The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
The Quarterly Review is published by the Research and Statistics Function of the Federal Reserve Bank of New \brk. Among the members of the staff who contributed to this issue are RICHARD G. DAVIS (on the recent performance of the commercial banking industry, page 1); LYNN PAQUETTE (on estimating household debt service payments, page 12); JOHN WENNINGER and LAWRENCE J. RADECKI (on financial transactions and the demand for M1, page 24); and ALLEN J. PROCTOR (on short-term borrowing by local school districts, page 30). Other staff members who contributed to In Brief—Economic Capsules are ROBERT N. McCAULEY (on large U.S. banks moving international activity off their balance sheets, page 42) and BRUCE KASMAN (on prospects for the U.S. international travel deficit, page 44). A quarterly report on Treasury and Federal Reserve foreign exchange operations for the period February through April 1986 starts on page 47. The Recent Performance of the Commercial Banking Industry Banks are the primary institutions through which the Federal Reserve conducts monetary policy. This is true both in respect to the central bank’s efforts to influence reserves, the money stock, and money market condi tions, and in respect to its efforts to underpin the stable functioning of the nation’s financial system. But banks are businesses, and to perform their critical role in the financial system and the policy process, banks must be profitable. A number of developments in recent years— some related to the overall economy and some specific to the markets where banks operate— have had impor tant effects on commercial bank profitability and the performance of the industry generally. This article brings together some of the main findings of a study under taken at this Bank on the recent performance of the commercial banking industry and the factors that have influenced it. The aim of the study is to understand the past rather than to attempt to foretell the future or to prescribe for it. Nevertheless, an understanding of commercial banking’s recent history is probably a nec essary preliminary in any effort to assess its future prospects or to evaluate policy options on matters that affect it. The recent slippage in bank profitability Bank profitability overall has clearly declined in recent This article is a slightly modified version of the concluding chapter of Recent Trends in Commercial Bank Profitability—A Staff Study, Federal Reserve Bank of New York (September 1986), Documentation and a more detailed discussion of the points raised in this article may be found in that volume. years (chart). For all insured banks, return on assets (ROA) declined in each year from 1980 to 1984 before recovering modestly in 1985. Even with the 1985 improvement, ROA remained close to its lowest level of the last 15 years. The recent performance of return on equity (ROE) for all insured banks followed a similar course, declining in each year from 1979 to 1984 wjth a modest recovery in 1985 to a still-low level. By 1985, ROE for all insured banks was about 11.4 percent, substantially below the 13.9 percent 1979 peak and, except for 1983 and 1984, the lowest since at least 1970. Much of the study focused on a group of 17 multi national bank holding companies, firms that hold some 37.9 percent of all bank assets and that are at the center of some of the structural changes hitting the banking industry. The profit performance of these 17 institutions has, of course, varied substantially from one firm to another. But for this group of holding companies as a whole, ROE slipped badly from its 14.4 percent peak in 1979 and 1980 to a low of 5.4 percent in 1984, recovering only to 11.2 percent in 1985, the lowest level for any year since 1976. Even excluding the two holding companies that lost money in both 1984 and 1985, ROE in 1985 was still some two and one-half percentage points below the 197 9-80 peak despite generally improved profits for most of these holding companies in 1985. In contrast to the multinational holding companies and insured banks as a group, the group of the 33 largest regional holding companies examined in the study does FRBNY Quarterly Review/Summer 1986 1 not show a recent downturn in profitability. For this group, ROE held at a high plateau of about 14.6 percent in each of the years 1979 through 1982. Profits did drop somewhat relative to this level in 1983 and 1984, but by 1985, ROE for the regionals as a group was essen tially back to peak levels. Overall, the 1980s have seen a slippage in profitability for most groups of banks— although from relatively high levels in terms of the postwar period as a whole. Obviously there is a great deal of diversity in the profit performance of individual banks and groups of banks. And despite the overall decline, the profitability of banking as a whole has shown greater stability than that of other financial industries and of many nonfinancial industries. The market share performance of commercial banks There has been some slippage in the overall role of banks in the credit markets over the last ten years or so. Moreover, there has been a marked decline in their share of one particularly important credit market—the wholesale loan market, as discussed below. But the banks have been able to hold their own or even increase their shares in most of the other markets they serve. According to flow of funds data, the commercial banks’ share of credit market debt claims against all nonfinancial sectors of the economy peaked at just short of 30 percent in 1974. Since then, this share has declined virtually year-by-year to 25.2 percent in 1985. However, the significance of this decline for the competi 2 FRBNY Quarterly Review/Summer 1986 tive position of the banks is hard to assess since the overall decline represents a reduced share of holdings of state and local and Federal debt. The banks’ share of holdings of domestic nonfinancial private debt was about unchanged on balance between 1974 and 1985. Indeed, in most of the private credit markets where banks compete, their share of the business has, as suggested above, either held about steady or risen in recent years. Thus the total share of home mortgages (including allocated mortgage-backed securities) held by commercial banks rose somewhat in the late 1970s and has held fairly steady in recent years in the 25 to 27 percent range. The thrift institutions have been the major losers in this market. The banks’ share of commercial mortgages has also risen over the last decade or so, from around 30 percent to around 36 percent in 1984. Over the same period, most of the other major participants in this market lost share to the banks. The banks’ share of consumer installment credit rose over the 1970s but then suffered a significant decline during the early 1980s due to a loss of share of auto loans to captive finance companies. The banks’ share of auto loans has been importantly weakened over the last several years as the major domestic car manufac turers have at times used reduced-rate financing as a marketing tool. On the other hand, the banks’ share of revolving consumer credit has risen almost continuously since the mid-1970s as the use of bank credit cards has spread at the expense of cards issued by retailers. On the other side of the balance sheet, there are various ways in which the market for deposits can be sliced. One obvious way is with respect to the market for “ transactions instruments.” The commercial banks’ share of deposits included in M1 (demand deposits, negotiable order of withdrawal accounts (NOWs), and Super NOWs) has of course declined somewhat in recent years as the thrifts have entered this market, from about 99 percent in 1978 to about 89 percent in 1985. A second approach to defining the deposit market might be in terms of the market for interest-bearing retail deposit accounts. This market includes NOW accounts, all savings and small time deposits, and money market fund shares. The banks’ share of this total seems to have suffered very little from the rise in the money funds in the 1978-82 period. Instead, the thrifts were appar ently the main losers in the rise of the money funds. With the introduction of money market deposit accounts (MMDAs) and Super NOW accounts around the end of 1982, the commercial banks’ share of this market has subsequently risen a few percentage points, mainly at the expense of the money funds. The share of the thrifts in this market has continued to drift down, but more gradually than in earlier years. The banks’ share in the markets for various financial services is somewhat harder to measure and to gen eralize about. Owing primarily to the purchase of a major discount broker by one of the multinational holding companies, the banks as a group now appear to have over half the revenue of discount brokerages that are New York Stock Exchange members. On the other hand, there has been a sharp fall in the banks’ share of pension fund assets under independent man agement. In the underwriting field, banks are permitted to underwrite municipal general obligation instruments, municipal revenue bonds for housing and higher edu cation, and private placements. These types of under writing account for about 42 percent of all domestic underwriting activity. Of these permitted activities, the banks’ market share has fluctuated between roughly 12 and 20 percent in the 1979-84 period, with no discern ible trend. Banks appear to have been advisors in only about 2 percent of the 100 largest merger and acqui sition deals of recent years. Data on smaller deals are quite limited. The decline of the wholesale loan market O ne im p o rtan t m arket in which the b a n k s ’ role undoubtedly has declined is the so-called “wholesale" market. Precise docum entation of this decline is impossible. For one thing, the concept of the wholesale market is somewhat vague. As generally used, the term means any large loan, whether to a national or multi national corporation or to a foreign business or gov ernment. However, “large” is not a precise concept. And at times the wholesale market seems to refer primarily to “good” large loans, i.e., to prime names. In any case, there are no data designed specifically to measure this market, so its dimensions have to be approached by approximation. The banks’ share of all nonfinancial business credit has actually risen over the last decade or more, from a little over 30 percent in the early 1970s to 35.2 per cent in 1985. However, this rise obscures the banks’ weakness in the wholesale market, in part because it reflects bank strength in commercial mortgage lending and in part because of the steadily rising share of short term business credit (in which the banks specialize) relative to total business credit over the past ten years. Since wholesale lending can be presumed to be con centrated at the larger banks, one way to construct a proxy for the decline in the banks’ share of this market is to measure the sharp decline in the weekly reporting member banks’ (WRMBs) share of all short-term non financial business credit from over 43 percent in 1974 to about 27 percent last year. This is obviously a major decline in the banks’ role. The slack seems to have been taken up mainly by the commercial paper market and by lending by branches and agencies of foreign banks. Because of statistical deficiencies, it is very difficult to factor in precisely the role of the foreign banks in the U.S. wholesale lending market, but it is obviously quite large. We estimate that lending by all foreign offices of foreign banks to U.S. commercial and industrial (C&l) borrowers could represent as much as 39 percent of such loans at WRMBs. Such limited data as we have, however, do not really make clear whether there has been any growth in the role of the foreign banks in our wholesale loan market in the 1980s. (And, of course, the role of our banks in foreign markets has also risen over the years.) The reasons offered by bankers and others for the decline in wholesale lending seem to go to the heart of the banks’ presumed comparative advantage as financial intermediaries, at least with regard to this particular market. Thus we are told that many prime corporate borrowers have as good or better credit ratings than all but a very few of the banks lending in this market. Moreover, it is argued that for large borrowers, the banks no longer have an informational advantage in assessing the credit-worthiness of potential borrowers in this market. Much of the relevant information is public and readily available to any potential purchaser of the borrower’s debt. Further, it is argued that the growing importance of large pools of funds with sophisticated management enables managers of these funds to diversify credit risk as readily as can banks. It is sometimes also argued that banks are at an inherent disadvantage in acting as intermediaries because, in addition to covering costs, they must also price loans to realize an “adequate” return on capital. This argument appears to be fallacious, however, since the required return on bank equity can be thought of as the needed reward for assuming credit and funding risk over and above any purely “actuarial” component. The “market” will also demand a reward for assuming such risk. Whether the market will be willing to fund a given credit at or below the level banks have to charge both to cover costs and to achieve their “required” rate of return on capital cannot be determined a priori but will depend upon the circumstances. The point seems to be that with respect to high-quality wholesale credits, at least, the market has in fact increasingly been able to outcompete the banks over the past decade or so. The decline of the wholesale loan market has obviously had profound implications for our largest wholesale lenders, but a question arises about how important it might be to the overall role of the banking system in our financial markets.1 We can approximate ’As an indication of the relative heft of the large banks most heavily FRBNY Quarterly Review/Summer 1986 3 Profitability and Price/Earnings Ratios in Financial Service Industries Industry Commercial b a n k in g .................... 17 multinational bank holding c o m p a n ie s ............................. Finance com panie s...................... Mortgage companies .................. Securities ...................................... Investment banks .................... Other s e c u ritie s ........................ Life in s u ra n c e ............................... Stockholder-owned .................. M u tu a l........................................ Property and casualty insurance . Stockholder-owned .................. M u tu a l........................................ Insurance brokerage Large firms ............................... Small firm s ................................. Diversified financial firms ........... Nonfinancial firms (S&P 400) Average after-tax return on equity, 1980-84 Average price/ earnings ratio, 1977-84 12.2 6.3 12.9 12.6 13.1 18.7 26.0 15.8 13.4 15.2 10.5 7.4 7.7 7.4 6.1 6.2 18.3 9.2 13.1 13.7 * 7.9 * * * 6.4 * * 7.1 12.8 * 8.3 9.6 For notes and sources, see Recent Trends in Bank Profitability— A Staff Study, Federal Reserve Bank of New York (September 1986), Chapter 14. *Not applicable. an answer by looking at the share of short-term nonfinancial business credit represented by C&l loans at WRMBs. As of December 1985, actual C&l loans at these banks were $255.2 billion or 27.3 percent of all short-term business credit. If the WRMBs had main tained their 1974 share of 43.5 percent, C&l loans at these banks would have amounted to $406.6 billion in 1985. With such a higher level of C&l loans, and given the actual 1985 total of outstanding debt of the nonfinancial sectors ($7,114.7 billion), such a figure would mean that the banks had held 27.3 percent of this credit total, or 2.1 percentage points higher than the share they actually did hold in 1985. Such a figure suggests what was implicit earlier in looking at the overall share of banks in credit markets generally: namely that the decline of the wholesale lending market has by no means been a body-blow to the banking system as a whole. Nevertheless, this decline should not be underestimated. It has been very important to the largest banks, banks that in some ways occupy a pivotal position in the banking system as a Footnote 1, con tin u e d involved in this market, the 17 m ultinationals had 37.9 percent of all bank assets as of last D ecem ber and 50.2 percent of all C&l loans. The C&l loans at these banks con stituted 10.1 pe rcent of total assets of all com m ercial banks. 4 FRBNY Quarterly Review/Summer 1986 whole. Moreover, the 2 percent figure could substantially understate the qualitative importance of the decline in wholesale lending for the role of the banking system as the backstop to our credit markets. For one thing, many bankers argue that the increasing competitiveness of the wholesale loan market has more or less forced the wholesale banks to reduce the average quality of their C&l book. Some bankers were able to offer internal data that support this argument. In addition, examination indicates that the weighted average “ betas” (a standard measure of stock price volatility) of the C&l loan portfolio of a sample of 47 of the largest banks are higher than those for the Standard and Poor’s (S&P) 500 and indicates a deterioration in loan quality since the mid-1970s. Deterioration in the quality of C&l loans seems to have been a significant drag on the stock market performance of the shares of both multinational and regional holding companies. The deteriorating status of less developed country (LDC) loans also seems to have been an important factor, especially for the multinationals. Profitability of banks relative to other financial firms As might be expected, there has been considerable diversity in the profit performance in the various non bank sectors of the financial industry that in at least some respects compete with banks. These include finance companies, mortgage companies, investment banks, securities brokers, the various components of the insurance industry, and large diversified financial firms. In comparing the profit performance of these groups with that of commercial banks, only a few generaliza tions can be made, and even then, only with qualifi cations. First, with the exception of the non-auto finance companies, the profitability of all segments of the non bank financial industry has tended to move through w ider— often much w ider— ranges during the past decade than has bank profitability. And the profit per formance of many of these groups shows substantially greater short-term variability than does that of the banks as a group. Second, virtually all the nonbank compo nents of the financial industry have shared the banks’ experience of deteriorating profit performance since peaks occurring around 1979-80—or at least this was true through 1984, the latest year for which complete data were available. The major exceptions to this recent downward trend have been the consumer finance and mortgage companies. The average profitability of the financial industry in the 1980-84 period has varied considerably from sector to sector (table). Over this period, average after-tax ROE was 12.2 percent for all insured banks and 12.8 percent for the multinational bank holding companies. Some other parts of the financial industry did considerably better than that, including large investment banks, other securities firms, and large insurance com panies. Average ROE in the period for finance companies, mortgage companies, and diversified financial firms was about the same as for banks or only a little better. A few sectors, notably property and casualty insurers, did substantially worse than the banks. On average, the reported after-tax ROE of nonfinancial firms in the S&P 400 was somewhat higher than that of the banks at 13.7 percent. Stock market treatment of banks and other financial firms For more than a decade, the stock market has priced the earnings of large bank holding companies at mul tiples well below those for corporate earnings in general. This is true both for the regional holding companies and the multinationals, but especially for the latter. As of late April 1986, the price/earnings (P/E) ratio of the S&P 500 was about 15.7, while that of a group of multinational banks was 10.2 (excluding banks with current losses) and 12.2 for a group of large regional bank holding companies. Bankers and financial analysts offer a wide range of explanations for the market’s relatively adverse treat ment of bank stocks over the past decade, and there seems no reason to insist on a single explanation to cover so long a stretch of time. One possible sequence of factors, for example, could be a sense early in the period that banks were simply “stodgy” investments with poorer earnings growth prospects than industry in gen eral. Subsequently, as noted below, there is reason to believe that as inflation heated up in the later 1970s, reported bank earnings became seriously overstated. Thus P/E ratios for earnings properly computed to take account of the overstatement of earnings due to inflation may have been a lot higher than P/E ratios computed simply from earnings as reported. For a time in the early 1980s, a difficult interest rate environment, as discussed below, may well have hurt bank stock performance. Most recently, concern about the quality of bank assets, and the LDC loan problem in particular, has probably been an important depressant. Some believe that the shrinking wholesale lending market has also been a pervasive factor in the stock market’s assessment of the earnings prospects, at least for money center banks, and therefore in the low P/E ratios it has accorded most of these banks relative to other firms. In assessing these possible explanations for the per sistently poor stock market evaluation of bank earnings, it may also be worth keeping in mind that with some notable exceptions, the market has also valued the earnings of most other segments of the financial industry at lower multiples than have been accorded to nonfinancial firms on average (table). Over the 1977-84 period, P/E ratios for nonfinancial firms, as represented by the S&P 400, averaged 9.6 as compared with 6.3 for a sample of 90 commercial banks. To be sure, the banks’ average P/E ratio was near the low end of the range even for financial firms. But except for insurance brokers, no nonbank financial group had P/E ratios that averaged as high as the average for the nonfinancial firms. We do not have P/E data for the large investment banks over comparably long periods of time since many of them have gone public only recently. Based on data for April 1986, the stocks of publicly-owned large investment banking firms show multiples well above those for the multinational banks, though still not especially impressive relative to stocks in general. Thus as of April, the average P/E ratio for fiv£ publicly-held large investment banks averaged 14.9, compared with the 10.2 average for the multinational bank holding companies cited earlier and 15.7 for the S&P 500. A number of bankers contacted in the course of the study were quick to point out the contrast between high mul tiples for investment banks and the much lower ones characteristic of major money center banks. However, some of the very high investment bank multiples they cited in 1985 (as high as 25) appear in retrospect to have been temporary spikes. Indeed these multiples have apparently declined further since the April data cited above. Overall, the low P/E ratios accorded most large banks relative to nonfinancial firms and even to many financial firms, of course, mean that equity capital is relatively expensive for these banks. The practical implication is that it is relatively difficult for banks to find new projects (or expansions of old projects) with yields high enough to justify the injection of new capital, and this is, of course, a deterrent to the longer run expansion of the industry. Macro versus structural causes of the deterioration in bank profits In examining the various possible causes of the recent deterioration in bank profitability, it is useful to make a distinction between causes related to the general eco nomic environment (“macro” causes) and causes spe cific to changing competitive conditions facing the commercial banking industry (“structural” causes). Such a distinction might at least provide a starting point for trying to determine whether pressures on bank profit ability are largely temporary or long-run in character. Thus macro developments, such as movements in the general level of interest rates, the rate of inflation, and real economic growth, tend to be associated with the business cycle and, therefore, tend to reverse them FRBNY Quarterly Review/Summer 1986 5 selves in tim e. Structural developm ents such as changes in an industry’s technology, relative costs, demand, the extent of competition, and the nature of regulation, clearly tend to be of longer-than-cyclical duration and may be fairly long-lasting. To be sure, the correspondence between macro and temporary on the one hand, and structural and long-lasting on the other, is far from perfect. Macro developments obviously can exhibit trends as well as cycles and can thus be longlasting. Similarly, structural changes are not necessarily irreversible. In any case, it seemed desirable to attempt a formal statistical analysis of the business cycle and trend components of movements in bank profitability. Not unexpectedly, regression equations using business cycle variables (such as actual GNP relative to trend) do in fact "explain” a statistically significant fraction of the variance of bank profits. Nevertheless, there has been a clear tendency for actual profits to fail short of “pre dicted" profits in recent years. Taken as a group, equations including both trend and cycle variables offer some limited evidence that there has been a downtrend in bank profits in recent years after allowing for the influence of cyclical variables. Of course, statistical evidence of a recent downtrend in profits over and above cyclical influences does not tell us which bank product lines may be responsible for depressing profits and, more generally, does not dis tinguish between a possible longer-than-cyclical per sistence of macro and structural problems. Perhaps most important, this evidence does not predict how long any downtrend in profitability might persist. Some identifiable macro influences on recent bank profits performance Whatever the relative roles of macro and structural impacts on the recent decline in bank profits, some effects of each kind can be clearly identified. Chrono logically, the first macro problem to hit the banks over the past decade was the sharply higher inflation rates of the late 1970s peaking in 1980-81. By reducing the real value of net worth, inflation means that book earnings overstate earnings for financial institutions, because, unlike industrial firms, they are large net holders of fixed-dollar-value assets. In theory, inflation need not also hurt inflation-adjusted bank earnings if nominal interest rates correctly anticipate inflation so that real after tax interest rates are unaffected by inflation. But it seems reasonably clear that the markets did not fully anticipate the extent of the acceleration in inflation that occurred. Thus there is a strong probability that the inflation of the late 1970s not only resulted in profits being overstated but also that the inflation actually reduced true profits. We made no attempt to 6 FRBNY Quarterly Review/Summer 1986 measure the extent to which bank profits were actually hurt by inflation, but we did estimate the extent to which they were overstated as a result of inflation. As Henry Wallich found in an earlier study of this problem,2 the overstatement appears to be sizable, and it alters the appearance of the statistical record in some important ways. First, inflation substantially overstated the average level of bank profits for most of the high inflation years of the 1970s and early 1980s. Adjusted ROE averaged only about 7 percent rather than the roughly 13 percent average shown by the raw data. Second, ROE at all insured banks appears to have been near its lowest level of the decade in 1979 and 1980 on an adjusted basis, not at its highest, as the reported data suggest. Adjusted ROE rose fairly sharply in 1981 and 1982. ROE in 1984 (the last year for which we have data on an adjusted basis) was close to the highest levels of the last decade, not at the lowest as the reported data suggest. Third, adjustment of profits for the effects of inflation substantially alters the appearance of bank profitability relative to that of nonfinancial corporations. On a reported basis, bank ROEs were roughly equal to average ROEs for nonfinancials until the 1980-84 period when reported ROEs of nonfinancials dropped very sharply. Adjusting ROE measures of both groups for inflation, however, puts banks’ ROEs below those of the nonfinancials (and often far below) in all but one of the last 12 years. Finally, adjusting earnings for the effects of inflation for both multinational banks and the S&P 500, P/E ratios for the banks were actually above the average for the 500 in the 1977-79 period, were close to the 500 in 1981, and were once again above the 500 in 1984. It seems entirely reasonable that a “rational,” “effi cient” stock market should “look through” the veil of inflation to price earnings on an inflation-adjusted basis. The numbers are at least consistent with the view that this is what happened. But it is somewhat curious that none of the bankers or bank stock analysts contacted in the course of the study included the distorting effects of inflation among the possible causes of consistently low reported bank P/E ratios. In any event, inflation should now be a much less important influence on bank P/E ratios than it may have been a few years ago, perhaps replaced, as suggested earlier, by the issue of credit quality. A second macro factor adversely affecting bank profits in the late 1970s through about late 1981 was the interest rate environment. The general uptrend in rates over this period had, at least in the short-term, an 2Wallich, Henry C., "Inflation is Destroying Bank Earnings and Capital Adequacy," Bankers Magazine (Autumn 1977), pages 12-16. adverse effect on profits because of a general tendency, that apparently increased over the period, for banks to have more short-term liabilities than short-term assets (defining “short-term” in a repricing sense). The exis tence of this short-term “repricing gap” made the flat tening and frequent inversion of the yield curve in this period a further drag on profits. Finally, the market cost of bank funds relative to market lending rates (as measured by the spread between certificates of deposit (CDs) and commercial paper rates) also became sharply more adverse over this period. It should be noted that the repricing effect of a onceand-for-all increase (for example) in interest rates on interest earnings can be measured precisely only with a complete knowledge of the maturity distribution of assets and liabilities. It is necessary to make do with a crude index based on the excess of liabilities repriced within a year or less over similarly short-term assets. This appears to capture the direction of effects but can provide only an index, not actual magnitudes. As a conceptual matter, it should also be noted that the earnings impact of a once-and-for-all rise in rates varies with the time horizon. It seems to be clearly adverse in the short run, but at some point it must turn positive. In the long run, the effect must be positive given the existence of some fixed-rate liabilities and equity in the banks’ balance sheets. The effect on the present dis counted value of the bank of a change in rates thus depends both on the time profile of earnings effects and on the discounting factor. In practice, the stock market seems to mark bank stock prices up or down inversely with interest rate movements. While the various aspects of the interest rate envi ronment were, as noted, adverse to profits through about late 1981, they have subsequently reversed and seem to have been positive on balance since then. The impact of the interest rate environment seems to be clearly a cyclical affair. There is no obvious reason to believe that there has been a permanently adverse change in this environment for the banks. One influence on profitability that could be related in part to macro and in part to structural developments is the increase in provisions for loan loss reserves. In a purely arithmetic sense, the increase in such provisions relative to assets has more than accounted for the deterioration of ROA at all banks and at all major com ponent banking groups. But the interpretation of this statistical fact presents some problems. In an ideal world where banks were able correctly to predict the overall rate of loan losses—though not which individual loans would go sour— loan loss provisions would be set to maintain loan loss reserves at a level equal to expected future losses. If a deliberate decision were made to increase the average riskiness of the loan portfolio, loan loss provisions would be raised accord ingly. Since the pricing of loans should, if market con ditions permit, include an allowance to cover “normal” loss experience, a rise in loan loss provisions relative to assets need not by itself indicate any decline in profitability. If the asset portfolio were to be shifted toward loans with a higher expected loss rate, the earnings figures should show higher net interest earn ings after the portfolio shift, partially or fully offset by a corresponding rise in the loan loss provision. ROA (and ROE) need not be significantly changed on balance. But in practice, actual default experience need not correspond at all closely to expectations. Many kinds of lending do not readily accommodate themselves to an “actuarial” analysis of risk. Moreover, average past experience will fail as a guide to future default rates in the face of adverse developments in the general eco nomic situation facing borrowers— just as mortality tables will understate mortality rates during an epi demic. The data indicate that fluctuations in loan loss provisions are in fact highly correlated with current charge offs. This suggests that these fluctuations are probably at least as much influenced by the failure of actual default experience to conform to expectations as they are by changes in the expected default rate resulting from deliberate changes in portfolio composition. On balance, it seems reasonable to conclude that much of the recent rise in loss provisions represents a deterioration in the quality of credits that was not anticipated in setting the levels of loan loss provisions in earlier years. To this extent, current rising loan loss provisions reflect unexpectedly adverse loan loss experience rather than a deliberate change in portfolio strategy toward loans both involving higher expected losses and offsetting higher interest earnings. Since it seems reasonable to expect that cyclical downturns would be a major cause of unpleasant surprises in loan loss experience, the overall cyclical influence on bank profits, identifiable in the regression results cited above, would manifest itself in the form of a corresponding cyclical influence on charge offs and thus on loan loss provisions. Such a pattern is in fact supported by the evidence. At the same time, there are other ways, not neatly tied to cyclical movements in GNP, that general eco nomic conditions could produce unexpectedly adverse loan loss experience with correspondingly adverse effects on profitability. Even though economic expansion has now been underway in the United States and the rest of the industrial world for about three and one-half years, substantial economic slack has rem ained. Moreover, there has been a persistent shift away from FRBNY Quarterly Review/Summer 1986 7 the inflation in commodity and asset prices of the late 1970s and early 1980s to an atmosphere of stable or declining prices. Further, over much of the period, both nominal and real interest rates have been at high levels relative to historical norms, and high levels of the dollar have been a problem for some domestic borrowers. Thus despite the resumption of overall economic expansion, the shift in the economic climate after the earlier inflation has left many groups of borrowers under continuing pressure. These, of course, include the LDCs, agriculture, natural resources (notably energy), and some sectors of heavy industry. So while loan losses in these areas may have persisted in the face of an improving overall economy, they would neverthe less appear to reflect macro-economic conditions, at least in substantial part. There is, however, a possible structural side to the story of rising loan loss provisions and their relation to profitability. The decline in the demand for bank credit by highly rated borrowers may well, as already sug gested, have left many banks, especially the largest, with a deterioration in the quality of their loan book beginning as long as a decade ago. If so, this essen tially structural development would show up as rising loan loss rates and as correspondingly higher loan loss provisions. So the rising level of provisions could reflect a structural development, at least in part. As noted above, if higher expected risks were fully priced in setting rates for these more risky customers, rising loan loss provisions would not necessarily reflect a corresponding drain on profitability. However, such a drain would occur if, in a highly competitive market, banks failed to charge adequately for the increased level of prospective defaults. Unfortunately, efforts to distinguish between macro and structural causes of rising loan loss provisions and their effects on profits are complicated by an inherent “collinearity” between the two. The reason is that higher loss rates arising from an increasingly risky portfolio are likely to be exacerbated by deteriorating overall macroeconomic conditions. Ideally, one would like to be able to determine whether actual loss experience has been worse than might be expected given the general macroeconomic climate and, if so, why. However, there is no neat way to do this. The evidence cited above does indicate that charge offs and loan loss provisions have been greater than expected, based solely on cyclical factors. However, the fact that this is true for all size classes of banks, and not just the wholesale banks, leaves open the question of why it has occurred. Structural factors Perhaps the most direct evidence of a structural change in the competitive conditions affecting bank profitability 8 FRBNY Quarterly Review/Summer 1986 would come from indications that the profitability of the traditional deposit-taking and loan-funding role of banks had declined at the expense of newer, essentially fe e -b a s e d a c tiv itie s . To docum ent such a shift, however, would require revenue and cost data by type of activity. For individual banks, determining the profitability of individual activities at a high level of detail is a difficult problem, despite access to internally generated data. At the level of publicly available data, there exists no direct information on the profitability of different banking activities, even over very broad classes of activities. As a result, it has been necessary to make estim ates which to some extent rest on arbitrary assumptions. The approach taken was to distinguish bank activities that involve deposit-taking and the funding of interestearning assets— i.e., “intermediary” activities narrowly defined— from all other, largely fee-based activities. Given this distinction between the two classes of activ ities, the problem is somehow to allocate revenues and costs between them. On the revenue side, net interest income can be ascribed to the intermediary category, but non-interest income arises from both intermediary and other activities in ways that have changed over time as bank pricing practices have changed. Consequently, it was necessary to make some alternative assumptions about the allocation of non-interest revenues between intermediary and other activities that seem to encom pass the range of possibilities. On the expense side, the only information available by activity comes from the Federal Reserve’s Functional Cost Analysis data, and for various reasons, these, too, present some problems in allocating costs between intermediary and other activities. While the analysis necessarily leaves the answers to many questions uncertain, some facts do stand out. First, under any reasonable set of assumptions, reve nues from intermediary activities at large multinational banks have been rising much more slowly than reve nues from other sources. Thus while revenues from intermediary activities appear to have risen on the order of 17 to 24 percent per dollar of total assets between 1980 and 1985, revenue from other sources per dollar of assets more than doubled over the same period. While revenues from these other activities constituted only about 30 percent or less of intermediary revenues at these banks in 1980, this fraction had risen to roughly 50 percent in 1985. Not surprisingly, the non-interest expenses (salaries, furniture, equipment, and occupancy and other operating expenses) of these "other” activities have also risen substantially more rapidly than intermediary-related, non interest expenses. Growth of non-interest expenses was about 35 percent for intermediary activities and around 75 to 95 percent for other activities over the 1980 to 1985 period. Since revenue and cost allocations have to be com bined to allocate profits between intermediary and other activities, the uncertainties in revenue and cost allo cations are compounded in making profitability esti mates. Hence the range of uncertainty is correspond ingly enlarged. Under a combined set of assumptions most favorable to the estimated profitability of inter mediary activities, profitability per dollar of assets showed no trend between 1980 and 1983 and then dropped by more than half in 1984 and 1985. Under this same set of revenue and cost allocation assumptions, the profitability of other activities (again scaled by total assets) rose irregularly between 1980 and 1983 and then rose sharply further in 1984 and 1985 to a level that was larger than the profitability of intermediary activities. Using a set of revenue and cost allocation assumptions unfavor able to intermediation, intermediary activities actually lost money in 1984 and 1985 while the profitability of “other” activities rose steadily between 1980 and 1985. Thus the results do depend significantly on the rev enue and cost allocation assumptions. And given the unusually sharp deterioration in intermediation profits and the sharp rise in other profits in the final two years of the period (1984 and 1985), it is difficult to draw hard conclusions about trends in the relative profitability of these two broad types of activities at the multinational banks over the longer run. Both the relative level and the trend of profits in the intermediary activities were importantly influenced in the 1980-85 period by the sharp rise in loan loss provisions, an item quite properly treated as a cost of intermediary activities. Excluding the loan loss provisions, there is no clear sign of a downtrend in the profitability of inter mediary activity over this period. In fact, favorable assumptions suggest an erratic but discernible upward trend while unfavorable assumptions suggest that intermediary profits before loan loss provisions have been generally unchanged. So for the profitability of intermediation, as for the profitability of banking overall, much depends on the extent to which recent increases in loan loss provisions prove to reflect a permanent rise in the level of such provisions beyond levels priced into loan spreads and the extent to which they prove to reflect a merely temporary effect of unexpected adverse economic conditions. One point should be made in the face of the agnos ticism forced on us by a strict adherence to what can be demonstrated from available data. The intermediary activities covered by the estimates involve the full range of deposit-taking and funding operations undertaken by these banks, not just the wholesale lending operations where profitability is widely believed to have declined. So the data in no sense conflict with this widely held perception about the wholesale market. Indeed the picture painted by the revenue data, at least, is entirely compatible with the generally received view about the wholesale lending market. But there is no evidence to suggest that the profitability of other kinds of bank lending activities (funding consumer credit, home and business mortgages, the middle and small business loan market) are declining and, indeed, none of the industry experts contacted in the course of the study suggested that they are. What may well be true, however, is that the natural market for these kinds of lending products (absent geographic expansion) falls greatly short of the deposit-gathering capabilities of large money center institutions that formerly used such capabilities to fund wholesale lending. One important structural development examined in the study was the effect of deposit interest rate deregulation on bank net interest margins— motivated in part by the superficially surprising fact that such margins have generally tended to rise over the years in which deposit rate deregulation has taken place. In looking at this problem, it is necessary to disentangle the adverse effect on interest margins of deregulation’s impact on relative rates, given the general level of interest rates, from the favorable effects on net interest margin of a rise in the general level of rates. As was suggested earlier, the long-run favorable impact of a rise in rates stems from the fact that as long as banks have some fixed rate liabilities (e.g., demand deposits) and equity, the long-run repricing effects of interest rate rises must be to increase net interest earnings. For example, shifts out of rate-regulated instruments such as demand deposits into market rate instruments such as super NOWs hurt interest rate margins. If at the same time, however, interest rates generally are rising, the remaining zero-rate demand deposits become more valuable. What the computations reveal is that rate deregu lation by itself hurt net interest margins significantly between 1977 and 1984, by about 0.37 percent of assets at multinationals, by about 0.56 percent at regionals, and by about 0.18 percent at other insured banks. A look at the year-by-year impact suggests that most of it was completed by the end of 1982 and that there has been little if any net deterioration in margins as a result of deregulation since then— at least for the 45 large bank holding companies we examined in detail. Apparently the adverse potential effects of MMDAs and Super NOWs was largely offset by an associated reduction of reliance on large CDs at these banks. According to our estimates, the effect of the overall rise in the level of interest rates from 1977 to 1984 was FRBNY Quarterly Review/Summer 1986 9 to raise net interest margins at 42 large holding com panies by 0.87 percent of assets and by 0.97 percent of assets at other insured banks. In other words, the rise-in-rates effect much more than offset the deregu lation effect between 1977 and 1984, accounting for much, but not all of the overall improvement in bank net interest margins over this period. By implication, a return of general rate levels to the 1977 levels would reveal the unfavorable effects of deregulation otherwise hidden by the general rise in rates through 1984. If one concentrates solely on the effects of deregu lation on net interest margin and ignores the non interest income and expense implications of deregula tion, the computed effects on profits seem to be large. Thus, for example, before-tax ROE of the multinationals averaged 15.8 percent in 1984. According to our com putations, net interest margin was reduced by deregu lation by 0.37 basis points, as noted above. Had it not been so reduced, these before-tax profits would have been a much larger 22.8 percent of equity in 1984. The problem with trying to translate these large effects of deregulation on net interest margin into profit terms, however, is that such computations ignore offsetting concomitant changes in income and expenses brought about by the onset of interest rate deregulation and rate competition. Thus explicit interest rate competition has meant reduced nonrate competition (and thus reduced non-interest expenses for branches, human tellers, etc.) and increased non-interest revenues in the form of explicit service charges. Hence the net effect on profitability of rate deregulation must be materially less than its effects on net interest margins alone would suggest. Implications of the study The most obvious question raised by a study of bank profitability is whether the longer-term position of the industry is deteriorating. The question is sometimes put somewhat differently: Is the profitability of traditional banking drying up so that the role and function of banks as we know them must undergo substantial change? Answers to such questions can be based on an assessment of what has happened or projections of what will happen. The present study obviously bears mainly on what has happened. The answers it gives are not unambiguous. On one side, there appears to be evidence that major upheavals in the macroeconomic climate have had, and are continuing to have, a significant adverse effect on bank profita bility. T h e re can be little doubt that a period of stable, low -inflation economic expansion, during which the credit problems generated by past eco nomic upheavals can be worked through, would do much to strengthen bank profitability. 10 FRBNY Quarterly Review/Summer 1986 On the other hand, some of the pressures on the banks clearly would not go away, even under the most favorable of macro-economic scenarios. Probably the most critical issues raised by the structural develop ments we have examined center on the decline of the wholesale loan market. The figures indicate that the decline in wholesale lending has already led to a modest but significant slippage in the banking system as a whole in the national credit markets. And the development has raised acute strategic issues for many of our largest banking institutions that have been most heavily involved in this market. Their success in forging successful new strategies in the face of the decline in this traditional market has so far been mixed. But perhaps more important than its impact to date, the most interesting question raised by the decline in wholesale lending is whether it may prove to be a par adigm for the future transformation of other traditional banking markets. The development that needs closest attention in this connection is the spread, or potential spread, of the process known as “securitization.” Securitization, mainly in the form of a huge expansion of the commercial paper market, made possible the decline of the wholesale bank loan market. Securiti zation has become a major factor in the mortgage market (though the banks themselves are important holders of mortgage-backed instruments), and securi tization of numerous other types of loans, especially consumer loans, appears to be in an active, if early stage of development. The question raised by the securitization process is whether a range of developments— including a widening ability of investors to assess credits and diversify risks and innovations in technology that facilitate the pack aging of securitized loans— are about to produce a broad-based erosion in the profitability of intermediating credit through the banks. Some analysts have already come to the conclusion that such an erosion is indeed in prospect. Some go further to argue that, as a con sequence, the banking system will eventually evolve to produce “banks” that are more like money funds, offering transactions instruments on one side of the balance sheet and holding essentially riskless money market instruments on the asset side, with traditional banking credits securitized out to the market or held by nonbank institutions. But the fact that an outcome is conceptually possible does not mean it will materialize. Banks retain tremen dous advantages as specialists in assessing and diversifying credit risks and in funding them through an array of highly attractive deposit instruments. The examination of past developments we have conducted does not tell us to what extent these advantages may erode in the future. It does suggest, however, that out side the wholesale lending market, there has been no substantial slippage to date. What is clear is that the profitability of the banking system, and hence its continued ability to play its present role in the credit markets and in monetary and financial policy, cannot be taken for granted. The prof itability of the banking system is likely to be a continuing factor in the consideration of a wide range of policy issues relating to banks and to financial institutions and markets more generally. Richard G. Davis This article is excerpted from a 15-chapter book entitled Recent Trends in Commercial Bank Profitability—A Staff Study recently completed by the Federal Reserve Bank of New York. For details, see the inside back cover. — _— .— „--------------------- ;...— ~ — — —— — .......... ..................— ..- FRBNY Quarterly Review/Summer 1986 11 Estimating Household Debt Service Payments Household debt rose very rapidly during the last several years, with the ratio of debt to disposable personal income reaching an all-time high of 0.74 in 1985-IV and 1986-1 (Table 1).1 This high level of household debt may cause concern for at least two reasons. First, to the extent that individuals may become subject to liquidity constraints, a high level of debt may reduce future consumer expenditures, aggregate demand, and real economic activity. Second, a high level of debt may increase consumers’ default rates and adversely affect the soundness of the financial system. However, liquidity constraints and default rates depend not only on the level of debt outstanding but also on the level of debt service payments.2 Because data on aggregate debt service payments are not collected, this article estimates a debt service payment series from 1975-1 through 1986-1.3 Aggregate debt service payments behaved quite dif ferently from debt outstanding in recent years. Although The author thanks James August, Paul Bennett, Christine Cumming, A. Steven Englander, Howard Esaki, Ellen Evans, Edward Frydl, Andrew Silver, and Charles Steindel for valuable comments and suggestions, and Michael Weitz for excellent research assistance. ’Federal Reserve Board, Flow of Funds. 2Of course, neither the debt outstanding measure nor the debt service payments measure takes into account such factors as demographics, wealth holdings, or the distribution of debt among . income groups, all of which would also affect the assessment of the consumer debt burden. •3See Goldman Sachs, Pocket Chartroom (November 1985), for alternative estimates of debt service payments on consumer installment debt. 12 FRBNY Quarterly Review/Summer 1986 the estimated ratio of home mortgage plus consumer installment debt service payments to income was higher in 1986-1 than in any of the previous ten years, it did not rise as rapidly over the past decade as the debtto-income ratio. While home mortgage debt service payments increased faster over the last decade, on average, than home mortgage debt, consumer install ment debt service payments increased much less than consumer installment debt. This article also analyzes why the debt and debt service payments ratios grew at different rates by examining how changes in loan extensions, maturities, inflation, and interest rates affect each measure. Cycles in the level of loan extensions generated the cyclical pattern of both debt and debt service payments. How ever, while the debt-to-income ratio surpassed previous levels in 1985 and 1986-1, extensions of most types of household loans were not higher, relative to income, during the current expansion than during the 1975-79 expansion. Rather, longer maturities on consumer loans and lower inflation rates contributed substantially to the recent increase in the debt ratio. Changes in maturities, inflation, and interest rates had different effects on debt service payments. Longer maturities on consumer loans decreased rather than increased the debt service payments ratio. In general, lower levels of inflation, if accompanied by lower nom inal interest rates, affect the debt service payments ratio much less than the debt ratio. However, interest rates on consumer loans remained high in recent years, rel ative to inflation, contributing to the growth of debt service payments. Estimating debt service payments Data on the aggregate debt service payments due on home mortgage and consumer debt outstanding are not collected. Therefore, debt service payments of the household sector from 1975-1 through 1986-1 were estimated for home mortgages, which consist of mort gages on one to four family homes, and the components of consumer installment debt—automobile, mobile home, revolving, and “ other consumer installment” credit. Debt service payments were not estimated for consumer noninstallment debt.4 The estimates of debt service payments are based on past levels of extensions, debt, and average maturities and interest rates at which loans were issued. In gen eral, for each type of debt, the stream of debt service payments due on loans issued during a given period is calculated based on estimates of the amount of loans extended during that period, and of the average interest rates and maturities at which the loans were issued. Prepayment rates are set so that quarterly changes in debt are equal to extensions of new loans minus esti mated repayments of principal. (For a detailed expla nation of the methodology used, see the box.) In addi tion, estimates of consumer installment debt outstanding are adjusted to account for any precomputed finance charges (that is, the interest component of the debt service payments due on debt outstanding) that finance companies may include in their reported levels of debt. This adjustment has very little effect on the growth of debt outstanding over the past decade, and reduces consumer installment debt by 6 percent in 1986-1. Table 1 Ratios of Home Mortgage and Consumer Debt Outstanding to Disposable Personal Income, 1975-86* Date 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986-1 . . . . . . . . . . . . Home mortage ratio Consumer installment ratio Other consumer credit ratio Total 0.40 0.41 0.43 0.45 0.47 0.47 0.47 0.46 0.46 0.48 0.51 0.50 0.15 0.15 0.16 0.17 0.17 0.15 0.15 0.14 0.15 0.17 0.19 0.19 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.58 0.60 0.63 0.66 0.69 0.66 0.65 0.64 0.65 0.69 0.74 0.74 . . . . . . . . . . . . *Debt is as of the end of the year. Disposable personal income is for the fourth quarter, at a seasonally adjusted annual rate. Data for 1986 are as of the first quarter, seasonally adjusted. Source: Federal Reserve Board, Flows of Funds. Chart 1 E s tim a te d Ratios of H om e M o rtg a g e and C o n s u m e r In s ta llm e n t D e b t and D e b t S e rv ic e P a y m e n ts to D is p o s a b le P ersonal In c o m e Results Chart 1 depicts the estimated ratio of required debt service payments on home mortgage and consumer installment debt to disposable personal income from 1975-1 through 1986-1, along with the estimated ratio of home mortgage and consumer installm ent debt to income.5 In general, the estimated ratio of debt service payments to income did not increase as much during this period as the debt-to-income measure. While the debt ratio was 22 percent higher in 1986-1 than in 1975-1, the debt service payments ratio was only 8 percent higher. Charts 2 and 3 depict the two estimated ratios by type of household debt (home mortgage versus consumer installment). Chart 2 shows that debt service payments in s ta llm e n t c re d it is c re d it sche duled to be repaid (or with the option of repaym ent) in two or more installm ents. N oninstallm ent cre dit is cre d it sche duled to be repaid in a lum p sum rather than through p e riodic de bt service paym ents. Home m ortgage and consum er installm ent de bt acco unt for 94 percent of the home m ortgage and consum er de bt owed by households at the end of 1985. E s tim a te d de bt service paym ents are also presented in Table B-2 of the box. 0.150 0.145 i i i l m l i i i l m lil o.i40 0.52 1975 76 77 78 79 80 81 82 83 84 85 86 D ebt s e rv ic e paym ents and d is p o s a b le pe rso n a l income are at annual rates. R atios include home m ortgages ow ed by h o useho lds, personal trusts, and n o n p ro fit o rg a n iza tio n s. S ources: Federal R eserve Board, Flow of Funds and Federal R eserve Bank of New Y ork s ta ff e stim ates. FRBNY Quarterly Review/Summer 1986 13 Methodology for Calculating the Debt Service Payments Ratio To estimate debt service payments for each type of consumer installment credit (excluding revolving credit), loans issued in each quarter are assumed to have an interest rate equal to the estimated average contract rate on new loans issued in that quarter, and a maturity equal to the estimated average maturity of new loans issued in that quarter. Under these assumptions, the required monthly debt service payment for loans issued in each quarter is calculated. For each type of debt, prepayment rates are set so that, at the end of each quarter begin ning 1974-1V, the amount of debt outstanding implied by the debt service payments calculations equals the amount of debt outstanding as estimated by the Federal Reserve Board in Statistical Release G.19. For a given type of loan, let E, = extensions during month i, Di = debt outstanding at the end of month i, rt = the interest rate on loans issued in month i, m, = the maturity of loans issued in month i, M, = the maximum (original) maturity on any loan outstanding in month i, Xl(t= debt service payment due in month t on loans issued in month i (assuming no prepayments), = E, * r, / 11 - (1 + r,) _m,]f for i < t « i + mi and = 0 otherwise. Rtt= principal portion of debt service payment due in month t on loans issued in month i (assuming no prepayments), = EiV < 1 + ri) " ‘-7 t(1 + rl)m> -1 ], for i < t ss i + m( and = 0 otherwise. p, = prepayment rate on debt in month t, Pw= percent of loans issued in month i that were prepaid prior to month t, t-1 = 1 - n (1 — p,) for t > i + 1 and j —i +1 = 0 otherwise. Then aggregate debt service payments due in month t equal t- 1 2 X,,t * (1 - PM), i = t-M , while aggregate required principal payments due in month t equal t- 1 T, = X Riit* (1 - P,t). i = t-M , FRBNY Quarterly Review/Summer 1986 Prepayment rates (p,) are calculated as follows. Let diit equal the estimated amount of principal remaining at the end of month t on loans issued in month i, t = E, - X R i/O -P y) j =i+ 1 t - 2 P, * idy., - R ,/(1 -P M)1, j = r+ 1 for t > i; = E, for t = i. That is, the amount of principal remaining at the end of month t on loans issued in month i equals the original amount issued minus the sum of required principal payments made through month t minus the sum of pre payments made through month t. Then, the amount of total debt outstanding as of month t equals t or = I dM. i = t - M, Note that D? is a function of the prepayment rates p,, j =s t. Prior to 1975, all values of p, are required to be equal (i.e., p = p, for all j < 1975). Then p is set so that, for t = December 1974, D, (actual debt) equals D^. At the end of each quarter beginning 1975-1, the monthly prepayment rate during that quarter is set so that t Dt - Dt_3 = 2 [Ej - T, - (D V, - Tj) *p,]f i= t-2 where T( and D?. , are functions of pt, for i = t-2 ,t. That is, the prepayment rate during each quarter is set so that the actual change in debt equals the estimated value of extensions during that quarter minus required principal payments minus prepayments. Estimating debt service payments using the method described above requires data on extensions (Et), debt (Dt), average (original) maturities (mt), and interest rates (rt) for each type of loan. The principal data sources for these series are listed in Table B-1. Unfortunately, data are not available for each series for all time periods. In particular, for mobile home loans, a maturity series is available only for credit on new homes at finance com panies, and only through 1982-IV. Maturities on all mobile home loans are assumed to equal maturities on loans made by finance companies, and after 1982, are assumed to increase at the same rate as the maturities on new car loans. (Between 1975 and 1982, average maturities on new car loans rose 21 percent, compared with 29 percent for new mobile home loans.) Also, the average maturity on all mobile home loans is assumed to equal 90 percent of the average maturity on new mobile home loans, since data reported in the American Methodology for Calculating the Debt Service Payments Ratio (continued) Bankers Association’s Retail Bank Credit Reports indi cates that maturities on used mobile home loans are lower than on new home loans. Data on average maturities on “ other consumer installment” loans are very limited. Through 1982-IV, finance companies reported the average maturity on loans for “ other consumer goods” (excluding autos, mobile homes, and recreational vehicles). This average maturity increased 44 percent between 1975 and 1982. While it corresponds to only a part of the “ other con sumer installment” category, this maturity series does suggest that average maturities on other consumer loans, like maturities on auto and mobile home loans, may have risen during the 1970s and early 1980s. On the basis of these data, maturities on other consumer installment loans are assumed to have increased at the same rate as used car loans (roughly 31 percent between 1975 and 1982). Data on extensions of consumer installment credit after 1982 were collected only from finance companies. Therefore, data on changes in outstanding debt are used along with estimated repayments (including prepayments) of principal, to estimate extensions of each type of debt in each quarter after 1982. The prepayment rates after 1982 are assumed to equal their average values during 1982. That is, Er = Dt - D,_, + Tt + p82*(D,_1 - Tt), where E? equals estimated extensions of debt in month t and p82 equals the average value of p, during 1982. While the estimated prepayment rates varied somewhat from quarter to quarter, they did not exhibit a trend over the 1970-82 period, and the averages during 1982 did not differ much from the averages during the preceding years. Predicting post-1982 prepayment rates using a more sophisticated moving-average process would not change the debt service estimates considerably. According to the Federal Reserve Board, finance companies generally include the interest component of debt service payments owed in their reported holdings and extensions of debt. Therefore, in calculating debt service payments on loans made by finance companies, all data on extensions and debt outstanding that finance companies report to the Federal Reserve Board are assumed to include not only the principal but also the interest component of the debt service payments owed. Actual extensions by finance companies (i.e., excluding Table B-1 Principal Data Sources For Debt Service Payments Calculations Debt ou tstanding — Federal Reserve Board, Flow o f Funds (home m ortgage debt). — Federal Reserve Board, S tatistical Release G.19 (consum er installm ent debt by holder and by type). Extensions — U.S. D epartm ent of H ousing and Urban Developm ent, “ Survey of M ortgage Lending A ctivity,” M onthly M ortgage Loan Transactions and Com m itm ents, 1970-79 (home m ortgage originations). — Federal Reserve Board, S tatistical Release G.19 (consum er installm ent extensions through 1982). — Federal R eserve Board, S tatistical Release G.20 (consum er installm ent extensions at finance com panies). M aturities — Federal R eserve Board, S tatistical Release E.4 (average m aturities on new auto loans at com m ercial banks through 1982). — Federal Reserve Board, Statistical Releases E.4 and G.19 (average maturities on new car and used car loans at finance companies). — Federal Reserve Board, Statistical Release E.10 (average maturities on mobile home loans and other consumer goods loans at finance com panies through 1982). — A m erican Bankers A ssociation, Retail Bank C redit Reports (distribution of m aturities on new car loans, most com mon maximum m aturities on other consum er loans at com m ercial banks). — Federal Home Loan Bank Board, "C onventional Home M ortgage R ates" (average m aturities on conventional new home and existing home m ortgages). Interest rates — Federal R eserve Board, S tatistical Release G.19 (rates at com m ercial banks on new car, personal, and m obile home loans, and on cre d it card plans. Rates at finance com panies on new car and used car loans). — Federal Reserve Board, Statistical Release E.10 (rates at finance companies on mobile home and other consumer goods loans, through 1982). — Federal Home Loan Bank Board, "C onventional Home M ortgage Rates” (average con tract interest rates on fixed-rate loans, percent of loans that have ad ju stable rates). Prepaym ent rates (hom e m ortgages) — Helen F Peters, Scott M. Pinkus, and David J. Askin, "Prepayment Patterns of Conventional Mortgages: Experience from the Freddie Mac P ortfolio," S econdary M ortgage M arkets (February 1984). — M ortgage Security P repaym ent Rate Profile, Salomon Brothers Inc., various issues. — Thomas N. Herzog and Dominick C. Stasufli, "Survivorship and Decrement Tables for HUD/FHA Home Mortgage Insurance Programs as of D ecem ber 31, 1983," U.S. Departm ent of Housing and Urban D evelopm ent (M arch 1984). FRBNY Quarterly Review/Summer 1986 15 Methodology for Calculating the Debt Service Payments Ratio (continued) these precomputed finance charges) are calculated based on the estimated average maturity and interest rate at which loans were issued. That is, x,' = EJ/m,, where Xf is the monthly debt service payment on loans issued by finance companies in month i, and E,' equals extensions as reported by finance companies in month i. Then “ actual” extensions by finance companies are calculated as E l = X<*[ 1 - (1 + r t) " mi] / r. Debt held by finance companies is adjusted as follows. Let Z, equal the “actual” amount of debt held by finance companies, and let D,' equal the amount of debt as reported by finance companies. Then Z, is estimated as D,' * k„ where kt equals the estimated ratio of principal payments remaining, in month t, on loans issued by finance companies, to remaining debt service payments. In general, these adjustments have a very small effect on the estimates of debt and debt service payments. Since not all finance companies include precomputed finance charges in their reported debt holdings, while some other holders of debt may, debt and debt service payments may be slightly understated or overstated. The method used to calculate debt service payments on home mortgage debt is very similar to that used to calculate payments on consumer installment debt. However, for home mortgage debt, prepayment rates are varied by year of origination and by age of the loan, based on average prepayment rates for Federal Housing Administration mortgages and for loans in various mort gage security pools. That is, the prepayment rate in month t [year y(t)] for a loan issued in year y(i) equals sy<i).y(t) * Qt> where sy(i)y(t) is the prepayment rate in year y(t) for mortgages issued in year y(i), and is estimated based on aggregate mortgage prepayment rate data. The adjustment factor q, varies by quarter and is set so that the estimate of debt outstanding implied by the debt service payment calculations equals the estimate of debt outstanding as reported in the Federal Reserve Board, Flow of Funds. That is, the variable q, in the home mortgage calculations is determined in a manner anal ogous to that of p, in the consumer installment calcu lations. The contract interest rate on adjustable rate mortgages was reset annually using the one-year Treasury bill rate. For revolving credit, required principal payments in a given month are assumed to equal a specified fraction of the amount of debt outstanding at the end of the previous month. For revolving credit held by banks and savings institutions, required principal payments are assumed to equal 5 percent of outstanding debt. For 16 FRBNY Quarterly Review/Summer 1986 credit held by retailers, this fraction is assumed to equal 8 percent, while for credit held by gasoline companies, this fraction is assumed to equal 20 percent. These assumptions are based on the required minimum pay ment schedules of various bank, retail, and gasoline company credit cards. Table B-2 Estimated Home Mortgage and Consumer Installment Debt Service Payments In billions of dollars Home m ortgage C onsum er installm ent Total 1 9 7 5 - 1 ....................... ........... 12.3 1975-11....................... ........... 12.6 1975-111 ............................... 12.9 1975-IV ........... 13.2 1976- 1 ........... 13.6 1976-11....................... ........... 14.0 1976-111 ............................... 14.3 1976-IV ........... 14.8 1977- 1 ........... 15.3 1977-11....................... ........... 15.8 1977-111 ............................... 16.4 ........... 17.0 1977-IV 1978- 1 ........... 17.5 1978-11....................... ........... 17.8 1978-111 ............................... 18.5 ........... 19.4 1978-IV 1979- 1 ........... 20.2 1979-11....................... ........... 21.1 1979-111 ............................... 22.0 1979-IV ...........23.0 1980- 1 ........... 23.9 1980-11....................... ........... 24.8 1980-111 ............................... 25.6 1980-IV ........... 26.5 ........... 27.4 1981- 1 1981-11....................... ........... 28.2 1981-111 ........... 29.2 1981 -IV ............................... 30.0 ...........30.8 1982- 1 1982-11....................... ........... 31.5 1982-111 ............................... 32.2 ........... 32.6 1982-IV 1983- 1 ........... 33.2 1983-11.................................. 33.8 1983-111 ............................... 34.7 1983-IV ...........35.9 1984- 1 ........... 37.2 1984-11....................... ...........38.6 1984-111 ........... 40.0 1984-IV ...........41.3 1985- 1 ...........42.4 1985-11....................... ........... 43.3 1985-111 ...............................43.9 1985-IV ...........44.8 1986- 1 ...........45.6 31.8 31.5 31.7 32.0 32.5 32.5 33.1 33.8 35.2 36.2 37.1 38.2 40.0 40.9 42.8 44.1 45.8 46.6 48.1 49.0 50.3 50.4 50.8 51.4 51.7 52.2 53.7 54.2 54.2 53.5 54.4 55.4 55.6 55.2 57.3 59.6 62.5 63.3 67.1 69.7 73.1 74.2 77.2 80.1 83 4 44.1 44.1 44.6 45.2 46.1 46.5 47.5 48.6 50.5 52.0 53.5 55.3 57.5 58.7 61.3 63.4 66.0 67.7 70.1 72.0 74.2 75.1 76.4 77.9 79.1 80.4 82.8 84.2 85.0 85.0 86.5 88.0 88.1 89.1 92.0 95.4 99.7 101.9 107.1 110.9 115.5 117.5 121.1 124.9 129.0 Quarter on home mortgages increased steadily over the past decade, and grew faster, on average, than home mort gage debt outstanding, particularly during the 1980-82 period. Chart 3 illustrates that consumer installment debt increased much more sharply, on average, than debt service payments. While the consumer installment debt ratio was 24 percent higher in 1986-1 than in 1975-1, the debt service payments ratio was 4 percent lower. The debt ratio was 11 percent higher in 1985 than in 1979 (its peak during the previous business cycle), while the debt service payments ratio was 3 percent higher in 1986-1 than at its previous peak. Factors that affect debt and debt service payments Changes in a variety of factors— amounts of loans issued, maturities of new loans, interest rates, and inflation—caused the debt service payment and debt ratios to grow at different rates over the past decade. Understanding how these various factors affect debt and debt service payments, both in the short and long run, will aid in the assessment of consumers’ current debt burden as well as in projecting how quickly debt and debt service payments are likely to grow in the future.6 To illustrate the effects of these factors, examples are presented based on the “ ty p ic a l” loan, where the required monthly debt service payment is constant (in nominal terms) over the loan maturity.7 That is, the principal portion of the monthly debt service payment increases over time as the interest portion decreases. Effects of an increase in loan extensions An increase in monthly consumer borrowing will lead, in the long run, to proportionate increases in aggregate debt service payments and debt outstanding, assuming that interest rates and m aturities remain constant. However, in the short run, debt will increase much more quickly than debt service payments. For example, assume that in each month prior to some month m0, $100 of new loans is issued at a 36-month maturity and C hart 2 Estim ated R a tio s of H o m e M o rtgag e D e b t and D e bt S e rv ice P a y m e n ts to D isposable P e rs o na l In c o m e R atio R atio 0 . 6 ------------------------------------------------------------------------------------- 0.08 Home m ortgage de bt 0 .5 ---------------------- ■*------- Scale ------------------------- - - o.07 0.4 l - V ------------------------------------------------------------- — 0.06 0 . 3 __________________Home m ortgage d e b t------------------- q .05 se rv ic e paym ents S c a le ------- ► 0 . 2 -------------------------------------------------------------------------------------- 0.04 0 ! 11 I I I I I I 1 1 1 . 1 1 L 1 I 1 1 1 I I I I 1 I I I I I I I I 1 I I I 1 I I I 1 I I I 1975 76 77 78 79 80 81 82 83 84 I I I p 0 3 85 86 D ebt se rvice paym ents and d is p o s a b le personal incom e are at annual ra tes. Ratios includ e home m o rtga ges ow ed by ho useho lds, personal tru sts, and n o n p ro fit o rg a n iz a tio n s . S ources: F e deral R eserve Board, Flow of Funds and Federal R eserve Bank of New Y ork staff estim ates. C hart 3 E s tim a te d Ratios of C onsumer In s ta llm e n t Debt and D e b t S e r v ic e Payments to D is p o s a b le Personal Incom e R atio 0.18 0.16 0.14 *The attributio n of changes in d e b t and de bt service paym ents to the factors listed above does not take into account the effect that one factor, such as interest rates, may have on another factor, such as loan extensions. Nevertheless, this exercise is useful in explaining why de bt and de bt service paym ents grow at different rates over time. Explaining how inflation and interest rates affect borrow ing p a ttern s is a sub je ct for future research. 0.12 0.10 0.08 7Most consum er loans are issued at a fixed interest rate for a given m aturity and require a de bt service paym ent that is constant over time. The main exce ptions are revolving cre dit, with required debt service paym ents that may de clin e over time, and adjustable rate loans, with de bt service paym ents that may vary over the life of the loan in response to changes in interest rates. Most of the exam ples presented below illustra ting the relation between de bt service paym ents and debt ou tstand ing generalize to cases where the interest rate varies over the life of the loan. Debt service paym ents and d is p o s a b le personal incom e are at annual rates. Source: F ederal R eserve Bank o f New Y ork sta ff estim ates. FRBNY Quarterly Review/Summer 1986 17 an interest rate of 0.5 percent per month, with no pre payments. Then prior to month mOJ the rate of amorti zation implied by the loan terms will generate an aggregate debt level of $1904, and aggregate debt service payments of $109.50 per month. Now suppose that beginning in month m0, the amount of new loans issued doubles to $200 per month. In the long run {i.e., after 36 months have elapsed), both debt and debt service payments will double as well. However, in the short run, for example after only six months have elapsed, debt service payments will have increased less than 17 percent, while debt outstanding will have increased over 33 percent (Table 2). That is, the ratio of the remaining principal on loans issued in the past six months to total debt is, in general, greater than the ratio of debt service payments on loans issued in the past six months to total debt service payments. Because much of the original principal on the “ older” loans has already been repaid, those loans account for a relatively small portion of debt outstanding. The debt service payment on the “ older” loans, however, remains con stant over the maturity of the loan. Therefore, recently issued loans account for a larger proportion of debt than of debt service payments, causing changes in the rate of borrowing to affect debt more quickly than debt service payments. Chart 4 presents the ratios of estimated extensions of home mortgages and consumer installment loans to income from 1975-1 to 1986-1.8 Extensions rose, relative to income, during the 1975-79 expansion, declined during the 1980-82 downturn, and then increased again. The cycles in extensions account for the cycles in debt and debt service payments. However, with the exception of revolving credit, the estimated ratios of extensions to income have not been higher, on average, in the current expansion than in the 1975-79 expansion. While Chart 4 depicts gross rather than net extensions, it neverthe less suggests that a trend in the rate of borrowing does not, at least by itself, explain why the debt ratio is cur rently at an all-time high. Table 2 also demonstrates that if the amount individ uals borrow each period increases, debt outstanding (as well as debt service payments) will increase at a decreasing rate over time. Thus, for example, even if borrowing were to continue at the relatively high level of the past year, the growth in the consumer installment •Data on extensions of consum er installm ent loans were collecte d after 1982 only from finance com panies. E xcept for revolving credit, extensions of loans were estim ated after 1982 based on changes in debt outstanding, estim ated required prin cip a l payments, and estim ated prepaym ents (box). Extensions of revolving de bt were not estim ated after 1982. C h a rt 4 Table 2 E s tim a te d R a tio s of H o m e M o r t g a g e and C o n s u m e r In s ta llm e n t Loan E x te n s io n s to D is p o s a b le P e rs o n a l In c o m e The Effects of Changes in Amount of New Loans and Maturity on Aggregate Debt and Debt Service Payments R atio 0.14 In percent C o n s u m e r in s ta llm e n t Increase in loans Issued* f \ r \ \\1 ji1 0.12 * / II i >* / \ / ^ I 1 /A ¥\\ 1/ i 0 .0 8 (e xclu d in g re v o lv in g ) H I i t r\ V ,/W \ \ 0 .0 4 \ Home m o rtg a g e lA v o ria in a tio n s \ 1 a * ' / \ Av v' / V » \* fl / ' ,V */ \ / \ / / V / S I 1 1 1 L. 1 L1 1 1_LL 11[„„l 1,1 1, 1 1.1 1,1 1 1 1 1 1 1 1 1 1 1 1J - l l .,1 i 11 1 1.1 i 1 1 1975 76 77 78 79 80 81 82 8 3 84 85 86 S o u rc e s : U.S. D epa rtm e nt of H ousing and U rban D e ve lo p m e n t (hom e m ortgage o rig in a tio n s ); F e deral R eserve B oard (c o n s u m e r in s ta llm e n t e xte nsions th ro u g h 1982); and Fe deral R eserve Bank o f New Y o rk s ta ff e s tim a te s (c o n s u m e r in s ta llm e n t e x te n s io n s a fte r 1982). 18 FRBNY Quarterly Review/Summer 1986 Between month 0 and month: S ix ............. Twelve . . . Eighteen . . Twenty-four. Thirty . . . Thirty-six Forty-two . Forty-eight . Increase in m aturityt Change in Change in Change in Change in debt debt service debt debt service outstanding payments outstanding payments 33.9 57.7 33.3 16.7 76.3 89.7 50.0 66.7 97.6 100.0 100.0 83.3 100.0 100.0 100.0 100.0 0.8 2.9 6.4 11.4 17.8 25.7 31.9 33.7 - 3.7 - 7.6 -1 1 .4 -1 5 .2 -1 9 .0 - 22.8 - 10.0 2.9 •Prior to month 0, $100 of new loans is issued in each month, at a maturity of 36 months and an interest rate of 0.5 percent per month. Beginning in month 0, $200 of new loans is issued per month. No loans are prepaid. fP rior to month 0. $100 of new loans is issued in each month, at a maturity of 36 months and an interest rate of 0.5 percent per month. Beginning in month 0, new loans are issued at a maturity of 48 months. No loans are prepaid. debt ratio (and eventually in the home mortgage debt ratio) would slow considerably, assuming maturities remain constant. Effects of changes in maturities Between 1975 and 1985, the maturities on certain types of consumer loans increased substantially. For example, the average maturity on new car loans issued by finance companies increased from 38 months in 1975 to 52 months in 1985 (Table 3). This lengthening of maturities may account for much of the difference between the growth patterns of consumer installment debt and debt service payments, since in the long run a change in maturity has a much larger effect on debt than on debt service payments. An increase in maturity at first decreases aggregate scheduled repayments of principal, thereby increasing aggregate debt outstanding. While aggregate debt service payments decrease in the short run, they may be slightly higher in the long run. To illustrate this point, consider a situation where issuances of new debt are constant over time, the interest rate is also constant at 0.5 percent per month, no prepayments occur, and the maturity at which debt is issued equals 36 months prior to month mD and 48 months in month mQ and thereafter (Table 2). Aggregate debt service payments at first decrease, because the debt service payment on each loan issued at a maturity of 48 months is less than the payment on each loan issued at a maturity of 36 months. However, debt issued at the longer maturity remains outstanding for a longer period of time. Thirty-six months after the increase in maturity, aggregate debt service payments begin to increase, since debt issued 36 months ago is not yet paid off. Forty-eight months after the increase in matu rity, aggregate debt service payments have reached their new long-run level, and are slightly higher than aggregate debt service payments in month m0, reflecting the fact that interest is owed on a larger amount of debt. In the more general case where prepayments occur, a change in maturity may have a different effect on debt and debt service payments than in the “ no prepay m ents” example, depending on how prepayments change. However, as in the example presented above, the effect on debt will be very different from the effect on debt service payments. To determine how much of the divergence between the consumer debt and debt service payment series has been due to increases in the maturities of consumer loans, the debt and debt service payments ratios were reestimated, assuming that maturities did not change after 1974.9 That is, household debt and debt service *Data on m aturities of m obile home and "o th e r consum er installm ent” loans are lim ited, and a num ber of assum ptions were made in con structin g the average m aturity series (box). Table 3 Average Maturities on Auto Loans Issued by Finance Companies In months Year New car loans Used car loans ..................... 38 1975 1976 ..................... 39 1977 ......................41 1978 ..................... 43 1979 ......................44 1980 ......................45 1981 ......................45 1982 ......................46 1983 ......................46 1984 ..................... 48 1985 ......................52 1 9 8 6 * .............................................. ......................51 29 30 31 33 34 35 36 37 38 40 41 43 ’ January through May. Source: Federal Reserve Board, Statistical Release G.19. payments were estimated under the assumptions that the amounts borrowed, and interest and prepayment rates, were equal to their actual levels, but that after 1974, maturities were equal to their 1975-1 levels. If maturities had not increased, the consumer installment debt ratio would have risen only 8 percent between 1975-1 and 1986-1 rather than 24 percent (Chart 5). Furthermore, the debt service payments ratio would have increased 3 percent, instead of falling 4 percent.10 Thus, increases in m aturities on consum er loans account for much of the difference between changes in the consumer installment debt ratio and changes in the debt service payments ratio over the past decade. The growth in maturities also explains, to some extent, why the consumer installment debt ratio was higher in 1985 and 1986-1 than during the 1975-79 expansion, while extensions of most types of consumer loans relative to income are estimated to be lower. If maturities had not increased after 1974, the consumer installment debt ratio would have increased only 2 per cent between 1979 and 1985 rather than 11 percent, while the ratio excluding revolving credit would have decreased 7 percent rather than increased 5 percent. Changes in maturity account for virtually none of the growth in home mortgage debt or debt service payments between 1975-1 and 1986-1, partly because maturities on home mortgages did not exhibit a pronounced trend 10By assuming that prepaym ent rates remain the same under the shorter maturities, the amounts prepaid decrease, since less debt is outstanding at a given time. As a result, aggregate debt service payments are slightly higher (rather than slightly lower, as in the example given above) under the shorter maturity. FRBNY Quarterly Review/Summer 1986 19 C h a rt 5 E s tim ated Ratios of C o n s u m e r In stallm ent D e b t and D e b t S e rv ic e P a y m en ts to D is p os ab le P ersonal In c o m e Ratio 0.18 0.16 0.14 0.12 0.10 0 .0 8 1975 76 77 78 79 80 81 82 83 84 8 5 86 Debt se rv ic e paym ents and d is p o s a b le p e rsona l incom e are at annual ra tes. S o u rce : F e deral R e s e rv e Bank of New Y ork s ta ff e stim a te s . during this period. Furthermore, since home mortgages are issued at long maturities, any change in the maturity on new loans will affect debt outstanding very slowly. Table 2 illustrates that the response of debt and debt service payments to an increase in maturity is gradual. As a result, recent increases in maturities will continue to affect the growth of consumer installment debt and debt service payments over the next few years. For example, if borrowing relative to income continues at the level of the past year through 1987-IV, and maturities remain the same, the consumer installment debt ratio will increase 4 percent between 1986-1 and 1987-IV. However, if maturities had not increased after 1974, the debt ratio would increase less than 1 percent.11 Effects of changes in inflation and nominal interest rates Large fluctuations in inflation and nominal interest rates over the past decade also affected debt and debt service payments differently (Table 4). An increase in inflation— holding real interest rates, maturities, and am ounts borrowed (in real term s) constant, and 11For this exam ple, interest rates on consum er loans were assum ed to fall roughly 100 basis points betw een 1986-1 and 1987-IV. P repaym ent rates were assum ed to remain at their average levels over the last fou r quarters, w hile incom e was assumed to grow roughly 6 pe rcent per year. 20 FRBNV Quarterly Review/Summer 1986 assuming no prepayments—decreases the real values of both debt and debt service payments in the long run. However, if nominal interest rates rise with inflation, real debt service payments decline much less than real debt.12 To illustrate this point, suppose that prior to month mD the inflation rate is zero, and beginning in month mD it increases to 0.5 percent per month (Table 5). The real interest rate is set at 0.5 percent per month and the maturity at 20 years. The amount borrowed is constant in real terms and no prepayments occur. In the long run, debt will decrease 25 percent, in real terms, while the real value of aggregate debt service payments will decrease only 10 percent. If the interest rate is adjust able rather than fixed, real debt service payments will rise considerably in the short run, before declining. In either case, the change in the rate of inflation affects real debt and debt service payments quite differently. To demonstrate how the level of inflation has affected the debt and debt service payments ratios over the past decade, the ratios were reestimated assuming that in every year after 1974, the inflation rate was 2 percent lower than its actual value. Prepayment rates, and the real values of interest rates, the amounts borrowed, and income were left unchanged.13 Under this “ lower infla tion” scenario, the growth of the home mortgage debt ratio between 1975-1 and 1986-1 increases from 21 to 32 percent, but the growth of the debt service payments ratio decreases from 36 to 32 percent.14 Thus, the high level of inflation experienced during the past decade slowed the growth of the debt ratio, while raising the debt service payments ratio. A change in the long-term inflation rate affects the consumer installment ratios much less than the home mortgage ratios, because consumer loans have shorter maturities. If inflation and nominal interest rates had been 2 percent lower after 1974, the consumer install ment debt ratio (excluding revolving credit) would have 12H olding the real interest rate constant, an increase in inflation will reduce the long-run level of real debt service paym ents by increasing the rate at w hich real paym ents on the de bt are m ade. That is, with no inflation, the required debt service paym ent on a given loan is constant in real term s over the life of the loan. With inflation, the real value of an in divid ual's m onthly de bt service payment decreases over time, since the nominal value is constant. Thus, the real value of the loan is paid off more qu ickly in the latter case, and therefore, the long-run level of ag gregate de bt service paym ents is lower in real term s. 13Nominal interest rates on loans issued after 1974 were reduced by 2 percent. Ex p o st real interest rates on loans issued before 1975 are increased by this simulation. 14By assum ing that prepaym ent rates remain the same under the "low e r inflatio n" scenario, the am ounts prepaid (in real term s) increase, since more debt is outstanding in real term s. Therefore, the decrease in inflation lowers the debt service paym ents ratio slightly, rather than increasing it slightly as in the exam ple presented above. Table 4 Interest Rates on Home Mortgages and Consumer Loans, and the Rate of Growth of the Consumer Price Index (CPI), 1975-86 In percent Interest rates* Home m ortgagesf Year Auto loans at commercial banks}: Personal loans at commercial banks§ Bank credit cards Rate of growth of the CPI// 1975 ....................... .................... 88 11.4 13.1 1976 ....................... .................... 8.8 11.1 13.0 17.2 17.1 4.9 6.8 1977 ....................... 7.0 .................... 8.8 10.9 13.0 16.9 1978 ....................... .................... 9.3 11.0 13.2 17.0 9.0 1979 ....................... .................... 10.5 12.0 13.9 17.0 1980 ....................... .................... 1 9 8 1 ....................... .................... 12.3 14.3 15.5 14.2 16.5 18.1 17.3 17.8 13.2 12.4 1982 ....................... .................... 14.5 16.8 18.6 18.5 1983 ....................... 1984 ....................... .................... 12.2 18.8 3.8 11.9 13.9 13.7 16.7 .................... 16.5 18.8 4.0 8.9 3.9 1985 ....................... .................... 11.1 12.9 15.9 18.7 3.7 .................... .................... 10.4 11.9 15.2 18.4 - 0.2 19861 ‘ Rates are annual averages of monthly data. Except for home mortgages, data are for midmonth of quarter only. fContract rate on fixed-rate mortgages for new homes. ^Before 1983 the maturity for new car loans is 36 months. Beginning in 1983 it is 48 months. §Loans with maturities of 24 months. //From December to December. Rate for 1986 is from December to June, annualized. fThrough the second quarter. Table 5 The Effects of Changes in Interest and Inflation Rates on Aggregate Debt and Debt Service Payments In percent Increase in the interest rate* Increase in the inflation ratef Percent change in debt outstanding Between month 0 and month: Percent change in debt outstanding Percent change in debt service payments S i x ........................................ 0.0 1.4 T w e lv e ................................. T w e n ty -fo u r....................... T h irty-six.............................. F o rty-e ig h t.......................... S ix ty .................................... One-hundred twenty . . . . One-hundred eighty . . . . Two-hundred f o r t y ............. 0.1 0.2 0.5 0.9 1.5 5.6 11.0 14.2 2.7 5.4 8.1 10.8 13.4 26.9 40.3 53.8 Fixed rate loans - 2.9 - 5.5 -1 0 .1 -1 4 .0 -1 7 .1 -1 9 .7 -2 5 .8 -2 5 .8 -2 5 .0 Adjustable rate loans - 1.9 - 3.8 - 7.2 -1 0 .2 -1 2 .8 -1 5 .0 -2 2 .2 -2 4 .6 -2 5 .0 Percent change in debt service payments Fixed rate loans - Adjustable rate loans 1.6 26.1 - 3.0 - 5.6 - 7.9 - 9.7 -1 1 .2 -1 4 .7 —13.8 -1 0 .4 23.7 19.4 15.4 11.8 8.6 - 3.1 - 8.8 -1 0 .4 'Prior to month 0, $100 of new loans is issued in each month, at a maturity of 240 months (20 years) and an interest rate of 0.5 percent per month. Beginning in month 0, the interest rate on new loans doubles to 1 percent per month. No loans are prepaid. fP rior to month 0, $100 of new loans is issued in each month, at a maturity of 240 months and an interest rate of 0.5 percent per month. Beginning in month 0, the rate of inflation increases from 0 to 0.5 percent per month. The real value of new loans issued and the real interest rate on new loans remain the same. Debt outstanding and debt service payments are measured in real terms. No loans are prepaid. FRBNY Quarterly Review/Summer 1986 21 Table 6 Interest Rates on Consumer Loans Minus Increases in the Consumer Price Index (CPI), 1975-86 In percent Interest rate minus average annual increase in CPI* Year 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 Auto loansf . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 3.8 1.0 - 1.0 0.6 5.8 10.6 13.2 11.1 10.3 10.2 8.9 Personal loanst Bank credit cards§ 7.1 6.8 4.8 0.9 0.8 5.2 11.6 14.7 13.5 12.8 13.5 12.4 10.9 11.2 10.5 7.0 2.9 5.8 9.9 14.7 14.3 15.2 15.7 16.2 'Interest rates are as of February. CPI increase is from February to February. fN ew auto loans at commercial banks. Prior to 1983, maturity on loan equals 36 months. Beginning in 1983, it equals 48 months. Increase in CPI is averaged over first three years of the loan. ^Twenty-four month personal loans at commercial banks. Increase in CPI is averaged over the two-year loan period. §lnterest rate on credit cards at commercial banks minus increase in CPI over the first year of the loan. increased an additional 1 percent between 1975-1 and 1986-1. The corresponding debt service payments ratio would have been virtually unaffected by the change in inflation. Fluctuations in the rate of inflation also explain why the home mortgage debt ratio is currently at an all-time high even though mortgage originations relative to income are not. Although home mortgage originations have been lower relative to income during the current expansion than during the 1975-79 expansion, the rate of inflation has been much lower as well. If, for example, inflation had been constant at 6 percent after 1975, with actual borrowing patterns left unchanged in real terms, the home mortgage debt ratio would have been 3 per cent lower in 1985-IV than at its previous cyclical peak, rather than 6 percent higher.15 Therefore, the high levels of borrowing during the late 1970s may have been, in part, a response to high levels of inflation, but the high inflation rates offset the impact of the borrowing on the debt ratio. 1sln this sim ulation, the spreads between interest rates and inflation w ere held con stant at the ir 1976 levels, w hile prepaym ent rates were kept at the ir actual values. 22 FRBNY Quarterly Review/Summer 1986 Effects of changes in real interest rates Some of the differences between the debt service pay ments and debt outstanding series may be attributed to the effects of changes in real interest rates, i.e., movements in nominal interest rates independent of movements in inflation or expectations of inflation. An increase in the interest rate at which debt is issued, holding the maturity, amounts borrowed, and inflation rate constant, will increase debt service payments as well as debt outstanding. Debt outstanding will increase because, at the higher interest rate, principal is paid back more slowly. That is, the higher the interest rate, the lower the first required principal payment on a given loan, and the higher the final principal payment. How ever, the effect of a change in interest rates on debt service payments is much larger than the effect on debt outstanding. For example, an increase in the interest rate from 0.5 to 1 percent per month— holding the maturity constant at 20 years, the amounts borrowed each period constant, and assuming no prepayments— will increase the debt service payments ratio 54 percent, and the debt outstanding ratio 14 percent, in the long run (Table 5). Determining how real interest rates on new loans have changed over the 1975-86 period is problematic, even on an ex post basis, since the real interest rate on a given loan depends on when the loan is repaid. How ever, very rough proxies for ex post real interest rates on certain types of consumer loans were calculated by subtracting the average increase in the Consumer Price Index (CPI) over the approximate term of the loan from the interest rate on the loan.16 The results of these calculations suggest that real interest rates on consumer loans have been rising in recent years (Table 6). These relatively high rates have contributed to the recent rise in the debt service payments ratio. Interest rates have a much larger impact on the home mortgage debt service payments ratio than on the consumer installm ent debt service payments ratio. Because home mortgages have longer maturities, prin cipal repayments make up a small part of mortgage debt service, and interest payments make up a large part. Lowering interest rates on all loans issued after 1974 by 200 basis points would reduce the home mortgage debt service payments ratio by 13 percent in 1986-1, while decreasing the consumer installment ratio by only 3 percent. Effect of the tax system on debt service payments Interest payments accounted for an estimated 45 per cent of aggregate debt service payments in 1986-1, compared with 31 percent in 1975-1. This growth reflects 16The inflation rate was assum ed to equal 3.2 percent (annualized) during the second half of 1986 and 4 percent during 1987. increases in nominal interest rates as well as maturities. Because interest payments are tax deductible, the increase in debt service payments overstates the increase in households’ debt burden. For example, if all households with debt itemized their deductions and faced a marginal income tax rate of 25 percent, the debt service payments ratio on a post-tax basis would have increased 3 percent between 1975-1 and 1986-1 rather than 8 percent. The effects of changes in maturity, inflation, and interest rates are different on a post-tax basis. A lengthening of maturity increases the proportion of aggregate debt service payments that represent interest, as does an increase in inflation and nominal interest rates. Therefore, in the examples given previously, assuming no prepayments, an increase in maturity would result in a larger percentage decrease in post tax debt service payments than in pre-tax payments in the short run, and, in the long run, would lead to a smaller percentage increase. Similarly, the percentage decrease in real debt service payments that results from an increase in inflation and nominal interest rates would be even larger on a post-tax than on a pre-tax basis, while the short-run increase in real payments on adjustable rate debt would be less. Conclusion Because changes in maturities, inflation, and interest rates affected debt and debt service payments differ ently, the household debt service payments ratio grew much less quickly than the debt ratio over the past decade. Increases in maturities on consumer loans kept debt service payments relatively low, and at the same time increased debt outstanding. Inflation increased the divergence between the home mortgage debt and debt service payments ratios, causing debt service payments to grow faster than debt. The debt and debt service payments ratios may change quite differently in the future as well. For example, an increase in nominal interest rates and inflation would reduce real debt but increase real debt service payments on adjustable-rate debt in the short run, assuming real borrowing remains constant. If, on the other hand, inflation remains at a low rate, and interest rates keep declining, the debt ratio will probably continue to grow much more quickly than the debt service payments ratio. The recent rise in the debt ratio to an all-time high is not simply due to unusually high levels of new loan extensions relative to income. Rather, longer maturities and lower inflation have contributed substantially to the increase. During the 1970s, consumers may have demanded longer maturities partly to offset the impact of higher inflation and nominal interest rates, since these factors increase the rate at which debt is amor tized, in real terms. In recent years, consumers may have used long maturities partly to offset the impact of high real interest rates on monthly debt service pay ments. In the future, if consumers choose shorter maturities in response to lower inflation and interest rates, the debt ratio may decrease as well, with the debt service payments ratio increasing slightly. Lynn Paquette FRBNY Quarterly Review/Summer 1986 23 Financial transactions and the Demand for Ml Over the past few years, trading volume in the financial markets has increased at a very rapid rate.1 At the same time, M1’s growth has been considerably stronger than would have been expected given the performance of GNP, creating doubt about the adequacy of GNP as a measure of the total dollar volume of transactions. In other words, GNP may understate the overall transac tions demand for M1 when trading in financial instru ments is increasing at a considerably faster rate than GNP. If this is the case, some economists suggest using the dollar volume of debits to checking accounts to approximate the transactions demand for money instead of GNP, b ecause debits to checking accounts occur for all types of transactions, financial and nonfinancial, and not just for sales of goods and services to final purchasers.2 To evaluate whether financial trading has significantly increased the demand for M1 by causing debits to checking accounts to grow much more rapidly than GNP, two links should be established. First, a relationship needs to be found between the volume of financial transactions and debits to transactions accounts; and 1This has been true not only for the established stock and bond markets, but also in relatively new markets such as options, swaps, and futures. For more detail, see the 1985 Annual Report, Federal Reserve Bank of New York, page 18, and “ Demystifying Money’s Explosive Growth,” Morgan Economic Quarterly (March 1986), pages 10-13. 2See, for example, John Wenninger, “ Reserves Against Debits," this Quarterly Review (Winter 1982-83). Also see, Ralph C. Kimball, “ Wire Transfers and the Demand for Money,” New England Economic Review, Federal Reserve Bank of Boston (March-April 1980); Charles Lieberman, "The Transactions Demand for Money and Technological Change,” Review of Economics and Statistics (August 1977); and Alexander J. Field, “Asset Exchanges and the Demand for Money, 1919-29,” American Economic Review (March 1984). 24 FRBNY Quarterly Review/Summer 1986 second, debits should be a better proxy than GNP for the transactions that affect money demand. If these points cannot be established, then it is somewhat less clear that financial transactions are affecting the growth of M1 to a large degree. In this article, we examine the data available on these two linkages. By and large, there has been little quarterto-quarter correlation between trading volume and debits to checking accounts. In addition, the data suggest that in the longer run, debits are not a significantly better measure than GNP of those transactions that matter for money demand. Finally, even in the case of 1985, when debits did track M1 growth b e tte r than GNP, it is uncertain whether financial transactions were the pri mary reason debits predicted M1 growth more accu rately. The growth rates of debits and GNP can diverge for reasons other than financial transactions. Trading volume, debits, and GNP Chart 1 shows the explosive growth of two readily available data series that are sometimes taken as indi cating the general growth of financial transactions: the dollar volume of transactions on the New York Stock Exchange and the dollar volume of trading by dealers in U.S. Government securities.3 Mirroring this explosive growth in financial trading volume has been the growth of debits to checking accounts. And the growth of M1 has generally been faster than expected since the early 1980s, when financial transactions and debits began to 3The trends in the trading volume of these segments of the financial markets, of course, may or may not parallel the growth in the volume of financial transactions in all markets—but the data on other financial transactions are rather limited and analysts have been forced to use these two series as an indication of what is happening more generally. accelerate sharply relative to GNP. At first glance, the similarity in these longer-run trends implies that the more rapid growth of trading volume in financial instru ments relative to GNP has contributed to more rapid growth in debits, and this in turn has increased the demand for M1. In theory, of course, an increase in financial transactions should add to the demand for M1, all other factors equal; but how important are financial transactions in practice? Charts 2 and 3 show that financial transactions may not explain much of M1’s growth. Chart 2 compares the long-run trend in the velocity of M1 measured two ways, using debits and GNR Since the late 1950s, velocity measured with debits has increased by a factor of 11, whereas velocity measured with GNP is only about two C hart 1 Growth of Financial Transactions, Debits, and M1 Index (1971-1=100) 800 times greater. Indeed, the extremely rapid (and accel erating) growth of velocity measured with debits sug gests that many financial transactions— such as arranging an overnight repurchase agreement— are undertaken specifically to reduce checking account balances. Since these transactions directly decrease the volume of M1, rather than adding to the demand for M1 as conventional transactions would, they cause velocity (as measured with debits) to rise. That is, these cash management transactions raise velocity by increasing the numerator in the debits/M1 ratio and decreasing the denominator at the same time.4 Another reason the increased volume of debits may not be increasing the demand for M1 can be seen from the components of M1. In recent years, the growth of total debits primarily reflected debits to demand deposit accounts, whereas M1’s growth has been dominated by increases in negotiable order of withdrawal (NOW) accounts (Chart 3). If a larger volume of financial transactions was increasing the demand for M1, we would expect the growth of debits and the greater demand for M1 to show up in the same component of M1, but this generally has not been the case for the 1982-85 period.5 In 1985, however, the demand deposits 4ln addition, increased em phasis on cash managem ent in general and tech nolog ical advances in m onitoring money balances have reduced the level of M1 relative to both the level of GNP and the volum e of debits over time. 5lt could be argued, of course, that the key issue is not w hich com ponent of M 1 con tributed the most to M1’s grow th, but rather w hether the larger volum e of debits to dem and deposits increased the dem and for these deposits beyond what it otherw ise w ould have been. C hart 2 A lternative Measures of Velocity Index 1946 = 1 400 --------------------------------------------------------------------------------------------------------------------------------- 2 5 20 / D ebits/M 1 / ------------ /- 15 / ------------- X ------ 10- G N P /M 1 50- o L llll ll. l- L ll. lll 1 ll 1111II ll 11 111 I ll II 1 1I I 11111111111111 l l l l l - L 1971 72 73 74 75 S o u rce : 76 77 78 79 80 81 82 8 3 84 85 F e d e ra l R eserve B ulletin. pi I I I l l I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I 1946 50 S ource: 55 60 65 70 75 80 85 F e d e ra l R eserve B ulletin. FRBNY Quarterly Review/Summer 1986 25 category did contribute a larger amount to M1 growth than in recent years, and financial transactions could have played a significant role in M1’s very rapid growth last year even though their importance in the longer run is uncertain. In the remainder of this article, we will show that financial transactions added about one per centage point to M1’s growth in 1985. But this estimate is subject to sizable error in either direction.6 Debits, GNP, and the demand for money In making this estimate, a conventional money demand equation (relating real money balances to real transsFor a more tec h n ic a l and d e tailed analysis of the transactions dem and for M 1 that arrives at sim ilar estim ates of the possible effects of fina ncial tra nsactio ns on M1 for 1985, see Peter E. Kretzm er and R ichard D. Porter, “ The Demand for the Narrow A g g regates— Is a Transactions A pproach S ufficient?" Board of G overnors of the Federal R eserve System (July 1986), unpublished. actions, a short-term interest rate, and lagged real money balances) was used. Real debits and real GNP were included as alternative measures of transactions.7 The left side of Table 1 shows that when debits and GNP are included together in the money demand equation at most one performs well. As alternative proxies for total transactions in the economy, they are competing to explain the movements in M1 and one of the two is redundant. Estimated from 1959 through 1973, the coefficient on GNP is significant and has the 7Only the coe fficients on the transactions variable are shown. These equations had all the well known problem s with sta b ility during the mid-1970s, and the results should be interpreted with caution as a result. Moreover, there are problem s with using the real volum e of debits as an alternative measure of transactions. That is, w hile it might be ap prop riate to use the GNP de flator to calcula te real M1 balances and real GNP, it may not be a p prop riate to calcula te real debits this way since de bits contain a com bination of GNP and non-GNP transactions. Chart 3 Growth of T o ta l D ebits and T ra n s a c tio n D e po s its By so u rc e F o u rth quarter to fourth q u a rte r P e rce n t 3 5 ------------------------------------------------------------------------------------------------------------G r o w t h of to t a l de b it s 30- I I Due to de b its to ------- NOW acco unts Due to d e b its to demand d e posits Total de bits 25G r o w t h of M1 t r a n s a c t io n d e p o s i ts (M1 les s c ur re nc y ) 20- | Due to demand d e posits 1982 S ource: 26 |j§ |f | Due to NOW accounts 1983 F e d e ra l R eserve B ulletin. FRBNY Quarterly Review/Summer 1986 M1 tra n s a c tio n s de p o sits 1984 1985 correct sign, whereas the debits variable has a negative coefficient not significantly different from zero. When estimated through 1982, the coefficients retain the same signs as in the earlier period but neither is significant. When the sample period is extended through 1985, the coefficient on debits becomes positive and significant, while the coefficient on GNP turns negative and insig nificant. In contrast, when either debits or GNP is included by itself (Table 1, right side), the estimated coefficient is statistically significant and has the correct sign across all sample periods, giving no clear indication of the better measure of transactions for money demand purposes. Evaluating the regression results over the longer run is complicated by the downward shift in the demand for M1 in the m id-1970s.8 Hence, we reestimated the money demand equation over the 1974 to 1984 period. During this period, both GNP and debits continued to have a strong correlation with M1 when included indi vidually. (Their t-statistics are over 6; Table 1, bottom right, memo.)9 Next, we use this equation to determine whether GNP or debits can track M1 growth more accurately in 1985. The first column of Table 2 shows that the out-of-sample projection based on GNP substantially underestimated M1 growth last year (four percentage points), whereas the projection based on debits (Table 2, far right column) missed by only about two percentage points. Therefore, it seems that financial transactions played a major role, adding roughly two percentage points to M1 growth. But there are reasons to believe that the actual contribution to M1 growth was somewhat less. First, there is an alternative reason why GNP under stated the demand for money. In the United States, a larger volume of goods and services was purchased in 1985 than was produced. If consumers are purchasing more goods, but those goods come from imports or inventories rather than from current production, the demand for M1 increases to make those additional transactions but GNP does not increase. Hence GNP (current production) understates the transactions demand for M1, and M1 appears unusually strong.10 When domestic final demand is substituted for GNP, the resulting error is about three percentage points (Table 2, center column). Therefore, after this adjustment, only about one percentage point of M1 growth remains to be attributed to financial transactions (and perhaps other factors as well). Second, the debits statistics may not adequately capture financial transactions. The link between trading volume and debits on a quarter-to-quarter basis has been weak. Table 3 shows the results of regressing the growth rate of debits on the growth rates of GNP and stock and securities trading volume. The only variable that is statistically significant (that is, has an estimated coefficient with a t-statistic greater than 2.0) is GNP.11 "T o investigate further why the tra ding volum e variables were insignificant, we looked at m onthly data to see if some of the correlation was masked by quarterly averaging. As it turns out, Table 1 Coefficients for Alternative Measures of Transactions in a Standard Money Demand Equation* Included together Years Included separately GNP Debits GNP Debits 1959 to 1973 0.152 (3.3) -0 .0 2 6 (1.7) 0.115 (36) 1959 to 1982 0.067 (1 8 ) -0 .0 1 4 ( 1.1) 0.027 (3 0) 0.041 (3.1) 0.008 (2.5) 1959 to 1985 -0 .0 0 5 (0 .2) 0.015 (2 0 ) 0.046 (4.3) 0.013 (51) -0 .0 3 0 (0 .8) 0.025 (3.1) 0.143 (6.4) 0.030 (8 6 ) Memo: 1974 to 1985 'Standard Goldfeld formulation where the In (real M1) is regressed on In (real GNP), In (short-term interest rate), and the In (lagged real money). Only the coefficients on the transactions variables are reported. Table 2 Money Demand Errors for 1985 Using Alternative Proxies for Transactions* Quarterly growth rates Gross domestic final demand Debits 1985-1.............................................. 3.0 2.1 1.4 •For more detail, see Stephen G oldfeld, “ The Case of the Missing Money,” B rookings Papers on E conom ic A ctivity (1976-111). 1985-11........................................... 2.9 1.8 1.0 1985-111........................................... 7.3 6.0 5.6 •Equations estim ated over this period also have larg er (in absolute value) coe fficients on the short-term interest rate. This helps them track the rapid M1 grow th in 1985 som ewhat better than those estim ated over the longer run. 19 8 5 -IV ........................................... 2.9 1.5 0.4 Average 2.9 2.1 10For more detail, see Law rence J. Radecki and John Wenninger, “ Recent Instability in M 1's Velocity," this Q uarterly Review (Autumn 1985). Quarter GNP .......................... 4.0 ‘ Equations estimated from 1974 to 1984. FRBNY Quarterly Review/Summer 1986 27 This result suggests that a large volume of financial instruments is purchased without debits to the trans actions accounts in M1, or that many financial trades completed during a given day by an individual firm are netted before its demand deposit account is debited. We also tried using trading volume directly in the money demand equation along with GNP. However, it probably is not appropriate to deflate financial trans actions by the GNP deflator (Footnote 7). To avoid this problem, the money demand equation was estimated in nominal terms. The resulting coefficients and the 1985 errors in projecting M1 growth are shown in Table 4. All three measures of transactions have significant coeffi- Table 3 Correlation Between Debits and Financial Transactions Quarterly growth rates Debits = 9.7 + 0.92 (GNP) (2 .8) (2.8) Fedwire activity Debits = 9.7 + 0.88 (GNP) + 0.02 (Stocks*) (2.7) (0.9) (2.8) Debits = 9.4 + 0.89 (GNP) (2.7) (2.6) + .01 (Stocks) + 0.006 (Securities!) (0.8 ) (0 .2) ‘ Dollar volume of transactions on the New York Stock Exchange. fD ollar volume of trading by Government securities dealers. Additional regressions using lags and seasonal dummies did not produce appreciably different results. Table 4 Results When Financial Transactions Are Included Directly in Money Demand Equation* Coefficients GNP ............................................................................... 0.100 (3.2) Stock v o lu m e .................................................................. 0.022 (38) Securities v o lu m e ........................................................... -0 .0 1 7 (2.4) 1985 errors (Quarterly growth rates) 1985-1................................................................................ 1.7 1985-11............................................................................ 3.1 1985-111............................................................................ 6.0 1 9 8 5 -IV ............................................................................ 2.1 Average 3.2 ......................................................................... 'Nominal rather than real values were used in this equation for money demand. As in the previous tables, only the coefficients for the variables being studied are reported. The sample period was from 1974 to 1984. Stock volume and securities volume are defined in the same way as in Table 3. 28 FRBNY Quarterly Review/Summer 1986 cients. The coefficients on GNP and on the volume of trading in the stock market also have the correct (posi tive) sign. The coefficient on the volume of trading in Government securities is negative, again implying that many trades are done for the purpose of managing money balances more efficiently (repurchase agree ments, for example). In any case, this equation does not track M1 growth in 1985 very well. Its average error was three percentage points, compared with an error of two percentage points when total debits were used and four percentage points when GNP was used (Table 2). Roughly speaking, the results in Table 4 are consistent with those in Table 2. That is, when a variable meas uring financial transactions is included in the equation, the 1985 average error is about one percentage point less than when GNP is used by itself. Hence, nonfi nancial transactions not captured by GNP probably account for another percentage point (Table 2, right column), leaving about two percentage points of the error in 1985 unaccounted for. Another proxy for the volume of financial transactions is the dollar volume of funds transferred over Fedwire.12 On a daily average basis, this volume has increased from $200 billion in 1978 to about $700 billion in 1985. Quarterly statistics on the volume of funds transferred over Fedwire are available only since 1977. Therefore, annual data were used to estimate money demand equations, and the sample periods were extended back to 1949 (Table 5). Estimated through 1974, the coeffi cient on the dollar volume of wire transfers is significant but has a negative sign (Table 5, equation 2), sug gesting once again that many financial transactions are made to manage money balances. Estimated through 1984, the coefficient on Fedwire volume remains neg ative, but declines in absolute value by about one-half (Table 5, equation 4), implying that its effect on M1 has not been stable over time. These results are difficult to interpret, however, because money demand equations have generally not been stable when the sample period is extended beyond 1974. Nevertheless, the negative coefficient on the dollar volume of Fedwire transfers does indicate that more rapid growth of financial trans actions over the longer run has not been associated with an acceleration in M1 growth. For financial transFootnote 11, continued debits were only slightly more likely to increase in any same month that trading volume increased. Over the past ten years, de bits and Governm ent securities trading volum e moved in the same direction in 52 percent of the 120 months, debits and stock m arkets volum e in 61 percent. 12For earlier work using the num ber of wire transfers (as a proxy for technological change) to explain unusual weakness in M1 in the mid-1970s, see Ralph Kimball, op. cit., pages 12-22. actions to explain the rapid growth of M1 in 1985, it would be necessary to find reasons why the historical relationship between financial transactions and M1 might have changed. One reason this relationship might have changed somewhat in recent years is increased concern on the part of banks about their potential exposure to corporate customers that engage in large dollar volumes of financial transactions. On any given day, some inflows that are expected by a corporation may not materialize (or some unexpected outflows may occur) resulting in an overnight overdraft. If the firm does not qualify for a line of credit that would cover the potential overdraft, it might be required to hold a larger balance at the beginning of the day. This could represent an indirect channel through which the rapidly growing volume of financial transactions could increase the level of demand deposits, but it is impossible to quantify the effect. Another indirect way financial transactions might affect M1 is through the higher level of demand deposits that firms are holding to compensate banks for the costs of making a larger number of transactions, i.e., higher compensating balances. Banks have also been moving toward more explicit pricing of transaction account services, and the growing number of financial trans actions may now have a more pronounced effect on M1’s growth as a result. Again, due to lack of data on the total number of financial transactions, this effect cannot be quantified. And working in the opposite direction, firms have been moving toward using fees to compensate their banks for transaction account services rather than holding balances. So it is not clear that the net effect of more explicit pricing has been to increase M1 balances.13 Conclusions In general, it appears that the more rapid growth of financial transactions is not having a very large effect on M1’s growth: • Many financial transactions are made explicitly to manage cash balances more efficiently. Such transactions would, of course, tend to reduce M1 13Of course, com pensating balances cou ld have increased for reasons other than the grow ing num ber of financial transactions. As interest rates fall, firm s must hold a higher level of balances to com pensate banks for the same level of transaction account services. W ithout any changes in banking practices, this effect should be picked up by the interest rate varia ble in the money dem and equation. However, as banks move tow ard more e xp licit pricin g of services, they may also enforce balance requirem ents more strictly. In turn, this could make com pensa ting balances more responsive to interest rate changes than in the past. Table 5 Volume of Wire Transfers Over Fedwire and the Demand for Money Annual observations 1949-84 1949-74 Coefficients GNP .................... . . Volume of wire tr a n s fe r s ............. Equation 1 Equation 2 Equation 3 Equation 4 0.46 (58) 0.58 (5.3) 0.11 (2 3) 0.33 (35) - 0.11 (2.2) -0 .0 5 (2.1) ‘ Not included. balances, not increase them, as other transactions would. • A large part of total financial transactions is done by investment firms and dealers. They are among the most sophisticated checking account managers, attempting to keep their balances at frictional levels almost regardless of the volume of transactions undertaken. Of course, as the volume of transac tions increases, these frictional balances are likely to increase at least somewhat because of unex pected cash flows or the desire to avoid costly overdrafts. • Many financial transactions can be completed without using the checking accounts in M1. Stock, bond, and money market transactions are often executed by using accounts at an investor’s broker. Moreover, individuals are likely to hold liquid assets suitable for investment in their money market deposit accounts or money market funds, not in M1. These accounts, with their limited transactions features, can be used to make purchases of financial instruments without the funds flowing through an M1 account. Likewise, the proceeds from sales of financial instruments would not need to be deposited in M1 accounts. Even though stronger-than-expected M1 growth has occurred during a period of rapid growth in volume of financial transactions, longer-run relationships do not confirm that there is a strong linkage between the two. John Wenninger and Lawrence J. Radecki FRBNY Quarterly Review/Summer 1986 29 Short-Term Borrowing by Local School Districts Short-term borrowing by school districts has undergone a dramatic change in New York State as well as else where in the nation. Traditionally, the purpose of such borrowing has been to finance temporary cash shortfalls that occur before property taxes are received. In the last several years, however, it has also been used to finance gaps created by state delays in payment of school aid. And even more importantly, many school districts have begun to use short-term borrowing to finance aggressive investm ent program s. These developm ents have increased the exposure of school districts to certain kinds of risk and have resulted in several districts’ incurring financial losses (box). Large increases in short-term borrowing and invest ment by school districts are readily apparent in the national statistics. But additional data to analyze the incentives for such aggressive financial management or to help prescribe effective remedies are not available at the national level. This article closely examines those factors that have led to widespread use of short-term borrowing in cash management by school districts in New York State, where such data are available. Based on this analysis of New York, the article also suggests possible ways to reduce the role of debt in school dis trict cash management. In particular, more flexibility to carry over revenues from one fiscal year to the next, changes in the schedule of state aid payments, and safer investment opportunities, such as state managed investment pools, would reduce the incentives for The author thanks Julie N. Rappaport for her research assistance in the preparation of this article. 30 FRBNY Quarterly Review/Summer 1986 aggressive borrowing and investment by the nation’s school districts. In the next section of this article, we analyze the cash management problems and financial profiles of school districts, focusing particularly on New York State. Cashflow projection models are then developed based on the alternative financial profiles that emerge. These models reveal why and to what extent New York school districts have responded to their cash management problems by borrowing to finance investment at higher yields. These models also help to quantify the success of state efforts to alleviate the need for such borrowing. Even with these efforts, however, the analysis shows that numerous cash management difficulties still remain. The cash management problems of school districts School districts across the country have increased their average investment activity and short-term borrowing. In 1961, the U.S. Advisory Commission on Intergov ernmental Relations (ACIR) encouraged all local gov ernments to invest more actively to generate additional income. In the first ten years following that recommen dation, school district interest earnings as a share of revenue doubled from one-half of 1 percent to almost 1 percent.1 From 1972 to 1982, this ratio tripled. During this 20-year period, short-term borrowing for cashflow and capital purposes remained the most rap idly growing portion of U.S. school district debt, which 1See U.S. Advisory Commission on Intergovernmental Relations, Investment of Idle Cash Balances by State and Local Governments (January 1961). National data are from the U.S. Bureau of the Census. totaled $36 billion by the end of fiscal year 1984. While total debt outstanding at year-end had grown at an average rate of about 4 percent per year since 1962, the short-term portion grew by 8 percent per year. The true growth of short-term debt exposure has probably been considerably greater than suggested by these figures because year-end measures exclude the unknown but large amount of cashflow borrowing repaid just before the close of each fiscal year. New York school districts have been in a sim ilar position: short-term borrowing and investment became widespread. During the fiscal year July 1, 1983 to June 30, 1984, over half of New York’s 732 school districts (excluding New York City) borrowed short-term for cashflow purposes (Table 1, column 1). On average, these districts issued $3.1 million per district in the form of tax or revenue anticipation notes (TRANs). For the U.S. Internal Revenue Service (IRS) to grant any TRAN tax-exem pt status, eligible issuers must be able to substantiate the likelihood of a cashflow deficit for at least one month of the fiscal year. Because tax-exempt TRANs have been issued by so many school districts, it must also be true that cashflow deficits occur for the majority of districts in New York during the course of the year. These cashflow deficits occur because school districts face inherent cash management problems. Their major expense is for personnel at a dollar cost that is gen erally fixed in advance with frequent disbursements in roughly constant amounts. As a result, districts have a fairly uniform monthly need for cash. At the same time, their major source of local revenue is property taxes, which are also generally fixed in advance but received (in contrast to disbursements) very infrequently. Usually paid once or twice a year after the fiscal year has begun, property tax payments can create large swings in school district cash balances and create the need for short-term borrowing.2 P e rso n n e l costs also can create cash managem ent problem s when labor contract negotiations are not coo rdina ted with the budget cycle. In particular, uncertainty con cernin g the size and effective date of salary increases will make it more d iffic u lt to an ticip ate cash shortfalls. In this article, it is assum ed that all costs are known in advance. It is hard to be precise about tim ing of tax receipts across the nation’s schools, but general observations are possible because state legislation usually establishes guidelines for how and when property taxes are to be collected. Table 1 Financial Profile of New York State School Districts Fiscal year 1984 Number of d is tr ic ts .............................. . . . . Average property t a x ........................... . . . . Average state a i d ................................. . . . . Non-city districts All districts* (1) Nassau (2) Suffolk (3) 732 54 73 69% 24% 51% 39% 45% 43% 41% 43% 88% 47% 59% $5.3 $7.1 $1.4 $4.3 $21 8 $19.9 $7.8 $28.4 (share of own budget) 48% 39% Other (4) 544 City districts* (5) 61 (percent of districts in category) District b o r r o w in g f.............................. . . . . 55% 91% (million dollars per borrowing district) Average b o r r o w in g f ........................... . . . . $3.1 (million dollars per district) Average expenditures ....................... . . . . $ 11.8 (average borrowing as a share of average expenses) Average debt d e p e n d e n c e !............. . . . . 26% 24% 36% 18% 15% 1.1% 2.1% 1.5% 2.2% 0.3% 1.8% 0.3% 1.3% (share of own budget) Interest p a y m e n ts f.............................. . . . . Interest receipts ................................. . . . . 0 .6% 1.8% 'Excluding New York City. |Tax and revenue anticipation notes. Sources: New York State Department of Education and New York State Office of the Comptroller. FRBNY Quarterly Review/Summer 1986 31 These swings in cash balances and the resulting need to borrow have often been magnified by the way the state disburses aid, the second major source of school revenues. Because aid to school districts is also a large part of New York’s budget, delays in payment of aid have been a common solution to state fiscal stress.3 3The im portance of state aid to school districts is even greater in states outside New York. State aid rose to over 45 percent of total U.S. school district revenues in the 1980s after having remained at about 38 percent for two decades. For examples of school district borrowing in general and of delayed payments of school aid in particular, see Joe Mysak, "Same Time Next Year,” Credit Markets (June 9, 1986), page 10, and Allen J. Proctor, "Tax Cuts and the Fiscal Management of New York State," this Quarterly Review (Winter 1984-85). Over the years, New York State has stayed within its own budget limitations by delaying payment of up to 75 percent of school aid until April, May, and June, the last quarter of the school fiscal year.4 Because districts cannot similarly delay their expenditures, a second, mid year deficit emerges (after tax proceeds have been spent but before most state aid arrives), and many districts need to borrow a second time each year. 4D istricts receive 8V3 percent of state aid per month in September, O ctober, and November. They receive 25 percent per month in A pril, May, and June under the regular aid program . S pecial program s for earlier paym ents are discusse d in a later section. Risks from Current Trends in School District Finance The end result of the trend toward more borrowing and investment has been an increasingly aggressive cash management style that seeks high net yields while exposing schools to several risks. One risk is that school districts that rely on short-term borrowing may have to cease operations temporarily if local lenders become unwilling to provide enough funds. The likelihood of this problem occurring is increasing. For example, in Iowa, banks have become reluctant or unable to supply dis tricts with all the funds they need. As a result, the state recently had to intervene to ensure school districts timely access to short-term financing.* At least as important is the risk that large interest rate swings may turn the process of borrowing and invest ment into a source of revenue losses rather than gains. This risk can be substantial because the decision to borrow, the actual borrowing, and the investment of the borrowed funds usually occur several months apart. If the school district incorrectly predicts interest rate movements, the cost of borrowing may substantially exceed the return on investment. This situation has become more common since 1979. School districts that have been locked into losses for months at a time can have difficulty finding additional revenues to replace the failed investment program. The search for higher returns has also led school districts to undertake investments that have placed their principal at risk. In 1984, the failure of two securities dealers, the Lion Capital Group and RTD Securities, turned these risks into losses for many investors, including school districts. In New York State, for example, 62 districts may face possible losses of up to $77 million as a result of insufficiently secured collateral for repur *See Joe Mysak, "S am e Time Next Year," C red it Markets (June 9, 1986), page 10. 32 FRBNY Quarterly Review/Summer 1986 chase agreements.t They have not yet recovered all their investment, and special state legislation has been enacted for the past three years allowing the districts to finance these losses until the funds are recovered.t High levels of short-term cashflow borrowing underlie all these risks, either directly or through encouragement of aggressive investment behavior. A process intended to improve the fiscal health of local government has evolved into a pattern of high-risk fiscal management. Recognizing the breadth of this problem, in 1985 the U.S. Advisory Commission on Intergovernmental Rela tions recommended for all local governments that “short term borrowing, both for operating and capital purposes, be strictly limited and regulated...” § fS ee New York State Assembly, Gambling with Public Funds: The Lion Capital Bankruptcy and Its Implications for Government Investment Practices (March 1985), pages 141-147. tT h e failure of professional dealers who were cau ght in an unanticipated interest rate sw ing made many school d istrict o fficia ls aware that aggressive cash m anagem ent could have sizable risks. To increase this awareness further, in D ecem ber 1984, the O ffice of the State C om ptroller of New York issued a detailed investm ent manual for localities that em phasized safety and liquid ity over yield as crite ria for investm ent decisions (Cash M anagem ent an d Investm ent Policies and P rocedures for Use by Local G overnm ent O fficials). Publication of these guidelines was follow ed by an extensive educational outreach program by both the State C om ptroller and the Federal Reserve Bank of New York. Inform ed state officials believe that investment practices in New York are now more cautious than in 1984. Similar prudential efforts are not obvious in other states, and an informal survey of the Southeast suggests that use of risky investments such as repurchase agreements is extensive (see B. McCrackin et a!., "State and Local Governments' Use of Repos: A Southeastern Perspective," Federal Reserve Bank of Atlanta, Economic Review (September 1985). §See Advisory Commission on Intergovernmental Relations, Bankruptcies, Defaults, and Other Local Government Financial Emergencies (March 1985). The overall effect of these property tax, state aid, and expenditure flows can be estimated using cashflow models that separately project and then combine the various flows into a cumulative cash balance projection. In particular, the cashflow models constructed for this analysis project monthly cash balances that are con sistent with the financial profiles in Table 1. While the exact situations of individual districts may vary consid erably, the cashflow models can reveal some of the types of cashflow situations that have led to the current financial practices of school districts. The basic differences in cashflow profiles across districts are closely related to the timing of property tax receipts. In contrast, the timing of conventional forms of state aid receipts and of overall expenditures is fundamentally alike for all districts. The schedule of property tax receipts, therefore, can be used to divide New York districts into four general cash man agement profiles.5 • Nassau county districts are highly dependent on property taxes but they receive no tax revenues until the fourth month of the fiscal year, and half the revenues are not received until the last quarter of the fiscal year. Combined with the fourth quarter receipt of most state aid, these factors substantially heighten the likelihood of cash shortages during the year. As a result, cashflow borrowing averages 24 percent of expenditures (Table 1, column 2). • Suffolk county d istricts have above-average dependence on property taxes that are received very late in the year. They must operate through December without tax revenues. Reflecting these circumstances, short-term borrowing on average finances 36 percent of their expenditures (Table 1, column 3). • City school districts typically receive taxes in the first through third months of the fiscal year and borrow relatively less than the other three types of districts, an average of 15 percent of expenditures (Table 1, column 5).6 • Non-city school districts (outside Nassau and Suf folk counties) generally receive property tax reve- 5The m odels used in this study assume that d is tric t receipts and paym ents are known with certainty. Obviously, unanticipated changes can occu r that may raise or lower m onthly cash balances from the projected levels. One area of uncertainty is the am ount of delinquent property tax paym ents. H igher de lin que ncies will reduce revenues and increase school d is tric t cash deficits. The effect of delinquencies on school revenues, however, is lim ited in duration in New York. For most school districts, the county governm ent assumes all delinquent school d is tric t property taxes by A pril (the beginning of the fourth quarter of the school fiscal year). At that point, the school d istrict receives all its levied taxes and the county undertakes collectio n efforts. 6The New York City school d is tric t is excluded in this study. The necessary data are unavailable because school cash m anagem ent is so intertw ined with the rest of the c ity ’s finances. We must necessarily generalize across other city school districts because their situations are governed by the separate charters of their respective cities. For exam ple, the average reliance on property taxes rises to 44 percent if we exclude Rochester, Buffalo, and Syracuse— fiscally dependent districts that also rely on sales tax revenues. The date of property tax receipts is the beginning of the fiscal year (July 1, except for Syracuse) for 12 d istricts and Septem ber for all other city districts. At least five city d istricts receive their property taxes in two installm ents and two d istricts in four installm ents throughout the year. In addition, some cities benefit from a special state aid program (Hurd aid) that can disburse assistance much earlier in the school fiscal year than the regular school aid program . FRBNY Quarterly Review/Summer 1986 33 C h a rt 2 S c ho o l D is tric t C u m u la tiv e Cash B a la n c e s M inim um b o rro w in g s tra te g y fo r a d is tr ic t w ith firs t and th ird q u a rte r d e fic its Thousands o f d o lla rs S ource: F e deral R eserve B ank of New Y ork sta ff e stim ates. C hart 3 School D is tr ic t C u m u la tiv e Cash B ala n c e s A lte rn a tiv e b o rro w in g s tra te g ie s T h o u sa n d s o f d o lla rs 4500 4000 3500 3000 2500 nues in the third and fourth months of the fiscal year.7 The slightly longer delay than for city districts increases the likelihood of insufficient cash in the early months of the fiscal year. Therefore, non-city districts borrow slightly more than city districts (Table 1, column 4). Because these non-city dis tricts account for 544 of New York’s 732 districts, they will be the focus of much of the following analysis. To illu stra te the general nature of the cashflow problem before going into more detail, Chart 1 shows an example of a non-city district’s cashflow profile for the July 1 to June 30 fiscal year. Property tax receipts create one revenue bulge in September and October, and state aid creates a smaller bulge in April through June. Expenditures are generally uniform across months. The net effect of these flows would be a period of sizable cash balances available for investment from September through January and during May. At the same time, the school district would need to borrow short-term in order to pay expenses in July, August, March, and April. By applying appropriate interest rates to these periods of borrowing and investment, it is possible to simulate school districts’ likely interest earnings and payments. By comparing actual and projected interest flows, it can be estimated how closely d is tricts’ borrowing and investment followed projected patterns based solely on routine cashflow needs. Any differences between the actual and projected values might be attributed to efforts to earn additional income through borrowing for invest ment purposes. Overall, projections of interest earnings and payments, based on fiscal year 1984 interest rates, are both con siderably less than the $157 million of total interest all New York districts earned and the $52 million of interest they paid in that year. This probably occurred because districts borrowed more than they needed for routine purposes and invested the additional proceeds at higher yields.8 2000 7State law allows counties to pass tax acts which, am ong other things, determ ine w hether school property taxes may be paid in up to six installm ents. Seven counties appear to perm it such installm ents. Moreover, a few counties are governed by spe cia l tax acts or charters. The Nassau and Suffolk County Tax A cts establish sp e cific tax paym ent dates for school districts in their jurisd ictions. These dates are unusual and create special problem s for their school districts. 1500 1000 S ource: F e d e ra l R e s e rv e B ank of New Y ork s ta ff estim a te s. 34 FRBNY Quarterly Review/Summer 1986 •Since the end of fiscal year 1984, interest rates have fallen. Similar cash m anagem ent practices today would be consistent with lower interest earnings and payments. The large diffe ren ce betw een actual and projected net interest earnings is consistent with efforts by school districts to earn above-norm al net yields through aggressive, and generally riskier, investm ent practices. In fact, the New York State School Boards A ssociation and the New York State A ssem bly Borrowing for the explicit purpose of investing at higher yields is generally referred to as arbitrage, and it is widespread among issuers of tax-exem pt debt throughout the United States. The next section of this article reviews the routine cashflow borrowing needs of non-city districts and uses cashflow models to analyze in more detail their opportunities and incentives to borrow for arbitrage purposes. Simulations of possible arbitrage programs are then used to estimate whether arbitrage has increased short-term borrowing in the four categories of New York districts summarized in Table 1. Arbitrage in New York school districts The average projected cashflow of a non-city district in New York has two potential deficit periods because of the timing of property taxes and state aid. As shown by the solid line in Chart 2, these deficits would require a borrowing program lasting at least several months each year. The school would need to borrow enough at the beginning of the fiscal year to finance July and August expenditures, and it would generally be able to repay that debt in September when property taxes began to arrive. Similarly, it could finance the third quarter deficit by borrowing again in March and repaying in May when most of its state aid would have arrived. The strategy of borrowing the smallest amount for the shortest period is shown by the dashed cash balance line in Chart 2.9 The incentive for arbitrage in New York Starting from this basic borrowing program, a school district could make some alterations that would reduce its costs. The school district could reduce the cost of issuing notes twice in the same year by extending the term of the TRAN it issues in the first quarter. Thus, in stead of repaying the July note in September, it would hold on to the funds until they were needed again in March. Keeping the debt outstanding through March has other cost advantages as well if interest rates remain rela tively stable. In the months when the funds are not Footnote 8, continued found that school districts tended at that time to favor yield over safety in pursuing their investment programs. See New York State School Boards Association, “ Survey of Local School District Investment Policies," draft report (April 1985), and New York State Assembly, Gambling with Public Funds (March 1985). In 1985, the state undertook an extensive investment education program which reportedly has led school districts to shift their emphasis from yield to safety. See the box for more information. •For some school districts this procedure may not be possible or it may incur excessive debt issuance costs. Such districts may borrow once a year for their entire year’s needs even if other incentives for extended-term borrowing do not exist. Also, two measures to reduce the size of first and third quarter deficits are discussed in the next section. Many non-city districts appear to use these measures to lower their cashflow deficits from the levels shown in Chart 2. needed to finance deficits, schools are able to invest the borrowed money at interest rates that are almost always higher than their borrowing rates. School districts can do this because they borrow in the tax-exempt market and invest in the taxable market, but are not subject to income tax themselves. The more the borrowing period is extended, the more the school district is able to take advantage of this favorable interest rate spread. In practice, the school district could, by borrowing early and repaying late, have borrowed funds available for investment for almost the entire year. This strategy would help districts reduce the net costs of financing recurrent cashflow deficits. Chart 3 illustrates how investment funds can be gen erated by extended borrowing. The solid area represents the amount of cash a district has available for invest ment when it borrows for as short a period as possible. By extending its borrowing to twelve-month maturities, the school district increases its investable cash balances by the amount of the gray area in the chart. This fullterm borrowing provides schools with investable sur pluses in July and April, which would otherwise be deficit months. It also enlarges the projected surpluses in September through February and in May. Comprehensive data are not available on how many of the over 400 New York districts that borrowed in 1984 used this arbitrage technique to reduce their net bor rowing costs. Announcements published by a smaller number of districts, nevertheless, suggest that full-term borrowing appears widespread. Specifically, of the 186 TRANs publicly announced in 1983 and 1984, 85 per cent were issued around the start of the fiscal year and 96 percent matured near the end of the fiscal year.10 Another way to increase arbitrage earnings is to increase the amount of borrowing beyond what is needed to finance cashflow deficits. For example, a school district with a cash shortfall of $1 million could borrow that amount at a tax-exempt rate of 6 percent for an annualized cost of $60,000. But if the district desired to reduce the overall cost of financing, it could borrow $2 million and invest the extra $1 million at a taxable yield of 8 percent. The school district’s annu alized cost of borrowing would rise by $60,000, but its investment income would rise by $80,000, for a net arbitrage profit of $20,000 and a reduction of its net annual interest expense from $60,000 to $40,000. As this example illustrates, from the standpoint of arbitrage earnings, the school district has an incentive to borrow as much as possible. The most important limits to arbitrage activity are IRS regulations that put a ceiling on the use of tax-exempt borrowing to finance 10These data were compiled by the Office of the New York State Comptroller from daily issues of the Bond Buyer. FRBNY Quarterly Review/Summer 1986 35 arbitrage.11 Nonetheless, IRS regulations still allow schools substantial leeway in arranging an arbitrage program. A school district that borrows up to the IRS maximum for the full term of the fiscal year will substantially increase its m onthly cash balances available for investment. The projected increase is represented by the hatched area in Chart 3. Provided the taxable/taxexempt yield spread is large enough, the school district will be able to use its additional investment income to pay all the interest cost of the additional arbitrage bor rowing and part or all of the interest cost of financing its recurrent cash deficits. "S c h o o ls and other local go vernm ents are perm itted to issue taxexem pt TRANs up to the sum of the ir largest monthly cum ulative cash d e fic it plus the follow in g m onth’s expenditures. When monthly exp enditu res are larg er than m onthly deficits, this rule allows localitie s to borrow more than tw ice their de fic it-fin a n c in g needs. C ashflow projection s sug gest that New York school districts ge nera lly face the latter situation. The IRS also requires that the borrow ed funds be spent w ithin several months of issuance, but this requirem ent is generally satisfied for schools when the July-A ugust shortfall is funded. C h a rt 4 E s tim a te s of S h o rt-te r m Borrowing B eyond R outine Cashflow Needs T h ousa nds o f d o lla rs per ten m illion d o lla rs of b u d g e t 5 0 0 0 ----------------------------------------------------------------------------O M axim um estim ated b o rro w in g a llo w a b le 4 5 0 0 — ♦ O b s e rv e d ave ra g e b o rro w in g FY 1984 A E s tim a te d ro u tin e b o rro w in g need 4 0 0 0 ----------------------------------------------------------------------------3 5 0 0 --------- O ------------------- O ---------------------------------------3 0 0 0 ----------------------------------------------------------------------------A A 2 5 0 0 ------------------------------- * --------------------------------------- 2 0 0 0 ----------* -------------------------------------------------------------------1 5 0 0 ------------------------------------------------------- O ---------------100 0 -------------------------------------------------------------------------------------------------------------------------- — ------------------------------------ A 5 0 0 ----------------------------------------------------------------------------- N assau n o n -c ity S u ffo lk O ther n o n -c ity n o n -c ity S chool d is tr ic t c a te g o ry In d e p e n d e n t city A verage b o rro w in g above routine levels su g g e s ts tha t a rb itra g e b o rro w in g is ta k in g p la c e . S choo l d is tric ts are groupe d by p ro p e rty tax levy dates. S ource: F e deral R eserve Bank o f New Y ork s ta ff e stim a te s. 36 FRBNY Quarterly Review/Summer 1986 Estimating the use of arbitrage A typical non-city school district in New York State appears to be able to offset virtually all the cost of financing a cashflow deficit by using arbitrage. The exact benefits of arbitrage will vary from district to dis trict, depending on various factors including the yields at which it can borrow and invest. Estimates suggest, however, that an interest rate spread of only 95 basis points would generally make arbitrage worthwhile in the sense of reducing overall costs, and larger spreads could result in net profits. For example, extending the maturity of debt from two months to 12 allows a district to reduce its net interest cost from an estimated $7,800 to only $460 based on a borrowing cost of 6 percent and an investment yield of 8 percent (Table 2). Full-term borrowing of an amount equal to the largest monthly cash deficit plus the fol lowing month’s expenditures (the maximum allowed by the IRS) would enable the district to pay all borrowing costs and earn an estimated profit of about $14,000. The prevalence of arbitrage borrowing in New York can be estimated by comparing average observed levels of borrowing with projections of routine and maximum amounts of borrowing. This is done by projecting monthly cashflows in the absence of borrowing, iden tifying the largest monthly cashflow deficit (which determines the routine level of borrowing necessary), and adding to this deficit the subsequent month’s expenditures (which determines the maximum allowable level of borrowing). Estimated routine borrowing needs and maximum allowable borrowing are compared with average bor rowing in fiscal year 1984 for all four categories of dis tricts (Chart 4). If observed average borrowing is near the estimated routine borrowing level, then it is likely that school districts in that group do not generally borrow primarily for arbitrage purposes. If borrowing exceeds the routine level, on the other hand, then some arbitrage borrowing is taking place. Borrowing near the estimated maximum level suggests that arbitrage is the principal motivation for short-term borrowing by most school districts in that group. This procedure reveals that city districts borrowed on average as much as IRS regulations would allow; that is, arbitrage played a significant role in their fiscal year 1984 short-term borrowing programs. In contrast, non city districts in Nassau and Suffolk counties seemed to borrow only about as much as required by their severe projected cashflow problems. Thus, while they borrow proportionately more than any other districts in the state, they do not appear motivated primarily by arbitrage. Arbitrage does seem to play a modest role, however, in the borrowing decisions of non-city districts in other counties. These districts on average borrow more than Table 2 The Use of Arbitrage to Reduce the Cost of Financing Cashflow Deficits* In dollars Without arbitrage With arbitrage Maturity of b o rro w in g ............................................................................... Amount of b o rro w in g ............................................................................... Minimum-term Routine Full-term Routine Full-term Maximum Cost of borrow ing...................................................................................... Income from investment of borrowing ................................................. Net cost of bo rrow in g............................................................................... 13,340 5,540 7,800 35,350 34,890 460 79,500 93,760 -1 4 ,2 6 0 *Based on simulations of a non-city school district receiving 50 percent state aid with an annual budget of $10 million. Interest rates are 8 percent for investments and 6 percent for borrowing applied to balances at the beginning of each month. The principal invested is the same as in Chart 5. Total school district investment income also includes $58,910 of interest earned from investment of unborrowed funds, principally unexpended property tax receipts. Full-term, maximum borrowing has estimated arbitrage gains exceeding costs for taxable/tax-exempt spreads exceeding 95 basis points in this simulation. their routine cashflow needs but not generally as much as IRS regulations would allow.12 Up to this point, the evidence shows that school dis tricts in New York borrow short-term both for routine cashflow and for arbitrage purposes. Estimates of probable cashflows suggest that this borrowing finances one or more temporary deficits that result from untimely receipt of property taxes and state aid. Once a school district does any borrowing, it has a strong incentive to borrow even more for longer periods in order to reduce its net cost of borrow ing. Since debt seems to encourage even more debt, the obvious solution is to reduce, if not eliminate, routine short-term borrowing needs. (Also see the box for additional reasons for reducing debt, including the risks and losses some districts have incurred.) State programs to reduce districts’ need to borrow New York has two laws that attempt to reduce or elim inate the cashflow deficits that many schools encounter. One provides schools with some scope to self-finance their first quarter deficits. The second provides earlier payment of state aid that may reduce third quarter cashflow financing needs for some school districts. First quarter deficits The majority of school districts are often short of cash 12These com parisons are based on averages for each group of school districts and, therefore, only suggest the motives underlying individual dis tric t borrow ing. The am ount of borrowing and the role of arbitrag e vary across districts. For exam ple, fiscal year 1984 borrow ing by the 61 city dis tric ts (excluding New York City) averaged 15 percent of expenditures but ranged from zero to 73 percent. Twenty-five districts did not borrow that year, twelve borrow ed about 10 percent of expenditures, twelve borrowed about 20 percent, nine borrow ed up to 30 percent, and three borrowed more than half their budgets. at the beginning of the fiscal year in July and August because they receive no property tax revenues or state aid before September. At the same time, the closing months of the fiscal year bring an estimated cash sur plus. If schools were able to carry some fourth quarter surplus over into the initial months of the next fiscal year, the need for first quarter borrowing might be reduced. In general, school districts are required to return to taxpayers any funds that are left over at the close of the fiscal year.13 Since 1977, however, New York state law has allowed schools routinely to retain up to 2 percent of their budgets as unreserved balances to finance the initial months of the subsequent fiscal year. Aggregate data provided by the State Department of Education suggest that school districts on average have taken full advantage of this program. But it is not enough. Cashflow projections suggest that even if the amount of permitted carryover were doubled, enough cash would be on hand to finance only July cash needs for the average district. About half of the first quarter borrowing would still be necessary to cover shortages in August. Third quarter deficits The projected deficits that many school districts encounter in the third quarter are caused by payment of most state aid in the closing months of the fiscal year. 13ln New York, the general fund balances of districts are called unreserved balances. The prohibition against accum ulating unreserved balances is intended to keep property tax rates as low as possible. Ironically, using tax rate changes instead of accum ulated balances to balance budgets over a business cycle may actually raise the overall burden on taxpayers. For additional discussion of this effect, see Allen J. Proctor, “ Tax Cuts and the Fiscal M anagement of New York State,” this Quarterly Review (Winter 1984-85). FRBNY Quarterly Review/Summer 1986 37 As the state increases its support (in percentage terms) of local education, the unfavorable effects on schools’ cash management problems (not necessarily their overall budgets) of this aid disbursem ent schedule become more pronounced. In particular, increased reli ance on state aid shifts a larger share of total school revenues into the last quarter of the school year. This shift has two adverse consequences for school districts’ cashflow profiles. Increased aid dependence reduces the projected amount of cash that schools have avail able for investm ent in the second quarter, and it increases the projected size of third quarter cash deficits (Chart 5).14 Lower balances in the second quarter will reduce investment income. The loss of investment income off sets some of the value of state aid to a district’s budget. As a result, a district’s taxes may be higher, expendi tures lower, or investments more aggressive than they would be if state aid were paid earlier. All other things equal, a school district financed 60 percent by state aid may have as much as two-thirds less cash available for investment than a school district financed only 40 per cent by state aid. 14D ependence on state aid has also cre ated cashflow uncertainty for d istricts in A pril and May because the state has not always made its paym ents on tim e (due to delays in approval of the state budget, w hich must be approve d before any paym ents can be made after A pril 1, the start of the state fiscal year). Schools, therefore, must take steps to ensure that they can continue operations if aid is late. The state has begun to fund an escrow account to ensure spring paym ents. At some point in the future, the amount in escrow may be sufficie nt to elim inate the risk of late A pril or May payments. The second effect of greater state support is that third quarter cash shortfalls may increase and schools may need to incur more short-term debt. For example, model projections suggest that the average-sized non-city school district will have cashflow deficits of about $1 million in March and April if it is 60 percent funded by state aid, whereas it would have cashflow surpluses if state aid were only 40 percent of its budget (Chart 5). In fiscal year 1984, state aid to non-city districts aver aged around 60 percent for districts in 20 counties, around 50 percent for districts in 19 counties, and 40 percent or less for districts in the remaining 18 counties. Recognizing the adverse impact on some school dis tricts of paying state aid in the spring, the state has a special payments program that moves some state aid payments from April and May into December, January, February, or March. The formula determining the size of the special payments is essentially based on the share of state aid in a school district’s budget. A large number of districts receive these more rapid payments. In the school year ending June 1984, 405 districts received March payments, 228 received Feb ruary payments, and 43 received January payments. No school district received aid payments in December of that year. The amount of special aid payments totaled about 6 percent of the state school aid budget with about 5 percent paid in March.15 15The am ount and tim ing of the special paym ents varies each year acco rding to the form ula. In fiscal year 1987, for exam ple, some aid is scheduled to be paid in D ecem ber and the total am ount of special aid represents 7 pe rcent of the state school aid budget. Table 3 Projections suggest that districts needed this share of aid each month... Projected average percentage of aid needed Independent c i t y ..................................................... . . . . Non-city O th e r ..................................................................... . . . . Nassau .................................................................. . . . . S u f f o lk .................................................................. . . . . December January February March 12.5 0 0 12.5 0 12.5 50 0 12.5 0 11.2 12.5 0 12.5 12.5 0 While the state paid early aid to this share of districts in fiscal year 1984... Percentage of districts In category Independent c i t y ..................................................... Non-city O th e r ...................................................................... Nassau .................................................................. S u f f o lk .................................................................. .... 0 2 18 74 .... .... .... 0 0 0 8 0 0 39 2 3 63 2 19 These estimates of needed aid are based on school districts which are heavily dependent on state aid in their annual budgets. The need for early aid would be lower in school districts where state aid accounts for less than half of their budgets. 38 FRBNY Quarterly Review/Summer 1986 The overall success of this more rapid payment pro gram can be assessed by comparing model projections of the monthly need for special aid (as measured by periods of cash deficits) to the special payments that were actually made. In fiscal year 1984, a total of 25 percent of aid was paid from Septem ber through November, 6 percent from January through March, and 69 percent from April through June. Table 3 provides estimates of a projected distribution of special aid that would have precluded the need for school districts to borrow from December through March. Comparisons of these estimates with aggregate data on the state special aid program suggest how successfully the program may be meeting individual district needs. • Independent city districts, on average, required up to 12.5 percent of their aid in December and again in March in order to avoid cash shortfalls. By and large, the bulk of city districts received March payments that year. None received December payments, which probably left a number of city schools with deficits to finance. • Most non-city districts (outside Nassau and Suffolk counties) have projected deficits in February and March that could be eliminated by special aid pay ments of up to 12.5 percent in each month. Districts receiving the average state aid allotment would need few if any special payments, but districts heavily dependent on state support are likely to run short of cash without a sizable amount of special aid. By and large, the state program addressed their February and March needs that year: the majority of non-city districts (outside Nassau and Suffolk) that were heavily dependent on state aid received February or March payments. • Non-city districts in Nassau and Suffolk counties have unusual cashflow problems that are not well addressed by the state program for earlier payment of aid. In particular, districts in these counties receive most of their state aid and property tax revenues in the last quarter of the fiscal year. As a result, their need for special aid payments is projected to be on average the most severe in the state, particularly in December and January.16 In 1984 (and currently), Nassau and Suffolk districts have not received December or January aid pay ments and only a small percentage have received payments in February or March. In sum, the New York program seems to address third quarter cashflow problems for the majority of school districts. Most city and non-city districts receive special aid payments that generally correspond to their probable needs. This program, however, does not serve many districts in Nassau or Suffolk counties, despite the likelihood of severe cashflow problems. Even if the program were applied to more districts in those two counties, however, it would be of limited value. In par ticular, the February and March payments generally provided by the program would not address their sizable projected cash shortages in earlier months. As a result, these districts would probably continue to require sub stantial third quarter borrowing. The combined effect of state programs Even with state efforts to reduce the need for short-term C hart 5 The R e la tio n s h ip B e tw e e n P ro je c te d Cash B ala n c e s and the Use of S ta te Aid By sh a re of sta te aid in scho ol d is tr ic t bu dgets T h o u sa n d s of d o lla rs pe r ten m illion d o lla rs of budget 3000A id as sha re of d is tric t revenue 2500- 40 p e rc e n t 50 p e rce n t 60 p e rc e n t S e p te m b e r-F e b ru a ry M a rch -A p ril P ro je c te d a ve ra g e m onthly ba la nce S e p te m b e r-F e b ru a ry b a la nces are a s o u rc e of 16State o fficia ls were aware of this problem when the spe cia l aid form ula was set up. The state form ula was con structe d to help d istricts whose third qu arte r needs were p rin cip a lly due to the state aid schedule. In contrast, the third qu arter needs of districts in Nassau and Suffolk counties are prin c ip a lly due to the late property tax schedules esta blished by their respective County Tax Acts. in ve stm e n t incom e. M a rc h -A p ril d e fic its re fle c t the n eed fo r e a rlie r state aid or s h o rt-te rm bo rrow in g. S ource: F e d e ra l R e se rve Bank of New Y ork s ta ff e stim a te s . FRBNY Quarterly Review/Summer 1986 39 borrowing by New York school districts, about 60 per cent of all city districts, 90 percent of non-city districts in Nassau and Suffolk counties, and about 50 percent of other non-city districts issued TRANs in fiscal year 1984. Some of this borrowing is unavoidable, and some appears to be for the purpose of generating arbitrage profits. Overall, it appears that the cashflow problems faced by most school districts in New York are primarily in the first quarter when one or two months’ operations need to be financed until property taxes are received. Addressing this problem would go far in reducing the need for borrowing by schools and, hence, would also reduce some of their arbitrage activity. The cashflow problems of districts in Nassau and Suffolk counties are different from those in the rest of the state. They are more severe and require a different approach than has been taken thus far. Potential improvements to school district finance While this analysis has focused on New York, cash management problems are an integral part of local school finance everyw here. The severity of these problems obviously varies across school districts and across states, yet one or more common sources of dif ficulty emerge: • Property tax schedules may not be coordinated with spending requirements. • State aid schedules may have adverse effects on the cashflow of schools. • Strong incentives may exist for schools to invest aggressively and to borrow more than necessary. Property tax payments ideally should begin at the start of the fiscal year. If rescheduling tax payments is not practical, an alternative would be to allow schools to accumulate sizable surpluses that could be carried from one fiscal year to the next. Permitting self-financing of cashflow deficits that occur at the start of the fiscal year would make a change in property tax schedules less critical. Under present conditions, existing limits on carryover may create an artificial need for schools to go into debt, with accompanying pressures to reduce the cost of borrowing through arbitrage. For school districts to carry over amounts large enough to eliminate firstquarter borrowing, it might be necessary to raise taxes temporarily to accumulate sufficient surpluses. This additional cost to the taxpayer is worthwhile only if it is offset by the benefits of lower debt. States should also reexamine the impact of school aid payment schedules on local cash management. New 40 FRBNY Quarterly Review/Summer 1986 York’s special rapid payment program meets the needs of a majority of school districts, but even this program seems insufficiently focused on the overall cashflow condition of individual districts. As with property tax payment schedules, aid payments ideally should closely parallel districts’ spending needs. States, of course, have their own cashflow problems, and the appropriate solution would strike a balance between the timing of local needs and the state’s ability to pay on an earlier schedule. For example, because of cash management problems, New York State currently finances the bulk of school aid with its own short-term borrowing, and, hence, it has an incentive to make the payments as late in the year as possible. This approach, in effect, means that both the state and the school dis tricts are often borrowing against the same aid dollar. New York State is borrowing in anticipation of tax rev enues, and the school districts are borrowing until state aid arrives. The overall expense of double borrowing to the taxpayer is obvious, and the state is already taking steps to reduce its reliance on short-term debt. But it would also be expensive for the state to pay aid even earlier in the year so that borrowing by school districts could be reduced or eliminated. Finally, there appears to be a need for increased prudential oversight of short-term borrowing by school districts. At a minimum, improved monitoring and reporting would enable public officials to anticipate financial problems before they became unmanageable. Some states might prefer more direct involvement in the cash management of schools. Three specific improve ments could be considered: • More information is needed on individual district’s cashflow problems and their borrowing practices. Presently the cashflow situation of school districts must be simulated using models such as those used in this study. With more complete data, state officials would be better equipped to develop local property tax laws and school aid programs that provide schools with adequate cashflow throughout the year. And, on the debt side, short-term bor rowing is not necessarily a good indicator of school district cashflow problems. It would be easier to use debt statistics to assess the financial condition of districts if enough information were available to separate, for example, arbitrage borrowing from other short-term borrowing. • Increased restrictions on how borrowed funds could be invested would help avoid some potential prob lems (box). Arbitrage provides strong incentives for aggressive investing to raise yields. This behavior exposes public funds to levels of risk that have already resulted in financial losses. One policy response would be to allow investment of borrowed funds only in instruments with the lowest credit and market risk. Such a requirement would sharply reduce the incentive for arbitrage borrowing. New York’s recent guidelines on permissible investments are a worthwhile move in this direction.17 17See Office of the State Comptroller, Cash Management and Investment Policies and Procedures for Use by Local Government Officials (Albany, New York, December 1984). • Requirements that state or regional short-term investment pools manage school districts’ invest* ments would insulate districts from pressures to maximize investment income. Without direct control over the yield from investing their borrowed funds, districts may be less inclined to borrow for invest ment purposes and, of course, would not be able to take unacceptable risks. Large pools also would provide other advantages, including lower man agement costs, greater portfolio diversification, and lower transaction fees. Allen J. Proctor FRBNY Quarterly Review/Summer 1986 41 In Brief Economic Capsules Are Large U.S. Banks Moving International Activity Off Their Balance Sheets? Recently market observers have drawn attention to the increased off-balance-sheet activity of banks.1 This article finds evidence that the top nine banks are doing more international business off their balance sheets. This shift by big U.S. banks is centered in the G-10 countries.2 Also, there is some evidence for the asso ciation of the shift with a new financial instrument. This financial innovation, the note issuance facility (NIF), is transforming international lending. Instead of arranging a medium-term credit facility with a syndicate of banks, many borrowers now arrange a NIF. Under this facility, the borrower can repeatedly issue short-term paper through an agent or by an auction. In the event that the paper does not sell below a contracted interest rate, expressed as a spread over an interbank reference rate, the underwriting or managing banks undertake to buy the paper or to make advances. Some NIFs permit borrowing from the committed banks with no prior attempt to sell notes. The contract, which typically has a life of three to seven years, limits the underwriters’ obligation to extend credit under specified circumstances that may include a deterioration in the borrower’s creditworthiness. Thus, the borrower may enjoy cheaper short-term funding with some assurance of its availability over the medium term. ’For example, see Bank for International Settlements, Recent Innovations in International Banking (1986). 2G-10, as used here, includes Belgium, Canada, France, Italy, Japan, the Netherlands, Sweden, Switzerland, West Germany, and the United Kingdom. 42 FRBNY Quarterly Review/Summer 1986 The innovation of the NIF has an ambiguous effect on U.S. banks’ commitments to lend. On the one hand, borrowers who have switched from medium-term syn dicated credits to NIFs may rely on U.S. banks only for the commitment to lend while depending on other sources— including nonbanks—for actual funding. On the other hand, borrowers who previously relied exclusively or mainly on large U.S. banks for lines of dollar credit may be broadening the nationality and increasing the number of their committed banks and so lowering the cost by arranging NIFs. Thus, the market for contrac tually committed lines of credit is growing, but large U.S. banks may be losing market share. U.S. banks do not report international commitments to lend per se, but these can be approximated using data from the Country Exposure Lending Survey (CELS). Loan commitments are taken to be all contin gent (off-balance-sheet) claims adjusted for guarantees, less letters of credit.3 An increase in the ratio of loan commitments to adjusted assets (on-balance-sheet claims) is taken to indicate a shift to increased offbalance-sheet banking by U.S. banks. The change in this ratio over three years suggests that large U.S. banks are moving their international activity off their balance sheets. The top nine U.S. banks4 have raised the ratio of loan commitments to adjusted claims from 35 percent to 43 percent between end-1982 and end-1985 for 22 OECD countries and five selected Asian borrowers— India, Indonesia, Malaysia, South Korea, and Thailand (table).5 The rise in commitments to lend in relation to assets is not uniform, however. The shift from on- to off- 3lt would be better if letters of credit adjusted for guarantees were available. 4The top nine banks’ aggregate reported in the CELS continues to include Continental Illinois. 5The most recent CELS data, from end-March 1986, show the same ratio, 43 percent. IN BRIEF—ECONOMIC CAPSULES International A ctivity of the Top Nine U.S. Banks: Ratio of Loan Commitments to Adjusted Assets (2 ) ( 1) End-1982 ratio R anking/C ountries 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 New Zealand . . . . Sweden .................... A u s tra lia .................... F in la nd ....................... D enm ark.................... T h a ila n d .................... N o r w a y .................... S w itz e rla n d ............. G reece .................... United K ingdom . . . In d ia ........................... N e th e rla n d s ............. I c e l a n d .................... F ra n c e ....................... M a la y s ia .................... I t a l y ........................... In d o n e s ia ................. C anada .................... B elg iu m -L u x............. West G erm any . . . A u s tria ....................... Ire la n d ....................... S p a i n ....................... Japan ....................... P o rtu g a l.................... Korea ....................... T u rk e y ....................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . W eighted averages: O v e r a ll.............................. . . G-10 + Sw itzerland . . . . . Other O E C D .................... . . East A s i a t ....................... (3) End-1985 ratio Ranking/C ountries 1.11 0.74 0.71 0.58 0.51 0.50 0.49 0.49 0.48 0.45 0.45 0.41 0.39 0.38 0.37 0.33 0.32 0.32 0.28 0.26 0.24 0.21 0.18 0.13 0.12 0.11 0.04 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Sweden .................... F ra n c e ....................... N o r w a y .................... United Kingdom . . . C anada .................... A u s tra lia .................... S w itz e rla n d ............. A u s tria ....................... N e th e rla n d s ............. B elgium -Lu x............. S p a i n ....................... Ire la n d ....................... F i n la n d .................... D enm ark.................... New Zealand . . . . I c e l a n d .................... I t a l y .......................... West G erm any . . . In d o n e s ia ................ In d ia .......................... Greece .................... T h a ila n d .................... K o r e a ....................... Japan ....................... T u rk e y ....................... P o rtu g a l.................... M a la y s ia .................... 0.35 0.33 0.45 0.25 W eighted averages: O v e r a ll.............................. . . G-10 + Switzerland . . . . . Other O E C D .................... . . East A s ia :):....................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Percent change, 1982-85* (4) Total NIFs arranged, 1983-85(b illion s of dollars) 1.69 0.66 0.63 0.60 0.55 0.54 0.53 0.45 0.45 0.43 0.43 0.38 0.36 0.34 0.26 0.26 0.26 0.26 0.24 0.21 0.20 0.18 0.16 0.13 0.12 0.11 0.08 129 72 30 33 75 -2 4 10 90 9 56 137 81 -3 9 -3 4 -7 6 -3 2 -2 0 0 -2 6 -5 4 -5 8 -6 5 51 4 249 -1 4 -7 8 9.6 5.0 1.4 4.5 1.1 11.6 1.2 0.6 1.3 0.5 0.6 0.3 1.3 1.4 2.9 0.2 1.7 0.3 0.5 0.4 0.0 0.2 0.9 1.4 0.0 0.6 0.3 0.43 0.47 0.40 0.18 25 40 -1 0 -2 8 Totals: 49.8 26.6 20.9 2.3 'F o r countries in colum n (2). fA rra n g e m e n ts with all banks, U.S. and non-U.S. ^In clu d e s India, Indonesia, Korea, Malaysia, and Thailand. Sources: Federal Financial Institutions Examination C ouncil, Country Exposure Lending Survey; Bank of England; and Interna tiona l Financing Review Data on NIFs include some arrangem ents with partial or without con tractu ally associated bank com m itm ents to lend; renegotiated fa cilities (for exam ple, New Z e aland's) are double-counted. balance-sheet exposure is concentrated in the G-10 countries, where the ratio of total commitments to total assets rose from 0.33 to 0.47. In the G-10, the overall growth of the market for loan commitments appears to be more than offsetting any loss of share in that market by U.S. banks. By contrast, the balance is shifting toward on-balancesheet financing in Asia. Whereas commitments to the five Asian borrowers stood at 25 percent of assets in 1982, they fell to 18 percent of assets in 1985. Since Japanese banks are well represented as underwriters of NIFs in Asia, U.S. banks may be losing market share there. The same loss of market share may lie behind the decline in off-balance-sheet exposure in the OECD outside of the G-10, especially in Australia and New Zealand. In general, the countries whose borrowers have been most active in the NIF market tend to rank higher in the ratio of commitments to assets (table). Swedish, French, Norwegian, and British borrowers have been active in the NIF market and hold the top four positions;6 Japa nese, German, and Italian borrowers have been rela tively less active. The relative rise of commitments to Canadian borrowers, even though Canadian withholding tax makes NIFs unattractive, shows that NIFs are not A u s tra lia n borrow ers have arranged a large sum of NIFs but have drawn more heavily on them than most NIF custom ers. As a result, the A ustralian ratio de clin ed over the three years. FRBNY Quarterly Review/Summer 1986 43 a necessary vehicle for the shift to off-balance-sheet banking. In summary, CELS data provide evidence that large U.S. banks are increasing their off-balance-sheet com mitments to lend cross-border in relation to their actual international assets. Further, these data locate the shift in the G-10 countries. Finally, the rise in off-balancesheet exposure appears to be associated with activity in the NIF market. Robert N. McCauley One important difference between U.S. travel and merchandise trade is their regional compositions. The Western Hemisphere’s share of U.S. travel is 53 per cent, considerably more than its 37 percent share of our merchandise trade (Table 2). Mexico alone accounts for 20 percent of overall travel trade, but only 6 percent of merchandise trade. In contrast, Japan accounts for a hefty 16 percent of U.S. merchandise trade, but only 7 percent of total travel expenditures. As explained later, these regional differences suggest that the recent decline in the dollar is effectively much less for U.S. travel than for merchandise trade. Historical perspective Prospects for the U.S. International Travel Deficit Travel is an important part of U.S. international trade that has posted record deficits in recent years. From a position of near balance in 1981, the travel and tourism account approached a $10 billion deficit in 1985 (Table 1). Analysis of past experience confirms that travel flows are heavily influenced by the same factors that deter mine merchandise trade—in particular, the deterioration in the U.S. travel balance during the 1980s is largely attributable to the strong dollar and rapid real growth relative to abroad. Though the dollar’s recent decline against the Japanese yen and major European curren cies should significantly lower the merchandise trade deficit over the next several years, it is likely to lead to only a modest improvement in the travel balance— probably no more than $3 billion from economic factors alone. This is because the dollar has not fallen against the currencies of our primary travel partners, Canada and Mexico. The travel account measures spending by foreign visitors to America and by U.S. travelers abroad on items such as food, lodging, transportation, and enter tainment. In 1985, foreign visitors spent a total of $14 billion here while U.S. travelers spent $24 billion abroad, amounts that represent about 7 percent of U.S. mer chandise exports and imports respectively. The travel deficit was about 8 percent of the 1985 U.S. current account deficit and accounted for almost half of the $20 billion deterioration in the U.S. services balance from 1981 to 1985. As with U.S. trade generally, changes in the travel bal ance are largely explained by relative income growth and exchange rate movements. Rising incomes increase demand for most goods and services, including travel.1 Travel expenditures are also sensitive to relative cost considerations and hence changes in exchange rates. A decline in the dollar tends to raise the cost of over seas travel for U.S. citizens and makes the United States more attractive to visitors from the rest of the world. Prior studies suggest that a 1 percent decline in the dollar can be expected to increase travel receipts ’ See, for instance, prior work by Jacques Artus, ‘A n E conom etric A nalysis of International Travel," IMF S taff Papers (1972), pages 579614; Jane S. Little, “ International Travel in the U.S. B alance of Payments," New E ngland Econom ic Review (M ay/June 1980), pages 42-55; and Jeffrey Rosensweig, "The Dollar and the U.S. Travel D eficit," Econom ic Review, Federal Reserve Bank of A tlanta (O ctober 1985), pages 4-13. Their find in gs suggest that a 1 percent rise in a cou ntry’s real GNP can be expected to raise its travel expenditures by som ewhat more than 1 percent. Table 1 International Travel and Passenger Fare Transactions In billions of dollars 1977 1981 1985 Total travel paym ents ( - ) Passenger fares .............................. Payments by U.S. travelers in foreign countries .................... 10.2 2.7 16.0 4.5 23.8 7.3 7.5 11.5 16.5 Total travel receipts Passenger fares .............................. R eceipts from foreign travelers in the United States . . 7.2 1.0 15.5 2.6 14.1 2.5 6.2 12.9 11.6 Net travel and passenger paym ent deficit 3.0 0.5 9.7 Memo: U.S. current account balance . . -1 4 .5 * 6.3 -1 1 7 .8 * 'In d ic a te s deficit. The author thanks B ernadette Alcam o for her research assistance in the preparation of this article. 44 FRBNY Quarterly Review/Summer 1986 IN BRIEF— ECONOMIC CAPSULES from abroad by somewhat more than 1 percent and to reduce travel expenditures by U.S. citizens by less than 1 percent.2 Examination of the experience of the past decade illu stra te s the response of the travel balance to exchange rate movements and real growth here and abroad (Chart 1). From 1977 to 1981 the travel deficit declined from $3 billion to $0.5 billion. Although the United States grew, on average, at a rate close to that of other industrial nations, the travel account improved, largely due to a 15 percent fall in the value of the dollar against the currencies of our major travel partners during this period.3 The deterioration in the travel deficit since 1981 can be attributed, in large measure, to the dollar’s appreciation (over 30 percent on both a traveland trade-weighted basis through mid-1985) and to faster real growth in the United States than abroad. From 1981 to 1985, total receipts from foreign travelers declined by almost 10 percent while expenditures by U.S. travelers abroad rose 44 percent. Developments in our bilateral travel flows since the late 1970s provide further evidence of how sensitive the travel balance is to exchange rate movements and real growth. Between 1977 and 1981, the U.S. travel deficit showed significant improvement with all major partners 2D ollar de p re cia tio n will norm ally lower the dollar value of travel paym ents less than proportionately. The resulting de clin e in real exp enditures by U.S. travelers is pa rtially offset by the rise in the dollar prices paid for food, lodg ing, and other items w hile abroad. On the other hand, the effect of increased real exp enditures by foreign travelers to the United States as a result of the falling dollar is reinforced by any rise in the dollar price s paid. 3The w eights for the tra vel-w eig hted do lla r index com puted for this article are bilateral shares in U.S. travel trade of eight major travel partners: C anada, Mexico, the United Kingdom , France, Germany, Italy, Japan, and the Bahamas. The index is in real terms, that is with exchange rates adjusted by indexes of relative national price levels. whose currencies appreciated against the dollar (Ger many, the United Kingdom, Japan, and Mexico). At the same time, our balance with Canada, whose currency depreciated against the U.S. dollar, deteriorated. The worldwide appreciation of the dollar over 198185 accompanied a deterioration in our major bilateral travel balances, as U.S. citizens stepped up their travel throughout the world. In contrast to the general pattern, however, our bilateral balance with Japan improved, partly because Japan grew somewhat more rapidly (on average) than the U.S. over this period, and because the yen depreciated less against the dollar than the major European currencies did. Our travel balance with Mexico is particularly sensitive to exchange rate movements because residents along the border can readily cross the boundary to shop. (Indeed, such spending amounts to over 50 percent of the two countries’ bilateral travel trade.) As a result, a real appreciation of the peso combined with rapid Mexican growth produced a $1.6 billion improvement in our bilateral balance from 1977 to 1981. The subse quent sharp rise in the dollar’s real value during 198283 led to a $2.6 billion decline in this balance. In the next two years our bilateral deficit changed little as an appreciation of the peso through mid-1985 apparently offset the effects of a weakening Mexican economy. Prospects Looking ahead, prospects for improvement in the U.S. travel deficit seem limited because the dollar’s depre ciation has been much less uniform than its prior appreciation. The dollar declined approxim ately 30 percent in real terms against the currencies of West Germany and Japan between the second quarters of 1985 and 1986. However, it has maintained its average value in relation to Western Hemisphere currencies; in Table 2 Regional Distribution of the U.S. Travel Balance: 1985 (excludes passenger fares) In billions of dollars Western H e m is p h e r e ........................... C anada ........................................... M exico ........................................... C aribbean and Central A m erica. W estern E u ro p e ..................................... Japan ..................................................... Travel paym ents Travel receipts Net balance 8.4 2.7 3.6 1.8 5.9 0.5 6.5 3.0 2.0 0.6 2.3 1.4 -1 .9 0.3 - 1.6 - 1.2 -3 .6 0.9 Percent Percent share of U.S. share of U.S. travel trade* m erchandise tra d e f 53 20 20 9 29 7 37 23 6 2 24 16 'B ila te ra l shares of U.S. travel flows (receipts plus payments, excluding passenger fares). tB ila te ra l shares of U.S. m erchandise trade (exports plus im ports). FRBNY Quarterly Review/Summer 1986 45 Chart 1 C h a rt 2 T h e U.S. T rav e l D eficit and the Dollar B illio n s o f d o lla rs Index, 1980=100 T ra d e -w e ig h t e d and T ra v e l- w e ig h te d Real V alu es of the Dollar Index, 1980=100 T5 0 -------------------------------------------------------------------------------- 1976 77 78 79 80 81 82 83 84 85 86 * T ra v e l-w e ig h te d inde x of the d o lla r’s re al v a lue (e xchang e rate ad ju sted by re la tiv e national p ric e levels) a g a in s t c u r re n c ie s of eigh t m ajor U.S. tra v e l p a rtn e rs . W eights a re b ila te ra l s h a re s o f U.S. travel. ^S e a s o n a lly a d ju s te d annual ra te s ; minus indica tes surplus. S ources: U.S. D epartm ent of C om m erce and Inte rn a tio n a l M onetary Fund, International Fin ancial S ta tis tic s . fact, the dollar has appreciated sharply against the Mexican peso. As a result, the real value of the dollar has fallen only 6 percent from last year’s average on a travel-weighted basis. This is significantly less than its decline during 1977-81 and reverses only about one-fifth of its rise since 1981. In contrast, the trade-weighted dollar has declined substantially, offsetting more than one-half of its prior rise as measured by most published indexes (Chart 2). The extent to which the travel balance responds to the dollar’s decline depends on the pace of real growth here and abroad. Assuming that the United States and other industrial nations grow at close to their potential rates (which implies increases in domestic demand growth rates abroad compared with the average of recent years), we could expect a modest decline of $1.5 to $3.0 billion in the travel deficit over the coming two years.4 Most U n d e rly in g this analysis is the assum ption that real travel e xp enditu res have an ela s tic ity of roughly 1.5 relative to incom e grow th and real exchange rate m ovements. 1976 77 78 79 80 81 82 83 84 85 86 * R e a l e ffe c tiv e tra v e l-w e ig h te d in d e x o f the d o lla r's value using b ila te ra l sh a re s of eigh t m a jo r U.S. tra v e l p a rtn e rs. ^M o rg a n G ua ran ty real e ffe c tiv e (b ila te ra l) index o f the d o lla r’s value a g a in s t c u rre n c ie s of eleven m a jo r U.S. m e rchan dise trade p a rtn e rs. S ources: M organ G uaranty T ru st C om pany and Interna tiona l M onetary Fund, Inte rn a tio n a l Financial S tatistics. of this decline should result from increased visits to the United States, particularly from Western Europe and Japan. Travel receipts might increase by as much as 20 percent annually over this period; spending by U.S. citizens abroad is likely to continue to increase, but at a slower rate than in recent years. Of course, the actual improvement in the travel bal ance may d iffe r sig n ifica n tly from this estim ate, depending on a variety of factors that are difficult to predict. For example, a pickup in foreign relative to U.S. growth would probably result in further improvement of the travel balance. Political factors also can be important influences on travel flows. In particular, recent events suggest that concerns over terrorism could sig n ifi cantly reduce U.S. travel abroad this year and next. If so, the decline in the travel deficit may be substan tially greater than economic factors alone would sug gest—although modest, in any case, in relation to the overall current account deficit (presently over $120 bil lion annually). Bruce Kasman 46 FRBNY Quarterly Review/Summer 1986 IN BRIEF— ECONOMIC CAPSULES February-April 1986 Report (This report was released to the Congress and to the press on June 4, 1986) Iteasury and Federal Reserve Foreign Exchange Operations The dollar moved substantially lower against all major currencies during the three months ended April, declining by more than 8 percent against most European currencies and by almost 12 percent against the Jap anese yen. The depreciation was a continuation of the trend that emerged in early 1985 and gained momentum after the September G-5 meeting. The U.S. authorities did not intervene for their own account in the exchange markets during the February-April period, and the dol lar’s decline proceeded without any new concerted intervention in the exchange markets. The decline took place against a background of downward-moving interest rates and narrowing interest rate differentials favoring the dollar. The market’s atti tude also seemed influenced by the implications of continuing large current account imbalances and by assessments of the varying inflation prospects in dif ferent countries as a result of oil price changes. Market participants were also influenced by perceptions of official views about the dollar, particularly in light of the January meeting of G-5 Finance Ministers and Central Bank Governors where they expressed satisfaction with the trend of exchange rates since the September meeting. As the period opened, the foreign exchange markets were in a sensitive phase. The dollar had already declined a substantial amount from its highs of the year before. Yet there was little evidence that the U.S. A report by Sam Y. Cross, Executive Vice President in charge of the Foreign Group at the Federal Reserve Bank of New York and Manager of Foreign Operations for the System Open Market Account. Willene A. Johnson was primarily responsible for the drafting of this report, assisted by Elisabeth S. Klebanoff. economy or its trade position had reaped enough benefit to allay concerns among market participants that U.S. authorities would wish to see further dollar depreciation. Indeed, statistics then being released indicated the U.S. deficit on merchandise trade had widened substantially during December, and the external sector continued to exert a significant drag on domestic production. The mid-February estimate of GNP growth for the final quarter of 1985 revealed a disappointing, downward adjustment. Statistics for January’s personal income and consumption were also worse than expected, leading many in the market to conclude that the outlook for 1986 was even less optimistic than they had believed earlier. Subsequent data were interpreted as indicating points of weakness, rather than strength, for the U.S. economic outlook. At the same time, concern faded about potentially adverse effects on prices of any continuing dollar depreciation. Inflationary expectations in the United States were being rapidly scaled back in response to further dramatic drops in prices for oil and some other commodities. Thus, the risk of an accelerating decline in dollar exchange rates was seen by many market observers to be less than before. In addition, through early April, interest rates in the United States continued to ease at a faster pace than those abroad as many market participants expected that the Federal Reserve might ease monetary policy considerably. As a result, interest differentials favoring the dollar continued to narrow. By April 16, interest differentials on long-term government securities fell to within 200 basis points visd-vis German and Japanese securities. Beginning in February, a heavy schedule of state FRBNY Quarterly Review/Summer 1986 47 ments and testimonies about U.S. economic policies, together with the approach of a number of important international meetings, provided opportunities for offi cials here and abroad to be questioned about their attitude toward exchange rate trends. Administration officials generally expressed satisfaction with the decline in dollar rates since the G-5 agreement in September. They noted that the dollar’s decline had been orderly and denied that there was any particular level or range of rates at which they expected or desired the dollar to trade. The President, in his annual State of the Union address, noted the problems that previous exchangerate fluctuations had caused for many Americans and asked Secretary of the Treasury Baker to determine whether it would be useful to hold an international C hart 1 T h e d o lla r c o n tin u e d to decline. P e rc e n t * J ap anese i A i Y\ x 'H i P < fri \ Pound ste rlin g \— r f V * S w iss French fra n c /\ \ G erm ar mark i ii i Im M J lu i la i d ii ill i i L u i i l i . i i J A S O N D J 1985 conference to discuss with other countries the role and relationships of currencies. Market participants and journalists tended to conclude from these statements that U.S. authorities would welcome a further depre ciation of the dollar, notwithstanding the repeated deni als by U.S. officials that they had a target for the dollar. In these circumstances, the market’s attitude toward the dollar was predominantly bearish. During February and early March, the dollar declined across the board without interruption. The focus of market attention initially was the yen. It was supported by mounting monthly trade surpluses and the view that declining oil prices would be particularly beneficial to the Japanese economy. The dollar, which had closed at ¥191.4 0 at the end of January, declined steadily throughout early February to breach the psychologically important ¥180 level and to reach ¥177.40, a seven and one-half year low, by February 19. The yen’s rise then stalled after Japan’s Finance Ministry confirmed that it was developing plans to ease regulations on capital outflows which could have the effect of raising the demand for dollars by Japanese institutional investors. At this point market participants shifted their focus to the dollar/mark exchange rate. Forecasts of German economic growth in 1986 were being revised upward toward 4 percent. Given the relatively lackluster growth performance in the United States during the last quarter of 1985 and the weakness in several U.S. business indicators early in 1986, Germany emerged as a likely y ’Sv I m 11111111 i I ll .l F M A 1986 U nited S ta te s m o n e ta ry a u th o r itie s did not in te r v e n e F e b ru a ry th ro u gh April. Chart 2 E x c h a n g e m a r k e t p a rtic ip a n ts fo c u s e d on th e policy im p lic a t io n s of p e rs is te n t t r a d e im b a la n c e s . B illio ns of U.S. d o lla rs Me nth ly tra d e bal ances * B illio n s of U.S. d o lla rs e q u iv a le n t 4 -------------------------------------------------------------------------------------------N et p u rc h a s e s of fo re ig n c u rre n c ie s 3 ----------------------- ------------------- --------------------------------------------2 ----------------------- --------------------------------------------- 1----------------- -------------------------------- o l_______2______ J ____________ 11.~ — - — :: •„_______Q_____ May 1985Jul 1985 Aug 1985O c t 1985 Nov 1985Jan 1986 Feb 1986Apr 1986 1 , ... Japan * P erce ntag e cha nge of w eekly ave ra g e ra te s fo r d o lla rs from the a v e ra g e ra te fo r the w eek ending A p ril 25, 1985. F ig ures c a lc u la te d from N ew York noon qu otatio ns. 48 FRBNY Quarterly Review/Summer 1986 G erm an y U n ite d S ta te s ♦ A v e ra g e o f th e se a s o n a lly a d ju s te d tra d e b a la n ce s fo r D e c e m b e r 1985 and Jan uary and F e b ru a ry 1986. candidate for relatively high growth. Moreover, the German trade surplus had been increasing during the previous few months. The continued drop in oil prices made it appear possible that Germany’s trade surplus for 1986 would be double the level of the previous year. Traders also anticipated a possible realignment within the European Monetary System (EMS) in which the mark would be revalued. With these expectations in the minds of market participants, the dollar declined against the mark until early March as far as DM2.1960. On March 6 and 7, the central banks of Germany, C h a rt 3 The de clin e in w orld oil prices. D o lla rs p e r b a rre l 35- 25- \ 20- B re n t oil n . \ WestTexasN^ V 15- \ ! i i i I i.i i 10L Nov D ec 1985 M i l l Jan / V " 1 1 1 l._ l...1 1 Q - L . l Feb M ar Apr 1986 1 Japan, the United States, France, and the Netherlands lowered their official interest rates. The central banks of the United Kingdom, Italy, and Sweden soon followed by cutting their official lending rates. These actions had little immediate impact on exchange rates since the reductions had been widely anticipated and international differentials in market rates were expected to be largely unchanged. But the concerted round of interest rate reductions underscored to the market the potential for further coordinated policy actions. For some time thereafter, most dollar exchange rates moved narrowly as markets reassessed the near-term outlook for the dollar and the prospects for other coordinated actions. Early in April, the dollar resumed its decline. Indica tions that world oil exporters were failing to agree on a plan to support oil prices kept alive expectations that inflation in the major countries would continue to slow. Thus dealers anticipated that the major central banks would act soon to cut interest rates again. Some market participants even thought that the Federal Reserve might cut its discount rate more than other central banks. In the United States, there was concern about the impact of lower oil prices on U.S. banks with expo sure to the oil-exporting developing countries and to the domestic energy sector. In Germany, the central bank had expressed doubts that a further cut in the discount rate would be appropriate given existing conditions in the domestic economy. Selling pressures against the dollar then emerged, C hart 4 showed th ro u g h in low er rates of in fla tio n . In te re st d iffe re n tia ls fa vo ra b le to the d o lla r narrow ed as long-term in te re s t rates declined in the United States more ra p id ly than abroad. P e rc e n t 12 ------------- ------------------C o n su m e r p ric e inde x T w e lv e -m o n th e rc e n ta g e c h a n g e P e rce n t 14 Y ie ld s on te n -y e a r m a tu ritie s 10 - Mi l , 12 S^ ' ^ SN ^ U .S . Treasu ry bonds F rance o----- —0-—_ 10 " U n ite d St at es Jap an G erm any ^ G erm an p u b lic -s e c to r bonds — Ja p a n e se G ove rnm ent bonds ---- — 2 L Nov J Dec Jan 1985 1............ 1.... Feb M ar 1986 _L Apr .... J L i F M 1 A i M i J 1 i i 1 J A S O 1985 i N i D _L_i J I I F M A 1986 FRBNY Quarterly Review/Summer 1986 49 especially against the yen. That currency had already appreciated significantly during the past year, and Jap anese exporters, particularly small- and medium-size firms, were facing a substantial drop in the demand for their products. Under these circumstances, Japanese officials began to voice concern whenever the yen advanced toward the level of ¥175 against the dollar. Dealers therefore remained wary that foreign exchange market intervention or some other official action might be taken to curb the exchange rate move. But the demand pressures on the yen were becoming so strong as to bring into question the Japanese authorities’ ability to resist its appreciation unilaterally, and the yen strengthened relative to all major currencies. Against the major European currencies, the dollar was influenced early in April by strains emanating from a realignment of exchange rates within the joint inter vention arrangement of the EMS. Market participants had long expected that some adjustment in exchange rates might occur soon after the March elections in France to adjust for inflation differentials that had existed between the participating countries since the general realignment in 1983. As such an event was thought to become more imminent, the dollar’s decline against the mark slowed. Some dealers were antici pating that the dollar might benefit from speculative reflows out of the mark after the realignment. But the realignment that occurred over the first weekend in April unleashed a strong demand for French francs against all currencies. French residents unwound long-standing commercial leads and lags and nonresidents sought quickly to build up investment positions in francs and to benefit from the relatively high interest rates in France. As a result, the dollar again came on offer against the European currencies after the realignment. By late April, the dollar had declined 10 percent against the Continental currencies and 13 percent against the yen from end-January levels. It touched ¥166.10 against the yen, a record low for the postwar period, and DM2.1520 against the mark, the lowest level against the German currency since April 1981. At this point the dollar was more than 37 percent below its highs vis-a-vis those currencies of about a year before. Meanwhile, the financial markets around the world were astir with talk of a renewed drop in interest rates in many major countries. The plunge in oil prices was beginning to show through in reduced inflation rates for a number of countries. Partly in sympathy, interest rates on U.S. long-term securities had declined almost to the level of the Federal funds rate, indicating strong market expectations that a further cut in the Federal Reserve’s discount rate was in the offing. Japanese bond yields fell to postwar lows. And in Europe, a number of coun tries that had not participated fully in the easing of interest rates prior to the EMS realignment were seen as having increased scope to cut rates. In fact, the moderation of global inflationary expec tations did provide the impetus for a lowering of official interest rates in a number of countries. The Federal C hart 6 O ffic ia l in te r e s t ra te s w e re lo w e re d . C h a rt 5 P e rce n t In th e United S ta te s , lo n g -te rm yields fell m o re ra p id ly than s h o rt-te rm yields. U nited K ingdom P e rc e n t .J " "L L Fra nee I-----------1 |___ L U nited S tate s I— ——■i Japan i_________ i G erm any I Sep Y ie ld s on a c tiv e ly -tra d e d is s u e s a d ju s te d to c o n s ta n t m a tu ritie s . 50 FRBNY Quarterly Review/Summer 1986 I O ct 1985 I Nov 1 Dec Jan 1 Feb 1986 1 M ar I Apr Net Profits ( + ) or Losses ( - ) on United States Treasury and Federal Reserve Current Foreign Exchange Operations In m illions of dollars Period February 2, 1986A pril 30, 1986 .................... Valuation profits and losses on outstanding assets and lia b ilities as of A pril 30, 1986* . . . . Federal Reserve United States Treasury Exchange S tabilization Fund - 0- - 0- + 962.9 + 1,031.4 Data are on a value -da te basis. 'V aluation gains represent the increase in the do lla r value of ou tstand ing c u rrency assets valued at en d-of-period exchange rates, com pared with the rates prevailing at the tim e the foreign currencie s were acquired. Reserve and the Bank of Japan both lowered their dis count rates by one-half of a percentage point, effective April 21. Central banks in the United Kingdom, France, and Italy lowered their official rates in one or two steps by one-half of a percentage point or more at about the same time. The German central bank, feeling more constrained by the weakness of the mark within the EMS, did not join in this second round of reductions of official lending rates. At the end of April, the dollar’s decline paused. Some foreign exchange and bond dealers had expressed concern that investors might be reluctant to acquire the additional dollar-denominated assets needed to finance the continuing U.S. current account and fiscal deficits. Market participants were also well aware of pressures on foreign governments to ease the pain for their own industries of too-rapid a fall of the dollar. Accordingly, they considered the possibility that some agreement to support the dollar might emerge from the discussions at the Economic Summit in Tokyo early in May. In addition, by late April, there was evidence that foreign participation in U.S. securities markets continued to be strong. The dollar therefore moved up somewhat from its lows by the close of the period under review. The EMS Realignment On April 6, the European Community announced a realign ment of the central rates within the EMS. This was the first overall realignment of EMS central rates in more than three years. The realignment was initiated by the new French government as part of its program to restore competitiveness in the French economy and dismantle exchange and other financial controls. It involved in effect a devaluation of almost 6 percent for the French franc’s bilateral central rates against the German mark and the Dutch guilder. Other adjustments against the mark and the guilder were much smaller, including 3 percent effective devaluations in the central rates of the Italian lira and the Irish pound and 2 percent for the Danish krone and the Belgian franc. Movements of the market exchange rates for the EMS currencies after the realignment were relatively small. The French franc depreciated 31/2 percent against the mark, substantially less than the 6 percent devaluation of its bilat eral rate with the German currency. In other respects the configuration of exchange rates within the EMS also showed only modest change—except for the German mark, together with the Dutch guilder, which moved from the top to the bottom of the narrow band. Heavy reflows of funds following the realignment were reflected in substantial changes in the foreign exchange reserves of several countries. German reserves declined by almost $4 billion equivalent during the two weeks following the realignment, although much of this decline was recouped later in the month. At the same time, there were large increases in the foreign-currency reserves of France and Italy during April. These two countries, as well as the other countries whose EMS central rates were effectively lowered against the guilder and the mark, took advantage of the relief from exchange market pressure late in April to add reserves, ease exchange controls, and lower interest rates. C h a rt 7 T he G e rm a n m a rk m o v e d to th e b o tto m of th e EMS a fte r th e re a lig n m e n t. EMS re a lig n m e n t P e rc e n t* Feb Mar A pr 1986 W ee kly a ve ra g e s of d a ily 9 a.m. ra te s. ♦ P erce ntag e d e v ia tio n o f ea ch c u rre n c y from its ECU c e n tra l ra te. D o tte d lin e s c o rre s p o n d to th e S y s te m 's 2 % p e rc e n t lim it on m ovem ent fro m b ila te ra l ce n tra l e x c h a n g e ra te s fo r all p a rtic ip a tin g c u rre n c ie s e x c e p t the Italian lira. The lira may flu c tu a te 6 p e rc e n t fro m its c e n tra l ra te s a g a in s t o th e r EMS c u rre n c ie s . FRBNY Quarterly Review/Summer 1986 51 Subscriptions to the Quarterly Review (ISSN 0147-6580) are free. Multiple copies in reasonable quantities are available to selected organizations for educational pur poses. Write to the Public Information Department, 33 Liberty Street, New York, N.Y. 10045 (212-720-6134). Single and multiple copies for United States and for other Western Hemisphere subscribers are sent via third- and fourth-class mail, respectively. All copies for Eastern Hemisphere subscribers are airlifted to Amsterdam, where they are then forwarded via surface mail. Multiple-copy subscriptions are packaged in envelopes containing no more than ten copies each. Quarterly Review subscribers also receive the Bank’s Annual Report. Quarterly Review articles may be reproduced for educational or training purposes only, providing they are reprinted in full, distributed at no profit, and include credit to the author, the publication, and the Bank. Library of Congress Catalog Card Number: 77-646559 FRBNY Quarterly Review/Summer 1986 A 15-chapter book entitled Recent Trends in Commercial Bank Profitability—A Staff Study was recently completed by the Federal Reserve Bank of New York. The book traces the performance of the banking industry over the past decade. Among the areas cov ered are competition from foreign and domestic banking institutions and nonbank financial institutions, as well as the effects of inflation, interest rates, loan loss provisions, and deregulation on bank earnings. It includes a review of recent literature on the subject and commentary by leading bankers and financial analysts on the condition of the banking industry. The new 379-page bound publication is available for $10 a copy from the Public Information Department, Federal Reserve Bank of New York, 33 Liberty Street, New York, N.Y. 10045. Checks or money orders should be made payable to the Federal Reserve Bank of New York. Payment will be accepted in U.S. dollars only; if ordering from outside the United States, please use a check or money order drawn on a U.S. bank or its foreign branch.