr o m p a r e n t s as a p e r c e n t a g e o f sales Source: U.S. Department of Commerce, Bureau of Economic Analysis; Federal Reserve Bank of New York staff estimates. Note: The t-statistic is in parentheses. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 57 manufacturing might account for a quarter of the $32 billion profit gap (Chart 1, bottom panel). ships and then summarize the results of a multivariate analy sis that is described in more detail in Appendix II. Even were this result accepted at face value, it could only help explain the low level of foreign firms’ profits. The Our factor-by-factor analysis overexplains the prof itability gap, reflecting overlap in the factors considered. puzzle of the declining profitability of foreign firms in the The acquisition-related factors of leverage and recency are United States would still remain. As a fraction of sales, inter most prominent, together accounting for five-sixths of the national trade with affiliated companies has fallen from 22 gap between the profits of foreign firms in the United States percent in 1977 to 16 percent in 1991. A declining propor and those of domestic firms. Possible transfer pricing tion of affiliate transactions does not accord with a widening accounts for about a fourth of the gap. The exchange rate profitability gap. effect is negligible. The m ultivariate analysis employs the variable RECENCY to measure the fraction of assets acquired in the A s s e s s i n g R e l a t iv e I m p a c t s We can now estimate how much each factor discussed in this previous three years. RECENCY captures a variety of influ article has contributed to the low profit of foreign firms in the ences: acquisition, selection, pricing, post-acquisition United States (Table 5). We have already considered the indi expenses, and eventual attrition. These acquisition-related vidual relationship between each of these factors and prof influences are more clearly distinguished from leverage in the itability. In this section, we return to these bivariate relation- multivariate analysis than in the bivariate analysis. The multiple regression analysis confirms the impor tance of those influences represented by the variable RECEN CY, yielding results remarkably consistent with the bivariate 7 able 5 E x p l a in in g t h e P r o f it a b il it y G a p B e t w e e n F o r e ig n F ir m s i n t h e U n i t e d S t a t e s a n d D o m e s t i c F i r m s in 1 9 9 0 age. Possible transfer pricing now explains only a fifth of the gap. Estimates (Billions o f Dollars) Based on M ultiple Regression Analysis Based on Bivariate Analysis Factor Return on Assets Return on Sales^ 12 .6 * 6 .8 f 10.1 6.5 3 4 .4 6.7* 26.0 20.6 32.1 3 2.1 3 2.1 Recency 1 1 .9 Leverage 14.5 D ollar exchange rate - Transfer pricing 8.0 Total 4 .0 Memo: Gap between profits o f U.S. and foreign firms analysis. Both regressions sharply downplay the role of lever * Regression 1 (Table B2) coefficient o f—.0 5 8 m ultiplied by .1 4 0 ( 1 9 9 0 value o f .222 [Table B l] less U.S. value o f .082) yields .0 0 8 1 o f assets o f $ 1 .5 5 tril lion, or $ 1 2 .6 billion. U.S. value o f .0 8 2 is estimated on the basis o f the com parative levels o f acquisition activity from 1 9 8 8 - 9 0 (M errill Lynch 19 9 2 ) as share o f 1 9 9 0 assets (Hobbs 19 9 3 ) for domestic and foreign firms. f Regression 1 (Table B2) coefficient o f- .0 8 1 m ultiplied by .0 5 4 ( 1 9 9 0 value o f .775 less U.S. value o f .7 2 1 , both from Hobbs 19 9 3 ) yields .0 0 4 4 o f assets of $ 1.5 5 trillion, or $ 6.8 billion. W e use Statistics o f Income data for foreign and domestic firms because we lack a domestic benchmark for Bureau o f Economic Analysis data outside o f manufacturing and other selected subsectors. | Regression 1 (Table B2) benchmark coefficient o f- .0 5 2 m ultiplied by .083 ( 1 9 9 0 value o fM P A R (Table B l) less 0) yields .0 0 4 3 o f assets o f $ 1 .5 5 tril lion, or $6.7 billion. X P A R was not used as it was statistically insignificant; its inclusion would have raised the impact o f transfer pricing to $ 7 .8 billion. § The impacts are calculated in the same manner as in the three prior notes, except that the coefficients for return on sales regression 7 are used instead: - . 0 6 1 for RECENCY, - .0 6 3 for LEV, and - .0 6 7 for M PAR. W e then m ultiply results by 1 9 9 0 sales o f $ 1 .1 8 trillion to arrive at dollar amounts. Digitized FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 58for FRASER The impact of the dollar was ambiguous in this analysis, so again we attribute no effect. Im p l ic a t io n s f o r T a x e s Observers have calculated how many more billions of dollars foreign-owned firms would earn if they were as profitable as comparable U.S. firms. Using the 1990 IRS data on return on sales by sector, they would arrive at $32.1 billion. They would then multiply this sum by the 34 percent top corpo rate tax rate to generate an estimate of potential additional tax receipts of about $10 billion.28 Were it simply a matter of transforming low-profit foreign firms in the United States into normally profitable firms, these calculations would hold. But roughly 60 percent of foreign firms report losses each year, while only half of U.S.-owned firms report losses.29 When solidly unprofitable firms become somewhat less unprofitable, they may not pay any more income taxes. That taxes paid in 1990 exceeded foreign firms’ aggregate pretax income underscores the importance of the distribution of profitable and unprofitable firms. Profitable foreign firms earned $29 billion while unprofitable ones lost $25 billion, so the firms earned $4 billion in aggregate. The $4 billion in aggregate earnings offers no clue to the taxes financing last year’s deficit adds to this year’s deficit. Owing to the poor returns of foreign companies in the United States, the U.S. economy is not yet paying the full cost of its external deficits. paid by foreign companies. It is revealing to compare taxes paid by foreign com future, the servicing costs so far avoided will begin to come panies with those paid by U.S. firms. As a fraction of sales, due and compound. The current account deficit would there taxes paid by foreign firms in the United States did not differ fore widen. On the home front, however, higher profits by much from the taxes paid by U.S.-owned firms in 1983-85 foreign firms would mean higher corporate tax payments and (Chart 15). Since then, a gap has opened up: if foreign firms a narrower fiscal deficit. In short, the maturing of foreign paid as high a fraction of their sales in taxes as U.S. firms, they affiliates will nudge the “twin” deficits in opposite directions, would have paid about $2 billion more in taxes in 1990. This with the external deficit widening by a m ultiple of the calculation, subject to several caveats,30 points to the conclu amount that the fiscal deficit narrows. If foreign firms achieve a more normal profit in the sion that the shortfall in taxes paid by foreign firms is much This prospect differs from that held out by analysts narrower than one might guess from the shortfall in aggre who interpret the poor profitability of foreign firms as a sign gate income. of tax evasion. In their view, if the foreigners can be made to own up to their success and pay taxes, the fiscal deficit would narrow. Higher profits would be reported because of smaller I m p l i c a t i o n s f o r t h e T w i n D e f ic it s Foreign firms building up their holdings in the United States helped to finance a decade’s excess of imports over exports. As foreign firms built up their holdings, rising payments of profit and interest should have widened the current account deficit. But the surprisingly low returns on those holdings short-circuited the compounding effect whereby the cost of The slowdown o f foreign acquisitions in the United States in 1991 —93, coupled w ith ongoing divestitures, points to a rebound in the profits o f foreign firm s in the years ahead. Chart 15 T a x e s P a id b y F o r e i g n F ir m s i n t h e U n i t e d S t a t e s B il l io n s o f dollars 10 payments to (or larger receipts from) foreign parents. Thus, this scenario would imply a narrower trade deficit matching the higher pretax profits of foreign firms here and would leave the current account deficit initially unchanged. But as for eign firms paid more taxes on their higher declared income, the current account deficit would narrow by the same amount as the fiscal deficit. 1983 84 85 86 87 88 89 90 Sources: Internal Revenue Service, Statistics of Income Bulletin, various issues; Federal Reserve Bank of New York staff estimates. Note: Hypothetical taxes represent revenues if foreign firms paid the same taxes as a proportion of sales as U.S. firms in each of the following sectors: manufacturing; wholesale and retail trade; finance, insurance, and real estate; and all other industries. This interpretation implies a win-win outcome on the twin deficits. In our interpretation, by contrast, the prospective improvement in foreign returns over time if for eign acquisitions remain modest implies a lose-win outcome: more borrowing from foreigners but also higher tax revenues from foreign firms. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r-F a ll 1994 59 This article has shown how the surge in foreign acquisitions goods, but it leaves services untouched and ultimately justi fies investment in U.S. manufacturing. of U.S. companies in the 1980s drove down the aggregate We find some evidence consistent with profit shift return on foreign direct investment in this country. Foreign ing. Foreign firms from countries whose fiscal systems offer ers acquired U.S. firms that barely turned a profit. These the greatest reward for shifting profits to tax havens show firms lost money just after acquisition and then gradually lower profits in the United States. And firms with greater recovered their profitability. Strong acquisition activity opportunities to transfer profits through the pricing of raised the share of foreign holdings at an early stage in this imports from their U.S. affiliates show weaker profits. C o n c l u sio n process. Foreign firms that sold or liquidated their money- Our interpretation suggests that the performance of losing holdings both contributed to the restoration of profits foreign firms in the United States will improve over the next and provided concrete evidence of disappointing returns. several years. The slowdown of foreign acquisitions in the Other proposed explanations carry some weight, but United States in 1991-93 (Chart 2), coupled with ongoing do not bear as much as this. A weak dollar works both ways. divestitures (Chart 10), points to a rebound in the profits of It hurts foreign firms principally engaged in selling foreign foreign firms. A p p e n d ix I: F o r e i g n I n v e s t o r s ’ L o s s e s i n U .S . R e a l E s t a t e Theodore Fischer and Robert N. McCauley Much of the publicity attending the surge in foreign direct billion in economic losses before taxes in 1990-92. The fur investment in the United States in the 1980s focused on for ther decline in real estate values in 1993 took cumulative eign purchases of commercial real estate. Official data show estimated losses to $27 billion. Thus, an industry that in that foreign investments in the real estate sector have per formed as poorly as those in other sectors. The Bureau of Eco 1990 accounted for 9 percent of the investment and 7 percent of the assets of all foreign direct investment in the United nomic Analysis (BEA) has reported that foreign-owned real States has done significantly worse than the reported data estate companies lost money every year after 1985, culminat indicate (U.S. Department of Commerce, Bureau of Econom ing in $ 10.1 billion in pretax losses in 1990-92 (U.S. Depart ic Analysis 1992, pp. 113-14, and 1993, Table A -l). ment of Commerce 1984-94).31 Questions have been raised about foreign firms’ use R e a l E st at e V a l u e s of excessive debt or transfer pricing to reduce their U.S. oper Both domestic and foreign-owned real estate companies have ations’ reported profits. But informed observers would agree seen their commercial property values crash since 1989- We that in real estate, reported losses actually understate true can assess this decline using the Russell-NCREIF index,32 a economic losses. The profitability of foreign real estate com measure of real estate value. According to the capital value panies in the BEA data reflects scheduled depreciation portion of this index, a building bought at the end of 1985 expenses, while an economic measure of net income must for $100 million would only have been worth $67.5 million incorporate changes in the market value of property hold at the end of 1993. Most of this decline occurred in the years ings. Our estimates of market value returns confirm this 1990-93, when commercial property values fell by 31 per hunch: foreign-owned real estate companies racked up $23 cent (Chart A l). Because real estate companies are highly 60 F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 A p p e n d i x I: F o r e i g n I n v e s t o r s ’ L o sse s i n U .S . R e a l E s t a t e (Continued) leveraged, their return on equity was even worse. We shall sold or in other special circumstances. As a result, foreign- assume that the Russell index tracks the price of U.S. prop owned real estate companies do not generally report unreal erty held by foreigners. Even if foreign investors did no ized gains or losses in the value of their commercial property. worse than U.S. investors buying at the same time,33 the for Still, the GAAP income of foreign-owned real estate showed eigners’ timing looks unfortunate in hindsight. considerable losses as early as 1990 (see table, column 4). The reported losses in 1990-91 must have arisen in A c c o u n t in g f o r C a p it a l G a in s a n d L o sse s large part from continuing operations. W hile realized gains Generally accepted accounting principles (GAAP) hide the or losses are included in net income, press accounts and BEA magnitude of the losses suffered by foreigners holding com data suggest that foreign investors increased sales markedly mercial property during the decline in property values from only as late as 1992. Thus in 1990-91, high vacancies and 1990 to 1993. The GAAP balance sheets collected by the falling rents combined to push rental income from commer BEA as part of its annual financial surveys only require com cial property below interest on debt and (noncash) deprecia panies to mark commercial real estate to market when it is tion expenses based on historic cost and GAAP. These depre- Cbart A 1 F o re ig n O u tla y s f o r U.S. C o m m e r c i a l R e a l E s t a t e a n d C o m m e r c i a l R e a l E s t a t e P r o p e r t y V a l u e s I n d e x : 1980= 100 --------------- 14Q B illio n s o f d o lla r s 12 ----------------------Other outlays British outlays 130 10 |] Canadian outlays ! Value o f commercial real estate Scale Japanese outlays 120 110 100 1980 1111 m 80 Sources: U.S. Department of Commerce, Bureau of Economic Analysis; Frank Russell Company. Notes: Investment outlays are the amounts spent by real estate companies to acquire commercial property. Expenditures on hotels and construction are not included. FRBNY Q u a r t e r l y R e v ie w / S u m m e r-F a ll 1994 61 A p p e n d i x I: F o r e i g n I n v e s t o r s ’ L o s s e s in U .S. R e a l E st a t e ( C ontinued) ciation charges, however, depend on the book value rather Russell-NCREIF index of the value of property (shown in than market value of property. As bad as GAAP income was column 1). The exercise makes the key assumption that this in 1990-92, economic returns were worse. index captures the market value of foreign-owned commer cial real estate: that is., foreigners did no worse than all investors. E s t im a t io n o f R e t u r n s We estimate economic returns for foreign-owned real estate We trace the value of foreign-owned real estate com firms, including unrealized capital gains and losses (see panies’ investments over time. BEA data on plant, property, table).34 The exercise combines the income and balance sheet and equipment, of which about 90 percent reflect commer data on real estate affiliates reported to the BEA with the cial property, measure the amount spent on ;acquisitions and • Es t im a t e d N e t In c o m e a n d R e t u r n o n E q u i t y f o r Fo r e i g n - o w n e d R e a l Est a t e C o m p a n ie s Russell Index o f Returns in Commercial Real Estate (Percent) Total Investment in PP&E by Real Estate Companies (Billions o f Dollars) Estimated Gains or Losses in Commercial Property (Billions o f Dollars) Published N et Pretax Income (Billions o f Dollars) Estimated Economic Net Pretax Income (Billions o f Dollars) Published Pretax Return on Equity (Percent) Estimated Economic Pretax Return on Equity (Percent) ( 1) (2) (3) (4) (5) (6 ) (7) 1977 N .A. 6 .1 0 * N .A. - 0 .1 5 19 7 8 6 .8 1 1.6 0 0 .3 8 - 0 .0 8 0.25 -4 .9 0 14 .6 4 1979 19 8 0 10 .8 0 2 .1 8 0 .7 6 - 0 .0 5 2 5 .9 1 7 .1 0 0 .8 3 3 .2 9 8 .4 2 1.22 0 .1 3 0 .1 7 0 .6 3 0.2 4 - 2 .0 4 9 .1 1 8 .3 0 1.4 6 9 .5 9 0 .3 2 - 0 .4 7 2.75 - 5 .7 0 6 .6 3 1 5 .0 7 19 8 2 0 .9 1 - 0 .3 4 -3 .8 2 1983 19 8 4 5.23 5.38 8 .2 9 7 .4 4 1.47 1.87 0 .0 7 1.84 0 .6 7 1 5 .2 0 2.88 2.45 - 0 .6 0 8 .6 3 9 .5 1 0 .9 9 - 0 .2 8 6 .6 0 2 .7 1 17 .5 4 1985 1986 0 .7 6 0 .3 4 - 0 .5 7 1987 -1 .4 7 1 1 .2 0 - 0 .7 8 - 0.66 - 0 .3 6 - 4 .0 5 —4.4 8 - 0 .2 3 - 1 .4 8 19 8 8 - 0.01 9 .2 5 - 0 .0 1 - 0 .6 0 - 0.12 1989 19 9 0 - 0 .4 8 13 .3 3 14 .0 1 - 0 .3 2 - 3 .7 5 - 0.12 - 0.20 -3 .7 1 -0 .6 0 - 0 .4 9 —0.66 -1 .8 7 - 3 .2 7 -1 2 .9 2 -1 1 .9 1 -3 0 .1 8 - 11.66 -9 .9 9 - 8 .9 4 - 4 .0 7 - 9.88 - 7 .7 0 9 .9 2 6 .1 8 - 4 .9 4 - 9.00 - 2 0 .2 5 - 3 2 .1 4 - 5 .8 2 5 .0 0 4 .1 6 -4 .0 0 -4 .0 0 - 1 6 .0 0 -1 6 .3 2 19 8 1 19 9 1 19 9 2 19 93e - 4 .9 9 -1 2 .3 4 N .A. 1.45 -0 .0 6 Cumulated using only years for which Bureau o f Economic Analysis data are available - 1 0 .8 0 N .A. 6 .9 1 Average annual retums+ 1978-92 15 .6 -1 6 .2 1 -1 1 .1 5 - 1 5 .8 4 - 5 .4 2 19 7 8 -8 8 57.8 6 .7 9 - 0 .9 6 7 .3 1 -0 .8 9 5.41 19 8 9 -9 2 -2 6 .8 - 2 3 .0 0 -1 0 .1 9 - 2 3 .1 5 -1 0 .5 1 - 1 7 .5 2 - 2 0 .3 8 -1 5 .1 5 - 1 9 .8 4 -6 .5 9 6 .7 9 - 2 7 .1 7 - 0 .9 6 7 .3 1 -0 .8 9 - 5 .5 5 5.41 -1 4 .1 9 - 2 7 .1 5 - 11.6 6 -1 7 .3 3 Cum ulated using estimated Bureau o f Economic Analysis data 19 7 8 -9 3 1978-88 8 .9 5 7 .8 198 9-9 3 - 3 1.0 - 4 .6 6 Average annual returns - Sources: Frank Russell Company; U.S. Department o f Commerce, Bureau o f Economic Analysis; Federal Reserve Bank o f New York staff estimates. Note: Data for 19 9 3 are all estimates except for the Russell return. * Includes the book value o f the holdings o f PP&E (plant, property, and equipment) as o f the start o f 19 7 7 . t W eighted by the book value o f owners' equity for the published return on equity and the market value o f owners' equity for the economic return on equity. Digitized 62 for FRASER FRBNY Q u a r t e r l y R e v ie w / S u m m e r-F a ll 1994 A p p e n d i x I: F o r e i g n I n v e s t o r s ’ L o sse s i n U .S . R e a l E s t a t e (Continued) construction by foreign-owned real estate companies (col economic returns make the investment in the late 1980s less umn 2) from 1977 onward. We assume these purchases occur anomalous, our calculation leaves foreign investors with a at market value (and that 1977 book value equals market $27 billion loss in 1989-93, probably twice the loss that BEA value, a fairly innocent assumption given the small scale of will eventually report.38 foreign investment in U.S. real estate then). The value of These results do not include commercial property these purchases then changes in accordance with the Russell- owned by industries other than real estate and so understate NCREIF index. The change in value (after accounting for the total impact of the fall in property values on foreign sales35) is the total mark-to-market gain or loss to the real investors. Foreign real estate companies only owned 51 per estate companies (column 3). cent of all foreign-owned commercial property in the United These capital gains and losses are added to operating States in 1991. Foreign-owned hotel firms, classified as ser income, namely, rental revenues less operating and interest vice firms, have the next largest share of commercial property, expense, to calculate a measure of economic income (column with 11 percent of the total foreign investment in U.S. real 5). Operating income is obtained by excluding depreciation estate (U.S. Department of Commerce 1994, Table D-10). On expenses and realized capital gains and revaluations from pre the basis of a 31 percent decline in the Russell index in 1989- tax net income.36 We compute the changes in the market 93, we estimate that including hotels would increase the value of owners’ equity as the sum of increases in external cap 1989-93 losses by an additional $6 billion, to $33 billion. ital and economic net income.37 A return on equity is then computed as the ratio of economic income to the cumulated market value of owners’ equity (column 7). Chart A 2 N e t In c o m e o f Fo r e ig n -o w n e d R e a l E st a t e C o m p a n ie s R e su l t s B il l io n s o f dollars Our estimates of economic returns for foreign investors in Estimated economic income U.S. real estate suggest that foreign firms have done at least as badly as they are reported to have done. In 1978-92, BEA reports a $11.2 billion loss; we compute a $15.8 billion loss , * Reported income ftV (columns 4 and 5). %\ % ft Our calculations for the late 1970s and early 1980s ...... ..... ..... ................ help make sense of the foreign buying since then, but our ..... . i t t f f ft estimates make the results of this recent buying worse. The 1 BEA data show that through the mid-1980s, net income flu ctu ated around zero, but economic returns were respectable if varying (Chart A2 and Table A1, columns 4 and ' ' ..... I 1977 ...... 1 .........i 80 t 1 i « 1 1 ft * ft f • L . 1 -....1 .-.-. i _ L ___t ----- 1____I - . » 'i.___1___ 1 85 90 93 6 versus 5 and 7). And although the Russell-NCREIF returns turn negative in 1986, our calculated returns remain better than the reported returns because the market price declines remain smaller than GAAP depreciation. While the earlier Sources: U.S. Department of Commerce, Bureau of Economic Analysis; authors’ estimates; Federal Reserve Bank of New York staff estimates. Notes: Economic income is the sum of operating income (rental revenues less operating and interest expenses) and calculated capital gains and losses. Data for 1992 are preliminary; data for 1993 are partly estimated. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 63 A p p e n d i x I: F o r e i g n I n v e s t o r s ’ L o sse s i n U.S. R e a l E s t a t e (Continued) Because other industries own the remaining 38 percent of some foreigners appear to have invested more heavily in loca commercial property, the total loss could be even higher, in tions and types of commercial property that suffered the the neighborhood of $50 billion. largest declines in value.39 At the same time, despite the scale of the losses in the United States, we make no claim that for C o n c l u sio n eigners’ investments turned out any worse than their invest In losing $27 billion over 1989—93, foreign-owned real ments in their home or third markets, since commercial estate companies have incurred an average annual return on property values have fallen worldwide (Bank for Internation equity o f-17 percent. The economic loss over 1989-93 is al Settlements 1993, pp. 155-81). likely to be $13 billion worse than the published loss. These are very conservative estimates. Both BEA data and market observers indicate that foreign investors have not liquidated their U.S. real estate Two important factors left out of our calculation holdings en masse. Yet in view of the price decline and lever would only deepen the losses. From the standpoint of the for age, many foreign investors in U.S. real estate must have lost eign investors, what matters is the rate of return in home cur all of their equity. In some cases, loan collateral may have rency. The substantial leverage associated with real estate extended beyond U.S. properties to home country property. investment still typically leaves the equity exposed to In other cases, banks and other creditors may have chosen to exchange rate changes. So for investors from the Continent restructure debt service rather than to foreclose. In the event and Japan, dollar losses have been compounded by the of restructurings or liquidations, the losses would have spread declines in the dollar against the home currency. Our esti from the equity investors examined here to creditors, includ mates also assume that foreign-owned commercial property ing foreign banks.40 performed no worse than the overall U.S. market. But at least A p p e n d i x II: M u l t ip l e R e g r e s s i o n A n a l y s i s D avid S. Laster The analysis in the text focuses on the bivariate relationships 1977 through 1991. Data on U.S. businesses acquired or between profitability and each of four factors: the recency of established by foreign investors are for the years 1980 acquisition of U.S. operations, leverage, exchange rates, and through 1992. The data are aggregates for each of the thirty- transfer pricing. In this appendix we use multiple regression five industry categories that together account for all nonbank analysis to assess the impact of each of these factors while foreign direct investment.41 holding the others constant. The analysis uses annual data from the Bureau of D e f i n i t io n s o f V a r i a b l e s Economic Analysis that fall into two general categories. Data To proxy for the m aturity of foreign holdings within an on the operations of foreign-owned affiliates cover the years industry, we define RECENCY for a given industry as the Digitized 64 for FRASER F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 A p p e n d i x II: M u l t i p l e R e g r e s s i o n A n a l y s i s (Continued) assets of affiliates acquired or established in the current year define M*E - (MPAR + MOTH)*E, and X*E = (XPAR + or previous two years in that industry, divided by assets of all XOTH)*E, where E is the de-meaned natural logarithm of affiliates currently in the industry. To measure the effect of the trade-weighted dollar, as measured by the Board of Gov leverage, the debt ratio LEV (total liabilities divided by ernors of the Federal Reserve System. Our expectation that an assets) is used. Two variables capture the effect of transfer appreciation of the dollar will benefit importers and harm pricing. MPAR is the ratio of imports from foreign parent exporters means that the coefficients of M*E and of X*E corporation to sales; XPAR is the ratio of exports to foreign should be positive and negative, respectively. parent corporation to sales. To control for any impact that trade activity, per se, has on profitability, we include two S u m m a r y S t a t ist ic s additional variables. MOTH is imports from other parties Table B1 provides sample means for the dependent and divided by total sales; XOTH is exports to other parties explanatory variables used in the analysis, in aggregate and divided by sales.42 While we have no expectations concerning by major sector.45 In aggregate, the profitability measures the signs of the coefficients of MOTH and XOTH, we expect were strongest during the first five years of the sample, 1977- for reasons discussed above that RECENCY, LEV, MPAR, 81; they were weakest in 1990 and 1991. Note that the aver and XPAR will all have negative coefficients. age affiliate debt ratio (LEV), steady from 1977 to 1984, The exchange rate is the most difficult factor to increased by 8 full percentage points from 1984 to 1991. model because it has widely diverse impacts on different industries. We assume, as a first approximation, that the sen R e g r e s s io n A n a l y s is sitivity of an affiliate’s profits to exchange rates is a linear A series of twelve pooled regressions were run to test the function of its import and export exposure. In particular, we impact of each of the factors discussed above on affiliate Table El C h a r a c t e r i s t i c s o f F o r e i g n F i r m s in t h e U n i t e d S t a t e s , 1 9 7 7 - 9 1 Sample Means for A ll Industries Annual Averages Average 1981 - 19 7 7 19 7 8 ROA (percent) 2 .9 4 2.83 3 .5 0 ROS (percent) RECENCY LEV MOTH 2 .7 9 N.A. 0 .6 5 9 0 .0 2 4 N.A. 19 8 1 19 8 2 19 8 3 19 8 4 1985 19 8 6 19 8 7 19 8 8 19 8 9 2.22 2 .8 0 - 0.02 1.3 1 1.83 0 .9 6 2.33 1.8 1 0 .9 8 2 .6 3 0 .5 1 0 .1 8 1.2 4 2 .2 7 0 .8 1 0.20 N .A. 1.3 1 0 .2 2 4 1.7 9 0 .2 8 0 0 .6 7 9 0.690 0 .0 3 5 19 7 9 1980 3 .1 4 3 .31 N.A. 0 .0 6 8 0 .6 7 8 0 .0 4 1 0 .0 6 6 0 .3 4 3 0 .6 6 5 0 .0 5 4 0 .6 7 2 0 .6 7 6 0 .2 9 9 0 .2 12 0 .121 0 .6 7 3 0 .6 5 5 0 .0 3 4 0 .6 5 6 0 .1 2 9 0 .6 6 7 0 .0 4 1 0 .0 3 7 0 .1 9 2 0 .6 7 7 0 .0 3 6 1990 1991 1.12 0 .16 - 0.66 1.05 1.2 6 - 0 .2 4 -1 .3 4 0 .8 5 0 .2 4 8 0.222 0 .7 3 5 0 .1 5 7 0 .7 3 8 0.22 0 .7 1 0 0 .0 3 6 0 .0 3 5 0.036 0 .0 3 7 0 .0 3 2 0 .0 4 7 0 .0 4 6 0 .0 4 8 0 .085 0 .0 8 3 0.05 0 .0 8 2 0 .0 4 0 .085 0.022 - 0 .16 0.022 0.021 - 0.20 0 .0 2 4 0.02 0.00 XOTH 0.038 0.032 0.036 0 .0 4 4 0 .0 4 7 0 .0 6 9 0 .0 4 2 0 .0 4 1 0 .0 3 8 0 .0 3 3 M PAR 0 .1 2 3 0 .0 7 7 0 .0 9 4 0.060 0 .0 5 6 0 .0 6 9 0 .0 6 7 0 .0 8 4 0 .085 0.0 3 0 .0 9 2 X PA R 0.022 0 .0 2 4 0 .0 2 4 0 .0 2 4 0.022 0 .0 2 3 0.020 0.021 0 .0 1 8 E - 0 .0 5 - 0 .16 - 0.21 - 0.21 - 0 .0 5 0 .0 7 0 .15 0 .0 2 3 0 .2 4 0 .0 8 9 0 .0 1 7 0.2 8 0 .0 4 - 0 .11 - 0 .0 9 -0 .1 9 19 9 1 0 .6 9 0 .0 4 0 .0 8 Sources: U.S. Department o f Commerce, Bureau o f Economic Analysis; Board o f Governors o f the Federal Reserve System. Notes: Statistics represent an average o f measures for component industries, weighted by the 19 8 7 gross product o f affiliates. RO A = net income/assets; ROS = net income/sales; RECENCY = assets o f U.S. affiliates acquired or established in the current year or previous two years/assets o f all U.S. affiliates; LEV = (current liabilities + long-term debt)/ assets; MOTH = imports excluding those from foreign parent/sales; XOTH = exports excuding those to foreign parents/sales; M PAR = imports from foreign parent/sales; X PA R = exports to foreign parent/sales; E = de-meaned log o f trade-weighted dollar. Fluctuations in average values o f MOTH and M PAR reflect changes in the set o f industries for which the variables’ values are available. N. A. = not available. FRBNY Q u a r t e r l y R e v ie w / S u m m e r-F a ll 1994 65 A p p e n d i x II: M u l t i p l e R e g r e s s i o n A n a l y s i s (Continued) profitability. Two dependent variables were used: return on Its estimated coefficient in aggregate equation 1, -.0 8 1 , assets and return on sales. Each regression was run for the means that each additional dollar of debt on an affiliate’s entire sample of thirty-five industries, for the nineteen manu books reduces its profits by 8.1 cents, consistent with a plau facturing industries, and for the five wholesale trade indus sible 8.1 percent rate of interest. The corresponding coeffi tries.44 The regressions were run using weighted least cient for manufacturing, —.048, while smaller, is still within squares. Each observation, representing the performance of a standard error of a credible value; the coefficient for whole affiliates in a given industry, had a weight proportional to the sale trade, -.163, seems too large. Equations 2, 4, and 6 sug 1987 gross product of affiliates in that industry. To control for gest that the variable is better at explaining differences in variations in macroeconomic conditions over the sample peri profitability across industries than at explaining variations in od, year dummies were employed in each regression. Finally, the profitability of a given industry over time. The reason each specification was run with and without industry dum could be that leverage ratios vary less within a given industry mies. When industry dummies are omitted, the regression than across industries. explains variations in profitability across industries. When industry dummies are included, the regression explains varia Ex c h a n g e R ates tions in profitability within each industry over time. Because transfer pricing and exchange rate effects are transac- The regression results are reported in Table B2. tions-based, the discussion concerning them focuses on the Because RECENCY and LEV are asset-based measures, their return on sales regressions 7-12. Five of these six regressions performance can best be discussed in the context of return on satisfy our expectation that the coefficients of M*E would be assets regressions 1-6. positive and those of X*E negative. X*E is most significant in the “all industries” regression; M*E is most significant for R ecen cy of A cq uisitio n of U.S. O perations A s predicted, RECENCY has a negative coefficient in all six wholesale trade. To interpret the magnitude of the parameter esti return on assets regressions; in five of these regressions, the mates, imagine two affiliates—one a “pure exporter,” the estimated coefficient is significant at the 5 percent level. The other a “pure importer.” The pure exporter sells all of its out coefficient of the first equation, —.058, implies that the differ put abroad; the pure importer has imports equal to sales. ence in the return on assets of an industry whose affiliates Were the dollar to depreciate by 10 percent, the pure exporter have been purchased or established in the past three years and would benefit. If the exporter held its export prices fixed in another all of whose affiliates have been operating under cur foreign currency terms, its profit margin would rise by 10 rent ownership for three or more years would be 5.8 per percentage points. The estimated coefficient for X*E in cent.45 While of a similar magnitude for each sector, the esti equation 7 states that the pure exporter’s profit margin actu mated effect has greatest sta tistic a l sign ificance for ally rises by less, by 5.6 percentage points. manufacturing affiliates. The pure importer, by contrast, would be harmed by a 10 percent dollar depreciation because it would raise the L everage price of its inputs, all of which are imported. In addition to In specifications without industry dummies, the leverage any compression of profit margins by exporters to the United variable LEV is negative and significant at the 5 percent level. States, our estimate of M*E in equation 7 implies that the 66 F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 A p p e n d i x II: M u l t i p l e R e g r e s s i o n A n a l y s i s (Continued) pure importer’s contemporaneous profit m argin would dummies are not used, the variable MPAR is significantly decline by 1.2 percent. negative at the 1 percent level for all industries and for manu facturing. Thus, reported profits are lowest for industries in T r a n sf e r P r ic in g which affiliates have the greatest opportunity to transfer Of the two proxies for transfer pricing, the variable XPAR price. Equation 7 associates each additional dollar of imports fails to be significant in every specification, suggesting that from the foreign parent with profits that are 6.7 cents lower. affiliates do not extensively manipulate transfer prices when This effect is more pronounced in manufacturing (30 cents), exporting. Imports are a different story. When industry and virtually nil in wholesale trade. TabIt B2 P r o f i t a b i l i t y o f F o r e i g n F i r m s in t h e U n it e d S t a t e s b y I n d u s t r y : 1 9 8 1 - 9 1 Regression Results Return on Assets A ll Industries Explanatory Variables RECENCY LEV X*E MOTH XOTH M PAR -0 .0 5 8 * * - 0 .0 3 3 * * (0.0 0 9 ) (0 .0 0 9 ) - 0 .0 8 1 * * 0 .3 3 1* * (0.0 7 8 ) Adjusted R-squared F-statistic (0 .0 19 ) 0 .3 2 4 * * (0.066) - 0 .4 5 0 * * (0 .17 3 ) (0 .15 2 ) All Industries Manufacturing 0 .1 5 4 * * - 0 .0 3 9 * * ( 0 .0 1 1 ) -0 .0 4 3 * * (0 .0 1 2 ) - 0 .0 7 8 * (0.0 56) --0 : )6 1 -* (0 .0 1 0 ) - 0 .0 2 5 * (0 .0 1 0 ) - 0 .0 3 6 ** (0 .038 ) - 0 .0 4 8 * 0 .0 0 5 - 0 .1 6 3 * * (0 .0 2 6 ) (0.033) —0.0 6 3 * * (0 .017 ) - 0.020 (0 .0 2 1 ) - 0 .1 1 3 * (0 .04 4) -0 .0 5 9 * * (0 .0 2 1 ) - 0 .2 10 (0.268) 0 .0 5 2 (0.257) 0 .5 15 * * (0.089) - 0 .0 6 8 - 0 .5 1 2 - 0 .0 0 4 (0.333) (0.320) (0 . 12 2 ) 0 .2 2 1 ** 0 .0 4 2 0 .0 9 7 0 .0 4 4 (0.045) (0 .0 5 8 ) (0.0 6 3 ) (0.0 7 3 ) (0.055) - 0 .0 6 4 - 0 .0 9 3 (0 .0 8 6 ) -0 .0 3 3 (0.062) - 0 .2 5 0 (0 .12 8 ) - 0 .0 4 1 (0.049) -0 .0 5 2 * * -0 .0 0 7 - 0 .3 1 2 * * - 0 .0 5 4 -0 .0 3 5 (0.089) N um ber o f observations - 0 .0 3 6 - 0 .3 9 8 * (0 .0 17 ) XPAR Return on Sales Wholesale Trade Wholesale Trade No Industry Industry No Industry Industry No Industry Industry No Industry Industry No Industry Industry No Industry Industry Dummies Dummies Dummies Dummies Dummies Dummies Dummies Dummies Dummies Dummies Dummies Dummies ( 10 ) (2 ) (4) (5) (6 ) (7) (8 ) ( 12 ) (3) (9) (1 ) (1 1 ) (0 .0 14 ) M*E Manufacturing (0.049) (0.053) 0 .0 5 0 0 .2 2 4 (0 .14 0 ) (0 .17 3 ) (0.0 8 4 ) - 0 .4 3 2 (0.280) - 0 .0 5 6 0 .5 0 9 * * (0 .091) -0 .0 7 7 ‘!U 33> 0 .3 0 9 * * os) mi.; (0.063) -0 .0 1 6 (0.098) 0 .0 1 5 -0 .0 8 7 (0 .0 3 1) (0.054) 0 .0 9 0 0 .1 5 1 (0.087) (0 .1 1 3 ) (0 .0 2 2 ) 0 .1 1 6 0 .1 3 8 (0.094) (0.0 73) io 5 - 0 .5 7 0 * * (0 .0 1 1 ) - 0 .1 2 8 (0.259) - 0 .1 0 8 - 0 .0 3 5 * * (0 .0 1 1 ) - 0 .0 2 3 (0 .0 14 ) -0 .0 1 3 - 0 .0 0 8 - 0 .0 5 1 * * (0 .0 1 2 ) - 0 .0 3 5 * (0.025) 0 .1 7 6 (0.2 42) - 0 .6 5 2 * 0 .1 4 5 * * (0 .0 3 1) -0 .0 0 7 (0 .0 19 ) (0 .0 15 ) 0 .1 4 4 * * (0.0 32) - 0 .0 2 3 (0 .04 6) |§|jg ‘ (0 .16 9 ) 0 .15 3 * * (0.054) 0 .0 4 3 0 .0 9 0 0 .0 4 0 (0.065) (0 .0 6 1) (0.0 69) (0 .0 19 ) (0 .03 7) - 0 .1 0 5 (0.096) 0.012 (0.0 59) - 0 .2 5 8 * (0 . 12 0 ) - 0.012 (0 .02 2 ) - 0 .0 1 5 (0 .03 4) - 0 .0 2 7 (0.0 54) -0 .3 0 0 * * -0 .0 6 7 (0.0 79) 0 .0 0 7 (0 .0 1 1 ) - 0 .0 2 8 (0 .0 5 1) 0 ,0 4 7 0 .2 4 2 (0 .15 6 ) (0 .16 7 ) 0 .0 3 0 (0.059) - 0 .0 3 6 (0 .10 7 ) (0 .3 2 1) (0.302) - 0 .3 8 4 (0.2 64) (0.0 43) 0 .0 9 6 * * 0 .1 4 1 * * (0 .0 19 ) 0 .0 2 8 0 .0 3 8 (0 .0 3 1) (0.039) 327 327 202 202 51 ' 327 327 202 202 51 51 0.32 0 .6 0 0.3 4 0 .5 8 0 .9 1 0.93 0.23 0 .5 8 0 .3 3 0.60 0.90 0.9 2 15 .1 * * 5.6** 10 .4 * * 2 .4 * 4 4 .5 * * 19.5* * 9 9 ** 2.7** 10 .2 ** 2.4* 38.0* * 15.5* * Sources: U.S. Department o f Commmerce, Bureau o f Economic Analysis; Board o f Governors o f the Federal Reserve System. Notes: Return on assets = net income/assets; return on sales = net income/sales; RECENCY = assets o f U.S. affiliates acquired or established in the current year or previous two years/assets o f all U.S. affiliates; LEV = total liabilities/assets; M*E = (total imports/sales) *E, where E is the de-meaned log o f the trade-weighted dollar; X*E = (total exports/sales) *E; MOTH = imports excluding those from foreign parent/sales; XOTH = exports excluding those to foreign parent/sales; M PAR = imports from foreign parent/sales; X P A R = exports to foreign parent/sales. Observations are weighted by 19 8 7 gross product o f U.S. affiliates. Standard errors are in parentheses. Constant term, industry dummies, and year dummies are not reported. Banking affiliates are excluded from the data. The F-statistic is a joint test o f the significance of the eight explanatory variables listed in the table. *Significant at 5 percent level. **Significant at 1 percent level. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 67 A p p e n d i x II: M u l t i p l e R e g r e s s i o n A n a l y s i s {Continued) C o n c l u sio n together with the dollar, while exporters’ profits move To summarize, the regression results closely conform to against it. And while exports to foreign parents prove expectations. Recent acquisition activity and high leverage insignificant, affiliates importing the most from their parent are associated with low profits. Importers’ profits move companies report the lowest profits. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 68 En dn otes 1. Figures cited in paragraph are from Zeile (1 9 9 4 , p. 154), and U.S. Note 8 continued Department of Commerce (1993 Appendix D, Tables 1 6 ,1 7 , 32, 33). between “foreign-controlled” and “other domestic” firms in the finance, 2. Fahim-Nader 19 9 4 , p. 50. “Acquisitions” and “establishments” in the within the sector. Second, focusing on the sector in which foreign firms insurance, and real estate sector, owing to differences in composition Bureau o f Economic Analysis data do not exactly correspond to purchases were most profitable in 19 9 0 minimizes the possibility o f mistaking a o f existing assets and greenfield investment, respectively. Many greenfield general decline in corporate profits for evidence o f post-acquisition costs. investments proceed within firms already operating here. The Japanese Indeed, foreign manufacturers were more profitable in 19 9 0 than in 19 8 6 - automotive companies, for example, have built up their manufacturing 89, so the estimate o f more than $4 billion for the post-acquisition decline capacity within extant sales operations. Conversely, establishments can in manufacturing profits is, if anything, understated. Data on the prof own preexisting assets. For example, a new partnership can hold an existing itability o f firms in the year before acquisition are from U.S. Department office building. o f Commerce (19 9 3 ) Appendix D, Tables 16, 17, 32, 33. The Internal Revenue Service data on firms recently incorporated differ slightly from 3. From a regression for 19 7 9 -9 1 that does not control for interest rates or the Commerce Department figures. The Internal Revenue Service data economic growth, Klein and Rosengren (1994, p. 382), report a slightly include newly established firms, but these are small relative to newly smaller elasticity of 1.56. Using International Trade Administration data acquired firms. In particular, for the years 19 8 7 -9 0 , foreign firms that for 19 7 7 -8 6 , Caves (1989) derives similar results for relative equity prices. were newly established generated in their first year o f operation only about His measure is significant for only some specifications. Given the strong a twentieth as many sales as did newly acquired firms in the year before trend of stock prices and acquisitions from Japan, it is not surprising that they were bought. Note also that some newly acquired firms do not get re without Japan, the R2 in our regression falls from .55 to .41. W hile foreign incorporated. and domestic equity prices retain their respective signs, the significance of foreign equity price declines to the .08 level and the significance o f U.S. 9. Manufacturers incorporated 19 8 7 -9 0 reported assets o f $ 1 0 9 billion on equity prices is lost. The last wave of the British acquisitions in the late their 19 9 0 tax returns; $ 1 0 9 billion x 10 .9 percent x 8 percent interest rate 1980s is hard to reconcile with the cost of equity explanation (McCauley yields $0.95 billion. andEldridge 1990). 10. Grubert, Goodspeed, and Swenson (19 9 3 , p. 2 5 1) reach a different 4. This measure is only indicative since outlays that are financed with debt to unaffiliated parties do not become part of the ownership stake. conclusion: “The date of incorporation variables reflect asset revaluations rather than operating start-up losses.” Curiously, these authors later (p. 258) embrace a “m aturation” effect as showing improved operating 5. Note that profitability is measured as a return on sales rather than on results over time. Note that the 19 8 6 Tax Act removed much of the oppor assets to eliminate the effect o f asset revaluations in an acquisition. tunity for acquirers o f firms to step up the value of assets for the purpose of taking larger depreciation expenses. 6. Fahim-Nader (19 9 4 , p. 58) provides return on sales data for foreign acquisitions in 19 8 7 -9 3 . Chart 6 focuses on 19 8 7 -8 9 to avoid any distor 11. The reporter explained that “development spending for new products tions owing to recession. The consistency o f the underperformance o f tar soared partly because Coca-Cola, in anticipation of a sale of the studios, had gets o f foreign acquisitions in the BEA data makes it odd that a sample of sharply curtailed production, and the new owners were left with only a 1 1 8 acquisitions showed no difference in profit between acquisition targets handful of projects.” An unnamed source was cited as saying that “operat and nonfinancial firms on Compustat (Grubert, Goodspeed, and Swenson ing profits never exceeded a hundred million dollars,” and that figure did 1993, p. 256). “The inconsistency may be due to differences in sample size not include the debt incurred in buying the studios for “the official price of and coverage, to purchases of parts or divisions o f a company, rather than $3.4 billion,” or “closer to six billion, considering all the collateral costs” the whole firm, and other differences” (Grubert 1993, p. 93). To the factors (Stewart 19 94, pp. 48, 51). cited may be added the possibility that the universe o f Compustat nonfi nancial firms may not offer an appropriate basis of comparison. 12. Canada was the exception, suggesting that geographic, linguistic, and cultural distance matters (Lupo, G ilbert, and Liliestedt, 1978). Year of 7. Swenson (1 9 9 3 , pp. 255-84) finds that price-earnings ratios o f foreign incorporation is a noisy proxy for age because new operations can be targets are 19 percent higher than those o f domestic targets. absorbed into, or made a subsidiary of, long-standing operations without a new incorporation. Long-standing operations can be reorganized and 8. This discussion confines itself to manufacturing for two reasons. First, freshly incorporated. doing so avoids the question o f the com parability o f interest expense N otes F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 69 E n d n o t e s (<Continued) 13- See Grubert, Goodspeed, and Swenson (1993, p. 256). The results are Note 2 0 continued not comparable since the firms in the fixed sample that were recently International staffing increased from approximately two hundred interna acquired in 19 8 0 would have benefited from the General Utilities step-up tional examiners in 19 8 0 to almost five hundred positions in 1 9 8 7 ” (State of assets for tax depreciation. ments of Rep. David E. Bonier and Patrick G. Heck, Assistant Counsel, ibid., p. 54. 14. The time that elapses between an acquisition and the sale or liquida tion that stems from a recognition o f its failure (perhaps after a manage 2 1. Foreign firms might also have other incentives for shifting income ment change) suggests that the (smaller) number of employees some years from the United States. Managers’ performance could, for example, be before the divestment would be a more appropriate denominator. Our evaluated more on the basis o f home market results than on the perfor attrition rate is therefore a conservative measure. mance o f foreign subsidiaries. Or firms might decide that it is in their ulti 15. Note that these data also include manufacturing results for Japanese Treasury. mate self-interest to pay taxes to local tax authorities rather than to the U.S. automotive firms whose primary activity was in wholesale trade. 22. Statements o f Fred T. Goldberg, Commissioner o f Internal Revenue, 16. Knetter (1994) recognizes that these studies overlook the fact that the and Charles S. Triplett, IRS Deputy Assistant Chief Counsel (U.S. Con U.S. importer is an affiliate of the foreign exporter. Ragnan and Lawrence gress, House Committee on Ways and Means, 19 9 0 , pp. 78 and 104). (1993) represent an exception on the export side; they find that although dollar export prices respond little to the dollar’s exchange rate, U.S. m ulti 23- In Britain, the integration of corporate and individual taxation allows nationals’ foreign affiliates do vary margins in response to exchange rate shareholders a tax credit for corporate income taxes paid to the Inland changes. Revenue on income earned in Britain. But many multinationals based in 17. By contrast, Grubert, Goodspeed, and Swenson (1 9 9 3 , pp. 256-7) They therefore must pay additional advance corporation tax on dividends force the profitability of all of wholesale trade and all o f manufacturing to paid. Britain lack sufficient British earnings and thus British income taxes paid. respond in the same manner to the dollars value and find a positive relation for wholesale trade and no relation for manufacturing. See Appendix II for a 24. But see Hufbauer and Van Rooij (1992, p. 132) for the view that major regression design that tests the interaction o f the dollar’s exchange rate countries other than the United States de facto tax on the territorial principle. with each industry’s net imports. 25. The 1992 loss for the credit countries reflects a loss of $3 .4 billion by 18. Offsetting this general relation to some extent is the greater import Canadian manufacturers in 19 9 2 , compared with a profit of $ 1.5 billion in propensity of foreign-owned manufacturing firms (Graham and Krugman 19 9 1. This swing is entirely explained by DuPont, in which the Canadian 19 9 1, pp. 67-70). firm Seagram’s holds a 24 .4 percent stake. DuPont swung from a $ 1 .4 bil lion profit in 19 9 1 to a $3-9 billion loss in 19 92 owing to $4.8 billion in 19- This test does not preclude the opportunity for foreign firms to be accounting charges for post-retirement medical benefits and for higher more leveraged at home, and therefore in the United States, than are U.S. deferred taxes (Moody’s 19 9 4 , pp. 1 1 2 6 ,4 1 8 5 ). firms. Moreover, note that if U.S. assets are accounted at closer to market value than are global assets because they are recently acquired, then even 26. Grubert, Goodspeed, and Swenson (19 9 3 , pp. 252-53) find no relation this standard permits effectively higher leverage in the United States than between affiliate profitability and total purchases from other firms (which in the consolidated global firm. at best proxies intrafirm transactions only crudely). W hen the relationship between reported profits and exports to parents was tested, no correlation 20. “IRS efforts in the transfer pricing area are o f relatively recent vintage, was found. as they relate to FCCs [foreign-controlled corporations}” (IRS Commis sioner Fred T. Goldberg, cited in U.S. Congress, Committee on Ways and 27. Recency of acquisition might also play a role. New affiliates are both Means, 19 90, p. 203). The international staffing at IRS can be interpreted less profitable and more reliant on intrafirm imports than are old affiliates. in various ways: “W hile the number of foreign-owned subsidiaries in this Accordingly, the multivariate regression controls for recency of acquisition country has soared in the last decade, the number of international examin as well. ers at IRS has remained flat, between 4 0 0 and 500, for the last four years.” Cfi: “The number o f foreign-owned corporate income tax returns filed 28. A similar calculation may have generated the estimate o f the tax rev increased from twenty thousand in 19 8 0 to forty-five thousand in 1987. enues to be raised by “preventing] tax avoidance by foreign corporations”: F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 70 N otes E n d n o t e s (<Continued) Note 2 8 continued 34. See Theodore Fischer, “Foreign Investors’ Losses in Real Estate: $9-0 billion in 19 93, $ 1 1 .0 billion in 1994, $ 11.5 billion in 19 9 5 , and A Methodological N ote,” Federal Reserve Bank o f New York working $ 13 .5 billion in 19 9 6 (Clinton 19 92, p. 22). paper, for details on the estimation. \ 29- The comparable figure for other U.S. firms was 47 percent in 19 9 0 3 5. The sales data do not include the property o f companies that are liqui (Hobbs 19 93, p. 131). dated or pass out o f foreign ownership, events akin to sales. The value of property, plant, and equipment of real estate affiliates that ceased to exist is 30. The Internal Revenue Service data on which the calculation is based do reported in a residual category labeled “restatements.” The value of liqui not include any adjustments resulting from audits. The comparability of dated or sold firms’ property could be as high as $2 billion and $5 billion in the activities o f foreign and domestic firms in wholesale trade has been 19 9 1 and 19 9 2 , respectively, as opposed to the sales o f $4.5 billion and questioned owing to the high import intensity of, and relatively low value $2.9 billion, respectively, for which we have accounted. Because such liqui added by, foreign firms (KM PG Peat Marwick 1994, p. 9). Offsetting this dations naturally occurred after most of the real estate price decline, includ bias is the failure of the calculation for manufacturing to account for the ing them in our estimates would decrease estimated 19 92-9 3 losses by no long-noted heavy mix o f more profitable, more research-and-development- more than $0 .6 billion. intensive industries among foreign firms here. 36. Depreciation is excluded from operating income since the Russell 31. Unless otherwise noted, “real estate” includes foreign affiliates princi index captures the drop in assessed value if needed capital improvements pally engaged in development and management of property. It excludes are not made, or their cost if they are made. Estimated realized capital gains hotels, which are grouped into services in the foreign direct investment and revaluations are excluded from operating income as it is used here since data gathered by the Bureau o f Economic Analysis (BEA). This appendix they are already accounted for in our measure o f capital gains based on the also excludes from the analysis commercial property owned by other kinds Russell index. of industries such as manufacturers, oil companies, or banks, as well as resi dential property used for personal purposes. The BEA also computes 37. Because consistent information is not available on the size o f divi income data in an annual analysis o f the balance of payments (BOP) flows dends, they are assumed to be zero. External capital infusions are obtained caused by foreign direct investment. The balance of payments income by taking the annual change in the BEA published value for owners’ equity includes that part of the operations net income that is earned by the foreign and excluding an estimated change in retained earnings o f all affiliates parent as well as net interest paid on intercompany debt. In 1 9 9 0 -9 2 , the (including unincorporated ones). BOP data show losses o f $3 .8 billion (U.S. Department o f Commerce, 1993a, p. 86). This number is smaller than the sum of the income reported 38. W e have projected that next year, BEA w ill report $2 .4 billion for in operations data because the foreign parents owned only 75 percent of 19 93 depreciation. The gap w ill be wider if this estimate proves too large. their real estate affiliates and were owed interest on a net debt of $ 17 billion in 19 9 1. Despite these differences, the general results of this appendix also 39. In its report 19 93 Japanese Disinvestment in U.S. Real Estate, Kenneth apply to the BOP concept of income. Leventhal & Company indicates that the Japanese investors put 50 percent 32. The Russell capital index tracks the assessed market value o f unlever percent of the Russell-NCREIF index (hotels only make up a very small aged property only, while the total index also includes rental income and part o f the Russell index). The Russell office index declined by 36.2 per operating costs. The capital index measures the change in appraised value cent over 19 9 0 -9 2 , while the overall index declined by only 26.7 percent. of their nonhotel investments into offices, while offices constitute only 32 less the cost of any capital improvements; it excludes the depreciation of However, there is no information on the sectoral distribution of all foreign- book values. Giliberto terms the index “the most widely used performance owned real estate. measure for real estate.” See Giliberto 19 94, p. 55, for the formula. 40. Losses on U.S. real estate can appear in odd places. In 19 93, Long Term 33. Because most domestic real estate companies are privately owned, Credit Bank’s Australian subsidiary reportedly made a A $ 45 million provi little data are available on their financial performance. Until quite recently, sion against a loan to the U.S. subsidiary of EIE Limited, a troubled Japan exchange-traded real estate investment trusts remained a small and proba ese real estate company that financed a New York hotel (“No Relief from bly unrepresentative proxy for real estate values. Moreover, like closed-end Bad Debts in Australia,” Asia Money, July-August 19 9 4 , p. 25). funds, such trusts trade at values that bear a varying relation to underlying asset values. A direct comparison between domestic and foreign real estate 4 1. Our thirty-five industries are: petroleum; food and kindred products; companies is therefore not attempted. industrial chemicals and synthetics; drugs; soap, cleaners, and toilet goods; N otes FRBNY Q u a r t e r l y R e v ie w / S u m m e r-F a ll 1994 71 E n d n o t e s 0 Continued') Note 4 1 continued Note 4 2 continued other chemicals; primary metal industries; fabricated metal products; M*E and X*E. machinery, except electrical; electric and electronic equipment; textile products and apparel; lumber, wood, furniture, and fixtures; paper and 43. To be consistent w ith the regression estimates, these averages are allied products; printing and publishing; rubber and miscellaneous plas weighted by the 1987 affiliate gross product of each industry. tics products; stone, clay, and glass products; transportation equipment; instruments and related products; other manufacturing; motor vehicles 44. Since six o f the explanatory variables used in the regression are trade- and equipment wholesale trade; metals and minerals wholesale trade; other related, and the finance sector is virtually uninvolved in trade, efforts to durable goods wholesale trade; farm-product raw materials wholesale apply the analysis to that sector yield few results and are therefore not trade; other nondurable goods wholesale trade; food stores; other retail reported. trade; finance, except banking; insurance; real estate; services; agriculture, forestry, and fishing; mining; construction; transportation; and communi 45. This estimate is much larger than the 2.2 percent gap implied by IRS cation and public utilities. The Bureau of Economic Analysis, for reasons of data. In 19 90, affiliates incorporated within the three previous years earned confidentiality, suppresses data for some industries in certain years. Thus, a return on assets o f- 1 .7 percent, while those incorporated more than three our “all industries” regressions have 327 observations o f a possible 385 for years earlier earned 0.5 percent. One reason for the discrepancy is that year the eleven-year sample period. The regressions are run on the observations o f incorporation is an unreliable indicator o f the true age o f a corporation. available. Thus, some newly acquired affiliates are misclassified as more mature ones and vice versa, downwardly biasing the gap between new and old affiliates. 42. A second reason for including these variables is that they, together For further discussion, see Hobbs 19 9 3 , pp. 13 2-34 . with MPAR and XPAR, serve as slope dummies for the interaction terms R e feren ces “A 25-Year Profile o f Mergers and Acquisitions.” MERGERS & ACQUISI TIONS, September-October 1990. Blair, Margaret M., and Robert Litan. 19 90. “Corporate Leverage and Lever aged Buyouts in the Eighties.” In John B. Shoven and Joel Waldfogel, eds., D ebt , T a x e s , Auerbach, Alan J ., and Kevin Hassett. 19 93. “Taxation and Foreign Direct an d C o r p o r a t e R e s t r u c t u r i n g . Washington, D .C .: Brookings, 1990. Investment in the United States: A Reconsideration of the Evidence.” In Alberto Giovannini, R. Glenn Hubbard, and Joel Slemrod, eds., S t u d ies in I n t e r n a tio n a l T a x a t io n . Chicago: University of Chicago Press. Bank for International Settlements. 1993. A n n u a l R eport . Basle: Bank for International Settlements. Browning, E.S. 19 9 3 . “Thomson’s RCA U nit No TV Bonanza.” W all S treet J o u r n a l , August 19. Cakici, Nusret, Chris Hessel, and Kishore Tandon. 19 9 1. “Foreign Acquisi tions in the United States and the Effect on Shareholder W e a lth .” J o u r n a l of F in a n c ia l M an a g e m e n t a n d A c c o u n t in g 3: 39-60. Bezirganian, Steve D. 19 93. “U .S. Affiliates of Foreign Companies: Opera tions in 1 9 9 1 ,” S u r v e y of C u r r e n t B usin ess 735 (May): 8 9 -112 . FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 72 N otes R e fe re n c e s (Continued) Caves, Richard E., and Matthew B. Krepps. 1993- “Fat: The Displacement of Nonproduction Workers from U.S. Manufacturing Industries.” B r o o k in g s P ape rs o n E c o n o m ic A c t iv it y : M icro eco n o m ics 2: 227-88. Graham, Edward M., and Paul R. Krugman. 1991- FOREIGN D irect In ve st m en t in the U n ite d S tates , 2d ed. Washington, D .C .: Institute for International Economics. Caves, Richard E. 1 9 8 9 . “E x ch an g e-R a te M o ve m e n ts and F o reig n D ire c t Grubert, Harry. 19 93. “Taxes an d F o re ig n -C o n tro lle d C o m p a n ies in th e In v e stm e n t in th e U n ite d S ta te s.” In D .B . A u d re tsc h an d M .P. C lau d on , U n ite d S ta te s.” In Fo r e ig n D irect I n vestm en t in the U n ited S tates : ed s., T h e I n t e r n a t i o n a l i z a t i o n of U .S . M a r k e t s . N e w Y o rk : N e w A n U pd at e . W a sh in g to n , D .C .: U .S. D e p a rtm e n t o f C o m m erce. Y o rk U n iv e rs ity Press. Grubert, Harry, Timothy Goodspeed, and Deborah Swenson. 1993. “E x p lain in g Cebenoyan, A. Sinan. GeorgeJ. Papaioannou, andNickolaos G. Travlos. 1992. th e L ow T axable In co m e o f F o re ig n -C o n tro lle d C o m p a n ies in the U n it “Foreign Takeover Activity in the U.S. and Wealth Effects for Target ed S ta te s .” In A lb e r to G io v a n n in i, R . G le n n H u b b ard , and J o e l S le m - Firm Shareholders.” F in a n c ia l M a n a g e m e n t 58-68. rod , eds., S tudies IN I n te r n a tio n a l T a x a t io n . C h icag o : U n iv e rs ity o f C h ica g o Press. Clinton, Bill. 1992. “Putting People First: A National Economic Strategy Harris, RobertS., and David Ravenscraft. 19 9 1. “The Role of Acquisitions in for America.” Foreign Direct Investment: Evidence from the U.S. Stock M arket.” Conn, R.L., and E Connell. 1990. “In te rn a tio n a l M ergers: R e tu rn s to U.S. J o u r n a l of F in an ce 46.3 (Ju ly): 825-44. an d B r itis h F ir m s .” J o u r n a l o f B u sin e ss F in a n c e & A c c o u n t in g Healy, Paul M., Krishna G. Palepu, and RichardS. Ruback. 1992. “Does Cor 1 7 :6 8 9 -7 1 1 . porate Performance Improve After Mergers?" J o u r n a l Dewenter, Kathryn L. 1992a. “Do Exchange Rate Changes Drive Foreign of F in a n c ia l E co n o m ic s 3 1 :1 3 5 -7 5 . Direct Investment?” University o f Chicago. Herr, Ellen M. 1988. “U.S. Business Enterprises Acquired or Established by _______. 1992b. “Do Foreign Investors Pay More than Domestic Investors for U.S. Acquisition Targets?” University o f Chicago. Foreign Direct Investors in 1 9 8 7 .” S u r v e y of C u r re n t B usiness 68.5 (May): 50-75. Encarnation, Dennis J . 1992. R ivals B e y o n d T r a d e . New York: Cornell University Press. Hobbs, James R. 1993. “Domestic Corporations Controlled by Foreign Per sons, 1 9 9 0 . ” I n t e r n a l R e v e n u e S e r v ic e S t a t is t ic s of In co m e B ulletin , 13.2 (Fall): 12 5 -4 1. Fahim-Nader, Mahnaz. 19 9 4 . “U.S. Business Enterprises Acquired or Established by Foreign Direct Investors in 19 9 3 .” S u r v e y of C urrent B usiness 74.5 (May): 50 -6 1. Hood, Neil, and Stephen Young. 19 86. “F o reign D irect In v e stm e n t in th e U .S . A u to m o b ile In d u s try .” In H . P e te r G ra y , e d ., U ncle S a m A s H o st , R esea rc h in I n t e r n a t io n a l B usin ess a n d F in a n c e , v o l. 5. G re e n Franks, Julian, Robert Harris, and Sheridan Titman. 1991- “The Postmerger Share-Price Performance of Acquiring Firms.” J o u r n a l of w ic h , C o n n e cticu t: S A I Press. F in a n c ia l E co n o m ic s 2 9 :8 1 -9 6 . Hooper, Peter, and Catherine Mann. 1989- “Exchange Rate Passthrough in the 1980s: The Case o f U.S. Imports of Manufactures.” B r o o k in g s Franks, Julian, and Robert S. Harris. 1989- “Shareholder Wealth Effects of Corporate Takeovers: The U.K. Experience 19 5 5 -1 9 8 5 .” J o u r n a l Pa pe r s in Ec o n o m ic A c t iv it y 1: 297-329. of F in a n c ia l Ec o n o m ic s 23: 225-50. Howenstine, Ned G. 19 87. “U .S . Affiliates o f Foreign Companies: Opera tions in 1 9 8 5 .” S u r v e y of C u r r e n t B usiness 67.5 (May): 36 -51. Froot, Kenneth, and P. Klemperer. 1989- “Exchange Rate Passthrough W hen Market Share Matters.” A m e r ic a n E c o n o m ic R e v ie w 79: 6 3 7 -5 1. Hufbauer, Gary Clyde, and Joanna M. Van Rooij. 1992. U.S. TAXATION OF I n te rn a tio n a l I n c o m e : B lu eprint for R e fo rm . Washington, D.C.: Giliberto, S. Michael. 1 9 9 4 . “The Inside Story on Rates o f R etu rn .” Institute for International Economics. J o u r n a l of R eal E state F in an ce 11 .1 (Spring): 51-54. N otes FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 73 R e fe re n c e s ( Continued) Kaplan, Steven N., andJeremy C. Stein. 19 93. “The Evolution o f Buyout and Financial Structure in the 1980s.” Q u ar te r ly J o u r n a l of Ec o n o m ic s Long, William E, and D avidJ. Ravenscraft. 1993. “Decade of Debt: Lessons from LBOs in the 19 80s.” In Margaret M. Blair, ed., T he D eal D ec a d e . Washington, D .C .: Brookings Institution. (May): 313-57 . Kaplan, Steven N., and Michael S. Weisbach. 1992. “The Success o f Acquisi tions: Evidence from Divestitures.” J o u r n a l OF F in an ce 47: 107-38. Lupo, L.A., Arnold Gilbert, and Michael Liliestedt. 1978. “T h e R e la tio n s h ip B etw een Age and R ate o f R e tu rn o f Foreign M a n u fa c tu rin g Affdiates o f U.S. M a n u fa c tu rin g P aren t C o m p a n ie s.” S u r v e y Kang. Jun-Koo. 19 93. “The International Market For Corporate Control: of C u r r e n t B usiness 58.8 (A u g u st): 60-66. Mergers and Acquisitions of U.S. Firms By Japanese Firms.” JOURNAL of F in a n c ia l E co n o m ic s (December): 34 5-71. Lutzy, John, and Randy Miller. 19 92. “C o n tro lle d F o reig n C o rp o ra tio n s, 19 8 8 .” In te rn a l R even u e S ervice S tatistics Kapler, Janis. 1994. “Measuring Real Economic Rates o f Return: TNCs vs. of I n co m e B ulletin 12:2 (Fall): 60-87. Domestic Firms.” Paper presented at the Eastern Economic Association Meeting, March 1994. Magenheim, Ellen B., and Dennis C. Mueller. 19 88. “Are Acquiring-Firm Shareholders Better O ff After An Acquisition?” In John Coffee Jr., Louis Kenneth Leventhal & Company. 19 93. 1993 J apanese D isin vestm en t in U.S. Lowenstein, and Susan Rose-Ackerman, eds., K n ig h t s , R a id e r s AND T a r g e t s . New York: Oxford University Press. R eal Estate . Klein, Michael W., and Eric Rosengren. 1994. “The Real Exchange Rate and Foreign Direct Investment in the United States: Relative Wealth vs. Marston, Richard. 1990. “Pricing to Market in Japanese Manufacturing.” J o u r n a l of I n te r n a t io n a l E co n o m ic s 29: 2 17 -3 6 . Relative Wage Effects." JOURNAL OF INTERNATIONAL ECONOMICS 36: McCauley, Robert N., and Dan P. Eldridge. 19 9 0 . “The British Invasion: 373-389- Explaining the Strength of U.K. Acquisitions o f U.S. Corporations.” In Knetter. Michael M. 1994. “E xch an ge R ates and C o rp o ra te P ric in g S tr a te g i e s .” In Yakov A m ih u d an d R ic h a r d M . L e v ic h , e d s ., E x c h a n g e I n te r n a tio n a l P rivate C a pit a l F l o w s . Basle: Bank for International Settlements. R ates a n d C o r p o r a t e P e r f o r m a n c e . N e w York: Irw in P rofessio n al McCauley, Robert N., and Steven A. Zimmer. 19 94. “E xch an g e Rates and P u b lish in g . In tern a tio n a l D ifferen ces in th e C o st o f C a p ita l.” In Y a k o v A m ih u d and Kogut, Bruce, a n d Sea J i n Chang. 19 9 1 . “Technological Capabilities and Japanese Foreign Direct Investment in the United States.” REVIEW OF Richard M. Levich, ed s., EXCHANGE RATES AND CORPORATE PERFOR MANCE. N ew Y o rk : Irw in P rofession al P u b lish in g . E c o n o m ic s a n d S tatistics 7 3 :4 0 1 - 1 3 . _______ . 1989- “Explaining International Differences in the Cost o f Krugman, Paul. 1 9 8 7 . “P r ic in g to M a r k e t W hen the Exchange Rate Changes.” In S v e n W. A r n d t an d J . D a v id R ic h a rd s o n , e d s., R e a l - C a p ita l.” F ederal R eserve B a n k of N e w Y o r k Q u ar te r l y R e vie w 14.2 (Su m m er): 7-28. F in a n c ia l L in k a g e s A m o n g O pen E c o n o m ie s . Cambridge, M ass.: Merrill Lynch Business Services. M e rgerstat R e v ie w . 1992. MIT Press. Landefeld, J . Steven, Ann M. Lawson, and Douglas B. Weinberg. 1992. “Rates o f Return on Direct Investment.” SURVEY OF CURRENT BUSINESS 72.8 Moody’s Investor Services. 19 93. M o o d y ’s I n d u st r ia l M a n u a l . N e w Y ork. M o o d y ’s. (August): 79-86. Okuro, Kenichi 1989- “Export Price Behavior o f Manufacturing: A U .S .Lipsey, Robert E. 1993. “Foreign Direct Investment in the United States: Changes Over Three Decades.” In Kenneth A. Froot, ed., FOREIGN Japan Comparison.” I n te r n a tio n a l M o n e t a ry F u n d S taff P ape rs 36: 550-79. D irect I n vestm en t . Chicago: University o f Chicago Press. Organization for Economic Cooperation and Development. 1 9 9 1 - T a x i n g _______. 19 9 1. “Foreign Direct Investment in the United States and U.S. Trade.” A n n als of the A m e r ic a n A c a d e m y of P olitical a n d S o c ial P rofits in a G lo b al E c o n o m y : D om estic a n d I n t e r n a tio n a l Issu e s . Paris. S cience 5 1 6 :7 6 -9 0 . F R B N Y Q u a r t e r l y R e v ie w / S u m m e r-F a ll 1994 74 N otes R e fe re n c e s (Continued') Quijano, Alicia M. 1 9 9 0 . “A Guide to BEA Statistics on Foreign Direct Investment in the United States.” S u r v e y of C ur r e n t B usiness 7 0 .2 (February): 2 9 - 3 7 . U.S. Congress. House Committee on Ways and Means. Subcommittee on Oversight. 1 9 9 0 . T a x U n d erpaym en ts b y U .S . S u b sid iar ie s of F o r e ig n C o m pa n ies : H e a r in g s , 1 0 1 s t C o n gress, 2 d Sess. Policy Economics Group, KPMG Peat Marwick. 1 9 9 4 . Review o f Internal U.S. Congress. Joint Committee on Taxation. 1 9 8 9 - F ed eral I n co m e T a x Revenue Service Statistics on Foreign Controlled Corporations, 1 9 8 6 A spects of C o r po r ate F in a n c ia l S t ru c t u r e s ( J C S - 1 - 8 9 ) , Ja n u a ry 18 . through 1990. U.S. Department of Commerce. Bureau of Economic Analysis. 1 9 8 4 - 9 4 . FOREIGN Porter, Michael E. 1 9 8 8 . “From Competitive Advantage to Corporate Strat egy.” H a r v a r d B usiness R e v ie w 4 3 - 5 9 - D irect I nvestm en t in the U n ited S tates : O peration s of U .S. A ffili ates of Rangan, Subramanian, and Robert Laurence. 1 9 9 3 . “The Responses o f U.S. F o r e ig n C o m pan ies : R evised 1 9 7 7 - 9 1 Estim ates . U.S. Department of Commerce. Bureau of Economic Analysis. 1 9 8 9 - “Foreign Firms to Exchange Rate Fluctuations: Piercing the Corporate V eil.” Direct Investment in the United States: U.S. Business Enterprises B r o o k in g s P ape rs o n Ec o n o m ic A c t iv it y 2 : 3 4 1 - 6 9 . Acquired or Established by Foreign Direct Investors 1 9 8 0 - 1 9 8 6 . ” Unpublished paper. Ravenscraft, D avidJ., and P.M. Scherer. 1 9 8 7 . M e r g e r s , S ell-O ffs , an d Ec o n o m ic E fficiency . Washington, D.C.: The Brookings Institution. U.S. Department of Commerce. Bureau of Economic Analysis. 1 9 9 2 . “F o reign D ire c t In v estm en t in th e U n ite d States: D e ta il fo r H isto ric a l-C o st Posi Remolona, Eli M. 1 9 9 0 . “Understanding International Differences in Lever age Trends.” Federal Reserve Bank of New York Q u ar t e r ly R e v ie w tio n a n d B a la n c e o f P a y m e n ts F lo w s , 1 9 9 1 . ” S u r v e y of C u r r e n t B usiness 7 2 . 8 (A u g u st): 8 7 - 1 4 4 . 1 5 . 1 (Spring): 3 1 - 4 2 . U.S. Department of Commerce. Economics and Statistics Administration. 1 9 9 3 . Scholl, Russell B. 1 9 9 4 . “The International Investment Position of the Unit ed States in 1 9 9 3 . ” S u r v e y of C u r r e n t B usiness 7 4 . 6 (June): 6 3 - 7 1 . Slemrod,Joel. 1 9 9 0 . “Tax Effects on Foreign Direct Investment in the Unit ed States: Evidence from a Cross-Country Comparison.” In Assaf Tazin F o r e ig n D ir e c t I n v e st m e n t in the U n it e d S t a t e s : A n U p d a t e . Washington, D.C.: Department of Commerce. U.S. Department of the Treasury and Internal Revenue Service. 1 9 8 8 . “A S tu d y o f In terco m p an y P ric in g .” S t a n d a r d F ederal T a x R eports 7 5 . and Joel Slemrod, eds., TAXATION IN THE GLOBAL ECONOMY. Chicago: University of Chicago Press. Weinberg, Douglas B. 1 9 9 4 . “U .S . In te rn a tio n a l T ran saction s, F irst Q u a rte r 1 9 9 4 . ” S u r v e y of C u r re n t B usiness (Ju ne): 8 6 - 1 2 7 . Stewart, James B. 1 9 9 4 . “Sony’s Bad Dream.” N e w Y o r k e r , February 2 8 . Zeile, William J . 1 9 9 4 . “F o reig n D ire c t In v e stm e n t in th e U n ite d States: Swenson, Deborah L. 1993- “Foreign Mergers and Acquisitions in the Unit ed States.” In Kenneth A. Froot, ed., F o r e ig n D ir ec t I n v e st m e n t . 1 9 9 2 B e n c h m a r k S u r v e y . ” S u r v e y o f C u r r e n t B u s in e s s ( J u ly ) : 15 4 -8 6 . Chicago: University o f Chicago Press. Digitized N otes for FRASER FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 75 The Baby Boom Generation and Aggregate Savings Richard Cantor and Andrew Yuengert uring the 1970s, household and national sav should rise with age, the impact on the aggregate saving rate ings rates fe ll sharply. A lth o u g h concern w ill be mitigated by a continuing high rate o f early retire about low saving rates remains widespread, ment and a rising share o f households headed by individuals many analysts have taken the optimistic po either over sixty-five or under thirty-five— households that D sition that the maturing o f the baby boom generation w ill restore aggregate savings to their earlier levels. Two argu ments have been offered in support o f this view: tend to have relatively low saving rates. The second argument, which stems from the popu lar notion o f baby boomer improvidence, is also flawed. Baby • As baby boomers reach their peak earning and boomers appear to have accumulated appreciable w ealth, saving years, aggregate savings w ill increase with both in comparison to their parents at similar ages and to tar this large generation’s rising share o f national get retirement wealth levels suggested by theoretical models income. o f optimal lifetime savings. Moreover, even if baby boomers • Baby boomers who spent freely and saved little as eventually find themselves behind schedule in saving for young adults w ill in their middle years find them their retirem ent (if not because o f their shortsightedness, selves unprepared for retirem ent. They w ill re then because o f a large cut in Social Security benefits), they spond by increasing their saving rates more ag may react to the shortfall by reducing planned bequests or gressively than earlier generations at a similar age. 1 taking other steps that do not have the effect o f increasing This article evaluates each argum ent in turn and aggregate saving rates. The uncertainty o f the baby boom finds little reason to expect a large increase in aggregate sav response to any savings shortfall gives us even less reason to ings. The demographic effects alleged in the first argument expect a surge in aggregate savings as the baby boomers w ill probably be small. W h ile baby boomers’ saving rates approach retirement. Digitized76for FRASER FRBNY Q u a r t e r l y R e v i e w /S u m m e r -F a l l 1994 L if e - C y c l e S a v i n g s , D e m o g r a p h i c incomes and saving rates tend to rise with the age of the T r e n d s, a n d A g g r e g a t e S a v in g s household head before retirement. The 1983 survey contains extensive data on the wealth of individual households as of If baby boomers raise their saving rates in line with increases observed in earlier generations at similar ages, will the aggre gate saving rate recover sharply? To answer this question, we year-end 1982. In 1986, 2822 households from the original group were surveyed again and asked for detailed information review the age profile of income and savings found in recent on year-end 1985 wealth and on annual income in 1983-85. consumer surveys and the demographic trends projected for The two Board surveys are particularly useful in analyzing the next few decades. average savings because they “oversample” the high-income households that account for a disproportionately large share A g e P r o f il e s o f I n c o m e a n d S a v i n g s of aggregate savings. A detailed explanation of our use of the Data from the Board of Governors’ Survey of Consumer survey data to calculate saving rates is contained in the box Finances allow us to examine the proposition that household below. C a l c u l a t in g S a v in g R a t e s f r o m th e S u r v e y o f C o n s u m e r F in a n c e s We calculate average annual income (in purchases of new homes. Capital gains on other property (sec 1982 dollars) from total household income data reported in ondary residences, farmland, undeveloped land, and invest the 1986 Survey of Consumer Finances for the years 1983, ment real estate) equal the 1982 value of the property multi 1984, and 1985. We also include in income several other plied by a price index for the type of property. Capital gains items reported separately in the survey, such as insurance set on direct holdings of stock and stock m utual funds are tlements and cash advances. assumed to equal the increase in the S&P 500 index over the A ve r a g e In c o m e . period multiplied by initial holdings. Because many house Total savings equals the total change holds hold IRAs, trusts, and thrift plan assets in stocks, we in wealth (in 1982 dollars) between year-end 1985 (drawn calculate capital gains on these assets as the gains in the S&P from the 1986 survey) and year-end 1982 (drawn from the 500 times the holdings of trusts, IRAs, and thrift plan assets 1983 survey), including capital gains but excluding “wind falls.” Windfalls account for roughly one-seventh of all sav times the aggregate share of these assets invested in stocks. ings in the sample and consist of net inheritances (inheri S a v in g R ates . tances received minus inheritances given by a deceased sonal saving rates. The total saving rate represents the average spouse), net support from other family members (received annual change in total wealth, including capital gains, minus given), and several other items. Like other researchers between year-end 1982 and year-end 1985, divided by aver in this area, we exclude windfalls in order to focus on planned age annual income. The personal saving rate is the average savings behavior. We also exclude changes in defined benefit change in wealth excluding capital gains, divided by income. C h an g e s in W ealth . We calculate both total saving rates and per pension wealth because they are not reported in the survey. E x c l u d e d O b se r v a t io n s . Even in carefully constructed The data permit the calculation of capital household surveys, wealth and savings are inaccurately mea gains on housing because they include the value of housing sured, and some household saving rates will be unreasonably wealth at year-end 1982 and year-end 1985 and sales and high or low. We therefore exclude from our sample certain C apital G a in s . FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 77 obvious outliers—households with particularly large positive or negative saving rates. The exclusion rule drops the bottom fer significantly from ours. Many of the differences are and top 0.5 period of the distribution of saving rates. House and the exclusion rules. Bosworth et al. exclude capital gains holds that experienced a change in marital status between the on housing, directly held stocks, and stock mutual funds, 1983 and 1986 surveys are also excluded. The final sample but not capital gains on other real estate IRAs, trusts, and contains 2494 households. thrift plans. Instead of excluding both large and small saving attributable to differences in the definition of capital gains rate outliers, as we do, they exclude only observations with Bosworth, Burt- large saving rates, greater than 1. They also exclude house less, and Sabelhaus (1991) calculated age-specific personal holds with significant net worth in unincorporated business saving rates from the Survey of Consumer Finances that dif es, which we do not. D if f e r e n c e s f r o m P r e v io u s R e s e a r c h . For the households sampled in the 1986 Board sur 1985, present a similar age profile (Chart 2). This measure of vey, average (mean) income rose, then fell, with advancing savings includes capital gains on real estate and all financial age (Chart 1). The highest average income, about $35,000 assets except employer contributions to pension plans, which per year in 1982 dollars, was earned by both the 35 to 44 and are not reported in the survey. Total savings rose more steeply 45 to 54 age groups. The distribution of incomes around this with age than did income, with the peak occurring at $8,200 peak is roughly symmetric, falling below $20,000 for house in the 55 to 64 age range. Savings in these years are large holds with heads either below 25 or above 75. The shape of because income is relatively high, family expenses are rela- this distribution is determined by variations in wage rates and labor force participation rates over the life cycle.2 Average total savings, defined as the average annual change in total wealth between year-end 1982 and year-end Chart 2 A v e r a g e A n n u a l T o t a l S a v in g s , b y A g e o f H o u s e h o l d H e a d 19 8 3 -8 5 T h o u san d s o f 19 8 2 dollars Chart 1 A v e r a g e A n n u a l H o u s e h o ld In c o m e , b y A g e o f H o u s e h o l d H e a d 1 9 8 3 -8 5 T h o u san d s of 1982 dollars Under 25 Under 25 25-34 35-44 45-5 4 55-64 65-74 75 and above Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; Federal Reserve Bank of New York staff estimates. Digitized FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 78for FRASER 25-34 35-44 45-54 55-64 65-74 75 and above Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; Federal Reserve Bank of New York staff estimates. Note: "Total savings" represents the change in total financial and real estate wealth, including capital gains. tively low, and the need for retirement savings is immediate ly apparent for most workers. because of the large expenses of child rearing and household formation. In addition to total savings, we are also concerned The age profile of saving rates o f course depends on with the age profile of households’ personal savings, a narrower conception of savings that excludes capital gains. Since capi the changes in both savings and income over the life cycle. tal gains are often passively earned, unplanned, and illiquid, Chart 4 depicts the average total saving rates by age. Before age 45, total saving rates are flat, around 12 percent, because households may treat them differently than other forms of income rises at about the same rate as savings until age 45. savings. In addition, capital gains income has been volatile in Thereafter, saving rates rise sharply to 20 percent between the past, and hence difficult to project accurately. Since, at ages 45 and 54 and peak at 31 percent between ages 55 and the aggregate level, capital gains do not lead directly to 64. The total saving rate declines only slightly after 65: increases in national investment, policymakers are generally although savings levels fall rapidly, income falls almost as more concerned about aggregate personal savings exclusive of quickly. Compared with the age profile of total saving rates, capital gains (Harris and Steindel 1991). Personal savings reach their highest level between the age profile of personal savings is more peaked (Chart 5). the ages of 45 and 54 (Chart 3). The peak occurs earlier than Between ages 45 and 64, personal saving rates average about the peak in total savings because capital gains are quite large 10 percent. Personal saving rates are very low for households for households with heads aged 55 to 64. Personal savings are extremely low, near zero, between ages 35 and 44, perhaps Chart 4 Chart 3 A v e r a g e T o t a l S a v in g R a te , b y A g e o f H o u s e h o ld H e a d 19 8 3 -8 5 A v e r a g e A n n u a l P e r s o n a l S a v in g s , b y A g e o f H o u s e h o ld H e a d 19 8 3 -8 5 Percent T h o u san d s o f 19 82 35 d ollars --------------------------------------------------------------------- 4 30 25 15 5 -2 Under 25 25-34 35-44 45-54 55-64 65-74 75 and above Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; Federal Reserve Bank of New York staff estimates. Note: "Personal savings" represents the change in total financial and real estate wealth less capital gains. 0 -i I I I I I I I I I I I I I I :— Under 25 25-34 i 1 1 1I B 35-44 45-5 4 I 1H1W81 I HHW IWBIP 55-64 65-74 75 and above Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; Federal Reserve Bank of New York staff estimates. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 79 headed by individuals between ages 35 and 44, and negative The actual steepness of mean life cycle saving rates is for those whose head is between ages 65 and 74. So far, our analysis has focused on means within age somewhat obscured in Chart 5 by the presence of early groups. In keeping with our interest in aggregate savings, the our sample, the age-savings profile of nonretirees rises more mean gives appropriate weight to high-income households. steeply with age and remains higher longer than the age pro A concern about the saving behavior of a typical household file for the full sample (Chart 7). This finding need not imply, would, in contrast, require a focus on median savings. More however, that early retirement reduces aggregate savings, over, popular perceptions about life cycle saving rates are because those that retire early may save more than others in probably based on observations of “typical” rather than the years preceding their retirement. (The age-savings pat “mean” household behavior. We therefore present, in Chart 6, terns in Chart 7 cannot be interpreted as the saving rate of a the median personal saving rates by age, also drawn from the household whose head is expected to retire at age 65, since survey sample. many of the households in the sample may be planning to A comparison of Charts 5 and 6 reveals that (1) retirees in age groups under 65. If we exclude retirees from retire earlier or later than 65.) median saving rates are much lower than mean rates, or alter natively, that high-income households account for a dispro D e m o g r a p h ic T r e n d s portionate share of aggregate savings; and (2) the age profile Before projecting the effect of baby boomer savings on the of the medians is less peaked than the age profile of mean aggregate saving rate, we must first assess the size of the baby saving rates. The absence of a strong life cycle pattern in median savings may be explained by the relatively modest amount of financial savings accumulated by most house holds and their near total reliance on pensions and Social Security in retirement. Chart 6 M e d ia n P e r s o n a l S a v in g R a t e , b y A g e o f H o u s e h o ld H e a d 19 8 3 -8 5 Percent Chart 5 A v e r a g e P e r s o n a l S a v in g R a te , b y A g e o f H o u s e h o ld H e a d 19 8 3 -8 5 Percent II 1 I ■ I 1 Under 25 25-34 35-44 45-5 4 55-64 65-74 ■■■■ i i i r: iI|z _ ................................... 1 1 75 and above Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; Federal Reserve Bank of New York staff estimates. Digitized F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 80 for FRASER i i Under 25 i 25-34 i 35-44 i 4 5 -54 i 55-64 i 65-74 i 75 and above Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; Federal Reserve Bank of New York staff estimates. boom demographic bulge and consider how its relative size can be expected to change over time. Chart 8 presents Census data on the age distribution of household heads for 1970 and 1990 and a forecast of the age distribution in 2010.3 That part of the age distribution represented by baby boomers is Chart 8 D ist r ib u t io n o f H o u se h o l d s b y A g e o f H o u se h o l d H e a d 1970, 1990, and Projection for 2 0 10 M i l l io n s 25 19 7 0 shaded in the chart. For simplicity, we have defined baby boomers as the generation born in the twenty-year period Baby-boom generation Other generations 20 after World War II, between 1945 and 1964. (The baby boom population is normally defined as individuals born 15 between 1946 and 1964.) In 1970, all baby boomers were less than 25 years old and headed only 7.6 percent of all households. By 1990, 10 Bem however, when baby boomers reached the ages of 25 to 44, . they headed 41 million households, roughly 44 percent of all households. In that year, baby boomers headed 50 percent more households than did members of the preceding genera tion (individuals between the ages of 25 and 44 in 1970). By _ 1 M Hi ill Hi i l |1| wm i|B : iililil IB B b HHI 1981 ■81 | | f* l____,.i ___i _____ i ___ 19 9 0 the year 2010, when baby boomers will be between 45 and 64 Chart 7 A v e r a g e P e r s o n a l S a v in g R a te , E x c lu d in g R e tir e e s , b y A g e o f H o u s e h o ld H e a d 19 8 3 -8 5 P ercent -10 Under 25 25-34 35-44 45-54 55-64 65-74 75 and above Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; Federal Reserve Bank of New York staff estimates. Under 25 25-34 35-44 45-54 55-64 65-74 75 and above Source: U.S. Bureau of the Census, Current Population Reports. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 81 years old, they will head about 46 million households, or roughly 39 percent of all households. The decline in the relative size of the baby boom share and its relative income (shown in Chart 1). The house hold population shares are assumed to change over time in keeping with the Census projections. Larger groups and generation between 1990 and 2010 from 44 percent to 39 groups with higher average household incomes have larger percent of all households will diminish its potential impact weights. (For a more exact description of the weighting pro on the aggregate saving rate. This decline in relative size is cedure, see Kennickell 1990.) due in part to the growing importance of two demographic The projected behavior of the broad saving rate mea groups whose low saving rates will offset the baby boom’s sure, inclusive of capital gains, is depicted in Chart 9- The increased savings. The generation following the baby demographic trends suggest that the total saving rate should boomers is expected to head 39 million households in 2010, have reached its low point at about 18.0 percent in 1990, will only 2 million fewer households than the baby boomers head rise to 19-4 percent in 2010, and peak at 19-8 percent in ed at the same ages. In addition, the share of elderly house 2015. These projected increases are expected to be quite holds in the population will increase significantly. gradual, with the first 140 basis point increase extending over twenty years and the next 40 basis point rise occurring T h e P r o je c t e d Im p a c t o f D e m o g r a p h ic T r e n d s on A g g r e g a t e S a v in g s over the following five years. The predicted behavior of the more narrowly defined Assuming that the age patterns in income and savings from personal saving rate, which excludes capital gains, is shown the Survey of Consumer Finances remain stable over time, we in Chart 10. Here demographic changes imply that the per can use the Census Bureau population forecasts to project the sonal saving rate should have reached its low point at about aggregate saving rate as the baby boom generation matures. 4.1 percent in 1990 and should rise about 80 basis points to Our aggregate saving rate projections are derived by averag peak at 4.9 percent near the year 2010. During the twenty ing household saving rates (shown in Charts 4 and 5), years between 2010 and 2030, the aggregate personal saving weighting each age group’s saving rate by its population rate should decline by about 100 basis points because the Chart 9 P r o je c t e d T o t a l S a v in g R a t e P ercent Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; U.S. Bureau of the Census, Current Population Reports; Federal Reserve Bank of New York staff estimates. FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 82 share of retirees in the population w ill be rising sharply. S e n s it iv it y A n a l y s is o f t h e A g g r e g a t e S a v in g s (Recall from Charts 4 and 5 that households headed by indi P r o je c t io n s viduals over 65 had low or even negative personal saving rates at the same time that capital gains kept this group’s total sav Our projections for aggregate savings are, of course, sensitive to our assumptions about the future age profiles of the popu ing rates fairly high.) lation, relative incomes, and saving rates. Other reasonable In conclusion, if the life-cycle increase in the saving rates of the baby boom generation is comparable to the increases for previous generations, the rise in aggregate sav ing rates over the next twenty-five years will be modest: 180 basis points (from 18.0 percent to 19-8 percent) in total sav ing rates and 80 basis points (from 4.1 percent to 4.9 per cent) in personal saving rates. These projected changes in the aggregate saving rates are not particularly large relative to previous fluctuations, reflecting the lim ited predictive I f the life-cycle increase in the saving rates o f the baby boom generation is comparable to the increases fo r previous generations, the rise in aggregate saving rates over the next twenty-five years w ill be modest. power of demographic models of savings. From the early 1980s to the present, the aggregate personal saving rate fell roughly 300 basis points, as compared with the 80 basis assumptions, however, do not reverse our basic finding that point improvement by 2010 implied by our projections. demographic forces are unlikely to boost the aggregate sav W hile the broader total saving rate measure of savings fell ing rate substantially. less sharply than the personal saving rate, the anticipated For example, if we shift the assumptions about pop improvement in savings attrib utab le to demographic ulation growth from the Census Bureau’s so-called middle changes cannot be counted on to produce a full recovery in projection to its high or low projections, we do not get signif either of these measures. icantly different results. Using the high population growth Chart 10 P r o je c t e d P e r s o n a l S a v in g R a t e Percent Sources: Board of Governors of the Federal Reserve System, Survey of Consumer Finances, 1983 and 1986; U.S. Bureau of the Census, Current Population Reports:; Federal Reserve Bank of New York staff estimates. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 83 projections decreases the projected 2010 aggregate personal saving rate by 4 basis points; using the low growth projection raises it by 10 basis points. Finances saving profile. Our main conclusion stands up to this sensitivity analysis: aggregate saving rates are not likely to be pushed up W e also tested the sensitivity of our projections to sharply by demographics if future age profiles o f saving rates changes in the ratio o f population to households (household are similar to those observed in the recent past. This is not to composition), which we assumed to be fixed from 1992 for say, however, that the aggregate saving rate would be insensi ward. If household size falls in line w ith various Census pro tive to an unexpectedly large increase in baby boomer saving jections, our projections o f aggregate personal savings change rates. If the age profile o f boomer saving rates turns out quite by no more than 2 or 3 basis points. different from what we’ve projected, the effect on aggregate Moreover, our assumptions about the age-income savings could be large. W e have already noted that in the year profile are, if anything, biased in favor o f finding a large rise 2 0 1 0 , 39 percent o f households w ill be headed by baby in aggregate savings over the next twenty years. W e may have boomers. Those households w ill be in their peak earning overestimated the baby boom generations’s impact on sav years and w ill account for 4 4 percent o f aggregate income. If ings by overestim ating its share o f aggregate income in the baby boomers’ personal saving rate in middle age exceeds future years. W e have assumed that boomers between ages 45 that o f previous generations by 2 percentage points (12 per and 6 4 w ill earn the same amount in relation to other groups cent, rather than 10 percent), the aggregate saving rate pro as did 45 to 64 year olds in the Survey o f Consumer Finances. jection would increase another 90 basis points beyond the 80 Because baby boomers are an unusually large labor cohort, basis point increase we are already projecting through 20 10 . their wages may be lower relative to other generations at each point in the life cycle. W elch (19 7 9 ) has documented that T he A lleged R etirem en t S a v in g s baby boomer entry level wages were 10 percent lower than S hortfall a n d A g g r e g a t e S a v in g s entry level wages for other cohorts, and Berger (198 5) claims Is there any reason to expect that the life-cycle increase in that boomer wages have remained relatively depressed. If baby boomer saving rates w ill be more pronounced than baby boomer incomes were assumed to be 10 percent lower those o f earlier generations? In the view o f some, the baby than the incomes reported in Chart 1, then our projected 80 boomers differ from earlier generations in the level o f their basis point increase in the aggregate personal saving rate retirement savings. The popular image in the 19 80s o f free- from 1990 to 2010 would be lower by about 20 basis points. spending baby boomers (“yuppies”), coinciding as it did w ith The most obvious risk in our forecast comes from a sharp fall in the aggregate personal saving rate, left the our assum ptions about the saving rates o f d ifferen t age impression that baby boomers do not save enough. If baby groups. Nevertheless, other studies, using other household boomers are not saving adequately for retirement, they may surveys and covering other tim e periods, have generally eventually try to make up the savings shortfall by raising found less steeply sloped age-saving profiles, also implying their saving rates more sharply than did previous generations. weak demographic effects on the aggregate saving rate. (See, In this section, we examine the available empirical evidence for example, Summers and Carroll 19 8 7 , Auerbach and Kot- on the adequacy o f baby boomer savings. W e concentrate on likoff 19 8 9 , Bosworth, Burtless, and Sabelhaus 1 9 9 1 , Ken- the older h a lf o f the baby boom (ages 35 to 4 4 in 1 9 8 9 ) nickell 19 9 0 , and Attanasio 19 93.) The most steeply sloped because younger baby boomers (ages 25 to 34 in 19 89) did age-saving profile among these studies is calculated from the not, as of 1989 , need significant retirement savings, so are not 19 7 2 -7 5 Consumer Expenditure Surveys by Bosworth, Burt yet interesting subjects o f study. less, and Sabelhaus (19 9 1). This profile, when used in our Exact statements about savings adequacy require projections, generates a 32 basis point rise in personal saving detailed information about the average household’s w illing rates from 1990 to 20 10 , compared w ith the 8 0 basis point ness to trade off current for future consumption (preferences increase projected using the 1 9 8 6 Survey o f Consum er for saving), the expected path o f future income, and expected FRBN Y 84 Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 future interest rates. Lacking this sort of data, analysts take tern is evident in the m edian com parisons (M anchester two approaches to the adequacy o f baby boomer savings. 19 9 3 ). From these rough measures, it appears that baby The first approach compares the accumulated sav ings o f the baby boomers to the savings o f previous genera boomers have managed to accumulate more wealth than their parents had at the same ages. tions at the same ages. The second approach uses estimates of These results may be misleading, however, because future earnings and interest rates, together w ith assumptions the 1963 and 198 9 surveys were successful to different about preferences for savings, to calculate “desirable” or degrees in measuring household wealth. Detailed compar “target” levels o f retirem ent savings for baby boomers at isons o f aggregate wealth as measured by the two surveys and every age. These savings targets are then compared w ith the the Federal Reserve’s Flow o f Funds Division suggest that actual savings o f baby boomers. W e find little compelling while aggregate wealth from the 1 9 8 9 Survey o f Consumer evidence o f retirement undersaving under either approach. Finances matched the Flow o f Funds aggregates closely, the 19 6 3 Survey o f Financial Characteristics o f Consumers cap C o m p a r iso n s w it h P r e vio u s G en eratio n s tured only 82 percent o f the Flow o f Funds aggregates .5 The The logic o f the first approach is simple: if the baby boomers Flow o f Funds data and the household surveys represent very have put aside as much as current retirees had at the same ages different methodologies o f wealth measurement; some dis and can expect similar levels of Social Security and pension parity in their findings is to be expected. Nevertheless, if the benefits in the future, then the baby boom generation is mak differences are due to mismeasurement in the household sur ing adequate provision for a comfortable retirement .4 veys, comparisons across the surveys may be flawed. O ur intergenerational comparisons focus on net The intergenerational wealth comparisons o f the w ealth relative to income because we need to adjust for two lines o f Table 1 are based on the extreme assumption that improvements in living standards and because we recognize all o f the differences between the household surveys and the that retirement savings are typically meant to replace a cer Flow o f Funds data are attributable to mismeasurement in tain fraction o f earned income. The first line o f Table 1 com the latter source. One way to check the sensitivity o f the pares the average wealth-to-income ratios o f baby boomers at cross-generational comparisons is to assume instead that all year-end 1988 w ith those o f their parents at year-end 1962 . differences between the two types o f survey stem from mis The second line compares medians. The ratios are calculated measurement in the household surveys and to benchmark the from the Board o f Governors’ 19 6 3 Survey o f Financial Char household surveys to the Flow o f Funds data. Benchmarking acteristics o f Consumers and the 19 8 9 Survey o f Consumer increases the 19 6 2 wealth measures by one-fourth. The 19 8 9 Finances. These survey measures o f wealth include housing household survey aggregates equal 97 percent o f the Flow of and most other assets except defined benefit and defined con Funds aggregates, so benchmarking increases 19 8 8 wealth tribution pension wealth. by 3 percent. In 19 8 8 , baby boomers held on average more assets Table 2 compares mean and median benchmarked than their life-cycle counterparts held in 19 62. The same pat- ratios across generations. By both measures, baby boomers have accumulated more assets than their parents did in 1962 , although the differences have shrunk from Table 1. Bench Table 1 S a v in g s o f the B a b y B o o m e r s a n d the P re ce d in g G en eratio n : E vid en ce fr o m U n a d ju st e d H o u seh o ld S u r v e y D ata marking the ratios thus leaves intact the finding o f higher wealth accumulation among baby boomers. Household Head Household Head 35 to 4 4 Years 35 to 4 4 Years Old in 19 6 2 O ld in 19 8 8 indicate that the baby boom generation is well prepared for Average wealth-to-income ratios 2 .0 7 2 .7 6 retirement. Unfortunately, simple comparisons o f nonpen Median wealth-to-income ratios 0 .8 7 1 .1 0 Sources: Survey o f Financial Characteristics o f Consumers, 19 6 3 ; Survey of Consumer Finances, 1 9 8 9 ; Federal Reserve Bank o f New York staff estimates. Cross-generational comparisons o f wealth, therefore, sion wealth ignore other factors that determine the need for saving: the baby boom generation faces challenges different F R B N Y Q uarterly R e v ie w /S u m m e r -F a l l 1994 85 from those its parents faced, challenges that may increase its baby boomer’s future earnings. Also built into the model are need for retirement savings relative to previous generations. assumptions about future changes in fam ily com position Three factors in particular point to an increased need and related expenses, the increased life expectancy o f baby for retirement savings for baby boomers. First, in spite o f an boomers, and the structure o f social security and pension im minent increase in the Social Security retirement age to benefits. 66 , baby boomers must save for a retirement that may last These technical assumptions and forecasts probably longer than that o f their parents. Second, the generation of affect the generated savings targets in important ways. More current retirees benefited from the inflation o f the 1970s, critical, however, are the authors’ assumptions about motives which sharply reduced its m ortgage liabilities. The baby for saving and their treatment o f sources o f retirement wealth boom generation cannot count on a similar windfall. Third, other than savings accumulated in financial assets. Bernheim the measures o f household wealth compared across genera and Scholz assume that people save for one reason only: con tions in Table 1 do not include the most important compo sum ption in retirem ent. B ut people also save to protect nents o f household retirem ent wealth: public and private pensions and Social Security. Although trends in these pro grams are uncertain at present, they may decrease the retire ment resources available to baby boomers. Cross-generational comparisons of wealth. . . indicate that the baby boom generation is well preparedfor retirement. B ern h eim a n d S c h o l z ’s S im u l at io n S tu d ies of S a v in g s A d e q u a c y In contrast to intergenerational comparisons o f wealth, the simulation studies o f Bernheim and Scholz (199 3) and Bern themselves against financial emergencies, to pay for their heim ( 19 9 2 , 19 9 3 , 19 94) attem pt to provide an absolute mea children’s education, or to leave their children an inheritance. sure o f savings adequacy that takes into account the particu Since we cannot separate observed savings into savings for la r econom ic c ircu m sta n c es faced by th e baby boom retirement and savings for other reasons, the targets are too generation. These researchers construct a model o f optimal low. In addition, the model assumes that baby boomers w ill savings behavior for baby boomers who would like to main receive no inheritances, and that they w ill not change hous tain after retirement roughly the same standard o f living they ing after they retire, using some portion o f home equity to enjoy just before their retirement. finance retirement consumption. If baby boomers expect to To calculate target retirem ent wealth at each age, receive inheritances and draw on their home equity, the Bern Bernheim and Scholz develop forecasts of future economic heim and Scholz targets are too high. Since these targets may conditions and assume a path for future interest rates. Con be either too low or too high, comparisons between the tar trolling for age and education, they forecast a representative gets and survey data may either under- or overstate retire ment preparedness. The simulation model generates target wealth (accu Table 2 mulated savings) for the head o f a household at every age. The S a v in g s o f t h e B a b y B o o m e r s a n d th e P r e c e d in g G e n e r a t io n : E v id e n c e f r o m H o u se h o l d S u r v e y D a t a B e n c h m a r k e d t o t h e F l o w o f F u n d s D ata change in target wealth as the head o f the household ages deter Household Head Household Head 35 to 4 4 Years 35 to 4 4 Years Old in 19 6 2 Old in 19 8 8 Average wealth-to-income ratios 2 .5 2 2.85 Median wealth-to-income ratios 1.0 6 1. 13 Q u a r t e r l y R e v ie w / S u m m e r - F a ll target savings for households earning more than $ 20,000 per year w ith the savings observed in M errill Lynch surveys o f such households. He concludes that in 1 9 9 1 and in 1 9 9 2 , Sources: Board o f Governors o f the Federal Reserve System, Flow o f Funds Accounts; Survey o f Financial Characteristics o f Consumers, 19 6 3 ; Survey of Consumer Finances, 19 8 9 ; Federal Reserve Bank o f New Y ork staff estimates. FRBN Y 86 mines annual target savings. Bernheim compares the annual 1994 baby boomers were saving at roughly 34 percent o f the rec ommended rate o f target (nonhousing) savings. Bernheim and Scholz obtain similar results by comparing saving rates that baby boomer saving rates were dangerously low in 1983 - from the 19 8 6 Survey of Consumer Finances w ith their target 85 and in 1 9 9 1 -9 2 , this generation’s accumulated wealth is rates, but they reveal that undersaving is concentrated among nevertheless in line w ith simulated retirement savings tar individuals w ithout college educations. gets. M oreover, high-incom e households, who are more Surprisingly, although these authors find that house holds’ saving rates fell far short o f the model’s savings targets im portant in the determ ination o f aggregate saving rates, seem to be putting aside savings on schedule. in 1983 - 85 , 1 9 9 1 , and 19 9 2 , their work also implies that boomers have done a reasonable job o f reaching their retire S o c ia l S e c u r it y a n d P e n sio n s ment wealth targets. In a 19 9 3 report, the authors compare Although we have so far kept to issues that are amenable to accum u lated w ea lth in the 1 9 8 6 S u rvey o f C onsum er empirical investigation, we recognize that aggregate savings Finances w ith target levels and find relatively high median may be substantially affected by other, more indefinite factors ratios, especially among college-educated households. Our such as the future o f the Social Security program, the impact own examination o f data from the most recent (198 9) Survey o f changes in pension coverage, and the various adjustments o f Consumer Finances reveals that the median ratio o f accu that baby boomers m ight make to retirement savings inade mulated wealth to the Bernheim-Scholz target level is 0.95 quacy. Baby boomers may receive lower Social Security bene for 35 to 4 4 year olds .6 That is, nearly h alf o f the sample fits than their parents as federal fiscal pressures force a choice report nonhousing assets greater than the authors’ targets. between increasing taxation and reducing benefits over the This discrepancy— that w ealth accum ulation ap pears on target while saving rates appear low— is not easy to next few decades. Moreover, the future generosity o f private and public pensions is uncertain. explain; however, we have somewhat greater confidence in Auerbach and K o tlik o ff (19 9 4 ) have analyzed the the wealth results. The inadequacy of saving rates was shown effects o f hypothetical Social Security restructuring on baby to hold only over a few years and hence may not be true for boomer wealth. They project that if social security benefits other years. Moreover, surveys o f changes in wealth (savings) are cut sharply in 2 0 0 9 , most baby boomers w ill face a stan probably suffer from proportionately more m easurement dard o f living worse than that o f the current generation of error than do surveys o f accumulated wealth. However, the results showing that the flow o f savings is too low, if true, would be quite troubling. If a shortfall in the flow o f savings This discrepancy— that wealth accumulation continues, it w ill inevitably lead to a shortfall in the stock of savings. appears on target while saving rates appear The Bernheim and Scholz approach is, however, not low— is not easy to explain; however, we have entirely appropriate for our purposes. Their analysis focuses on the retirem ent preparedness o f the median, or typical, somewhat greater confidence in the wealth results. household. Since our interest is in aggregate savings, the mean ratio is also o f interest. This ratio gives a greater weight to the savings o f high-income households, which account for a disproportionately large share o f aggregate savings. The retirees. To date, potential shortfalls in the Social Security mean ratio o f wealth to the Bernheim targets in 19 8 8 is 2.07, and Medicare Trust Funds have been remedied by increases in suggesting that those households whose savings have the taxation, w ith few cuts in benefits. Only the increase in the greatest im pact on aggregate savings are com fortably on normal retirement age, enacted in 198 3 , and the more recent track for retirement. taxation o f the Social Security benefits o f higher income In summary, although the sim ulation studies o f Bernheim (199 3) and Bernheim and Scholz (199 3) suggest recipients can be unambiguously construed as a decrease in benefits. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 87 Trends in public and private pensions are more diffi their retirement funds for any reason, would they rapidly step cult to evaluate. Although private pension coverage has increased over the last three decades, the nature of that cover up their saving rates in response to it? To answer this ques age has changed. In the last twenty years, the mix of plans has approaching retirement, discovers the inadequacy of his or changed: defined benefit coverage has been falling while her retirement provision. tion, one must examine the options facing a person who, defined contribution coverage has been rising (Chart 11). It is The first option is a large increase in saving rates. not yet clear which of these plans is the better vehicle for Such a response will, of course, boost aggregate saving rates as retirem ent savings, from the perspective of either the baby boomers approach retirement. The second option is a employer or the employee. W hile some evidence suggests large decline in consumption in retirement. This response that the most popular defined contribution plans stimulate will also lead to an increase in aggregate savings, as retiring savings (see, for example, Poterba, Venti, and Wise 1992), baby boomers dissave (eat into their retirement savings) at a there is growing concern that employees do not invest lower rate. (Note, however, that under this scenario, the defined contribution assets appropriately for retirement. increase in saving rates will come later, after baby boomers retire.) A third reaction is to postpone retirement and extend A d ju s t m e n t s t o S a v in g s S h o r t fa l l s working life. Both the income and savings of baby boomers Although there is no conclusive evidence of a baby boomer w ill rise as they work longer. A fourth possibility is an savings shortfall, the risks to the social security benefits of increase in net intergenerational transfers to retirees, either baby boomers cannot be ignored. The risks are admittedly from an increase in the generosity of Social Security benefits speculative—we do not know whether the government will or from a decrease in inheritances. In light of the looming move to cut benefits substantially. Nevertheless, the question shortfalls in Social Security, an increase in the generosity of arises, if baby boomers were to face a substantial shortfall in Social Security seems unlikely; a decrease in bequests by baby boomers will entail a decrease in retiree savings (increases in dissaving) and an increase in the savings of younger genera Chart 11 P e n s io n C o v e r a g e , b y P r im a r y P l a n P e r c e n t o f labo r force tions, who will expect smaller inheritances. (Empirical work in Weil 1994 suggests that the expectation of bequests low ers savings.) A review of these options suggests that even if there is a retirement savings shortfall, the baby boom generation may not increase its savings in response to it. No clear evi dence on the reaction of households to savings shortfalls cur rently exists. Should a surge in baby boomer savings occur, the effect on aggregate savings will very likely be large, but the surge itself is uncertain. C o n c l u sio n Policymakers concerned about low aggregate saving rates should not rely on the aging of the baby boom generation to restore aggregate savings to earlier levels. We draw this con clusion from (1) a review of life-cycle increases in saving rates and demographic trends and (2) an assessment of the prospect that baby boomers w ill accelerate their saving rates in Source: Employee Benefits Research Institute. 88 F R B N Y Q u a r t e r l y R e v ie w / S u m m e r-F a ll 1994 response to a retirement savings shortfall. First, as baby boomers move more fully into middle benefits, the evidence that baby boomers are not saving ade age, the life-cycle increases in their saving rates w ill lead to quately for retirem ent is unconvincing. To date, boomers only a modest increase in aggregate saving rates. Survey evi have generally accumulated more wealth relative to their dence from the m id- 1980 s suggests that boomers’ own sav incomes than their parents had by the same ages. Moreover, ing rates w ill roughly double as they move into middle age. the stocks o f w ealth that boomers have accum ulated are Partially offsetting this positive effect on savings, however, roughly consistent w ith target levels recommended by opti w ill be a rise in the relative population shares o f low-savings mal lifetime savings models. These healthy stocks o f savings groups (people above 75 and below 35) over the next two are, however, hard to reconcile with the reportedly weak flow decades. Our projections indicate that demographic trends o f savings observed in survey data. Nonetheless, even if baby should push up the aggregate personal saving rate in total boomers do face a shortfall— perhaps one resulting from a only about 80 basis points by the year 2010 . decrease in Social Security benefits— it is not clear that they Second, barring a major reduction in Social Security would dramatically increase their savings in response. FRBNY Q u a r t e r l y R e v i e w /S u m m e r -F a l l 1994 89 En d n o tes 1. These arguments have been analyzed by a number of authors. Summers Note 3 continued and Carroll (1987), Bosworth, Burtless, and Sabelhaus (1991), and Attana- Bureau in Current Population Reports, Population Projections of the United sio (1993) examine the role o f demographics in the fall o f the aggregate sav States, by Age, Sex, Race, and Hispanic Origin: 19 9 2 to 20 50, P2 5 -10 9 2 . ings rate in the early 1980s. Auerbach and KotlikofF (1989) and Kennickell (1990) examine the effect of baby boomer aging on future saving rates. 4. Several analysts have explored this approach. Easterlin, Schaeffer, and Auerbach and K otlikoff predict a modest increase in saving rates; Kennick- Macunovich (1993) compare income and wealth across generations at sim ell predicts a small increase. Arguments supporting the view that baby ilar points in the life cycle and find a general improvement across genera boomers are undersaving are presented in Bernheim (1993) and Bemheim tions. Manchester (1993) concentrates on wealth and finds that the median and Scholz (1993); opposing views are presented in Manchester (1993) and wealth position o f baby boomers is modestly better than that o f earlier Easterlin, Schaeffer, and Macunovich (1993). generations. 2. The age-income cross-sectional distribution in Chart 1 is measured at a 5. For comparisons of the 19 63 Survey o f Financial Characteristics of Con point in time and is not representative o f the age-earnings profile that a sumers with the Flow o f Funds data, see Avery, Elliehausen, and Kennick- typical household can expect to earn over time. The latter profile would ell (1988). W e are indebted to Rochelle Antoniewicz of the Federal Reserve probably rise more steeply with age, because young workers will typically Board Flow of Funds Division for pointing out a mistake in the compar earn more when old than old workers currently earn. (For further discus isons in this source and for providing preliminary comparisons o f the 19 8 9 sion, see Easterlin, Schaeffer, and Macunovich 1993.) This “bias” in the Survey of Consumer Finances with the Flow of Funds Accounts. cross-sectional age-earnings profile is also present in the age-savings pro file. The age profile of saving rates, however, w ill not necessarily be biased. 6. The analysis was carried out on 35 to 4 4 year olds. Bemheim (1993) presents targets for individuals who earned $ 3 0 ,0 0 0 or more at age 35. We 3- Historical data on the age distribution o f households are published use his estimates of age-earnings profiles, combined with his method of annually by the U.S. Department of Commerce, Bureau of the Census, in imputing income to spouses, to estimate earnings at age 35 for the 19 8 9 Current Population Reports, “Household and Family Characteristics,” Survey of Consumer Finances sample. Interpolating where necessary, we Series P20. Projections of the future age distribution of households were calculate target savings (excluding household wealth) for each household. obtained by assuming that (1) within ten-year age bands, the current ratios Target savings are taken from Table 1 of Bernheim (1993). Estimated age- o f households to population remain constant over time and (2) population income profiles are taken from Bemheim (1992). cohorts grow at the “middle series” estimates published by the Census R eferen ces Attanasio, Orazio P. 1993. “A Cohort Analysis o f Saving Behavior by U.S. _______ . 1994. “The United States Fiscal and Saving Crises and Their Households.” National Bureau o f Economic Research W orking Paper Implications for the Baby Boom Generation.” Merrill Lynch, New York. no. 4454. Avery, Robert B., Gregory E. Elliehausen, and Arthur B. Kennickell. 19 8 8 . Auerbach, Alan J ., and Laurence J . Kotlikoff. 1990. “Demographics, Fiscal “Measuring Wealth with Survey Data; An Evaluation of the 19 8 3 Sur Policy, and U.S. Savings in the 19 80s and Beyond.” In Lawrence H. vey o f Consumer Finances.” R e v ie w Summers, ed., T a x P o l ic y ber); 339-69. a n d the E c o n o m y . Cambridge, Mass.; MIT of I n co m e a n d W ealth (Decem Press. Digitized F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 90for FRASER N otes R e fe re n c e s (Continued) Berger, Mark C. 1985. “The Effect o f Cohort Size on Earnings Growth: A Harris, Ethan S., and Charles Steindel. 19 91- “The Decline in U.S. Saving Reexamination o f the Evidence.” JOURNAL OF P o l it ic a l ECONOMY and Its Implications for Economic Growth.” F ederal R eserve B a n k (June): 56 1-74 . N e w Y o r k Q u ar t e r ly R e v ie w (Winter): 1-19. Bernheim, B. Douglas. 19 92. “Is the Baby Boom Generation Preparing Ade quately for Retirement? Technical Report.” Merrill Lynch, New York. _______. 1993- “Is the Baby Boom Generation Preparing Adequately for Retirement? Summary Report.” Merrill Lynch, New York. of Kennickell, Arthur B. 19 9 0 . “Demographics and Household Savings.” Board of Governors of the Federal Reserve System, draft. Manchester, Joyce. 1993. "Baby Boomers in Retirement: An Early Perspec tive.” Washington, D.C.: Congressional Budget Office. _______ . 1 9 9 4 . “The M errill Lynch Baby Boom Retirem ent Index.” Merrill Lynch, New York. Poterba, James M., Steven F. Venti, and David A. Wise. 19 92. “401(k) Plans and Tax Deferred Saving.” National Bureau of Economic Research Working Paper no. 4 1 8 1 . Bernheim, B. Douglas, andJohn KarlScholz. 1993. “Private Saving and Pub lic Policy.” In James M. Poterba, ed., T a x P o licy a n d the Ec o n o m y . Cambridge, Mass.: MIT Press. Bosworth, Barry, Gary Burtless, and John Sabelhaus. 1991- “The Decline in Saving: Some Microeconomic Evidence.” B r o o k in g s P ape r s ON ECO Summers, Laurence, and Chris Carroll. 1987. “W h y is U.S. National Savings So Low?” B r o o k in g s P a pe r s on Ec o n o m ic A c t iv it y 2 :6 0 7 -3 5 . Weil, David N. 1994. “The Saving of the Elderly in Micro and Macro Data.” Q u arterly J o u r n a l of Ec o n o m ic s (February): 55-82. NOMIC ACTIVITY 1: 18 3 -2 4 1. Welch, Finis. 19 7 9 . “The Effects o f Cohort Size on Earnings: The Baby Easterlin, Richard A., Christine M. Schaeffer, and Diane J . Macunovich. 1993. “W ill the Baby Boomers Be Less W ell O ff Than Their Parents? Income, Boom Babies’ Financial B u st.” J o u r n a l OF P o l it ic a l E c o n o m y (October, pt.2): s65-s98. Wealth, and Family Circumstances Over the Life Cycle.” University of Southern California and Williams College, draft. Digitized N otesfor FRASER FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 91 Mortgage Security Hedging and the Yield Curve J u lia D. Fernald, Frank Keane, and Patricia C. Mosser steepening of the yield curve in response to tighter Treasury yield curve did not flatten after policy was tight monetary policy this past spring puzzled many market ana ened: the 125 basis point increase in the federal funds rate lysts and economists. Most of the explanations of this phe from February through May of 1994 was accompanied by a nomenon have focused on macroeconomic issues such as mar 133 basis point increase in the ten-year Treasury rate. Third, ket expectations of higher inflation or higher future interest after the change in policy direction in February, the yield rates. This article offers an additional, markets-based expla curve became more hump-shaped: the two-year Treasury nation that examines hedging activity—particularly the yield rose 175 basis points, but the thirty-year yield increased hedging of mortgage-backed securities—and its effect on the only 96 basis points. 1 rih Although the steepening after February was extreme short-run dynamics of the yield curve. When interest rates rise, both the duration and the by historical standards, Cohen and Wenninger (1994) have expected m aturity of a mortgage-backed security (MBS) increase. If market participants seek to counteract the increased price risk in MBSs by taking short positions in sim Chart 1 ilar duration Treasury securities, the increase in MBS dura T r e a s u r y Y ie l d C u r v e s tion should cause participants to move their short Treasury Percent positions out the yield curve, effectively increasing the “sup g --------------------------------------------------------------------- --- ply” of long duration Treasuries. Thus the hedging of mortgage securities in the Trea April 20 ,19 9 4 sury market may—in the short run—magnify any increases in long-term rates that accompany policy tightening. To the extent that such hedging activity has become a standard fea ture of the marketplace in the last few years, it may have per manently altered the short-run dynamics of the yield curve and thus changed the transmission of monetary policy. R e c e n t M o v e m e n t s in t h e T r e a s u r y Y ie l d C u r v e Chart 1 highlights several significant movements in the yield 2 I__ I_I__ I______ I____________________________ I FF* 2 3 5 curve since the fall of 1993- First, long rates began to rise in October 1993, well before monetary policy tightened. Sec ond, despite little or no observable inflation pressure, the Digitized FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 92 for FRASER 10 30 Maturity in years Source: Bloomberg L.P. *FF is federal funds rate. I n B rief noted that since the mid-1980s, the short-run responsiveness How MBS of long rates to changes in the federal funds rate has increased S te e p e n t h e Y ie ld C u rv e sharply. A 100 basis point increase in the funds rate caused When long-term interest rates rise (because of policy tight (on average) only a 15 basis point increase in the ten-year yield in the early 1980s, but a 40 basis point or more increase in the ten-year yield more recently (Chart 2).1In other words, ening or other factors such as higher expected inflation), households, in aggregate, refinance and prepay their mort gages more slowly. For a typical mortgage pool, slower pre when monetary policy is tightened, the yield curve now flat payments mean that the future mortgage principal will be tens less. Indeed, in the latest episode, it did not flatten at all. repaid more slowly, thus extending the expected maturity of Interestingly, the change in yield curve dynamics the MBS and increasing its duration. This is extension risk: coincided with large-scale structural changes in financial slower prepayments increase the sensitivity of MBS prices to markets, in particular the development of new financial rising yields (see the box below). H e d g in g U s in g T r e a s u r ie s C o u ld instruments and the widespread securitization of home mort Dealers in MBSs and collateralized mortgage oblig gages.2 In 1983, less than 20 percent of the stock of residen ations (CMOs) hold inventories of these securities, which tial mortgage debt was securitized; by 1993, nearly 50 per they attempt to hedge against such extension risk. One com cent was securitized. Increased securitization has led to mon hedging strategy used by dealers is to offset long MBS increased use of mark-to-market accounting of mortgage positions by taking short positions in combinations of Trea debt, making owners of mortgage assets more sensitive to suries that approximate the mortgage security’s duration.4 short-run rate movements.3 To the extent that mortgage Thus as rates rise and prepayments fall, dealers must increase securitization caused quicker adjustments of mortgage port the duration of their Treasury hedges to roughly match the folios to changing market conditions and thus brought closer increasing duration of their MBS portfolios. For example, a links between mortgage and Treasury markets, it may have dealer hedging its MBS portfolio with short positions in two- contributed to the change in yield curve dynamics. to five-year Treasuries might change to a combination of five- Chart 2 S e n s i t i v i t y o f T e n - Y e a r T r e a s u r y Y i e l d s t o C h a n g e s in t h e F e d e r a l F u n d s R a t e R e g r e s s io n c o e f f ic ie n t 1968 70 72 74 76 78 80 82 84 86 88 90 92 94 Source: Cohen and Wenninger 1994. Note: Chart plots coefficients from five-year rolling regressions o f monthly changes in ten-year Treasury yields on monthly changes in the federal funds rate, that is, |3 from AlOyr = a + pAFF. Shaded areas indicate periods designated recessions by the National Bureau of Economic Research. I n B rief FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 93 to seven-year Treasuries; a short position in five- to seven-year Treasuries might be changed to a short position in seven- to ten-year securities. (Thirty-year bonds are not generally used to hedge MBSs.) sury hedges. Thus MBS hedging could cause positive feed back or a “multiplier” effect that would further steepen the yield curve.5 Just as dealers may use Treasury securities to hedge As durations rise and market participants attempt to their MBS positions, portfolio managers who hold mortgage change their hedges simultaneously, the increase in short securities often use the government securities market to positions of long m aturity Treasuries should cause their adjust their portfolios in an attempt to maintain a target yields to rise by more than the yields of shorter maturity duration. From a market dynamics standpoint, such portfolio bonds, and the yield curve should steepen. The higher long rebalancing is analogous to dealer hedging of mortgage term yields could, in turn, reduce refinancing and prepay inventories: as interest rates rise and mortgage securities ment rates even further, again increasing MBS duration (and extend, the portfolio’s duration increases. Longer duration in its price sensitivity) and causing additional changes in Trea turn prompts managers to sell longer maturity Treasuries to M o r t g a g e S e c u r it y S t r u c t u r e : C all a n d E x t e n sio n R isk Fixed rate mortgages give homeowners the option to prepay lengthen, then the MBS is subject to “extension risk.” Slower part or all of the mortgage loan, at any time and for any rea prepayments mean mortgage principal is repaid later, thus son, before the final maturity date. The prepayment option extending the expected maturity of the MBS. Longer maturi can dramatically affect the price sensitivity of the mortgage ty also means that the security’s price becomes more sensitive security because the tim ing and amount of prepayments to rising yields. As the mortgage extends with the rising rates change the actual life of the security. and slower prepayments, its price falls more than it would have Although the underlying mortgage loans often have thirty-year terms, an MBS is never viewed as a thirty-year had the prepayment speed remained constant (see Chart 3). Both mortgage pass-throughs and collateralized instrument. Mortgage market participants constantly fore mortgage obligations (CMOs) are subject to call and exten sion risk. By construction, however, some CMO tranches are cast future prepayments to predict the security’s expected life. If fast prepayments (usually due to a decline in interest rates) cause the maturity (or duration) of an MBS or CMO to substantially more sensitive to such risks than the underlying pass-through, and other tranches are less sensitive. This split shorten substantially relative to expectations, that security is may have increased aggregate hedging related to mortgage subject to “call risk.” securitization. Call risk is analogous to the risk to an owner of a In a nutshell, call risk forces investors to reinvest in callable bond: as interest rates fall and the price of the bond fallin g rate environm ents, and extension risk exposes rises, the issuer (here a household) can exercise its option and investors to escalating price risk in rising rate environments. call the bond at par. While there is no loss of principal in this Hedging activity is probably greater with extension risk case, the owner of the security must reinvest the proceeds at because rising price risk and outright losses require a quicker lower market interest rates. adjustment of hedge positions than does the opportunity cost If, on the other hand, slow mortgage prepayments of lower reinvestment returns. (due to rising interest rates) cause the duration of an MBS to 94 for FRASER F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 Digitized I n B rief regain their portfolio’s target duration. If many portfolio 1.92 percent. More realistically, the increase in rates could managers attempt such duration adjustments simultaneous slow prepayments from 15 percent to 10 percent per year—a ly, the excess supply of longer maturity Treasuries will steep en the yield curve. move from point A to point C—and cause a larger 3-75 per M o r t g a g e P r e p a y m e n t s a n d t h e S h if t i n cent price decline. Alternatively, a fundamental change in the direction of future rates could produce a large drop in prepay ments to, say, 5 percent per year—a move from point A to M o n e t a r y P o l ic y point D—and a large drop in the MBS price.6 Although long rates began rising in late 1993, MBS dealers With slower prepayments, the duration of the MBS were slow to lower their forecasts of MBS prepayment rates is longer as well. In Chart 3, the duration of the MBS depends (see table below). After the change in policy direction in Feb on the slope of the line connecting point A to one of the three ruary, however, estimates of expected future prepayment rates other points—whichever reflects the expected prepayment dropped sharply, presumably because the policy change sig speed. The effect of prepayment assumptions on duration is naled that further declines in interest rates were unlikely. In sizable, even for relatively small interest rate changes. A 50 fact, in 1992 and 1993, mortgage prepayment rates had been basis point increase in rates with no change in prepayments much higher than most dealers had anticipated, and so the (point B) will give this MBS a duration of 3.28, or approxi policy shift may have caused a particularly large drop in mately that of a four-year Treasury. If the same 50 basis point assumed prepayment speeds. increase in rates is accompanied by a drop in prepayment Chart 3 illustrates how critical prepayment expecta tions are for mortgage security prices and durations. In the speed from 15 percent to 5 percent (point D), duration increases to 5.36, roughly that of a seven-year Treasury. chart, we consider the relationships among price, yield, and The table above shows that actual declines in pre prepayment speed for an 8 percent coupon thirty-year con payment speeds and increases in effective duration during ventional MBS currently yielding 9Vi percent at a prepay early 1994 were indeed quite large—of the same order of ment speed of 15 percent (point A). As interest rates change, different assumptions about prepayment response to the change in yield determine both the price and the duration of Chart 3 an MBS. For example, a 50 basis point increase in yield that P r ic e - Y ie l d - P r e p a y m e n t S u r f a c e f o r a T y p ic a l does not affect the speed of mortgage prepayments would M o r t g a g e - b a c k e d S e c u r it y lead to a move from point A to point B and a price decline of D e a l e r P r e p a y m e n t F o r e c a s t s a n d E f f e c t iv e D u r a t i o n s FNMA 7.5 Percent Coupon Thirty-Year Conventional MBS “Effective” Duration (Years) Prepayment Forecast (Percent) Ten-Year Yield (Percent) Fed Funds Rate (Percent) 3 .4 0 5.71 3.00 3.00 February 9 , 1 9 9 4 21.8 20.8 20.8 5 .17 3.49 3.52 5.91 3.25 March 2 3 ,1 9 9 4 4 .81 1 1 .9 6 .4 9 3.50 A pril 2 0 ,1 9 9 4 5.27 9.5 3.75 May 1 7 , 1 9 9 4 5.41 9.0 7.03 7.04 Date October 1 5 ,1 9 9 3 January 26, 19 9 4 4.25 Source: Bloomberg L.P. Notes: Prepayment forecasts are dealer medians quoted in PSA and converted to conditional prepayment rates in percent. Effective durations are calculated with dealer median prepayment forecasts using Bloomberg analytics. Dealers include First Boston Corporation, DLJ, UBS Securities, Paine W ebber, Bear Stearns, Sm ith Barney, Prudential Securities, M errill Lynch, Lehman Brothers, and Salomon Brothers. I n B rief Note: Example is 8 percent coupon thirty-year conventional mortgage-backed security. FRBNY Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 95 magnitude as those in Chart 3. Duration (for a FNMA IV2 flattening o f the long end o f the yield curve would result, percent coupon MBS) was basically unchanged from October however, if relatively more ten-year securities were being sold to February, rem aining at about 3 lA years, and then rose to hedge mortgage securities. sharply after the policy change in February to nearly 5 Vi years In contrast, m ortgage security hedging does not by mid-May. Hedging this particular MBS would call for a explain the bulge in the Treasury curve from two to five years short position in four-year Treasuries in February but a short over the same period. Indeed, MBS hedging should have put position in seven- to eight-year Treasuries by May. However, downward pressure on two-year Treasury yields in particular. because most hedging is done w ith new-issue Treasuries, in It seems likely that widely cited macroeconomic factors such practice such an MBS would be hedged with three- to five- as expectations o f higher inflation and higher future interest year Treasuries in February and five- to ten-year Treasuries by rates dominated movements at the short end o f the yield May. (Issuance o f seven-year Treasury debt was discontinued curve. after A pril 19 9 3 , making it a less likely hedging vehicle.) A lthough the yield curve evidence is ambiguous, W h ile this example o f a particular MBS illustrates daily price correlations between MBSs and Treasuries show the duration and hedging issues, in practice, market partici consistently stronger relationships between MBSs and longer pants hedge entire portfolios o f MBSs and CMOs, not indi m aturity Treasuries in early 19 9 4 . A m inimum condition for vidual mortgage securities. Unfortunately, we have no infor mortgage hedging to have affected the yield curve is that m ation on the com position o f these p o rtfo lio s, so it is MBSs should have behaved more like ten-year securities and impossible to say exactly how durations and hedges o f MBS less like shorter term securities from October to May. In fact, portfolios may have actually changed in late 1993 and early price correlations between different m aturity Treasuries and 1 9 9 4 .7 Thus we turn to more indirect evidence to try to iden the IV2 percent FNMA in Chart 4 show just such a pattern. tify links between the mortgage and Treasury markets. W hen long rates began to rise in October and November, the correlations between prices o f two-year Treasuries and m ort E m p ir ic a l E vid ence gage securities fell, while the correlations between the MBS Some market participants have estimated that from October and five-year Treasuries rose slightly. 19 9 3 to A p ril 19 9 4 , aggregate dynamic hedging o f m ort After the policy tightening in February, correlations gage extension risk by dealers, portfolios managers, and other between the MBS and two-year Treasuries dropped again, investors resulted in Treasury market sales of more than $ 3 0 0 and correlations between the MBS and the five-year Treasury billion in ten-year Treasury equivalents. A lthough this figure dipped slightly. In contrast, the ten-year Treasury/MBS cor is impossible to verify, we can provide some circumstantial relations were stable or rising in February and March. Fur evidence that mortgage security hedging using Treasuries ther, the tim ing o f the changes in price correlations corre had a significant, although probably not dominant, eflfect on sponded quite closely to that o f the increases in MBS duration the Treasury yield movements in late 19 9 3 and early 19 94. in the table on page 95. In late 19 9 3 , this MBS behaved like a Because mortgages are usually hedged w ith Trea five-year lather than a two-year Treasury. By March and A pril suries up to ten years in maturity, but not with thirty-year 1 9 9 4 , its duration increased enough that the MBS price Treasuries, the flattening o f the Treasury curve between ten behaved more like that o f a ten-year than a five-year Treasury. and thirty years seen in Chart 1 provides some evidence sup W h ile the price correlations are consistent w ith a porting a link between MBS hedging and the Treasury yield relationship between MBS hedging and Treasury prices, they curve. In particular, the spread between the ten- and thirty- cannot tell us if such activity was actually occurring. One year securities fell from nearly 6 0 basis points in early Febru obvious question is whether MBS activity was really large ary 19 9 4 to less than 2 0 basis points by early May, a flattening enough to affect Treasury prices. Chart 5 suggests that it was. that is very hard to explain by expectations o f higher inflation New five- to ten-year Treasury supplies (lower right), which or higher short-term rates over the next year or two. Such a are most likely to be used for hedging purposes, were about Digitized 96 for FRASER FRBN Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 I n B rief $45 billion a quarter during 1993, while dealer inventories of MBSs, both pass-throughs and CMOs, were $50 billion or lateral market for a particular Treasury issue because the holder of the collateral pays the repo rate. more in late 1993 and early 1994.8 Comparing the outstanding amounts of Treasury and mortgage securities at year-end 1993 provides further Chart 5 evidence that the mortgage market was large enough to affect the Treasury market. In fact, private holdings of Treasury D e a l e r P o s i t io n s in M o r t g a g e - b a c k e d S e c u r it i e s a n d C o l l a t e r a l iz e d M o r t g a g e O b l ig a t io n s marketable debt maturing in two to ten years were smaller ($964 billion) than outstanding securitized agency mortgage B i l l io n s o f d o l l a r s 70 MBS Pass-throughs debt ($1,350 billion). CMOs Perhaps the most direct information on the demand for Treasury securities for hedging purposes comes from the repurchase agreement market for Treasury collateral—the “repo” market. The holder of Treasury collateral pays the repo rate to the party seeking to borrow the collateral (often for delivery against short sales in the cash market). A short seller Quarterly Issuance of Treasury Debt (Approximate) usually borrows a specific Treasury issue to meet its cash mar ket delivery obligations. Two-year Three-year Five-year Ten-year The repo rate for general collateral (that is, any maturity Treasury) is similar to other overnight interest rates. $51 $17 $33 $ 12 billion billion billion billion m in im u m However, imbalances between supply and demand for specif M ic Treasury issues are reflected in issue-specific repo rates.9 1993 For example, low repo rates reflect excess demand in the col Source: DRI/McGraw-Hill. M 1994 Chart 4 P r ic e C o r r e l a t io n s : M o r t g a g e - b a c k e d S e c u r it ie s a n d T r e a s u r ie s FNMA 7.5 Percent Coupon w ith Two-, Five-, and Ten-Year Treasury Securities C o r r e l a t io n 1993 1994 Source: DRI/McGraw-Hill. Note: Chart plots twenty-day lagged rolling price correlations. I n B rief F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 97 Repo rates for particular maturities are commonly presented as spreads relative to the rate for general collateral. A high repo spread (that is, low repo rate) can be interpreted as the financing premium that a short seller must pay in order to borrow a particular maturity Treasury security overnight. (Chart 8) but no clear pattern that can be related to MBS hedging. Furthermore, it is difficult to extract information about hedging activity from movements in the open interest for the thirty-year bond contract because daily trading vol ume is particularly high relative to open interest.14 The high Repo spreads for the most recently issued or “onthe-run” Treasuries are shown in Chart 6.10 Spreads for the first four months of 1994 are consistent with high demand for Chart 7 progressively longer dated Treasuries, presumably stemming U .S . T r e a s u r y F i v e - a n d T e n - Y e a r N o t e F u t u r e s Open Interest from efforts to counteract mortgage security extension risk. Spreads widen first for five- and seven-year maturities and T h o u sa n d s o f c o n t r a c t s 350 ,'V~' \ then for the ten-year maturities.11 Further evidence of increased hedging activity can / 300 /. I be seen in Chart 7, which shows open interest in the five- and Ten-year ten-year Treasury futures market from the beginning of 1994.12 These data support the repo data: open interest V 250 \ •K> \ S * V. \ /V' <w -N. increased first for the five-year contract and then for the tenyear contract as rates continued to rise.13 Moreover, the 200 Five-year increase in open interest for the ten-year contract corre sponded closely to the high and sustained financing premium 150 L M in the ten-year repo market through April. M 19 94 In contrast, open interest for thirty-year bond futures shows a m ild upward trend during the period Sources: Chicago Board of Trade; DRI/McGraw-Hill. Notes: Each contract has a $ 100,0 00 face value. Chart 6 . S e l e c t e d T r e a s u r y “R e p o ” S p r e a d s B a s is p o in t s 350 1994 Source: Federal Reserve Bank of New York. Note: Chart shows the general repurchase agreement rate minus the rate on each specific maturity. 98 F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 I n B rief Chart 8 in mortgage security hedges and realignments of portfolios in U.S. T r e a s u r y response to longer MBS durations had a significant effect on T h ir t y - Y e a r B o n d Fu t u r e s Volume and Open Interest T h o u san d s o f c o n tr a c ts 800 --------------------------------------------------- the Treasury yield curve, particularly after the change in monetary policy direction in February 1994. Although MBS hedging certainly cannot explain all the shifts in the yield curve in early 1994, some macroeconomic evidence does sup port the relationship: the flattening of the ten- to thirty-year spread in early 1994 and the increased (short-run) sensitivity of long rates to changes in short rates. In addition, estimates of mortgage prepayments and durations, evidence on MBS and Treasury prices and volumes, and information from the repo and futures markets all suggest that the hedging of mortgage security extension risk was widespread and had a significant impact on the short-run movements of the Trea 1994 sury market, particularly the ten-year market. Sources: Chicago Board of Trade; DRI/McGraw-Hill. Although there is no evidence that hedging activity Notes: Each contract has a $ 10 0 ,0 0 0 face value. has affected the long-run relationship between long-term and volume reflects a high level of intraday trading and hedging, short-term interest rates, this latest episode is further evi which is unlikely to be related to the MBS market. dence that the short-run dynamics of the yield curve have changed over the last decade. As a result, the transmission of Sum m ary monetary policy from short-term interest rates to the real The circumstantial evidence presented above, as well as wide economy via long-term interest rates has probably changed spread reports from market participants, suggests that shifts as well. I n B rief F R B N Y Q u a r t e r l y R e v ie w /S u m m e r -F a ll 1994 99 1. In addition to using the simple regression evidence presented in Chart 2, Note 5 continued Cohen and Wenninger (1994) estimate more complicated time series mod duration bonds, putting more downward pressure on long-term yields in els of the term structure to show an increase in the short-run sensitivity of the short run. long rates to short rates. 6. The slowing in prepayments exacerbates the effect that rising rates have 2. The m id -1980s change in yield curve dynamics may also have been a on the price of the MBS because the repayment o f mortgage principal delayed reaction to the 19 7 9 change in Federal Reserve policy regime occurs over a longer period. toward a stronger anti-inflation stance. 7. In addition, some CMO tranches, by construction, contain substantial 3. Mortgage securitization may have contributed to the greater sensitivity ly more extension risk than MBS pass-throughs and involve more compli o f long-term interest rates to short-term interest rates by moving residen cated relationships between yield changes, prepayments, and duration than tial housing finance away from financial intermediaries and directly into is suggested by Chart 3- For such securities, Chart 3 and the table on page financial markets. Before mortgage securitization, a rise in short-term rates 95 may underestimate changes in durations and thus changes in hedges. hurt the cash flows o f financial intermediaries who held mortgages. But because mortgages were not marked to market, intermediaries were proba 8. We focus on dealer inventories of mortgage securities because they are bly slow to adjust their asset portfolios to reflect the decline in mortgage the most likely to be dynamically hedged. values. This slow portfolio adjustment meant that any feed-through to long-term interest rates tended to be indirect and slow. W ith the advent of 9. W hen the repo rate for a specific Treasury issue diverges from the repo mortgage securitization, however, the majority of mortgages are no longer rate for general collateral, it is said to be “on special” or “special-” held on bank balance sheets but in MBSs, which are marked-to-market daily and, in many cases, dynamically hedged. Further, portfolios contain 10. On-the-run Treasuries provide the best liquidity for hedgers. ing mortgages are adjusted more quickly, and as a result, the adjustment of long rates to short rates is probably quicker as well. 11. Because the last seven-year Treasury was issued in A pril 19 9 3 , the “seven-year Treasury” in Chart 6 is actually a six-year security during this 4. Dealers who attempt to hedge MBSs using offsetting Treasury positions period. are, by definition, using imperfect hedges. Because o f the implicit pathdependent optionality and negative convexity of MBSs, hedges must be 12. Open interest is the net number of outstanding futures contracts. adjusted dynamically as market conditions change. See the box on page 94. 13. Increases in open interest suggest that market participants have estab 5. This process is probably somewhat symmetric. W hen interest rates lished more permanent positions, and thus these increases may be inter fall, durations and maturities of MBSs shorten and MBSs are subject to preted as evidence o f greater hedging activity within the futures market. call, or refinancing, risk (see the box on page 94). To hedge such call risk, market participants could sell shorter duration Treasuries and buy longer 14. For the five- and ten-year futures, daily volume is one-third to one-half o f open interest. Cohen, GeraldD., andJohn Wenninger. 1994. “The Relationship between the Federal Funds Rate and Economic Activity.” Federal Reserve Bank of New York Research Paper no. 9406. Digitized F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 100 for FRASER N o tes Has Excess Capacity Abroad Reduced U.S. Inflationary Pressures ? James A . Orr W 7 t ttt r? W tllL Z U.S. manufacturing capacity utilization has U.S. demand, used their available capacity to expand produc been rising in recent years, capacity utilization in the manu tion w ithout increasing their prices significantly.1 Import facturing sectors o f the major foreign industrial economies prices that remained flat or rose more slowly than the prices of has declined. Falling utilization rates abroad have given for domestically produced goods would slow U.S. inflation in eign firms the potential to expand production without incur two ways: directly, as the imports entered into U.S. consump ring significant cost increases. This article investigates tion, and indirectly, as the imports siphoned off increases in whether sizable slack abroad, which has helped to slow for U.S. demand and thus restrained price increases by compet eign inflation, could also have eased U.S. inflationary pres ing U.S. firms.2 A significant amount o f excess capacity exists in the sures by preventing the prices o f imports from rising as fast as the prices o f U.S.-produced goods. manufacturing sectors o f Japan, Canada, Germany, France, The analysis shows th at grow ing foreign excess Italy, and the U nited K ingdom , countries that together capacity has provided only a lim ited amount o f protection account for more than one-half o f all U.S. imports. Although against domestic inflationary pressures. Although inflation aggregate foreign manufacturing capacity utilization in these abroad has been lower than U.S. inflation, exchange rate economies increased slightly in the first half o f this year, it changes have exceeded the inflation differentials and exerted declined more than 10 percent between the end o f 19 9 0 and significant and varied influences on import prices. For exam the beginning o f 1 9 9 4 (Chart 1). In two previous downturns, ple, dollar depreciation against the yen has erased the depres capacity utilization abroad reached even lower levels, but the sing effects on prices o f significant excess capacity in Japan, decline for the past several years is p a rticu larly notable w hile dollar appreciation against the Canadian dollar has because it occurred as U.S. capacity utilization was rising greatly enhanced the effects o f moderate excess capacity in sharply. Canada. Aggregated across all sources, import price growth The available evidence shows that growing excess has roughly kept pace with U.S. inflation. Moreover, in U.S. m anufacturing capacity abroad, among other factors, has manufacturing industries nearing full capacity, imports have exerted downward pressure on the prices o f foreign manufac not seized an increasing share o f the market, a development tured products expressed in local currency.3 Foreign producer that one would expect if foreign excess capacity were going to prices have risen more slowly than U.S. prices. In fact, pro influence U.S. pricing decisions significantly. Consequently, ducer prices in Canada and W estern Europe have lagged the analysis concludes that the mere presence of excess capacity growth in U.S. producer prices by roughly 2 percent since the abroad has not greatly restrained U.S. inflationary pressures. end o f 1 9 9 0 (Chart 2). During the same period, Japanese prices have lagged U.S. price growth by almost 10 percent.4 S l a c k C a p a c it y A b r o a d a n d I m po r t P rices Nevertheless, although excess capacity abroad has Excess capacity abroad would relieve inflationary pressures in helped to lower the local currency prices o f foreign manufac the U.S. economy if foreign suppliers, responding to growing tured goods relative to U.S. prices, other factors affect the I n B rie f F R B N Y Q u a r t e r l y R e v i e w /S u m m e r -F a l l 1994 101 U.S. dollar price of imports—specifically, changes in the exchange rate and the extent to which these changes are passed through by foreign suppliers to U.S. consumers.5 ments (Chart 3). Even with only part of the change in nomi nal exchange rates being passed through into import prices, exchange rate movements largely undercut the potential ben Movements in the nominal value of the dollar against the cur efits of relatively lower inflation rates abroad. From 1990 rencies of key industrial countries have far exceeded the mod through the second quarter of 1994, dollar prices of U.S. erate slowing in their producer prices relative to U.S. pro manufactured imports from Japan rose roughly 6 percent ducer prices. Since the end of 1990, the dollar has appreciated compared with the prices of U.S. manufactured goods. This more than 15 percent against the Canadian dollar and 16 per rise was consistent with the combination of a 10 percent fall cent against an average of Western European currencies, in Japanese local currency prices relative to U.S. prices and a while depreciating 25 percent against the Japanese yen.6 20 percent nominal appreciation of the yen. The exchange Dollar appreciation against the Western European and Cana rate movement thus overwhelmed the potential benefits of dian currencies has thus augmented their modestly lower Japan’s excess capacity for U.S. inflation. The dollar prices of inflation rates; by contrast, dollar depreciation against the manufactured imports from Western Europe and Canada fell yen has more than offset Japan’s sizable decline in producer roughly 7 percent against the prices of U.S. manufactured prices relative to U.S. producer prices. goods, a decline that was much more than the relative fall in Direct evidence on the prices of manufactured imports from industrial countries compared with U.S. pro their local currency producer prices but consistent with their nominal currency depreciations of more than 10 percent. ducer prices between the end of 1990 and m id-1994 bears Overall, prices of imports from industrialized coun out the significance of exchange rates for import price move tries grew only 1 percent less than U.S. prices between 1990 Chart 1 M a n u f a c t u r in g C a p a c it y U t il iz a t io n In d e x : 1 9 9 0 = 10 0 P ercent Notes: The index of capacity in foreign industrialized economies is an import-weighted average of the utilization rates in Japan, Germany, France, the United Kingdom, Canada, and Italy. The 1994 utilization rates for France and Italy are estimates. 102 F R B N Y Q u a r t e r l y R e v ie w / S u m m e r -F a ll 1994 I n B r ie f and m id-1994.7 Slack capacity alone, without any changes in slack capacity abroad is easing inflationary pressures where exchange rates, would likely have caused import prices to fall' U.S. capacity strains are most concentrated. It is in these perhaps 5 percent more than U.S. prices— a difference that industries that we would expect to see import price growth was roughly equal to the gap between foreign and U.S. pro lagging U.S. price growth even if it did not show up in aggre ducer price inflation. However, largely because of exchange gate import prices. rate changes, the rise in the relative dollar price of imports The results of the import price comparisons for these from Japan offset a large part of the decline in the dollar price industries are mixed. Prices of imported primary metals and of imports from Western Europe and Canada and hence lim industrial machinery have not risen as fast as U.S. producer ited overall relative import price declines. prices in these industries (Chart 4). The steady decline in the relative price of imported primary metals since early 1992 is I m p o r t P r ic e s in S e l e c t e d In d u s t r ie s broadly consistent with the continued expansion of excess Several U.S. manufacturing industries that reached or were capacity abroad and the increased utilization of U.S. manu nearing previous peak capacity utilization rates during the facturing capacity. However, unlike primary metals, indus second quarter of 1994 have also had a large share of imports trial machinery imports exhibited a sharp decline in relative from industrialized countries. These industries include auto price in the first half of 1991, when significant foreign excess m obiles, prim ary m etals, and electrical and industrial capacity was only beginning to appear and U.S. capacity uti machinery.8 Import prices that have been flat or falling rela lization was not particularly tight. Since the end of 1991, the tive to U.S. prices in these industries would indicate that price of imported industrial machinery has kept pace with Chart 2 T rends in R e l a t iv e P r ic e s I n d e x 1990: Q 4 = io o and D o lla r E x c h a n g e R a tes I n d e x 1990: Q 4 = io o r........................ ..................... N ominal Dollar Ex :hange Rates ^ Japan A w \ A • 1 r m * v \ V Canada V. _ \- / ........ L 1990 i 91 i *»'' L i - J ___ 92 - > - Western Europe \ \. _ 1 93 1 1 94 Notes: The left panel plots the ratios of foreign wholesale price indexes (the producer price index for Japan) to the U.S. producer price index for finished goods. The right panel plots the indexes of the foreign currency price of the U.S. dollar: a rise signifies dollar depreciation. The index of relative prices and exchange rates for Western Europe is an import-weighted average of relative prices and exchange rates in Germany, France, Italy, and the United Kingdom. I n B r ief F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 103 the price of domestically produced industrial machinery, sug inflationary pressures by enabling imports to capture an gesting that growing foreign excess capacity has not eased increasing share of growing U.S. domestic demand. Import inflationary pressures in this industry. market shares in autos have remained constant or declined No significant declines have occurred in the prices of slightly since 1990 (Chart 5). The same pattern holds for pri imported road vehicles, largely autos, or electrical machinery mary metals, despite the steady fall in the relative prices of compared with U.S. prices to date. In fact, in these industries, these imports. Although import market shares in both cate import prices have risen slightly more than U.S. domestic gories of machinery have increased since 1990, neither the prices since 1990. Japan is an important supplier of both timing nor the magnitude of the rise in import market share products, and the failure to observe any relative price declines appears linked to the growth in excess capacity in foreign could be related to the yen’s appreciation since 1990. industrialized economies since 1990. Rather, the rise in the These same four industries offer little broad-based past several years essentially continues the trend increase in evidence that foreign excess capacity has held down U.S. import penetration in these industries that appeared before 1990. This longer term rise in import penetration, particu larly in the electrical machinery industry, in part represents the steady expansion of capacity in developing economies. Chart 3 R a t io of Im p o r t P r ic e s to Therefore, despite the relative importance of imports in U .S . P r i c e s domestic consumption in these four sectors, the data do not I n d e x 1990: Q 4 = 100 106 ••• • • 104 show that growing excess capacity abroad has caused a signif Japan icant increase in the penetration of the U.S. market by foreign • • • • • 102 100 • rA +• % Growing excess capacity in industrialized countries since * » 98 • suppliers in the past several years. 1 ♦ A t •• •M \A tW V 96 Foreign industrialized — economies \ 94 from inflationary pressures in the U.S. economy. The price of U.S. imports from industrialized countries relative to U.S. Western Europe ' v '" 1990 has provided at best a limited amount of protection prices has declined only modestly as lower inflation abroad V, has been offset by exchange rate changes. Dollar depreciation in the case of Japan has rendered excess capacity there basi 92 Canada 1 90 1990 1 91 1 1 1 92 1 1 1 cally ineffective against U.S. inflationary pressures. 1 93 1 1 1 1 94 Source: U.S. Department of Labor, Bureau of Labor Statistics. Notes: Chart plots the ratios of import price indexes to the U.S. producer price index for manufactured goods. The index for the group of foreign industrialized economies is the average, weighted by U.S. imports, of the individual indexes for the European Community nations plus Japan, Canada, Australia, New Zealand, and South Africa. The Western Europe index includes the twelve economies of the European Community. 104 F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a l l 1994 In the years ahead, the strengthening expansions in the United Kingdom and Canada as well as the beginnings of recovery in continental Europe and Japan will increase the utilization of foreign manufacturing capacity. Consequently, the likelihood that excess capacity abroad will offer future inflation relief appears limited. I n B r ie f Chart 4 R Im p o r t P r i c e s : S e l e c t e d I n d u s t r i e s e l a t iv e I n d e x 1990: Q 4 = i o o I n d e x 1990: Q 4 = io o 104 A A * i / ^ -v k / ---- v / l / ' K « \ 102 \ / 1 00 f t t j I \ \ \> Road vehicles / J --J•J % 1/ / I 4.... / 1 \ \ \ \ . _____ i ____ 98 ....... 96 1 1 1 1 1 ............i \ Primary metals / y 94 _ Jl j ___ 1 1991 _ : L I -i 92 > :£ ll ! j 93 1 94 I 1 Notes: Chart plots the ratio of import prices to the U.S. producer price index for each industry. Chart 5 Im p o r t M a r k e t S h a r e : S e l e c t e d In d u s t r ie s P ercen t Percen t 35 30 \ 1 V 25 20 15 \ \ / ■» V /St... V 10 \r % Road vehicles / '\ J I ^ / * i * *\ •J \ «% i \ 1 K I MI I ! 1986 ¥ Primary me tals 1 1 1 1 111 87 89 _ L J J !..i L i ’-': *112. Li. l..J_ I I I 92 94 93 90 91 Notes: Chart plots the ratio of imports to domestic demand. Domestic demand is defined as shipments plus imports less exports. I n B r ie f F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 105 E ndno tes 1. The threat of a surge in imports could also cause workers to hold back wage demands or firms to hold off price increases. Note 5 continued 10 percent dollar depreciation would raise import prices by roughly 6 to 7 percent. 2. Imported consumer goods account for roughly 8 percent of U.S. con sumption expenditures. 6. Between the end of 1990 and m id-1994, the dollar appreciated 12 percent against the deutsche mark, 10 percent against the French franc, and more 3. Phillips curve models of inflation use measures of the level or change in than 20 percent against the currencies o f Italy and the United Kingdom. the degree of resource utilization as one factor explaining price movements; other factors include changes in energy and raw materials prices. 7. Assuming imports account for roughly 8 percent o f total consumption expenditures, this 1 percent decline in the growth of import prices relative 4. Producer price inflation in Italy and the United Kingdom has exceeded to the growth of domestic prices since 1990 would have had a negligible U.S. producer price inflation since the end of 1990, while producer price impact on U.S. inflation, slowing price growth by less than 0.1 percentage inflation in France and Germany has trailed U.S. inflation over the same point over 1990-94. period. 8. Capacity utilization rates in these industries were below 1975-80 peak 5. Most model estimates show that a change in foreign producer prices rates but near or above 1983-90 peaks. Slower auto output in second-quar- changes the prices of U.S. imports by roughly an equal amount; that is, a ter 1994 did reduce capacity utilization 9 percentage points from its high 10 percent drop in prices abroad would cause a 10 percent decline in U.S. first-quarter rate, but the rate is still within 2 percentage points of its import prices. However, only about 60 to 70 percent of a change in nomi 1983—90 peak. Foreign capacity utilization rates for these industries are nal exchange rates is estimated to be passed through to import prices, so a not available. Digitized1 for 0 6 FRASER F R B N Y Q u a r t e r l y R e v i e w / S u m m e r - F a ll 1994 N otes Recent Trends in the Profitability of Credit Card Banks Andrea Meyercord card lending has traditionally been a highly average annual return on assets. Chart 1 traces the weighted profitable line of business for banks. Increased competitive average return on assets over the years 1989 to 1994 for all pressures in recent years, however, have prompted many card credit card banks with total assets of $200 million or more.2 issuers to reduce interest rates, lower or waive annual mem The data through 1993 measure the returns as of year-end; bership fees, provide program enhancements, and offer rebate the 1994 return is estimated by annualizing first-quarter fig programs to make their plans more attractive. Despite these ures. The chart shows that the average return on assets for developments, the profitability of “credit card banks,” or credit card banks has been relatively high over the years banks specializing in credit card operations,1 not only examined. As a point of contrast, whereas the average return remains high relative to the rest of the banking industry but for credit card banks ranged from 1.9 percent to 3.4 percent also continues to grow. during the 1989-93 period, all other U.S. commercial banks This article analyzes recent trends in credit card within the same asset size category had an average return on bank profitability and the factors underlying them. Evaluat assets ranging from 0.4 percent to 1.1 percent in the same ing the trends only in the aggregate, however, would ignore period. some interesting differences in profitability between two dis Chart 1 also reveals that the average return on assets tinct groups of institutions. The first group consists of credit for credit card banks has been steadily increasing since 1991. card banks owned by bank holding companies (BHCs), a rel This rising trend in profitability is particularly notable given atively well-established market segment. The second group that the spread between card issuers’ lending rates and their comprises credit card banks owned by nonbank firms, a more funding costs has been shrinking since early 1992. Compar recent and fast-growing component of the credit card market. ing the average credit card rate of U.S. card issuers3 with the The article finds that despite growing competition one-year Treasury note rate— a conservative benchmark for from nonbank-owned credit card issuers, the return on assets issuers’ cost of funds— indicates that, overall, issuers’ mar of the more “traditional” issuers— those owned by BHCs— gins decreased from 14.1 percent in February 1992 to 12.7 increased significantly over 1992-93 and continues to exceed percent in February 1994. Although both lending rates and that of nonbank-owned credit card banks. BHC-owned insti funding costs fell over this period, the decline in lending tutions have proved more profitable than their nonbank- rates exceeded the decline in funding costs, resulting in a owned counterparts largely because better asset quality contraction in the overall funding margin. trends in recent years have enabled them to maintain a lower level of provisions for loan losses. P r o f it a b il it y a t B H C - o w n e d a n d N o n b a n k - o w n e d C r e d it C a r d b a n k s A g g r e g a t e P r o f it a b il it y T r e n d s o f The aggregate trends presented in Chart 1 conceal some C r e d it C a r d B a n k s notable differences between the levels of profitability exhib The profitability of credit card banks can be measured as an ited by BHC-owned credit card banks and those observed for Digitized I n B rfor ie f FRASER FRBNY Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 107 nonbank-owned credit card banks. Such differences are of credit card issuers (ranked on the basis of total credit card particular interest since these two groups are direct competi loans outstanding at the credit card subsidiaries of BHCs, tors in the credit card business. thrifts, and nonbank diversified financial services companies), Increased competition from nonbank-owned issuers the share of outstandings held by nonbank-owned issuers is a recent and important development in the credit card mar grew from approximately 24 percent in 1991 to 37 percent in ket. Although the Bank Holding Company Act of 1956 pro 1993 (Chart 2). In part, this trend reflects the recent entry hibited nonbank companies from owning banks, in the early into the credit card market of firms like AT&T, General Elec 1980s several nonfinancial firms found that they could con tric Capital Corporation, First USA, ADVANTA Corpora duct credit card business by acquiring so-called nonbank tion, and Ford Motor Company. The number of such non banks. Since nonbank banks limit their operations to either bank-owned institutions figuring among the twenty-five deposit taking or lending, they did not legally meet the Bank largest issuers rose significantly around this time, increasing Holding Company Act’s definition of a “bank” as an institu from five in 1990 to ten in 1993. In addition to making tion that engages in both activities, and thus were not subject entry-related gains, nonbank-owned issuers have also cap to the act’s restrictions on bank ownership. The Competitive tured market share through increased lending, both by origi Equality Banking Act of 1987 addressed this exception by nating credit card loans themselves and by acquiring the amending the definition of a “bank” and banning new non credit card portfolios of other issuers. bank bank charters. However, the 1987 act exempted credit Decomposing the data for all credit card banks card banks and a few other special purpose banks from the reveals a difference in the general level of profitability of the new definition of a “bank.” Thus, since 1987, the number of BHC-owned and nonbank-owned subsets. Chart 3 adds to nonbank-owned commercial banks that specialize in credit the aggregate trend shown in Chart 1 two more series depict card lending has surged.4 ing the profitability trends of these subsets of the credit card Over the past few years, nonbank-owned issuers bank population.5 Despite the competitive challenges posed have gained substantial market share at the expense of more by nonbank-owned issuers, the average return on assets for traditional BHC-owned issuers. Among the top twenty-five Chart 2 M a rket Share Chart 1 A ggregate T r e n d s in C r e d it C a r d B a n k P r o f it a b il it y Per c en t of N o nbank C r e d it C a r d Is s u e r s o f o u t s t a n d in g s 40 ----------------------------------------------------------------------------------Per c en t 5 ----------------------------------------------------------------------------------- i i_________ ! _______ i__________________i---------------------------- 1----------------------------1---------------------------- 1--------------------------- 1989 90 91 92 93 94e Source: Federal Financial Institutions Examination Council, Reports of Condition and Income. Notes: Chart shows the weighted average return on assets for all credit card banks with total assets of $200 million or more. The data through 1993 measure the returns as of year-end; the 1994 return is estimated by annualizing first-quarter figures. Digitized108 for FRASER F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 1988 89 90 91 92 93 Source: American Banker. Notes: Market share figures are based on the total credit card loans out standing at the commercial bank and thrift subsidiaries of the top twentyfive credit card issuers. The data for 1991-93 include securitized credit card loan figures derived from the Reports of Condition and Income. Banks began reporting securitization figures in the September 1991 Call Reports. I n B r ief BHC-owned credit card banks was 4.1 percent in 1993, the average return on assets of the nonbank-owned subset in almost double the level reported by nonbank-owned credit 1994 drops from 4.1 percent to 2.9 percent. card banks. Moreover, the returns of BHC-owned institu The primary factor driving the higher level of pro tions grew by 1.3 percentage points from 1992 to 1993, a fitability exhibited by the BHC-owned segment has been a larger rise than the increase exhibited by nonbank-owned significantly lower level of provisions for loan losses relative issuers. First-quarter 1994 results suggest that the average to the nonbank-owned subset. Chart 4 tracks provisions as a return on assets for BHC-owned credit card banks remains percent of total loans from 1989 to 1994 for both BHC- higher than that for nonbank-owned institutions. owned credit card banks and nonbank-owned credit card Chart 3 includes a fourth series describing the banks. Although the ratio of provisions to loans has followed returns of all nonbank-owned credit card banks except the a similar trend at both subsets, peaking in 1992 and subse Greenwood Trust Company, a sizable institution that largely quently falling, this ratio has consistently been higher at the drives the results of this subset. This series demonstrates that nonbank-owned subset in the time period examined. In although the average returns of nonbank-owned credit card 1993, the provisions ratios of BHC-owned credit card banks banks appear to have increased substantially from the end of and of their nonbank-owned counterparts were 2.9 percent 1993 (and by more than the increase exhibited by BHC- and 5.5 percent, respectively. First-quarter data for 1994 owned credit card banks), most of this trend is actually indicate that provisioning remains significantly higher at the attributable to a marked increase in the returns of Greenwood nonbank-owned subset. Underlying the provisioning behavior of BH C- Trust Company.6 If Greenwood Trust Company is excluded, owned credit card banks in recent years are more favorable loan quality and loan loss trends. The BHC-owned subset has faced a smaller percentage of past-due loans and net chargeChart 3 offs relative to total loans than have nonbank-owned credit S u bsa m ple T r e n d s in C r e d it C a r d B a n k P r o f it a b il it y card banks since 1991 (see table on next page). One possible Percen t 4.5 BHC-owned Chart 4 P r o v is io n s as a Percen tage C r e d it C a r d B a n k s and at of N T o ta l Lo an s o n b a n k -o w n e d at B H C -o w n ed C r e d it C ard B a n k s P ercen t 8 1.0 I_____________ I_____________ I_____________ 1_____________ I_____________ I_____________ I 1989 90 91 92 93 94e Source: Federal Financial Institutions Examination Council, Reports of Condition and Income. Notes: Chart shows the weighted average return on assets for all credit card banks with total assets of $200 million or more. The data through 1993 measure the returns as of year-end; the 1994 return is estimated by annualizing first-quarter figures. Most of the increase in returns exhibited by nonbank-owned credit card banks from 1993 to 1994 was driven by Greenwood Trust Company, the second largest institution within this subset in terms of assets. A series has been added to show the return on assets for nonbank-owned credit card banks excluding Greenwood Trust Company. I n B r ief 2 i____________ i____________ l____________ 1____________ I____________ I____________ I 1989 90 91 92 93 94e Source: Federal Financial Institutions Examination Council, Reports of Condition and Income. Notes: The data through 1993 measure provisions as of year-end; the 1994 value is estimated by annualizing first-quarter figures. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 109 reason for the more severe credit quality problems of the non “bank issuers are continuing to lose share to nonbank players bank-owned issuers is that in pursuit of market share, these [because] many banks continue to lack a differentiated con institutions may have marketed their credit card accounts sumer status with their cards”; in contrast, nonbank issuers more aggressively by, for example, offering better terms. such as ADVANTA, American Express, and Dean Witter, Consequently, they may be attracting a greater proportion of Discover and Company have “aggressive pricing/balance high-credit-risk customers than are BHC-owned issuers. transfer/credit line product offerings.” Other research (Can- Although an analysis of customer risk characteristics ner and Luckett 1992) suggests that “nonbank firms have at BHC-owned and nonbank-owned credit card banks is garnered significant market shares, in part by differentiating beyond the scope of this article, some anecdotal evidence sup their plans by forgoing annual fees or by offering rebates on ports the idea that nonbank-owned issuers may be pursuing a purchases or discounts on selected services.” Finally, a recent more aggressive strategy to attract customers. For example, Report to Congressional Requesters by the U.S. General one set of market analysts (Liss and Rhei 1994) argues that Accounting Office (1994) contends that “nonbank entrants may be more comfortable than depository institutions with P a st - D u e L o a n s T otal Lo a n s at and N et C h a r g e -o f f s 1993 1994 Percen tage of C r e d it C a r d B a n k s Past-Due Loans NonbankBHCowned owned (Percent) (Percent) 1989 1990 1991 1992 as a 4.78 5.25 6.19 5.17 4.29 4.18 7.42 6.68 5.14 4.23 3.56 3.34 lower interest rates and earnings because their primary reason for issuing credit cards was to increase volume in their tradi Net Char#e-offs NonbankBHCowned owned (Percent) (Percent) 2.93 4.42 5.08 6.05 3.89 3.49 3.27 3.60 4.20 4.47 3.19 3.40 Source: Federal Financial Institutions Examination Council, R eports o f Condition and Incom e. Note: The 1994 past-due loan values are first-quarter figures. The 1994 net charge-off values are estim ated by annualizing first-quarter figures. 110 F R B N Y Q u a r t e r l y R e v i e w / S u m m e r - F a ll 1994 tional lines of business.” In sum, notable differences in the levels of profitabil ity at credit card banks exist on the basis of ownership struc ture. A comparison along these lines suggests that growing competition from nonbank-owned issuers has not yet signifi cantly curtailed the profitability of BHC-owned credit card banks, which has continued to rise over the past two years. As nonbank issuers evolve into more mature, established players in the credit card market, they may pose a more serious com petitive threat to BHC-owned institutions. I n B r ie f E ndno tes 1. This analysis defines a “credit card bank” as an institution allocating 90 Note 4 continued percent or more of its total loan portfolio to credit card loans. Because the After the act was passed, the number of nonbank-owned credit card banks Federal Financial Institutions Examination Council s Reports of Condition (commercial banks) roughly doubled, to twenty-six banks. and Income do not break out all expenses and revenues by specific product lines, this definition allows one to use the reports to assess the profitability 5. The BHC-owned subset ranges in size from thirteen to seventeen insti of banks established primarily to issue and service credit card accounts. tutions from 1989 to 1994. The nonbank-owned subset ranges in size from nine to seventeen institutions over the same period and includes both non 2. The returns are calculated using on-balance-sheet assets. The sample bank banks and commercial banks. ranges in size from tw enty-four to thirty credit card banks over the 1989-94 period. 6. Greenwood Trust Company is a $6.7 billion nonbank bank that issues 3. The average rate is reported in Federal Reserve Statistical Release G. 19, accounts for roughly a quarter o f the total loans held by the nonbank- a quarterly survey of the most common rates charged by a sample of com owned credit card bank subset. The increase in Greenwood Trust Compa the Discover Card for Dean Witter, Discover and Company and that alone mercial banks during the first week of each midquarter month. ny’s return on assets from the end o f 1993 to the first quarter o f 1994 stemmed from a 72.6 percent rise in net income (driven primarily by an 4. Before the passage o f the Com petitive Equality Banking Act, only increase in noninterest income) and a 24.2 percent drop in total assets twelve credit card banks (nonbank banks) were owned by nonbank firms. (reflecting a high volume of loans sold in the first quarter of 1994). R eferen ces Canner, Glenn B., and Charles A. Luckett. 1992. “Developments in the Pric U.S. General Accounting Office. 1994. U .S . C r e d it C a r d I n d u s t r y : C o m ing of Credit Card Services.” F e d e r a l R e ser v e B u l l e t in 78 (Septem p e t it iv e ber): 652-66. C ongressional R equesters G A O /G G D -9 4 -2 3 . W ashington, D .C .: D e v e l o p m e n t s N e e d t o B e C lo se ly M o n i t o r e d . Report to GPO, April. Liss, Samuel G., and Mary A. Rhei. 1994. “Credit Card Issuers - Compara tive Look at the Public Proxies.” U n it e d S t a te s E q u i t y R e s e a r c h . Salomon Brothers, Financial Services, January 4. N o tes FRBNY Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 111 Regional Employment Trends in the Second District Charles Steindel and Lois Banks It is well known that growth in the Northeast, including manufacturing employment in each. Taken as a whole, aggre the Second Federal Reserve District, has trailed that in the gate payroll em ploym ent in these m etropolitan areas nation as a whole over the last several years. What is less often accounts for more than 90 percent of that in the District as a recognized is that while some parts of the Northeast continue whole. to lag the nation, other parts, including sections of the Sec In general, job growth in the District has not kept ond District, are now growing close to or in line with the pace with national trends,1 and in some important regions national average. This tendency to overlook movements in such as Rochester and Long Island, employment has recently smaller areas is perhaps especially marked in the Second Dis faltered. Some signs of progress have emerged, however: trict, where discussions of the regional economy frequently employment has risen throughout most of the District, and in emphasize New York City. Although New York City com northern New Jersey and Albany, the gaps with the national mands attention as the major employment and population data have been either small or narrowing. center of the Second District, the economy of the District is far more than the city writ large. N e w Y o r k C it y This article attempts a more balanced assessment of The city’s economy has improved somewhat over the last economic developments in the District by considering New year. Employment finally bottomed out in m id-1993, and York City alongside three other important regions. We focus year-to-year gains in jobs moved up to the neighborhood of on job growth, giving particular attention to the factors that 0.5 percent in early 1994 (Chart 1). Recent job growth, how have given rise to diverse employment trends within the Dis ever, falls far short of that in the nation as a whole and only trict. Our analysis draws on the Bureau of Labor Statistics’ begins to counter the city’s losses since 1989- In June 1994, nonfarm payroll employment series, a source notable for for instance, payroll employment in the city was 3.3 m il comprehensive and timely data at the local level. lion— 8 percent less than its 1989 high. In addition to New York City, the large regions The unemployment rate in New York City has examined are 1) the environs of New York City, consisting of recently declined more rapidly than in the nation as a whole. Long Island, the northern metropolitan area, and northern The city’s rate fell from 11.4 percent in the first quarter of 1993 New Jersey; 2) midstate New York— the large region stretch to 9 percent in the second quarter of 1994, while the national ing north from the city suburbs to the Canadian border and rate dropped from 6.9 percent to 6.2 percent over the same west past Syracuse; and 3) Western New York (Figure 1). period. A shrinking labor force, however, has contributed sig Since developments in the agricultural sector contribute only nificantly to the city’s declining unemployment rate. modestly to overall employment trends in the heavily urban The city’s recent employment gains have also been ized Second District, we focus on the major metropolitan unevenly distributed across industries. M anufacturing areas in the four regions. The table lists these areas and gives employment, which still accounts for about 9 percent of the the 1993 size of aggregate nonfarm payroll employment and city’s job total, has continued falling at a moderate pace 112 F R B N Y Q u a r t e r l y R e v i e w / S u m m e r - F a ll 1994 I n B r ie f (Chart 2). Encouragingly, however, changes in manufactur Figure 1 ing employment in the city over the last year more closely R e g io n s o f t h e Seco nd D is t r ic t matched national developments rather than trailing them badly, as had been the case for a number of years. Thus, the manufacturing sector is currently not a source of weakness for the city relative to the nation. (See Moss 1994 for a discussion of recent developments in the city’s manufacturing sector.) A large portion of the recent payroll growth in the city can be traced to the securities sector and to health, social, and busi ness services. But with the city’s employment recovery just starting to take root and advance beyond Wall Street, reports that some early 1994 trading losses were leading securities firms to reevaluate expansion plans or cut staff (Raghavan 1994) suggest that the city is still some distance from a fullfledged expansion. N e w Y o r k C ity E n v ir o n s 2 The region surrounding New York City can be subdivided into three segments: Long Island; the northern metropolitan C o m p o s it io n of Pa yro ll E m plo y m en t , 1 9 9 3 T h o u sa n d s United States New York City New York City Environs Northern Metropolitan Area* Nassau-Suffolk MSA (Long Island) Northern New Jerseyf Total Manufacturing Percentage of Total Jobs in Manufacturing 110521 18005 16 3275 290 9 672 90 13 1054 2287 122 388 12 17 Chart 1 T r e n d s in N e w Y o r k C it y P a y r o l l E m p l o y m e n t Twelve-Month Percent Changes Percen t United States S Midstate New York Dutchess County MSA (Poughkeepsie) Binghamton MSA Utica-Rome MSA Syracuse MSA Albany-Schenectady-Troy MSA 104 113 124 330 426 18 27 21 50 44 18 24 17 15 10 Western New York Rochester MSA Buffalo-Niagara Falls MSA 515 528 130 90 25 17 Note: MSAs are m etropolitan statistical areas. *C om prises Putnam , Rockland, and W estchester counties and the Stam ford-Norw alk m etropolitan statistical area. tC om prises the N ew ark, Bergen-Passaic, Middlesex-Som erset-Hunterdon, and Jersey City m etropolitan statistical areas. In B r ie f f 4 New York City 1 /- V '~ s J 11 II 1 1 1 1 1 1 1 ........................M l 1991 _L1 1 i 1 1 1 ! ! 1 1 ............... 92 93 94 Sources: U.S. Department of Labor, Bureau of Labor Statistics; state employment security agencies. Note: Chart shows data through June 1994. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 113 area of Rockland, Westchester, Orange, and Putnam counties ation by smaller firms has largely offset cutbacks by large in New York and Fairfield County in Connecticut; and the corporations. twelve counties of northern New Jersey included in the Sec Retail employment has risen markedly in most parts ond District (Figure 2). Taken as a whole, the environs of of the region. Much of this increase stems, of course, from New York City have a considerably larger population and even employ more people than the city proper. In fact, the region contains four metropolitan statistical areas (MSAs) Figure 2 E n v ir o n s with 1990 populations above 1 million.3 of N ew Y o r k C it y Recent employment trends have diverged across the three segments of this region. In the spring of 1994, overall job growth on Long Island and in the northern metropolitan area actually turned negative on a year-to-year basis, while northern New Jersey growth maintained a pace only moder ately below the national norm (Chart 3)- These differences in part reflect differences in the manufacturing sector. Jo b cuts— some attributable to corporate restructuring, others to layoffs at defense contractors— continue to be significant at Long Island and northern metropolitan manufacturers (Chart 4). Northern New Jersey, by contrast, has experienced fewer manufacturing job losses, apparently because job creChart 3 Chart 2 T T r e n d s in N ew Y ork r e n d s in Pa yro ll E m plo y m en t in N ew Y o r k C it y E n v ir o n s Twelve-Month Percent Changes C it y M a n u f a c t u r in g Pa yro ll E m plo y m en t Twelve-Month Percent Changes Percen t Percen t Northern New Jersey Uni ted States r '- * / ,—' ^ Nrw York City — r / / ✓— / v i t 1 1 —s' 111111111111 i / A r i 1 M M ... ............... 1991 Northern metropolitan area */ s-jfjr *» C *-+ Long Island / A f ' * United States 1991 11111111111 11111111111 92 93 i i i i i i i i i i i i i i i i i i i i i i 92 93 111111 94 Sources: U.S. Department of Labor, Bureau of Labor Statistics; state employment security agencies. 1 1 .1.1 1 J 94 Sources: U.S. Department of Labor, Bureau of Labor Statistics; state employment security agencies. Note: Chart shows data through June 1994. Digitized 114for FRASER F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 Note: Northern New Jersey comprises the Newark, Bergen-Passaic, Middlesex-Somerset-Hunterdon, and Jersey City metropolitan statistical areas. Long Island refers to the Nassau-Suffolk metropolitan statistical area. The northern metropolitan area comprises Putnam, Rockland, and Westchester counties and the Stamford-Norwalk metropolitan statistical area. Chart shows data through June 1994. I n B r ief economic recovery and increased household spending. In those in other parts of the District. In 1990, the largest mid addition, several parts of the region have recently seen the state MSA, Albany-Schenectady-Troy, had a population of introduction of large “warehouse” retail outlets of national only 861,000, placing it ninth among the District MSAs. chains. Although this restructuring of the retail sector may The employment picture has varied sharply around have no permanent positive effect on employment, the con the region. The Dutchess County MSA (Poughkeepsie), struction and staffing of the new stores have generated jobs exemplifying the mid-Hudson Valley area, and the more over the short term. westerly Binghamton MSA are faring quite poorly (Chart 5), largely because corporate restructuring moves have depressed M N id s t a t e Y ew manufacturing employment (Chart 6).5 The pace of job loss o rk This large region (Figure 3), dominated by the Catskills and in the Poughkeepsie area, however, did moderate somewhat Adirondacks, resists generalization.4 The metropolitan areas in the spring of 1994. The area along the Mohawk River west in the region are dispersed along the lakes and rivers sur to Lake Ontario— here illustrated by data for the Syracuse rounding the mountains. Nevertheless, discussing this and Utica-Rome MSAs— shows somewhat more strength. region as a unit rather than treating each MSA separately Syracuse employment has been flat, while Utica job growth makes some sense given the small size of the MSAs relative to has recently moved in line with the national trend. Scheduled job cuts at a major military installation in the Utica-Rome MSA, however, will hurt that labor market. Manufacturing Chart 4 employment, though not rising, has held up better in Syra T rends in M a n u f a c t u r in g P a y r o l l E m p l o y m e n t in cuse and Utica than in the more southerly parts of the region, N e w Y o r k C it y E n v ir o n s Twelve-Month Percent Changes Percent Figure 3 Unit ed States M Northern New Jersey \ \ I V A id st a t e N ew Y ork < t » •* t r* * /" 1 4 I Northern metropolitan area Long Island j f V y ! J ' 'f ! 1 i : ! i ! 1 1 1 1 1 II 1 1 1 1 1991 92 . .. . . . . . . 1 1 1 1 II j.1 93 94 Sources: U.S. Department of Labor, Bureau of Labor Statistics; state employment security agencies. Note: Northern New Jersey comprises the Newark, Bergen-Passaic, Middlesex Somerset-Hunterdon, and Jersey City metropolitan statistical areas. Long Island refers to the Nassau-Suffolk metropolitan statistical area. The northern metro politan area comprises Putnam, Rockland, and Westchester counties and the Stamford-Norwalk metropolitan statistical area. Chart shows data through June 1994. I n B r ie f F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 115 and some service industries, such as wholesale and retail area have been recent increases in employment by both the trade, have been growing. state and the local government. The strongest portion of the midstate region, how ever, is probably the Albany MSA. O f all the major parts of W the Second District, Albany alone consistently experienced We define Western New York as the fourteen counties in the job growth ahead of the national pace from early 1993 Buffalo Branch area of the Second District (Figure 4). The through April 1994. The apparent slippage in Albany’s year- economy of Western New York is dominated by the large to-year job gains in May and June of 1994 may simply reflect Buffalo and Rochester MSAs. The two areas have fared some comparisons with unusually strong months in 1993, rather what differently in recent years, and their economies tend to than a recent deterioration in that area. M anufacturing belie their popular reputations. Buffalo, like many other employment in the Albany area has proven more resilient large cities along the Great Lakes, has gradually transformed than in some other parts of the midstate area, such as Bing its economy from one based on durable goods manufacturing hamton and Dutchess County, although manufacturing is to a more diversified, service-centered one, though manufac clearly not an important source of job growth in Albany. turing remains an important sector (the early stages of this Employment growth in the Albany area has been centered in transformation were described in Doolittle 1985-86). A spe health care, retailing (with most months in early 1994 seeing cial factor aiding Buffalo in this transition was the U.S.- ester n N ew Y o rk job gains in this sector in the 3 to 5 percent range on a yearto-year basis), and the diversified service sector. The develop ment of new large retail outlets has been especially pro nounced in the Albany MSA. Adding to the growth of this Chart 6 T r e n d s in M id s t a t e N ew Y o r k M a n u f a c t u r in g Payro ll Em plo ym en t Twelve-Month Percent Changes Percen t Chart 5 T r e n d s in M id s t a t e N Twelve-Month Percent Changes ew Y o rk Pa yro ll E m plo y m en t Percen t 0 -2 -4 Poughkeepsie ................Ml 93 Sources: U.S. Department of Labor, Bureau of Labor Statistics; state employment security agencies. Sources: U.S. Department of Labor, Bureau of Labor Statistics; state employment security agencies. Note: Chart shows data through June 1994. Note: Chart shows data through June 1994. F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 116 I n B r ief Canada Free Trade Agreement of 1989, which encouraged Figure 4 Canadian firms to locate operations in the relatively low-cost W ester n N ew Y ork Buffalo area and Canadian shoppers to patronize the area’s stores. Subsequently, Buffalo had a slow exit from the national recession (Chart 7), partly because the decline of the Canadian dollar starting in the second half of 1991 weak ened the incentives for Canadians to shop and set up business in the Buffalo area. More recently, a leveling-off in manufac turing employment (Chart 8) and the stimulus provided by new retail outlets have helped job growth to resume in the Buffalo MSA, though retail sales to Canadian buyers still appear to be weak (Linstedt 1994). Buffalo is today primarily service-oriented, but it is still perceived as a blue-collar town. In contrast, Rochester, often considered a white-collar city, has an extraordinary 25 percent of its MSA payroll employment in manufacturing! O f course, many of these m anufacturing jobs are highincome ones in the headquarters and research and develop ment units of m ajor high-tech corporations. Recently, responding in part to intensified foreign competition, several of these firms have undertaken sizable restructuring opera tions and have been reducing their employment in the Rochester area. The deterioration in manufacturing employ Chart 8 ment has contributed to an erosion in overall job growth in T r e n d s in W ester n N ew Y o r k M a n u f a c t u r in g Payro ll Em plo y m en t Twelve-Month Percent Changes P ercen t Chart 7 T r e n d s in W e s t e r n N Twelve-Month Percent Changes ew Y o rk Payro ll Em plo y m en t 1 « 1 1 I 1 4 1 United States Rochester 2 \ 0 'J -4 ^ D 1 1 /— * ^ Buffalo X. o 1 1 I I 1 1 1 i 1 i 1 1 1 1 1 1 1 1 1 1 i 1 92 i i i i i i i / V !| y / V 1991 / > I J i i i i 93 i i i 94 i i ...................M U 1991 4 1 1 1 1 1 M M1 1 92 1 1 1 1 II 1 1 1 1 1 93 Sources: U.S. Department of Labor, Bureau of Labor Statistics; state employment security agencies. Sources: U.S. Department of Labor, Bureau of Labor Statistics; state employment security agencies. Note: Chart shows data through June 1994. Note: Chart shows data through June 1994. I n B r ief Buffalo / \ / \ f\ * j A \ Q ' / v f \ < W\ S \' n f c r ° ~ l » f —* \ / rf\ ff b /\ \ -2 V 1 1 United States 4 Percen t F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 111111 94 117 Rochester; by the middle of 1994, the growth rate was barely A second factor that has contributed to the differential experi positive. To date, segments of Rochester’s service sector such ence of areas within the District is the performance of the ser as business, health, and retailing have expanded sufficiently vice sector. New York City’s financial sector dominates the to offset the increased drag from manufacturing. District’s service economy, and job gains in this industry have been a particular spur to the city’s economy. Service employ C o n c l u s io n ment in other areas, most notably the outskirts of New York Employment trends have differed across regions and metro City and the Albany MSA, has benefited from new large-scale politan areas in the Second District. The Binghamton and operations in retailing. Because the warehouse-type chains Poughkeepsie areas continue to experience significant job have been relatively late coming to the Northeast, hiring by losses, and growth has faltered on Long Island, in the north these outlets will probably continue to be stronger in the Dis ern metropolitan area, and in the Rochester and Syracuse trict than in other parts of the nation. MS As. New York City and the Buffalo MSA, by contrast, are Despite the regional differences detailed in this arti seeing modest but real job growth, and the job gains in cle, the Second District as a whole shows some signs of recov northern New Jersey and the Albany and Utica areas have ery (Rosen and Wenninger 1994). Although job growth in been quite respectable. the District continued to trail national trends through the Developments in manufacturing to some extent middle of 1994, aggregate employment has been expanding. explain these differing employment patterns. Areas that con The areas in the District that are lagging are mainly those tinue to face substantial cutbacks in manufacturing employ feeling the effects of corporate restructuring and defense cuts. m ent— Long Island, the northern m etropolitan area, While these ills will continue to be felt in the short run, over Binghamton, Poughkeepsie, and Rochester— have tended to the medium and longer term they should abate, setting the trail much of the rest of the District and the nation as a whole. stage for healthier growth. 118 F R B N Y Q u a r t e r l y R e v i e w / S u m m e r - F a ll 1994 I n B r ie f Endno tes 1. Comparing District and national trends is complicated by the inclusion Note 2 continued of an explicit “bias adjustment” in the national data for the period since on Long Island— have little to do with short- or medium-term develop March 1993- This adjustment, designed to account for job growth at new ments in New York City. firms not yet included in the monthly payroll employment survey, current ly adds about one percentage point to the national payroll annual growth 3. The MSAs are Newark, Bergen-Passaic, and Middlesex-Somerset-Hun- rate. Since the data from New York and New Jersey do not include such terdon— all in New Jersey— and Nassau-Suffolk in New York. These areas explicit bias adjustments, it is conceivable that recent D istrict employ plus the Monmouth-Ocean MSA in New Jersey and the bulk of the north ment trends are being understated (assuming the national bias adjustment ern metropolitan area are included in the “New York Consolidated M SA.” is roughly correct, as is usually the case). Nevertheless, District employ ment data are modified to reflect some information about hiring at new 4. It also lacks an accepted name. The obvious term “upstate” includes firms and job counts from ongoing unemployment insurance tax reports, so Western New York, a region treated separately in this article. Another log the understatement is probably small. ical choice is “central New York,” but this term is used by the New York State Labor Department to refer more narrowly to the area around Syracuse. 2. We use the term “environs” rather than “suburbs” to denote this region both because significant parts of it— for instance, Newark, New Jersey, the 5. The Census Bureau includes Dutchess County in the large New York north fork o f Long Island, and the Delaware Water Gap— are hardly “sub Consolidated MSA. The significance of the sharp drop in manufacturing urban” in appearance in any ordinary sense o f the word, and, more funda employment for the Hudson Valley would be obscured, however, if we were mentally, because certain key sectors of the region’s economy— for exam to consider Dutchess County a part of the northern New York City suburbs. ple, the pharmaceutical industry in New Jersey and the aerospace industry R eferen ces Doolittle, FredC. 1985-86. “A djustm ents in Buffalo’s Labor M arket.” F e d e r a l R e s e r v e B a n k o f N e w Y o r k Q u a r t e r l y R e v i e w 10 Raghavan, Anita. 1994. “Bear Market Ax Again Hits Wall Street.” W a ll S t r e e t J o u r n a l , August 9- (Winter): 28-37. Rosen, Rae, andJohn Wenninger. 1994. “Second District Update: A Moderate Linstedt, Sharon. 1994. “Canadian Shoppers Are Staying Home.” BUFFALO N e w s , Ju ly 17. Recovery is in Progress.” F e d e r a l R e ser v e B a n k terly of N ew Y o r k Q u a r R e v ie w 19 (Spring): 45-53. Moss,-Mitchell L. 1994. “Made in New York: The Future o f Manufacturing in New York City.” U rban Research Center, N ew York U niversity (August). N o tes F R B N Y Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 119 Fu rth er R R ecent e a d in g FRBNY U n p u b lis h e d R e s e a r c h P a p e r s 9331. Malz, Allan M. 9412. De Kock, Gabriel, a n d Thomas DeLeire. N ew V a r ie t ie s o f F o r e ig n C u r r e n c y O p t io n s . T h e R o le o f t h e E x c h a n g e R a te in t h e M o n e t a r y December 1993. T r a n s m is s io n M e c h a n is m : A T im e -S e r ie s A n a l y s is . 9407. Akhtar, M.A. M o n e t a r y P o l ic y E f fe c t s o n L o n g - T e r m I n t e r e s t A u g u st 1994. 9413. Femald, Ju lia , Frank Keane, an d M artin Mair. R a t e s : A C r it ic a l S u r v e y o f t h e E m p ir ic a l T h e M a r k e t f o r C o l l a t e r a l iz e d M o r t g a g e L it e r a t u r e . July 1994. O b l ig a t io n s (C M O s ). S e p te m b e r 1994. 9408. Cohen, G eraldD ., an dJoh n Wenninger. 9414. Chang, P.H. Kevin, a n d Carol Osier. T h e R e l a t io n s h ip b e t w e e n t h e S p r e a d a n d t h e E v a l u a t in g C h a r t - B a se d T e c h n ic a l A n a l y s is : F u n d s R a t e . July 1994. T h e H e a d -a n d - S h o u l d e r Pa t t e r n in F o r e ig n 9409- Steindel, Charles. A s s e s s in g R e c e n t T r e n d s in M a n u f a c t u r in g . J u l y 1994. 9410. Blomberg, S. Brock, an d G eraldD . Cohen. S c o r in g P o l it ic a l E c o n o m y M o d e l s : A M u l t ipl e E q u il ib r ia A p p r o a c h . July 1994. 9411. Femald, Ju lia , Frank Keane, an d Patricia C. Mosser. E x c h a n g e M a r k e t s . S e p te m b e r 1994. 9415. Estrella, Arturo, D arryll Hendricks, J o h n Kambhu, Soo Shin, an d Stefan Walter. O p t io n s P o s it io n s : R is k M e a s u r e m e n t a n d C a pita l R e q u ir e m e n t s . S e p te m b e r 1994. 9416. Clark, John. T he Str u ctu re, G ro w th , and R ecent P erfo r L a t in A m e r ic a n B o n d M a r k e t . M o r t g a g e S e c u r it y H e d g in g a n d t h e Y ie l d m ance of the C u r v e . August 1994. S e p te m b e r 1994. FRBNY Q u a r t e r l y R e v ie w / S u m m e r - F a l l 1994 120 Single copies of these papers are available upon request. Write Research Papers, Public Information Department, Federal Reserve Bank of New York, 33 Liberty Street, New York, N.Y., 10045. F u r t h e r R e a d in g T he Q uarterly The F ederal R Q R eser v e Bank of N ew Y ork e v ie w uarterly R e v ie w is published by the Research and Market Analysis Group of the Federal Reserve Bank of New York. The views expressed in the articles are those of the indi vidual authors and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. Single copy subscriptions to the Q u a r t e r l y R e v ie w (ISSN 0147-6580) are free. M ultiple copies are available for an annual cost of $12 for each additional subscription. Checks should be made payable in U.S. dollars to the Federal Reserve Bank of New York and sent to the Public Information Department, 33 Liberty Street, New York, N Y 10045-0001 (212 720-6134). Single and multiple copies for U.S. sub scribers are sent via third-and-fourth class mail. Subscrip tions to foreign countries, with the exception of Canada, are mailed through the U.S. Postal Service’s International Sur face Airlift program (ISAL) from John F. Kennedy Interna tional Airport, Jamaica, New York. Copies to Canadian sub scribers are handled through the Canadian Post. Q uarterly R e v ie w subscribers also receive the Bank’s Annual Report. Q uarterly R e v ie w articles may be reproduced for educa tional or training purposes, provided that they are reprinted in full and include credit to the author, the publication, and the Bank. Library of Congress Card Number: 77-646559 FRBNY Q u a r t e r l y R e v ie w / S u m m e r - F a ll 1994 121 F ed era l R eserv e B ank o o of N ew Y ork O 0 4^ 1 O o i/i »n crq a> o c p < -* p 3 n> CL