The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Vol. 29, No. 3 ECONOMIC REVIEW FEDERAL RESERVE BANK OF CLEVELAND ECONOMI C REVIEW 1993 Quarter 3 Vol. 29, No. 3 Capital Requirements and Shifts in Commercial Bank Portfolios 2 by Joseph G. Haubrich and Paul Wachtel Since 1989, U.S. commercial banks have shifted their portfolios away from commercial loans toward government securities. Using data for in dividual banks, the authors document this shift and test for whether it can be attributed to the imposition of risk-based capital requirements. Their results indicate that these requirements may indeed account for part of the portfolio shift. FDICIA’s Emergency Liquidity Provisions 16 Economic Review is published quarterly by the Research Depart ment of the Federal Reserve Bank of Cleveland. Copies of the Review are available through our Public Affairs and Bank Relations Depart ment. Call 1-800-543-3489, then immediately key in 1-5-3 on your touch-tone phone to reach the pub lication request option. If you pre fer to fax your order, the number is 216-579-2477. Coordinating Economist: William T. Gavin by Walker F. Todd The Federal Deposit Insurance Corporation Improvement Act of 1991 (FDICIA) made a potentially significant change in the standards for Federal Reserve discount window access by nonbanks. In exploring the background of this issue, the author contends that although most of the legislation retrenched the federal financial safety net for under capitalized insured depository institutions, the provision effectively ex panded the safety net for uninsured nonbanks, irrespective of their capital or net worth positions. Efficiency and Technical Progress in Check Processing 24 by Paul W. Bauer Cost functions can provide valuable insights into the efficiency and technological constraints faced by firms. Using panel data for 47 Federal Reserve offices from 1983:IQ to 1990:IVQ, the author examines the cost of providing check-processing services by esti mating a multiproduct cost function using an econometric frontier approach. The article demonstrates how the Federal Reserve’s unit cost measures of performance can be decomposed into separate effects related to differences in cost efficiency, output mix, input prices, and environmental variables to provide a much richer un derstanding of the sources of relative office performance. Estimates of technical progress are also presented. Advisory Board: Jagadeesh Gokhale Erica L. Groshen Joseph G. Haubrich Editors: Tess Ferg Robin Ratliff Design: Michael Galka Typography: Liz Hanna Opinions stated in Economic Re view are those of the authors and not necessarily those of the Fed eral Reserve Bank of Cleveland or of the Board of Governors ot the Federal Reserve System. Material may be reprinted pro vided that the source is credited. Please send copies of reprinted material to the editors. ISSN 0013-0281 Capital Requirements and Shifts in Commercial Bank Portfolios by Joseph G. Haubrich and Paul Wachtel Introduction A dramatic and virtually unprecedented shift in the portfolio structure of U.S. commercial banks has taken place since 1989. Specifically, govern ment securities as a share of total loans has risen from 15 percent in 1989 to more than 22 percent today. This portfolio shift has coincided with an important change in the financial regulatory struc ture. Bank regulators around the world agreed to a common set of risk-based capital requirements in mid-1988. These requirements were phased in gradually in the United States and became fully effective this year. Some have suggested a connection between the regulatory changes and the portfolio shift, although this claim has not been substantiated. In this paper, we will present some rather strong evidence that the portfolio shift is consistent with regulatory change, which has increased the attractiveness of government securities as an asset.1 The evidence comes from an exami nation of the quarterly “call report” data on commercial banks from the Federal Financial Institutions Examination Council (FFIEC). Joseph G. Haubrich is an eco nomic advisor at the Federal Re serve Bank of Cleveland, and Paul Wachtel is a research profes sor in the Department of Econom ics at the Leonard N. Stern School of Business, New York University. The authors thank Robert Avery, Allen Berger, and James Thom son for helpful comments. Though bankers and their regulators find the portfolio shift interesting in itself, it also has broader implications. Our results provide evidence that regulation matters— a point of considerable debate for capital requirements in particular (Keeley [19881) and for public poli cies in general (Stigler [19751). The reason is that bank portfolio risk strongly affects the chance of financial collapse and an associated government bailout. Concerns about this possi bility motivated the risk-based capital standards in the first place. Furthermore, by altering the credit available to businesses and consumers, a shift in bank portfolios may slacken the pace of economic recovery. The new risk-based capital requirements classify bank assets. Government securities are deemed to be riskless and therefore have a zero weight w hen the bank determines its re quired capital.2 Thus, a bank that finds it diffi cult to meet its capital requirements can do so by shifting its asset portfolio away from loans and other high-risk-weighted assets toward government securities. ■ ■ 1 For some other interesting approaches to the same problem, see Furlong (1992), Jacklin (1993), and Hancock and Wilcox (1992). 2 U.S. government securities have a zero risk weight because there is no default risk. However, they are subject to interest-rate risk, and the new capital requirements have been criticized for ignoring this component. FIGURE 1 Growth in Loans and Securities for Commercial Banks, 1973-93 Billions of dollars SOURCE: Board o f Governors o f the Federal Reserve System, statistical release G.7. FIGURE 2 Growth in Government Securities, C&l Loans, and Real Estate Loans for Commercial Banks, 1973-93 Billions of dollars There are, of course, other plausible reasons why bank portfolios have shifted toward gov ernment securities. First, the large loan losses of the 1980s made business lending appear more risky and less attractive. Second, the busi ness slowdown that coincided with the intro duction of risk-based capital requirements weakened loan demand. The decline in loan demand was exceptionally large in the recent recession because of the boom in business and consumer leverage in the mid-1980s. Thus, the shift toward government securities could also be the result of these factors. We submit that the changes in portfolio com position are strongly related to the introduction of risk-based capital requirements. Specifically, banks with the largest increases in government securities holdings tend to be those with the low est capital-asset ratios when the new capital re quirements were introduced. The conclusion is unaffected when we control for the weakness of the bank’s loan portfolio.3 Thus, the change in bank portfolios does not seem to be the result of this weakness. I. Aggregate Trends in Bank Assets The composition of commercial bank portfolios has changed dramatically over the years. For the first two postwar decades, banks reduced the proportion of their assets in securities and increased the proportion in loans. Part of the reason was the need to liquidate large holdings of government securities accumulated during World War II. Moreover, the development of highly liquid and active money markets as sources of funds reduced the precautionary need to hold both government securities and cash assets (see Boyd and Gertler [19931). These secular shifts in banks’ activities were completed by the early 1980s. Some dramatic changes have taken place more recently, however. Figure 1 shows the growth of total loans and total securities since 1973- After rapid gains beginning in 1973, the outstanding stock of bank loans has been con stant for the last three years. Total securities hold ings expanded less rapidly through the 1980s and began to speed up in the last three years. SOURCE: Board o f Governors o f the Federal Reserve System, statistical release G.7. ■ 3 It is more difficult to control for the influence of loan demand on portfolio shifts because we lack any bank-specific measures of the strength of demand. Still, while other factors may explain part of the portfolio shift, they do not overturn the importance of the new capital requirements. F I G U R E 3 Government Securities, C&l Loans, and Real Estate Loans as a Share of Total Loans and Securities, 1973-93 Percent SOURCE: Board o f Governors o f the Federal Reserve System, statistical release G.7. More detail is provided in figure 2, which shows three critical categories— government securities, commercial and industrial (C&I) loans, and real estate loans. The rapid increase in government se curities holdings since the late 1980s has clearly co incided with a substantial decline in the volume of C&I loans outstanding. Finally, figure 3 presents the proportions of these three critical categories in total loans and securities. The share of C&I loans began to de cline around 1984. Real estate loans as a percent of the total began to increase in 1986 and then leveled off around 1990. Most important, the pro portion of U.S. government securities in total loans and securities rose dramatically in the three years following 1989- The bank portfolio shifts of the last decade thus occurred in two stages: Banks initially turned from C&I loans to real es tate loans, but then shifted from loans to U.S. government securities in recent years. and secondary capital to assets. U.S. and for eign regulators agreed in 1988 to implement risk-based capital requirements. The new re quirements were phased in gradually begin ning in 1990 and became fully effective at the end of 1992 (see Saunders [19931). U.S. commercial banks are now required to have a minim um ratio of total (Tier 1 + Tier 2) capital to risk-adjusted assets of 8 percent.4 In order to calculate risk-adjusted assets, each as set is assigned to one of four risk categories and given a weight of 0, 20, 50, or 100 percent. U.S. government securities are in the first cate gory, with a risk weight of zero. C&I loans and most real estate loans (except securitized mort gage pools and regular residential mortgage loans) are assigned a weight of 100 percent. Risk-adjusted assets are thus simply a weighted average of the bank’s portfolio of assets. In ad dition, the entire portfolio faces a leverage re striction: Total capital must be 4 percent of total assets (unweighted).5 Thus, a commercial bank that moves its as set holdings from loans with a full 8 percent capital requirement to government securities with no capital requirement eases the associ ated regulatory burden. Clearly, banks that are inadequately capitalized have an incentive to increase the proportion of their assets in gov ernment securities. Our central hypothesis is simply that the large changes in bank balance sheets observed in the last three years repre sent a response to these incentives. One alternative hypothesis is that the shift into government securities was an effort to avoid risk as bank asset portfolios weakened in general. Because banks found it more diffi cult to manage their risky asset portfolios, they viewed government securities more favorably, and there was a flight to quality. This hypothe sis is credible in light of the documented dete rioration in the condition of commercial bank portfolios in the 1980s. ■ II. Changes in Bank Portfolios Regulation mandating commercial-bank capital requirements has evolved over the years. In 1985, regulators established a required ratio of book value of equity (primary capital) to assets of 5.5 percent. There was also a total capital re quirement of 6 percent for the ratio of primary 4 The minimum ratio of Tier 1 capital (primarily common stock equity) to risk-adjusted assets is 4 percent. Tier 2 capital includes certain types of preferred stock and subordinated debt. The details of the new rules are published in the Federal Register, January 27,1989, pp. 4186-221. ■ 5 The Basel agreements themselves specify only Tier 1 risk-based and total risk-based ratios. Outside the United States, banks face only those capital requirements. U.S. banks have an additional constraint: minimum leverage. While the capital guidelines implementing the Basel Accord specified a constraint of 3 percent, the prompt corrective action guidelines of the FDIC Improvement Act of 1991 (FDICIA) mandated a constraint ot 4 percent, except for banks with a regulatory CAMEL rating of 1. For a discussion, see Huber (1991), chapter 15, or Carnell (1992). F I G U R E 4 Nonperforming Assets and Net Charge-offs as a Share of Totals, 1983-92 Percent This downtrend is illustrated in figure 4. Nonperforming assets as a percent of total as sets and net charge-offs as a share of total loans and leases both began to rise in the mid1980s. However, this portfolio deterioration preceded the change in asset composition by several years. The shift of assets into govern ment securities started in the late 1980s when banks’ condition began to recover. The increas ing net charge-off rate and nonperforming loan rates in 1990 and 1991 stemmed not from an upturn in bad loans, but rather from a decline in total loans outstanding. As a final check, we control for the effects of loan quality in section V. Though we cannot rule out this factor, our results do indicate a risk-based capital effect independent of loan quality. An alternative explanation is that the change in bank portfolios was related to loan demand and overall economic conditions. Indeed, cycli cal changes in bank portfolio preferences are quite common. For example, when monetary policy eased in the mid-1970s and again at the end of the 1980-82 recessions, the government securities proportion of total loans and securi ties headed upward. The episode in the early and mid-1980s is similar to the current situation. Although monetary policy eased, banks were reluctant to boost lending, and the proportion of government securities in their portfolios in creased. At that time, the debt crisis in less developed countries influenced bank behavior. In both of the earlier cases, however, a grow ing economy generated loan demand and the run-up in government securities holdings lasted only about two years. In the recent episode, the rise in government securities holdings has continued for almost four years without any sign of abatement, plac ing the proportion of these securities in com mercial bank portfolios at unprecedented levels. This situation may be unique because the recovery that began two years ago has been particularly sluggish. Despite an expan sionary monetary policy, the persistently weak economy has held dow n loan demand, and as a result, banks continue to augment their hold ings of government securities. Although it is dif ficult to distinguish between the effects of weak loan demand or risk-based capital requirements on bank holdings of government securities, a cyclical response to demand is unlikely to be entirely responsible for the enormous portfolio shifts observed. A third alternative is that government securi ties became more profitable in the late 1980s. The combination of a steep yield curve and a large supply of government securities, driving prices down, may have made banks eager cus tomers. This term-structure argument requires more justification than is usually given: Many bank loans are long term, and thus could also be profitable for banks. The story appears to rest on some shift not in the term structure, but in the risk structure, between Treasury bonds and bank loans. This is less obvious than the initial statement, how ever. Perhaps the explosion in government debt drove down the price of Treasuries (though this point itself is controversial). In either case, such general factors should not affect individ ual banks differently. Therefore, our strategy of comparing well-capitalized and weakly capital ized banks is not sensitive to this shift. The lower prices for Treasuries may explain the portfolio shift of well-capitalized banks. III. Relationship between Capital and Portfolio Shifts Our hypothesis indicates that a bank’s incentive to satisfy the newly introduced risk-based capi tal requirements by adjusting its portfolio is larger if the institution initially fails those new requirements. That is, banks that will be capi tal constrained under the new standard if they T A B L E 1 Asset Allocations for Commercial Banks by Size Class, March 1990 (percent) r» ,• Size Class C Proportion of Total Assets Held as: 1 2 3 4 5 6 Cash assets Total securities Treasuries (book value) 8 30 10 6 30 9 6 26 8 7 19 6 8 19 5 10 15 4 Total loans C&I loans Real estate loans Mortgages (1-4 family) Consumer loans 51 9 30 13 11 54 11 30 14 12 59 13 30 14 14 65 16 28 11 17 65 18 24 65 21 19 7 14 9 17 SOURCES: Federal Financial Institutions Examination Council, Quarterly Reports on Income and Condition; and authors’ calculations. TABLE 2 Capital Ratios by Size Class, March 1990-September 1992 (percent) do nothing will thus take greater actions to comply. In this section, we present the data used to examine the relationship between the initial risk-based capital ratio of individual banks (in either 1988 or 1990) and the banks’ portfolio changes. Our data source is the quarterly call reports on all U.S. commercial banks. In addition, we show that a bank’s in centive to hold government securities increases even if it was not initially capital constrained. Although the risk-based capital requirements were announced in July 1988 and began to be implemented in March 1989, the call reports were not revised to reflect the new definitions until March 1990. Prior to 1990, however, it is possible to approximate the risk-based capital ratio of the bank from the available data. In both instances, we use algorithms developed at the Federal Reserve Board by Avery and Ber ger (1991) to derive the risk-adjusted assets of the bank. Thus, we will be able to look at the changes in bank portfolios over two periods: from June 1988 (when the new capital require ments were announced) to September 1992, and from March 1990 (when the phase-in of the new capital requirements began) to Sep tember 1992. Our data set consists of 12,187 commercial banks divided by asset size (as of March 1990) as follows:6 1. Less than $50 million 2. $50-100 million 3. $100-500 million 4. $500 million-1 billion 5. $1-5 billion 6. More than $5 billion Size Class 1 2 3 4 6 5 A. Capital Ratios and Changes Total capital/ risk-adjusted assets, March 1990 18.38 16.41 13.95 Change, 1990-92 Tier I capital/ risk-adjusted assets, March 1990 0.12 8.67 1.41 1.45 2.71 17.20 15.27 12.77 10.04 9.38 6.79 1.28 2.37 Change, 1990-92 0.12 1.14 11.44 10.82 1.09 1.27 1.21 1.37 B. Distribution of Banks by Capital Class, March 1990 Capital Class 0-4% 4-8% 8-10% 10-14% >14% 6.8 2.2 3.7 23.7 63.6 0.5 2.1 5.8 33.6 58.0 0.6 3.6 12.9 44.9 38.0 1.0 10.9 28.8 46.7 13.5 6,558 2,685 2,350 229 260 105 0.0 14.6 37.7 40.4 0.0 32.4 7.3 1.9 53.3 11.4 SOURCES: Federal Financial Institutions Examination Council, Quarterly Table 1 shows that commercial bank asset allocations differ according to bank size. For example, the smallest banks had only 9 per cent of their assets in C&I loans, while the pro portion for the largest banks was 21 percent. However, the asset allocation changes that oc curred over the two-and-a-half-year period be ginning in March 1990 were common to all sizes of banks (the very smallest were some times an exception). Holdings of securities, par ticularly Treasury securities, rose and loans (except for real estate loans) decreased. The top part of table 2 shows the ratios of capital to risk-adjusted assets at the start of the period and the change for banks in each size class over the sample period. O n average, Reports on Income and Condition; and authors’ calculations. ■ 6 Banks were removed from the sample it the data seemed to be erroneous, if extreme outliers were present, or if the banks had greater than 50 percent capital. T A B L E 3 Bank Adjustment to Risk-Based Capital Requirements: Portfolio Shifts, Growth, and Raising Capital Size Class Capital Class 0 1 2 3 4 2 1 4 3 5 6 Portfolio Shift P -0.08 - 0.11 - 0.06 - 0.01 0.06 0.59 - 0.10 - 0.10 -0.03 0.01 - 0.20 -0.13 - 0.06 -0.03 0.01 — — — -0.14 •-0.08 -0.02 0.07 - 0.10 -■0.08 -•0.06 0.30 -0.08 -0.07 0.00 - 0.10 Size Shift TA 0 1 2 0.87 0.09 0.24 3 4 0.25 0.31 0.70 0.01 0.38 0.28 0.19 - 0.20 0.15 0.20 0.22 0.19 — 0.04 0.19 0.16 0.36 — 0.17 0.20 0.20 0.10 — 0.00 0.10 0.27 0.28 Capital Shift C 0 1 2 3 4 48.55 0.89 0.47 0.32 0.24 12.01 0.46 0.48 0.31 0.20 2.42 0.48 0.30 0.28 0.21 — 0.35 0.31 0.20 0.24 — 0.56 0.30 0.22 0.15 — 0.34 0.30 0.30 0.29 SOURCES: Federal Financial Institutions Examination Council, Quarterly Re ports on Incom e and Condition; and authors’ calculations. banks of all sizes were sufficiently well capital ized; the minimum total capital requirements were 8 percent. Finally, in every size class, banks augmented capital in this period. To explore the relationship between portfo lio changes and capital requirements, we classi fied banks by total capital to risk-adjusted asset groups at the start of the period. The capital re quirement classes and the distribution of banks by size class are shown in the bottom part of ta ble 2. Most smaller banks had very high capitalasset ratios, although there were a significant number of exceptions. As bank size increases, the proportion of banks with capital ratios under 8 percent rises as well. When we reach the largest size class, very few banks exceeded the minimum capital requirement by a comfortable margin. Under this classification scheme, banks that are severely undercapitalized (0 to 4 percent capital ratio) or moderately undercapitalized (4 to 8 percent capital ratio) must meet the new requirements to stay in business. They may downsize, raise new capital, or rebalance their portfolios to take advantage of the different risk weights. The explicitly undercapitalized banks are not the only ones facing incentives to increase their capital, however. Regulators require banks to hold capital well in excess of the minim um requirements in order to expand or to be able to acquire new entities or busi nesses.7 A bank that just satisfies the 8 percent minimum capital ratio and wishes to sell m u tual funds, for example, would probably need to increase its capital ratio before obtaining regulatory permission. To assess how banks responded to the new capital requirements, we explore the nature of capital. Capital satisfies the following identity: Capital = (capital/risk-weighted assets) x (1 ) (risk-weighted assets/total assets) X total assets. In other words, C = R x P x TA, where C = capital, R = the risk-weighted capital ratio, P = the portfolio factor, and TA = total assets. Using the standard circumflex notation for a AC proportionate changes (C = -77-), we get A A A A O C =R +P + TA , or (2) A r - A A A c - P - TA. Because the risk-adjusted capital requirements are a constraint on R, we see that equation (2) descriptively allocates the adjustment of banks to three possible courses of action: raise capi tal (increase C), adjust the portfolio (lower P), or shrink total assets (lower TA ). Table 3 re ports this breakdown. Three patterns stand out in table 3- Banks did shift their portfolios in a way that reduced their capital requirements. Furthermore, this shift was more pronounced for undercapitalized banks at every size level. Banks likewise responded by raising capital, although the well-capitalized banks apparently raised more. Finally, on aver age, banks did not shrink, and in fact grew over this period in every size and capital class. These patterns confirm our primary emphasis on the portfolio effects of the new capital requirements. ■ 7 FDICIA directs bank regulators to use the risk-based capital re quirements in making supervisory decisions. The Act established five categories based primarily on the bank’s capital position. To be consid ered well capitalized, a bank would have to exceed the minimum capital requirements by a substantial margin. We caution the knowledgeable reader that the capital classes we use are not FDICIA prompt-correctiveaction zones. B O X I Why ANOVA? Though commonly used in many areas of statistics, analysis of variance (ANOVA) is less popular among economists, who gen erally prefer regression analysis. For evaluating bank portfolio shifts, however, ANOVA has several advantages. First, it does not require assumptions about the nature of the functional form of the statistical relation: In particular, it does not impose a linear relation between capital and portfo lio shifts. A difference in the response of well-capitalized and undercapitalized banks assumes a nonlinear response by defi nition. The different degrees of capital constraint (for exam ple, deeply undercapitalized, barely capitalized) coupled with our ignorance about the correct form of the relation (linear, logarithmic, quadratic) make the ANOVA specification particu larly attractive. ANOVA might also be called “comparison of means.” It sta tistically estimates the effects due to various factors (here, they are size and capital class) and then allows comparison of those effects— analyzing how and why the cells of table 6 differ from each other. ANOVA has a further advantage in that it facilitates the esti mation and interpretation of interaction effects. Our analysis con siders two m ain effects, size and capital. Accounting for each one separately may not provide the whole story: The main ef fects may not be additive, and there may be interaction effects. For example, undercapitalized large banks may receive more scrutiny from the regulators or find it easier to invest in certain markets, and so may adjust their portfolios differently. Banks had another reason to adjust portfolio shares. The new requirements changed the re turns on different types of investments. Relative to business and commercial real estate loans, government securities became more profitable because they required less capital backing. A simple calculation shows that the difference can be substantial. The standard way to approach these issues is with a version of the Miller (1977) debt model as extended to banks by Orgler and Taggart (1983). Banks have two sources of funds: deposits and equity. Deposits have a tax advantage in that banks may deduct inter est paid as a business expense, but cannot de duct dividends paid on equity. Deposits have an additional cost of reserve requirements, but in general banks would prefer to raise funds using debt. Banks cannot fund themselves ex clusively with deposits, however, because they face a constraint on their funding, namely a capital requirement that the ratio of debt (for example, deposits) to equity not exceed a limit If we denote the return on deposits as rd and the return on equity as re, the marginal cost of raising funds, r, is given by (3) r= re/ ( l - t) + ^ r a 1 + S (1 - P ) where t is the corporate tax rate and p is the reserve requirement. The bank lends until the return on the loan equals the cost of funds needed to fund the loan.8 The capital requirements impose a different C, on different assets, and thus induce a different rate of return. As an example, consider a return on equity re of 10 percent, a return on deposits rd of 4 percent, a corporate tax rate t of 28 per cent, and required reserves p of 12 percent. A U.S. Treasury bond has a £ of 24 (a zero risk weighting and the 4 percent leverage require ment that becomes a debt-to-equity ratio of 0.96/0.04), while a C&I loan has a £ of 11.5 (a 100 percent risk weighting). In this case, the cost of raising funds internally (r7) to buy a Treasury bond is 4.9 percent, while the cost of raising funds internally to make a loan is 5.4 percent. The relative cost of loans has increased, making their inclusion in a portfolio less attractive.9 IV. Analysis of Variance The relationship between portfolio changes and the risk-adjusted capital ratio prior to the intro duction of risk-based capital requirements is ex amined with an analysis of variance (ANOVA, detailed in box 1). We investigate the relation ship for the asset categories outlined in table 4. ■ 8 A little more intuition on the exact form of equation (3) can be gained as follows. Assume that the bank wishes to raise one dollar as cheaply as possible. The bank would like to use debt, for which it pays rd , but it faces a capital constraint, so it can raise only a fraction of the funds us ing debt. It also must raise equity, and must pay more than rg because of corporate income tax. This explains the first term in the numerator. Because the bank raises money from two different sources, the actual cost is a weighted average of the cost of funds from those sources, and a little algebra shows that the 1/1 + C, and £ /1 + £ terms provide the proportion of equity and debt to total assets. Finally, some of the debt must be invested in re quired reserves, so to invest one full dollar, the bank must raise slightly more than that, which accounts for the p term in the denominator. ■ 9 The general situation is more complicated, of course. For exam ple, some banks can meet their capital requirement by increasing their Tier 2 capital. This includes subordinated debt, which despite being more expensive than deposits avoids the corporate tax penalty of equity. 9 TABLE 4 Proportion of Total Assets (percent) March 1990 Cash assets Total securities Treasuries Other securities September 1992 7.3 6.0 28.7 31.7 9.3 10.3 19.3 21.4 Total loans 53.8 54.2 C&I loans 11.0 9.3 Mortgages (1-4 family) 13.2 14.5 29.6 30.4 Other loans (includes other real estate) SOURCE: Authors’ calculations. TABLE 5 Analysis of Variance Results— Probability that Observed Effect Is Due to Chance Difference across Classes by: Size and Capital Size Capital Asset Changes, 1988-92 Cash assets Total securities Treasuries Other securities Total loans C&I loans Mortgages Other loans 0.003 0.0 0.0 0.0047 0.0 0.0 0.178 0.0 0.0017 0.0 0.0 0.0003 0.0 0.0 0.708 0.0 0.0384 0.8049 0.3661 0.7267 0.0072 0.0005 0.9200 0.0894 0.2578 0.0 0.0 0.006 0.0 0.0 0.9709 0.0 0.0 0.0 0.0 0.0 0.1272 0.0 0.0215 0.6090 0.0176 0.2811 Asset Changes, 1990-92 Cash assets Total securities Treasuries Other securities Total loans C&I loans Mortgages Other loans 0.1553 0.0 0.0003 0.0017 0.6553 0.0270 The ANOVA was performed for the change in the ratio to total assets for each category for two time periods. The first period begins in June 1988, just before the risk-based capital require ments were announced, and the second one starts in March 1990, the first available data after the requirements were phased in. This process shows whether the changes in the asset ratios dif fer significantly across size or capital classes. The ANOVA F-tests for the effects of size and capital class are summarized in table 5, which presents the probabilities at which the null hypothesis of no significant effect can be rejected. That is, it gives the probability that all effects of the given type are zero. The first col um n provides the overall test on all the effects and interactions. The next two columns are tests that depend on the ordering of the vari ables. The second column tests for the signifi cance of the size effects alone. This test is based on the sum of squares, putting the size effect in the estimation first. The third column is a stringent test for the significance of the capital class effects. It is based on the sum of squares when the capital class is added last; it tests the significance of the additional effect of this variable, having already controlled for the size and interaction effects.10 In most instances, there are significant differ ences in asset changes among banks of various size classes. This reflects a wide divergence in portfolio allocations between large and small banks. More important, the differences across capital classes are significant even at the 5 per cent level for only a handful of asset categories. For the asset changes between 1988 and 1992, there are substantial differences across risk-adjusted capital ratio classes for only cash assets and C&I loans. The changes in Treasuryto-total-asset ratios do not vary much across capital ratio classes (p = 0.3661). However, when we examine changes from the introduc tion (rather than the announcement) of the risk-based capital requirements, 1990 to 1992, additional significant changes arise. For this period, the changes in the Treasury-to-asset ra tios vary widely by capital ratio class ( p = 0.0176). In addition, there are substantial differences at the 5 percent level for cash assets, C&I loans, mortgages, and other loans. The ANOVA results indicate a strong rela tionship between the initial capital ratio and SOURCE: Authors’ calculations. ■ 10 For a theoretical background, see Searle (1971); lor a discus sion of the tests, see the SAS/STAT User's Guide (1990), chapters 9 and 24. The SAS system refers to the last two columns in table 5 as type I and type III tests. T A B L E 6 Change in Selected Asset Ratios, 1990-92 Size Class Capital Class 1 2 4 3 5 6 0.00 0.03 0.02 0.00 0.03 0.02 0.02 Government Securities 0-4% 0.04 0.04 4-8% 8-10% 0.03 10-14% 0.01 >14% -0.01 0.03 0.03 0.02 0.02 0.00 0.09 0.03 0.02 0.02 0.01 0.00 0.02 0.03 0.02 -0.02 0.03 0.01 -0.03 Cash 0-4% ' 4-8% 8-10% 10-14% >14% -0.01 -0.01 -0.01 -0.01 -0.01 0.00 --0.01 --0.01 --0.01 --0.01 0.01 -0.01 -0.01 -0.01 -0.01 _ _ _ 0.00 -0.01 -0.02 -0.02 -0.01 -0.01 -0.01 -0.04 -0.01 -0.01 -0.01 -0.07 C&I Loans 0-4% 4-8% 8-10% 10-14% >14% -0.05 -0.03 -0.03 -0.02 -0.01 --0.02 --0.03 --0.03 --0.02 --0.01 -0.05 -0.05 -0.04 -0.03 -0.02 __ -0.04 -0.03 -0.03 -0.01 __ -0.04 -0.04 -0.03 -0.01 __ -0.02 -0.03 -0.02 -0.02 Mortgages 0-4% -0.00 4-8% 0.01 8-10% 0.01 10-14% 0.01 >14% 0.01 0.02 0.02 0.01 0.01 0.01 -0.02 0.02 0.02 0.01 0.01 __ -0.03 0.01 0.02 0.01 _ __ 0.00 0.02 0.01 0.01 0.02 0.01 0.03 0.00 SOURCES: Federal Financial Institutions Examination Council, Quarterly Re ports on Incom e and Condition; and authors’ calculations. bank portfolio changes. In particular, the changes emerge more clearly when the phasein of the new regulations began rather than at the time they were announced. Two reasons account for this delay: First, risk-based capital requirements represented a radical change in U.S. banking regulation, so a period of learn ing about their consequences is not surprising. Second, if portfolio changes were made to im prove banks’ capital position, they were not necessary until the phase-in began. In addition, government security portfolios can be changed quickly and easily. The ANOVA significance tests suggest that there are important differences across asset ra tio categories, but do not imply any particular direction in the relationship. For the four asset categories with significant differences across capital-ratio classes, we show the actual mean changes in each capital class for the two-and-ahalf-year period after the introduction of riskbased capital requirements in table 6. The evidence is clear for both government securities and C&I loans. The extent to which the ratio of government bonds to assets in creased diminishes as the initial capital position of the bank improves. In fact, in four of the six size groups, the extremely well-capitalized banks (capital ratios greater than 14 percent) did not even boost their holdings of govern ment securities. The evidence for C&I loans is equally compelling. Banks in all categories de creased their portfolio share in C&I loans. In each size class, the fall in the C&I loan share was larger for the poorly capitalized banks. For banks with initial total-to-risk-adjusted capital of less than 8 percent, the share of gov ernment securities in total assets increased on average by 4 percentage points, and the share of C&I loans in total assets decreased by 4 per centage points. Thus, there is a strong indica tion that poorly capitalized banks responded to the new capital requirements by shifting from C&I loans to government securities. Be cause the movement away from C&I loans is at least partially due to the deteriorating quality of loan portfolios, it is important to see if the results are robust when we hold the quality of the portfolio constant. The mortgage results are more ambiguous, as expected. With a 50percent risk weight, they fall between commer cial loans and Treasury securities. Tables 5 and 6 do not completely make the case that a greater portfolio shift took place among undercapitalized banks. The F-test sug gests that the means differ, and the means themselves show greater portfolio shifts for un- D TABLE 7 Tukey Multiple Comparison Tests for Differences in Means A. Total Securities Alpha = 0.05, Confidence = 0.05, Degrees of Freedom = 10861, Mean Square Error = 0.009487, Critical Value of Studentized Range = 3.858. Capital Classes Com pared Simultaneous Lower Confidence Limit Difference betw een Means B. Treasury Book Alpha = 0.05, Confidence = 0.95, Degrees of Freedom = 10861, Mean Square Error = 0.005965, Critical Value of Studentized Range = 3.858. Simultaneous U pper Confidence Limit Capital Classes Com pared Simultaneous Lower Confidence Limit Difference betw een Means Simultaneous U pper Confidence Limit 0.044850 0.057620 0.063316 0.083068 0 0 0 0 1 2 3 4 -0.028772 -0.018189 -0.011794 0.003935 0.013774 0.023062 0.028919 0.044573 0.056321 0.064313 0.069632 0.0852113 0.008804 0.014403 0.020778 0.040624 0.062459 0.033228 0.037635a 0.0571923 1 1 1 1 0 2 3 4 -0.056321 -0.005640 0.001777 0.017660 -0.013774 0.009287 0.015145 0.030799 0.028772 0.024215 0.0285123 0.0439373 -0.033228 -0.057620 -0.004188 0.016125 -0.014403 -0.005599 0.006375 0.026220 0.004421 0.046423 0.016937 0.0363163 2 2 2 2 0 1 3 4 -0.064313 -0.024215 -0.002518 0.013506 -0.023062 -0.009287 0.005857 0.021511 0.018189 0.005640 0.014233 0.0295163 1 0 2 4 -0.037635 -0.063316 -0.016937 0.014214 -0.020778 -0.011973 -0.006375 0.019846 -0.0039203 0.039369 0.004188 0.0254773 3 3 3 3 0 1 2 4 -0.069632 -0.028512 -0.014233 0.011188 -0.028919 -0.015145 -0.005857 0.015654 0.011794 -0.0017773 0.002518 0.0201203 1 0 2 3 -0.057192 -0.083068 -0.036316 -0.025477 -0.040624 -0.031819 -0.026220 -0.019846 -0.024055 0.019429 -0.0161253 -0.0142143 4 4 4 4 0 1 2 3 -0.085211 -0.043937 -0.029516 -0.020120 -0.044573 -0.030799 -0.021511 -0.015654 -0.0039353 -0.0176603 -0.0135063 -0.0111883 0 0 0 0 1 2 3 4 -0.062459 -0.046423 -0.039369 -0.019429 -0.008804 0.005599 0.011973 0.031819 1 1 1 1 0 2 3 4 -0.044850 -0.004421 0.003920 0.024055 2 2 2 2 1 0 3 4 3 3 3 3 4 4 4 4 a. Significant at the 0.05 percent level. SOURCE: Authors’ calculations. dercapitalized banks, but neither approach in dicates which means differ from which other means. To do so properly requires a multiple comparison procedure, which introduces a complication. The significance level (say 0.05) of the standard t- and F-tests applies only to that particular test, and not to a series of tests. Thus, it would be inappropriate to use the standard t-test to determine if the mean of capi tal class 1 a n d capital class 2 differed from the mean of capital class 6. The standard statistic is further inappropriate if the comparison is sug gested by the data, say comparing the highest and the lowest means. For example, in compar ing the highest and lowest means, with six classes the standard 5 percent test is in fact a 60 percent test (Neter and Wasserman [1974], section 14.2). Table 7 corrects for these problems by using the Tukey method for multiple compari son, which is based on the studentized range dis tribution (see Neter and Wasserman [1974], section 14.3, and SAS/STAT User’s Guide [1990], volume 2, chapter 24). For example, the first line of table 7 compares the mean change in the proportion of total securities for capital class 0 with the same mean for capital class 1. KB TABLE 7 (CONT. Tukey Multiple Comparison Tests for Differences in Means D. C&I Loans Alpha = 0.05, Confidence = 0.95, Degrees of Freedom = 10861, Mean Square Error = 0.002579, Critical Value of Studentized Range = 3-858, C. Total Loans Alpha = 0.05, Confidence = 0.95, Degrees of Freedom = 10861, Mean Square Error = 0.008611, Critical Value of Studentized Range = 3-858. Capital Classes C om pared Simultaneous Lower C onfidence Limit Difference betw een Means Simultaneous U pper C onfidence Limit Capital Classes C om pared Simultaneous Lower Confidence Limit Difference betw een Means Sim ultaneous U ppe r C onfidence Limit 0 0 0 0 4 3 2 1 -0.103050 -0.071166 -0.046760 -0.038886 -0.054225 -0.022252 0.002801 0.012231 -0.0054013 0.026663 0.052363 0.063349 0 0 0 0 4 3 1 2 -0.0656967 -0.0514908 -0.0409112 -0.0398953 -0.0389768 -0.0247219 -0.0129366 -0.0127722 -0.01225693 0.0020471 0.0150380 0.0143508 1 1 1 1 4 3 0 2 -0.082242 -0.050543 -0.063349 -0.027364 -0.066457 -0.034483 -0.012231 -0.009430 -0.050672a -0.0184233 0.038886 0.008504 1 1 1 1 4 3 2 0 -0.0346787 -0.0205744 -0.0096505 -0.0150380 -0.0260402 -0.0117853 0.0001644 0.0129366 -0.01740163 -0.00299613 0.0099792 0.0409112 2 2 2 2 4 3 0 1 -0.066644 -0.035116 -0.052363 -0.008504 -0.057027 -0.025053 -0.002801 0.009430 -0.0474093 -0.0149903 0.046760 0.027364 2 2 2 2 4 3 1 0 -0.0314679 -0.0174567 -0.0099792 -0.0143508 -0.0262045 -0.0119496 -0.0001644 0.0127722 -0.02094113 -0.00644263 0.0096505 0.0398953 3 3 3 3 4 0 2 1 -0.037339 -0.026663 0.014990 0.018423 -0.031974 0.022252 0.025053 0.034483 -0.0266083 0.071166 0.0351 l6 a 0.050543a 3 3 3 3 4 1 2 0 -0.0171911 0.0029961 0.0064426 -0.0020471 -0.0142549 0.0117853 0.0119496 0.0247219 0.01131873 0.02057443 0.01745673 0.0514908 4 4 4 4 3 0 2 1 0.026608 0.005401 0.047409 0.050672 0.031974 0.054225 0.057027 0.066457 0.037339a 0.1030503 0.0666443 0.0822423 4 4 4 4 3 1 2 0 0.0113187 0.0174016 0.0209411 0.0122569 0.0142549 0.0260402 0.0262045 0.0389768 0.01719Ha 0.03467873 0.03146793 0.06569673 a. Significant at the 0.05 percent level. SOURCE: Authors’ calculations. The difference between the means is positive, but the confidence limits include 0, so we can not reject equality of the means. The results in table 7 confirm the significance of the portfolio change. The undercapitalized banks shifted toward securities and away from loans more than did the adequately capitalized and well-capitalized banks. But another possibility is yet unaccounted for. Low-capitalized banks might have different portfolio shifts even without a change in capi tal requirements. For example, suppose a bank has low capital because of takedowns of loan commitments that had been funded by pur chased money. That is, the bank ends up with an unexpectedly high proportion of loans. Over time, the bank might lower its loan level to re store the desired balance between loans and securities. We wish to demonstrate that lowcapital banks do not normally increase their securities holdings in the years following a change in requirements. To provide some evidence on this, we com pare the behavior of banks from 1988 to 1990 with their behavior from 1990 to 1992. Specifi cally, we compare the portfolio changes in low- TABLE most of the differences (even between nega tive and positive terms) are not statistically sig nificant, even at the 10 percent level. 8 ANOVA Comparison of Portfolio Shifts between Periods Asset Components 1990-92 1988-90 Capital Class Cash 0-4% 4-8% 8-10% 10-14% >14% 0.001 -0.015 -0.012 - 0.016 -0.019 0-4% 4-8% 8-10% 10-14% >14% -0.003 -0.003 -0.001 -0.004 -0.014 (0.044) (0.047) (0.035) (0.045) (0.050) -0.003 -0.010 -0.010 -0.013 -0.015 (0.031) (0.047) (0.039) (0.042) (0.050) Government Securities (0.037) (0.048) (0.042) (0.046) (0.065) 0.046 0.032 0.032 0.017 0.001 (0.099) (0.068) (0.057) (0.067) (0.085) C&I Loans 0-4% 4-8% 8-10% 10-14% >14% -0.038 (0.069) -0.017(0.062) -0.019 (0.059) -0.012 (0.051) -0.002 (0.040) -0.047 -0.035 -0.035 -0.023 -0.008 (0.072) (0.060) (0.061) (0.056) (0.046) Mortgages 0-4% 4-8% 8-10% 10-14% >14% 0.013 0.007 0.007 0.005 0.007 (0.065) (0.042) (0.050) (0.043) (0.039) -0.006 0.014 0.013 0.012 0.014 (0.062) (0.067) (0.065) (0.053) (0.051) NOTE: Standard deviations are in parentheses. SOURCE: Authors’ calculations. capital banks from 1988 to 1990 with portfolio changes in all other banks from 1990 to 1992 and with low-capital (as of 1990) banks from 1990 to 1992. By using this method, we control for portfolio shifts due to both macroeconomic effects and low capitalization. Table 8 reports these results. Capital require ments certainly appear to have had an impact. Across each capital class, banks reduced their C&I loans more from 1990 to 1992 than from 1988 to 1990. Low-capital banks even de creased their bond holdings in the earlier pe riod, but raised them in response to capital requirements from 1990 to 1992. A large caveat goes along with this work, however, in that V. Regression Analysis We examine the influence of deterioration in the quality of the loan portfolio on bank portfo lio allocation changes with a regression model that is a simple extension of the ANOVA frame work. The regression equation includes dummy variables for each of the size and capital classes and a measure of the quality of the i th bank’s loan portfolio: A asset ratiof = a + X (3; size dummiesf. + X yk capital dummiesi + 5 loan quality,.. The charge-off ratio (as of March 1990) — the ratio of net charge-offs to assets — is used to measure loan quality. A summary of the regression results for the asset ratio changes between 1990 and 1992 for each category is presented in table 9. The charge-off rate has a significant influence on each asset category. The largest effects of poor loan quality are on the increase in Treasury se curities and on the decrease in real estate loans. In both of these instances, a 0.5 percentagepoint increase in the charge-off ratio (which is about equal to the increase in the aggregate ra tio over the 1980s, as shown in figure 4) re sults in an absolute change in the asset ratio of about 0.01 percentage point. Significant differ ences between size classes and capital classes appear in all but one category. Finally, the re gressions explain only a small proportion of the interbank variation in asset ratios. The bottom part of table 9 shows the esti mated coefficients for the capital dummies. They represent differences from the omitted category: banks with initial risk-adjusted capi tal ratios in excess of 14 percent. The relation ship between the initial capital position and the extent to which the bank increased govern ment securities holdings and reduced loans is still substantial. That is, even with the influ ence of the quality of the loan portfolio held constant, poorly capitalized banks made large portfolio adjustments away from both C&I and real estate loans and toward holdings of gov ernment securities. TABLE 9 VI. Summary of Regression Results Cash assets Total securities Treasuries Other loans Total loans C&I loans Mortgages Other loans Coefficient and t-statistic F-test Probability Charge-Off Ratio Size Capital Dummies Dummies 0.051 (2.1) 4.60 (9.2) 2.78 (7.0) 1.83 (3.7) -5.06 (10.6) -0.86 (3.3) -0.009 (3.4) -0.03 (8.1) 0.3592 0.0499 R2 0.002 0 0 0.028 0 0 0.021 0.0002 0.6557 0.004 0 0 0.068 0 0 0.038 0.635 0 0.163 0 Conclusion The evidence presented here strongly suggests that bank portfolio changes since 1990 are at least in part a response to the introduction of riskbased capital requirements. Qualitatively, at least, the regulations succeeded. Comprehending the changes improves our general understanding of the effects of bank regulation. The particular ef fect of capital requirements on bank portfolios merits special interest. The shift in bank portfo lios can affect their overall risk, and therefore the risk of financial collapse and the liability of the federal government acting as the lender of last re sort. O n the other hand, the reduction in loans may (under the “credit view”) have macroeco nomic consequences and reflect on overall eco nomic growth, income, and unemployment.11 0.002 0.032 Capital Dummy Coefficients (Difference from omitted category — Ratio >14%) 0-4% 4-8% 8-10% 10-14°/ 0.01 0.00 0.00 0.00 Total securities Treasuries Other loans 0.03 0.04 -0.01 0.03 0.02 0.01, 0.02 0.02 0.00 0.02 0.01 0.00 Total loans C&I loans Mortgages Other loans -0.05 -0.04 -0.02 0.01 -0.05 -0.02 -0.02 -0.03 -0.04 -0.02 -0.00 -0.02 -0.03 -0.01 -0.02 -0.01 Cash assets NOTE: Standard deviations are in parentheses. SOURCE: Authors’ calculations. ■ 11 The credit view argues that changes in bank lending— and in credit more generally— have an important effect on the aggregate econ omy above and beyond any effect on the money supply. References Avery, Robert B., and Allen N. Berger. “RiskBased Capital and Deposit Insurance Re form,” Jo u rn a l o f B anking a n d Finance, vol. 15, nos. 4/5 (September 1991), pp. 847-74. Boyd, John, and Mark Gertler. “U.S. Commer cial Banking: Trends, Cycles, and Policy,” National Bureau of Economic Research, Working Paper 4404, July 1993Camell, Richard Scott. “The FDIC Improve ment Act of 1991: Improving Incentives of Depository Institutions’ Owners, Managers, and Regulators,” Ohio State University, working paper, August 1992. Furlong, Frederick T. “Capital Regulation and Bank Lending,” Federal Reserve Bank of San Francisco, Economic Review, 1992 no. 3, pp. 23-33. Jacklin, Charles J. “Bank Capital Requirements and Incentives for Lending,” Federal Re serve Bank of San Francisco, Working Pa per No. 93-07, February 1993Hancock, Diana, and James A. Wilcox. “Bank Capital and Portfolio Composition,” Board of Governors of the Federal Reserve System, working paper, December 1992. Huber, Stephen K. Bank Officer’s H andbook o f Government Regulation, 2d ed., Cumulative Supplement No. 1. Boston: Warren, Gorham, & Lamont, 1991. Keeley, Michael C. “Bank Capital Regulation in the 1980s: Effective or Ineffective?” Federal Reserve Bank of San Francisco, Economic Review, 1988 no. 1, pp. 3-20. Miller, Merton H. “Debt and Taxes,” Jo u rn a l o f Finance, vol. 32, no. 2 (May 1977), pp. 261-75. Neter, John, and William Wasserman. Applied Linear Statistical Models-. Regression, A naly sis o f Variance, a n d Experimental Designs. Homewood, 111.: Richard D. Irwin, 1974. Orgler, Yair E., and Robert A. Taggart, Jr. “Impli cations of Corporate Capital Structure The ory for Banking Institutions,” Jo u rn a l o f Money, Credit, a n d Banking, vol. 15 (May 1983), pp. 212-21. SAS/STAT User’s Guide. Version 6, 4th ed. Cary, N.C.: SAS Institute, Inc., 1990. Saunders, Anthony. M odem F in an cial Institu tions. Homewood, 111.: Richard D. Irwin, 1993 (forthcoming). Searle, Shayle R. Linear Models. Toronto: Wiley, 1971. Stigler, George J. “The Tactics of Economic Re form,” in The Citizen a n d the State: Essays on Regulation. Chicago: University of Chi cago Press, 1975, pp. 23-37. FDICIA’s Emergency Liquidity Provisions by Walker F. Todd Is there any reason why the Am erican people should be taxed to guarantee the debts o f banks, any more than they should be taxed to g u a ra n tee the debts o f other institutions, including the merchants, the industries, a n d the mills o f the country? Senator Carter Glass (1933)1 Introduction The Federal Reserve Banks’ discount window advances to failing depository institutions have become an increasingly controversial issue with in the last 20 years or so. This debate culminated in congressionally mandated limitations on Re serve Banks’ advances to undercapitalized banks in the Federal Deposit Insurance Corpo ration Improvement Act of 1991 (FDICIA), pre viously the subject of a Federal Reserve Bank of Cleveland Economic Commentary.2 In a comparatively little-noticed amendment of the Reserve Banks’ lending authority, FDICIA made potentially significant revisions to the emergency liquidity provisions of the Federal Reserve Act. In particular, the Act now permits all nonbank firms — financial or otherwise (called “nonbanks” here for simplicity) — to Walker F. Todd is an assistant gen eral counsel and research officer at the Federal Reserve Bank of Cleve land. Helpful comments and sug gestions were provided by Melvin Burstein, Joseph Haubrich, Owen Humpage, William Osterberg, Robin Ratliff, Mark Sniderman, James Thomson, and two anony mous referees. borrow at the discount window for emergency purposes under the same collateral terms afforded to banks. Ironically, while the princi pal thrust of FDICIA was to limit or reduce the size and scope of the federal financial safety net, at least as applied to insured depository in stitutions, this provision effectively expanded the safety net. This article describes the histori cal and theoretical backgrounds of the Reserve Banks’ emergency lending authority for non banks and analyzes the changes made by FDICIA that affect that authority. ■ 1 See Smith and Beasley (1972), p. 357. Senator Glass offered these remarks during the Senate debate on the Banking Act of June 20, 1933 (Glass—Steagall Act), which established, among other things, the first plan of federal deposit insurance. ■ 2 See Todd (1992a). FDICIA is Public Law No. 102-242 (Decem ber 19,1991). The provisions of FDICIA principally affecting Reserve Banks’ discount window operations are Sections 131-133 (prompt cor rective action) and Sections 141-142 (least-cost resolutions, systemicrisk exceptions, and lending limitations). On prompt corrective action, see Pike and Thomson (1992); on systemic risk, see Wall (1993); on lending limitations, see Todd (1992a, 1993a). I. Background of Emergency Lending Provisions for Nonbanks Since the creation of the first central banks in Western Europe in the seventeenth century, parliaments have often asked them to rescue enterprises sponsored by the state or sover eign, favored private-sector enterprises, and even, occasionally, the state itself.3 In the United States, Congress understood quite early that it should avoid the expediency of direct funding of the Treasury by borrowings from the central bank.4 This maxim of fiscal propri ety (“central banks should not undertake fiscal activities”) also makes theoretical sense and has been explained as follows regarding the central banks of developing countries: Fiscal activities [such as implementing selective credit policies or recapitalizing insol vent financial institutions] involve expenditures that reduce central bank profits and may even produce losses. If central bank losses are not met from government budget appropriations, they must eventually lead to an expansion in central bank money and the abandonment of any monetary policy goal of price stability.5 Fiscal and monetary authorities in the United States generally followed this view of the divi sion of their responsibilities during peacetime from 1791 until sometime during 1931-33-6 The extension of governmental credit di rectly to nonbank enterprises historically has been a fiscal operation in the United States, not a monetary policy operation of the type or dinarily undertaken by a central bank.7 For ex ample, the original Federal Reserve Act of 1913 provided for the extension of Reserve Banks’ credit directly to member banks, but did not allow for such credit to or for the account of the Treasury, nonmember banks, or nonbanks. Borrowing by member banks was governed by the applicable sections of the Federal Reserve Act (originally, Section 13), and borrowing by other entities simply was not permitted. The Federal Reserve Act was enacted in an era in which peacetime federal budgets regularly were in surplus, and it apparently was intended that the Reserve Banks’ money-creating powers should not be substituted for explicit congres sional decisions on the Treasury’s funding. During the presidential election year of 1932, economic pressures generated by the Great De pression caused President Herbert Hoover to propose changing the previously indirect credit relationship between Reserve Banks and non banks (the Reserve Banks could lend to banks, but only banks could lend to nonbanks) to a more direct one. Although he had vetoed a prior version of the Emergency Relief and Con struction Act that summer because it would have authorized the former Reconstruction Fi nance Corporation (RFC) to make loans di rectly to individuals,8 Hoover allowed Section 13 (3) to be added to the Federal Reserve Act as part of a road construction measure de signed to relieve unemployment. Subject to cer tain restrictions, Section 13 (3) authorized Reserve Banks, “in unusual and exigent circum stances,” to extend credit directly to “individu als, partnerships, and corporations.”9 Section 13 (3) proved to be so restricted that it did not open the floodgates of Reserve Banks’ ■ 6 See Todd (1993b). Beginning in early October 1931, President Hoover proposed that the Reserve Banks expand their lending authority to include the rescue of insolvent banks during peacetime, but the princi pal proposals for use of the Reserve Banks' lending authority for fiscal purposes were not enacted until the early months of the New Deal, after March 4,1933. The most noteworthy of those proposals was the Thomas Amendment to the Agricultural Adjustment Act of May 12,1933, revised on May 27,1933, and on many subsequent occasions, added as Sec tions 14 (b)(3) and 14 (h) of the Federal Reserve Act (expired in 1981). See Moley (1966), pp. 3 0 0 -0 3 ; and Hoover (1952), pp. 395-99. ■ 7 See Todd (1992b) and Martin (1957), pp. 76 8 -6 9. ■ ■ 3 See, for example, Fry (1993), Bordo (1992), Todd (1988), and Humphrey and Keleher (1984). ■ 4 Our Founding Fathers were well aware of the problems created by Treasury borrowings from central banks. Alexander Hamilton, the first Secre tary of the Treasury, recommended, and Congress later passed, a bill provid ing that the First Bank of the United States, our first central bank, should be prohibited from lending more than $50,000 to the Treasury or to any state or foreign prince without the prior, explicit consent of Congress. When the Sec ond Bank of the United States was chartered in 1816, this limit was raised to $500,000. See Hamilton (1967), pp. 31-32,34. ■ 5 See Fry (1993). 8 On the RFC, see generally Todd (1992b), Keeton (1992), and Olson (1988). Strange as it may seem to modern readers, banks’ lending to individuals (as opposed to farmers or business associations) before 1933 was commonly regarded as either a kind of speculation more appro priate for investment bankers than for commercial bankers or a charitable act more appropriate for mutual savings banks or benevolent societies than for commercial banks. For a colorful account of this phenomenon, see Grant (1992), pp. 7 6 -9 5 ,2 6 7 -6 8 . ■ 9 The text of the Emergency Relief and Construction Act of July 21, 1932, Public Law No. 72-302, is found in Federal Reserve Bulletin, vol. 18 (August 1932), pp. 520-27. Section 210 of that Act [Section 13 (3)] is at p. 523. The Board's circular authorizing emergency discounts under Section 13 (3) for six months beginning August 1,1932, is at ibid., pp. 518-20. liquidity to the general public in 1932. At least five members of the Board of Governors (the “Board,” which then included six regularly ap pointed and two ex officio members) had to vote affirmatively to find that “unusual and exi gent circumstances” warranting implementa tion of Section 13 (3) existed. The collateral offered by borrowers had to consist of “real bills” and certain Treasury obligations “of the kinds and maturities made eligible for discount for member banks under other provisions of [the Federal Reserve] Act.”10 In essence, the only acceptable collateral would have been near substitutes for cash. The final statutory re striction required the Reserve Banks to find evi dence that the borrower was unable to “secure adequate credit accommodations from other banking institutions.”11 These restrictions made it unlikely that many nonbanks could qualify for emergency advances from Reserve Banks. In fact, due to these restric tions and the availability of credit elsewhere,12 the Reserve Banks “made loans to only 123 busi ness enterprises [from 1932 until 1936] aggregat ing only about $1.5 million [under Section 13 (3)]. The largest single loan was for $300,000.”13 In 1935, the Board requested, and Congress approved, an amendment of Section 13 (3) in tended to make nonbanks’ borrowing somewhat easier. Despite that statutory change, no such loans actually have been made since the amend ment became effective in 1936.14 Prior to the 1935 amendment, a borrower had to satisfy two relevant conditions: a satisfactory endorsement ■ 10 Real bills, for the purposes discussed here, are “notes, drafts, and bills of exchange arising out of actual commercial transactions,” with remain ing maturities of not more than 90 days [therefore, self-liquidating], “issued or drawn for agricultural, industrial, or commercial purposes, or the pro ceeds of which have been used, or are to be used, for such purposes,” as distinguished from “speculative,” investment, or working-capital pur poses. See Section 13 (2) of the Federal Reserve Act (12 U.S.C. Section 343) and Hackley (1973), pp. 37 and 129. ■ 11 See Hackley (1973), pp. 127-28. Under another provision of the Federal Reserve Act, Section 13 (13)(12 U.S.C. Section 347c), added in 1933, nonbanks may borrow directly from Reserve Banks without a finding of financial emergency (“unusual and exigent circumstances") by the Board, but only on the security of the U.S. government or (since 1968) U.S. government agency obligations. ■ 12 In particular, after 1934, the Federal Reserve was authorized to mount a rival program to extend credit directly to individuals, partnerships, and corporations “for working capital purposes” under former Section 13b of the Federal Reserve Act (expired in 1958). However, the operations of the RFC expanded greatly after 1933 and displaced the direct credit extension role earlier foreseen for the Reserve Banks under Sections 13 (3), 13b, and 13 (13). Regarding former Section 13b, see discussions in Schwartz (1992), pp. 61-62, and Hackley (1973), pp. 133-45. http://fraser.stlouisfed.org/ ■ 13 Hackley (1973), p. 130. Federal Reserve Bank of St. Louis (by either the borrower or a third-party surety) on the borrower’s own note pledged to the Re serve Bank, a n d security (eligible collateral) for the borrower’s discounted note or notes. After the 1935 amendment, either an endorsement or additional security for such notes was required. This change made it easier for a borrower to dis count his own note. After 1935, however, borrowers had a clear choice between the distinct concepts of eligi ble collateral (what security could be pledged to secure the Reserve Bank’s advance) and eli gible purpose (the use to which the Reserve Bank’s advance would be put). That is, non banks could borrow for any purpose as long as they pledged eligible collateral. Failing that, they could borrow on their own notes against any satisfactory collateral, including ineligible collateral, as long as they had eligible purposes for their borrowings. Securities firms, mutual funds, and insurance companies, the greater part of whose asset portfolios included ineligible collateral, could not be said to have eligible purposes for bonowing to fund those particular assets. The payment of an ordinary business firm’s general operating expenses could qualify as an eligible purpose for borrowing from a Reserve Bank, but eligi ble expenses normally included such things as the payment of utility bills, regular taxes, pay roll, and the purchase of raw materials. Activi ties deemed speculative, such as the purchase of a portfolio of common stocks or investment securities generally (other than government se curities), or the financing of permanent fixed investments with instruments maturing in more than 90 days, were ineligible purposes.15 As the principal historian of the subject explained this point, ■ 14 The Board has reactivated Section 13 (3) rarely since the 1930s, but this emergency lending authority has not actually been used since 1936. It was activated for savings and loan associations, mutual savings banks, and nonmember commercial banks in 1966 and 1969 (Hackley [1973], p. 130). Its use also was contemplated for assistance to New York City (said to be a “municipal corporation") in 1975. The potential use of Section 13 (3) for depository institutions became unnecessary when the Monetary Control Act of 1980 added Section 19 (b)(7) to the Federal Reserve Act (12 U.S.C. Section 461) to authorize routine advances of Reserve Banks’ credit to “any depository institution in which transaction accounts or nonpersonal time de posits are held." Such routine advances are secured by any satisfactory as sets (not limited to eligible collateral) and are available at nonpenalty rates, even for nonmember depository institutions. Thus, there has been no need for emergency discounts for those institutions that could be secured only by collateral that was a near substitute for cash. ■ 15 See generally Hackley (1973), pp. 34—38. At p. 129, he dis cusses the use of a borrower's own note under Section 13 (3). [T]he reason why the Reserve Banks were prohibited from extending credit on stocks and bonds [under Section 131 was that the [Re serve] Banks were intended to assist commer cial banking and not investment banking. Paper eligible for discount was confined to self-liquidating paper arising out of commer cial rather than investment transactions.16 While securities firms and other nonbank fi nancial firms could borrow for the eligible pur pose of funding the types of current operating expenses described above, their liabilities for such expenses normally would constitute only a small fraction of their balance sheets. In con trast, their loans to carry customers’ accounts invested in securities (other than government securities) are ineligible purposes but poten tially require much greater funding than the proportion of their assets related to eligible pur poses. It apparently was the intent of Congress to remove these ineligible collateral/ineligible purpose restrictions on nonbanks’ borrowings from Reserve Banks that underlay the 1991 amendment of Section 13 (3). II. Amendments of Section 13 (3) in FDICIA Section 13 (3) has been discussed very little since the 1930s, so it might seem unusual to find Section 473, amending Section 13 (3), in serted in the final stages of the congressional deliberations on FDICIA in November 1991. Increasingly, however, since the stock market crash of October 1987, some policymakers had been discussing the potential use of the Re serve Banks’ discount windows to relieve non bank financial firms’ liquidity crises directly. Procedurally, there were enough obstacles to such use of the discount window to discour age financial firms from relying on Section 13 (3) to rescue them in a liquidity crisis: The pro cedural starting point always was an emer gency declaration approved by at least five members of the Board. Also, the practical ob stacles appeared insurmountable: For borrow ings secured by eligible collateral, nonbank financial firms typically held comparatively few unpledged assets that would qualify, and bor- ■ 16 Hackley (1973), p. 38. Depository institutions may, however, obtain extensions of Reserve Bank credit under Section 10B (12 U.S.C. Section 347b) even on ineligible stock or bond collateral (“any satisfac tory assets"), but the amounts available might be limited under Section 11 (m)(12 U.S.C. Section 248 [ml), added in 1916. rowings said to be for eligible purposes typi cally would be quite limited. Another issue that was not, but probably should have been, raised explicitly during con gressional deliberations on FDICIA was that any consideration of altering the Reserve Banks' col lateral or purpose of borrowing standards to ac commodate nonbanks’ asset portfolios under Section 13 (3) clearly would shift a portion of the risk of loss previously borne by the nonbanks’ creditors onto the Reserve Banks and, thus, indi rectly onto the taxpayer.17 One of the poten tially troublesome aspects of the FDICIA amendment of Section 13 (3) is that it appears to reflect a motive or spirit that contradicts that of the FDICIA provisions intended both to limit Reserve Banks’ loans to undercapitalized depository institutions and to make it more dif ficult for the Federal Reserve to treat an institu tion as too big to fail. If the amendment was intended to provide a vehicle for possible Fed eral Reserve treatment of a failing securities firm as too big to fail, then it arguably consti tutes a contradictory extension of the same fed eral safety net that was retrenched in other parts of FDICIA and apparently enlarges the moral hazard problem of deposit insurance. O f the issues just identified regarding the amendment of Section 13 (3), only restrictions based on the types of collateral that nonbank borrowers could offer were discussed explicitly during the congressional deliberations on FDICIA in 1991. It appears that, having satisfied itself that the risks from expanding the collateral limits were minimal and that it might prove helpful to provide the Reserve Banks with this additional, liquidity-maximizing policy tool for a financial emergency, Congress adopted the revisions of Section 13 (3) as Section 473 of FDICIA without extensive discussion or debate, leaving a rather sketchy legislative history for this statute. How ever, by altering the collateral standards explic itly, FDICIA implicitly rendered Section 13 (3)’s purpose of borrowing restrictions largely super fluous because the prior standards for eligible purposes were binding only on nonbanks that could not pledge eligible collateral. ■ 17 The Reserve Banks’ operations create an indirect gain or loss for the taxpayer because the operating profits are rebated to the Treasury as a miscellaneous receipt offsetting part of the federal government’s op erating expenses. In fiscal year 1992, those receipts were $27.1 billion, of which the Reserve Banks contributed $22.9 billion (Council of Economic Advisers [1993], p. 437). Losses incurred on Reserve Banks’ operations would reduce those receipts. While material losses for Reserve Banks have been rare since World War II, they are not inconceivable for central banks that attempt to subsidize fiscal operations on their balance sheets. See Fry (1993). The actual statutory language change made by Section 473 of FDICIA was comparatively minor. The restrictive phrase in quotation marks below was deleted from the part of Section 13 (3) that described the collateral acceptable for emergency discounts for nonbanks. Prior to the change, a Federal Reserve Bank could discount for any in dividual, partnership, or corporation any notes, drafts, and bills of exchange when these instru ments were endorsed or otherwise secured to the satisfaction of the Reserve Bank and, when en dorsed, were “of the kinds and maturities made eli gible for discount for member banks under other provisions of this Act ....’’ It generally was under stood that this reference was primarily to the types of financial instruments meeting the eligible pur pose standards as illustrated in Section 13 (2), but also included instruments described in other parts of Sections 13 and 14 of the Federal Reserve Act. Since FDICIA, Reserve Banks’ emergency ad vances to nonbanks may be based on the types of collateral acceptable for depository in stitutions under an entirely different provision of the Federal Reserve Act, Section 10B, which permits “advances ... secured to the satisfaction of ... [the] Federal Reserve Bank,” or “any satis factory assets.”18 Because nonbanks’ emer gency borrowings need not be secured by eligible collateral, eligibility of purpose of bor rowing has become moot. The only collateral test remaining under revised Section 13 (3) is “satisfactory security,” the same test that ap plies to borrowings by depository institutions under Section 10B. III. Analysis of Potential Ram ifications The changes made by FDICIA expanded emer gency discount window access for nonbanks of all types, not merely securities firms, because any satisfactory assets (not just marketable securities, for example) may be pledged to secure the bor rower’s own note. Whether these changes will have practical consequences is an open question. After all, Section 13 (3) is an emergency lending provision that has been and presumably will con tinue to be invoked very rarely and that requires the affirmative vote of five Federal Reserve Board governors. It is important to keep in mind that nonbanks’ behavior depends in part on howr they expect the Federal Reserve to manage its emer gency lending powers. ■ 18 See Hackley (1973), pp. 109-12, and Eccles (1951), pp. 171-73. The few, scattered public statements regarding congressional intent with respect to Section 473 of FDICIA do indicate that the intended benefi ciaries were securities firms, and no other type of nonbank was mentioned explicitly.19 Although a brief reference was made during the FDICIA delib erations to the absence of any discounts under Sec tion 13 (3) since 1936, the potentially increased taxpayer risk from alteration of the collateral and purpose standards was not discussed.20 How could a new element of taxpayer risk arise? One possible source is derived from the m oral hazard aspects of the increased availa bility of Reserve Banks’ loans to nonbanks dur ing financial emergencies. Nonbanks lacking eligible collateral or eligible purposes for bor rowing must manage their affairs and conduct their relations with creditors and clients so as to be able to survive financial market emergen cies. Now, with increased potential for assis tance during emergencies, nonbanks’ managers might have less incentive to avoid recourse to the Federal Reserve. Although nonbanks still have strong incentives to run their firms pru dently, their managers now have potential ac cess to another funding source during financial crises. Whether this potential access alters non banks’ business decisions — so as to make their calling upon that funding source more likely — remains to be seen. More troubling, however, are the macro im plications of these incentive changes. The ex tension of the federal financial safety net to nonbanks may increase the probability of mar ket liquidity crises that appear to require Fed eral Reserve emergency lending. This could happen during periods of market stress if the costs of risky investment and funding strategies are not fully borne by the managers and share holders of nonbank firms, but instead are per ceived as being partially or fully underwritten ■ 19 During the floor debate in the Senate on the version of FDICIA that was enacted, Senator Christopher Dodd of Connecticut spoke as fol lows in support of the bill: It [FDICIA] also includes a provision I offered to give the Federal Re serve greater flexibility to respond in instances in which the overall financial system threatens to collapse. My provision allows the Fed more power to provide liquidity, by enabling it to make fully secured loans to securities firms in instances similar to the 1987 stock mar ket crash. See Congressional Record (1991), p. S18619. For similar legal interpre tations of Section 473 of FDICIA, see FDICIA (1992), pp. 37 and 92. See also Holland (1991). ■ 20 See U.S. Senate Report No. 102-167 (October 1,1991), pp. 2 0 2 -0 3. by U.S. taxpayers.21 Self-correcting market forces that help to insulate financial markets from macroeconomic shocks could be eroded by what nonbanks regard as implicit taxpayer guarantees of nonbank losses and, thereby, in crease the probability that a real-sector shock would become translated into a financial crisis. A certain amount of adverse selection also might compound the Federal Reserve’s difficul ties: It becomes increasingly likely that bettercapitalized firms would remain outside the Reserve Banks’ lending net (in order to avoid the perceived stigma of borrowing). It also is likely that only the worst-capitalized firms could not raise adequate funds during financial market emergencies. The other main source of taxpayer risk from the revision of Section 13 (3) is derived from the accounting principles that would be used in evaluating the collateral offered for emergency loans. Nonbanks’ previously ineligible assets, including corporate equity securities and mort gages on real estate in the case of securities firms and institutional investors, tend to be illiq uid under the market emergency conditions that would conceivably give rise to the Board's authorization of Section 13 (3) loans. In an emer gency, whatever market value satisfactory (but formerly ineligible) assets that nonbanks already had could undergo severe downward market pressures, triggering wide gaps between par and market collateral valuations. Although all dis count window advances are expected to be ex tended against collateral that is thought to be both sound and ample, there is reason to be con cerned about accurate valuation of nonbanks’ as sets in periods of intense financial distress. The expansion of the collateral limits for Re serve Banks’ extensions of credit under Section 13 (3) might appear to be somewhat at odds with the principal thrust of the other discount window provisions of FDICIA, Sections 141 and 142, which, together with the prompt corrective action provisions, Sections 131-33, were in tended to reduce taxpayers’ potential risk of loss due to loans to insured banks. The lending criteria applicable to undercapitalized depository institutions were tightened, and more exacting and publicly accountable procedures for such lending decisions were established. In Section 141 of FDICIA, provision for a “systemic risk” ex ception to normal supervisory intervention and closing requirements was limited to circumstances ■ 21 Comparable perverse incentives tor insured depository institu tions’ behavior are described in the deposit insurance literature. See Barth and Brumbaugh (1992), pp. 7-12; National Commission (1993), pp. 62-68; and Kane (1989), pp. 95-114. http://fraser.stlouisfed.org/ Federal Reserve Bank of St. Louis in which both two-thirds of the Board and two-thirds of the FDIC’s Board of Directors ap proved the exception, with the further concur rence of the Secretary of the Treasury, after consultation with the President.22 The clear objective of that provision was limiting the tax payer’s potential exposure to loss through in creased procedural hurdles that had to be over come to invoke the exception. IV. Conclusion The removal of the collateral barriers for Re serve Banks’ extensions of credit under Section 13 (3) seems to conflict with the spirit of the other discount window provisions of FDICIA, Sections 141 and 142. These provisions, along with the Act’s prompt corrective action provi sions, Sections 131-133, were intended to less en taxpayers' potential exposure to loss resulting from loans to insured banks. In contrast, Section 473, by removing the eli gible collateral threshold, may have marginally increased taxpayers’ potential risk of loss. This risk could arise from the moral hazard associated with the perceived availability of the equivalent of a federal guarantee for nonbanks. Conse quently, increased access to the discount window by nonbanks carries with it some of the same kinds of risks that arose during the savings and loan debacle: Adverse selection and misaligned agency incentives could increase, together with the probability of use of the emergency lending facility and the implicit underwriting of nonbank losses by taxpayers. The increased degree of discount window access for nonbanks was not accompanied by some of the safeguards normally applicable to discount window access, such as annual exami nations by the federal bank supervisory authori ties, maintenance of required reserves and clear ing balances at Reserve Banks, and requirements to meet minimum regulatory capital adequacy standards. Moreover, by extending a component of the federal safety net, the Reserve Banks’ dis count windows, to nonbanks without limitations on too-big-to-fail rescues, Section 473 of FDICIA contradicts the spirit of the limitations on the toobig-to-fail doctrine enacted for depository institu tions in FDICIA. ■ 22 See Todd (1992a). Systemic risk, as described in Section 141 of FDICIA, is a condition in which the closing of an insured institution, without redemption of uninsured claims at par, “would have serious ad verse effects on economic conditions or financial stability.” The connec tion between systemic risk for banks and for securities firms is made strikingly and explicitly in Wall (1993), p. 10. Finally, it is unclear that there was a real (as opposed to a perceived) need for revision of Section 13 (3). Section 473 of FDICIA appar ently was intended to deal primarily with situ ations like the aftermath of the stock market crash of October 19, 1987, in which securities firms, mutual funds, and other nonbank hold ers of large investment portfolios consisting of ineligible collateral would have found it help ful to obtain credit from Reserve Banks instead of from banks, insurance companies, invest ment banks, and other usual providers of funds to nonbank financial firms. Normally, financial markets treat eligible col lateral as high-quality instruments that are close substitutes for cash. Firms holding large, unpledged amounts of such collateral ordinar ily could be expected to be able to obtain suffi cient extensions of credit without having recourse to direct loans from Reserve Banks, even during market conditions approximating financial emergencies, as long as financial mar kets had adequate supplies of liquidity that the Federal Reserve could ensure through openmarket operations. In fact, aggressive use of open-market operations in October 1987 pro vided sufficient aggregate liquidity to prevent the stock market crash from generating sub stantive harm to the economy. The changes effected by Section 473 of FDICIA should prove quite harmless if the stat ute is implemented in a straightforward, riskaverse manner. However, perverse incentives, continued observance of a too-big-to-fail doc trine (in this case, for nonbanks), and the ab sence of adequate procedural safeguards could increase Reserve Banks’ and, ultimately, taxpay ers’ losses from Section 13 (3) lending activities in the future. Furthermore, greater potential ac cess to the federal financial safety net could boost the risk-taking incentives for nonbanks, thereby increasing the probabilities that they will request discount window lending during financial emergencies. References Barth, James R., and R. Dan Brumbaugh, J r . “De pository Institution Failures and Failure Costs: The Role of Moral-Hazard and Agency Prob lems,” in Peter Dickson, ed., Rebuilding Public Confidence Through Financial Reform. Ohio State University, College of Business, Confer ence Proceedings, June 25, 1992, pp. 3-19Bordo, Michael D. “The Lender of Last Resort: Some Insights from History,” in George G. Kaufman, ed., Research in F in an cial Serv ices: Private a n d Public Policy, vol. 4. Greenwich, Conn.: JAI Press, Inc., 1992, pp. 1-20. Congressional Record, vol. 137, no. 37 (March 5, 1991), 102nd Congress, 1st Session. Council of Economic Advisers. Economic Re port to the President, 1993• Washington, D.C.: U.S. Government Printing Office, 1993. Eccles, MarrinerS. Beckoning Frontiers: Public a n d Personal Recollections, Sidney Hyman, ed. New York: Alfred A. Knopf, 1951. FDICIA. Legal symposium publication, The Fed eral Deposit Insurance Corporation Improve ment Act o f 1991, William P. Bowden, Jr., et al., eds. Englewood Cliffs, N.J.: Prentice Hall Law & Business, 1992. Fry, Maxwell J. “The Fiscal Abuse of Central Banks.” International Monetary Fund, Work ing Paper No. 93-58, July 1993Grant, James. Money o f the M ind: Borrowing a n d Lending in America from the Civil W ar to M ichael Milken. New York: Farrar Straus Giroux, 1992. Hackley, Howard H. Lending Functions o f the Federal Reserve Banks: A History. Washing ton, D.C.: Board of Governors of the Fed eral Reserve System, 1973. Hamilton, Alexander. “Report on a National Bank (December 13, 1790),” in M. St. Clair Clarke and D.A. Hall, eds., Legislative a n d Docum entary History o f the B ank o f the United States (1832). New York: Augustus M. Kelley, 1967 (reprint), pp. 15-35. Holland, Kelley. “Limits on Fed’s Discount Loans Prompt Fears,” Am erican Banker, Decem ber 31, 1991, p. 1. Smith, Rixey, and Norman Beasley. Carter Glass: A Biography (1939). New York: Da Capo Press, 1972 (reprint). Todd, Walker F. “Lessons of the Past and Pros pects for the Future in Lender of Last Resort Theory,” Proceedings o f a Conference on Bank Structure a n d Competition, Federal Reserve Bank of Chicago, May 11-13, 1988, pp. 533-77. Hoover, Herbert. The Great Depression: 19291941 (vol. 3 of Hoover’s memoirs). New York: Macmillan Co., 1952. ________ . “FDICIA’s Discount Window Provi sions,” Federal Reserve Bank of Cleveland, Economic Commentary, December 15, 1992a. Humphrey, Thomas M., and Robert E. Keleher. “The Lender of Last Resort: A Historical Per spective,” Cato Journal, vol. 4, no. 1 (Spring/ Summer 1984), pp. 275-318. ________ . “History of and Rationales for the Re construction Finance Corporation,” Federal Reserve Bank of Cleveland, Economic Re view, 1992b Quarter 4, pp. 22-35. Kane, Edward J. The S&L Insurance Mess: How D id It Happen? Washington, D.C.: Urban In stitute Press, 1989. ________ . “New Discount W indow Policy Is Important Element of FDICIA,” B anking Pol icy Report, vol. 12, no. 5 (March 1, 1993a), pp. 1, 11-17. Keeton, William R. “The Reconstruction Finance Corporation: Would It Work Today?” Federal Reserve Bank of Kansas City, Economic Re view, First Quarter 1992, pp. 33-54. Martin, W illiam McChesney, Jr. “Problem of Small Business Financing,” Federal Reserve Bulletin, vol. 43, no. 7 (July 1957), pp. 767-69. Moley, Raymond. The First New Deal. New York: Harcourt, Brace & World, Inc., 1966. National Commission on Financial Institution Re form, Recovery, and Enforcement. Origins an d Causes o f the S&L Debacle: A Blueprintfo r Reform, A Report to the President and Con gress of the United States. Washington, D.C.: U.S. Government Printing Office, July 1993Olson, James S. Saving Capitalism: The Recon struction Finance Corporation a n d the New Deal, 1933-1940. Princeton, N.J.: Princeton University Press, 1988. Pike, Christopher J., and James B. Thomson. “FDICIA’s Prompt Corrective Action Provi sions,” Federal Reserve Bank of Cleveland, Economic Commentary, September 1, 1992. Schwartz, AnnaJ. “The Misuse of the Fed’s Dis count W indow ,” Federal Reserve Bank of St. Louis, Review, vol. 74, no. 5 (September/ October 1992), pp. 58-69. ________ . “The Federal Reserve Board before Marriner Eccles (1931-1934).” Paper pre sented at Western Economic Association In ternational Conference, South Lake Tahoe, Nev., June 23, 1993b. United States Senate Report No. 102-167. “Com prehensive Deposit Insurance Reform and Taxpayer Protection Act of 1991,” Report of the Committee on Banking, Housing, and Ur ban Affairs, U.S. Senate, to accompany S. 543 (October 1, 1991), Calendar No. 245. 102nd Congress, 1st Session. Washington, D.C.: U.S. Government Printing Office. Wall, Larry D. “Too-Big-to-Fail after FDICIA,” Federal Reserve Bank of Atlanta, Economic Review, vol. 78, no. 1 (January/February 1993), pp. 1-14. Efficiency and Technical Progress in Check Processing by Paul W. Bauer paui w gauer ¡s an economist at the Federal Reserve Bank ot Cleve land. For their comments, sugges tions, and encouragement, the author would like to thank Randall Eberts and Mark Sniderman. Introduction By lowering the transaction costs associated with barter, a payments system greatly facilitates the exchange of goods and services.1 Although vastly improved over the years, the process of transferring funds remains costly, and the evolu tion of the payments system has been at least partially determined by efforts to trim these costs further.2 Increasing the productivity of the pay ments system improves economic welfare both by releasing resources to other sectors of the economy and by lowering the effective purchase price of goods and services. In addition to its roles as the nation’s central bank and as the primary federal regulator of state member banks and bank holding companies, the Federal Reserve System is also a major pro vider of payment services. Ordered by the Fed eral Reserve Act of 1913 to ensure the efficiency of the payments system, the central bank has ■ 1 The payments system refers to such activities as the provision of currency and coin, processing and clearing of checks, providing for set tlement of checks and other types of payments, and wire transfers of funds. See Board of Governors of the Federal Reserve System (1984). ■ 2 See Garbade and Silber (1979), Niehans (1971), and Brunner http://fraser.stlouisfed.org/ and Meltzer (1971). Federal Reserve Bank of St. Louis directly participated in the market since its in ception. Initially, it provided a national mech anism for clearing and settling checks — two major components of payment services — and instituted regulations that eliminated the incen tive for the circuitous routing of checks.3 Prior to passage of the Depository Institu tions Deregulation and Monetary Control Act (MCA) of 1980, the Federal Reserve did not charge fees for its payment services and pro vided them only to member banks. Conse quently, it faced little competition from private providers serving nonmember financial institu tions. Starting in 1981, the MCA required the Federal Reserve to make its services available to all depository institutions and to charge fees that would recover its costs. The goal was to foster a more efficient payments system by giv ing private providers of payment services the opportunity to compete. Given this new competitive environment, it became even more important for the Federal Reserve to be able to track the performance of its various offices. Over the years, an extensive accounting system has been developed to iden tify costs associated with each of its services. ■ 3 See Garbade and Silber (1979) and Humphrey (1980). This has allowed unit cost performance meas ures (total service costs divided by service vol ume) to be calculated for each service offered. This article examines the costs of providing check-processing services at 47 Federal Re serve offices (District Banks, branch offices, and regional check-processing centers) from 1983:IQ to 1990:IVQ by estimating a multiprod uct cost function using an econometric frontier approach. After briefly discussing the advantages and disadvantages of the Federal Reserve’s unit cost measures, I demonstrate how they can be de composed into separate effects related to differ ences in cost efficiency, output mix, input prices, and environmental variables (these control for various site-specific characteristics) using estimates derived from the cost function.4 The cost-function approach provides much more complete informa tion about the sources of office perfonnance than do unit cost measures, but it is more difficult and time-consuming to calculate. In order to explore how the cost frontier may have shifted over time in response to technologi cal and regulatory changes, the article also pre sents estimates of technical progress, as measured by whether the cost of producing a given level of output declines over time. This technique pro vides valuable insights into the technological con straints faced by the Federal Reserve. It should be remembered, however, that re search such as this is a continuing process and that a more complete understanding of the pro duction and cost efficiencies associated with check processing will require multiple investiga tions. Consequently, the numerical estimates pre sented here must be interpreted with caution, understood in the context of stated caveats, and viewed as only a partial effort to model one as pect of the payments system. Section I describes the central bank’s provision of check-processing services and summarizes some previous studies of the payments system. Section II then discusses how the econometric frontier approach is used to estimate the multi product cost function, and explains how a unit cost measure of performance can be decom posed into its various components. After describ ing the data employed in the study, I analyze ■ 4 Output mix includes the effects of scale economies, whether aver age cost rises or falls as output expands, and the effects of the relative production of the various outputs. Cost efficiency determines how closely firms operate to the cost frontier. ■ 5 Under same-day settlement, banks will have access to funds on the same day they are deposited, as long as the checks are presented be fore 8:00 a.m.- Electronic check truncation refers to sending only an elec tronic image of the check, rather than the check itself, through the http://fraser.stlouisfed.org/ settlement process. Federal Reserve Bank of St. Louis estimates of cost efficiency, scale economies, and technical change. Unit costs are decomposed for each office using the estimated multiproduct cost frontier. The final section considers the future of Federal Reserve check processing in light of new technologies, such as same-day settlement and electronic check truncation.'’ I. Background Description of Check Processing Check processing is, in some ways, a fairly straightforward operation: A payor writes a check to a payee, who deposits it at his bank or other depository institution. This is all most of us ever think about, and if the payor and the payee are customers of the same bank (which occurs about 30 percent of the time), this is almost the end of the story. For these “on-us” items, the only step left is for the bank to debit the payor’s account and credit the payee’s account. But if both parties have accounts at different banks, then the payee’s bank must forward the check to the payor’s bank — a situation that occurs roughly 45 billion times a year. For these items, a bank can send checks . directly to the payor’s institution or route them in directly through a local clearinghouse, a corre spondent institution, or a Federal Reserve office. The Fed processes about 35 percent of these interbank checks. In the relatively rare event that the check is re turned for insufficient funds (less than 1 percent of checks), the process repeats itself, only in re verse. The return process is more labor intensive and costly. In contrast to forward volumes, the Federal Reserve handles the vast majority of pay ment system return items. This lack of privatesector competition suggests that the Fed’s prices for handling returned checks may be too low, a subject discussed in more detail below. Thus, the central bank provides two types of check-processing services, forward items and return items, and has a separate price schedule for each. Although the end result is the same for all checks, this description fails to reveal the myriad products offered by a typical processing center. Items can be differentiated by the location of the payor bank, the times of presentment and settlement, and the amount of presorting performed by the institution sub mitting the checks. Costs can vary significantly as a result of these product characteristics. Fine-sort items, for instance, are fully presorted by the submitting m FI GURE 1 Check-Processing Volumes Index, 1981:IIQ = 100 180 140 100 60 20 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 SOURCE: Author’s calculations. While not changing the physical process of check clearing in the United States, the MCA altered the institutional environment profoundly. Fed eral Reserve payment services prior to passage were available at no charge, but only to mem ber banks. The MCA required the Fed to begin charging for its services and to offer them to all depository institutions, including those that are not members of the System.6 Based on guide lines established by the Board of Governors, prices for each payment service are designed to re cover direct and indirect costs as well as a markup (known as the Private Sector Adjustment Factor [PSAF]) that imputes other costs typically incuned in the private sector. In check process ing, each Federal Reserve District offers a slightly different mix of products, and District Banks have some flexibility in pricing. Although the MCA increased the number of institutions that could employ the Federal Re serve’s payment services, a large drop in volume was expected because fees were imposed on previously “free” services. When pricing was implemented, the Fed’s share of interbank check processing fell from approximately 45 percent in 1981 to 38 percent in 1982. Cunently, the System processes about 35 percent of all interbank items. The drop in volume that immediately fol lowed pricing can be seen in figure 1. In the first year, systemwide and Fourth District proc essing volumes plunged 15 and 18 percent, re spectively. However, not all Fourth District offices experienced similar declines: In Pitts burgh, check volume dived almost 40 percent and has grown relatively little since, yet in Co lumbus, check volume recovered within the first year and expanded rapidly thereafter.7 Even with this overall drop in Federal Re serve volume following the onset of pricing, the national allocation of resources improved because banks that already owned their own reader-sorters frequently found it less expen sive to process more of their own checks and even to offer the service to others. Pricing boosted the efficiency of resource allocation in ■ ■ institution and use only the Federal Reserve’s transportation, settlement, and adjustment serv ices, meaning that they cost very little to handle. At the other extreme, an item can be submitted without any presorting during the peak period of check processing (in Cleveland, from 10:00 p.m. to 1:00 a.m., but this varies significantly across offices), when the check reader-sorters are oper ated at close to maximum capacity. The incre mental costs of these items are much higher. Sorting checks and forwarding them to payor institutions (or returning them to depositing insti tutions) involves a variety of resources, or inputs. Transit (transportation and communication) is re quired to get the items to the processing site and on to their final destination once sorted. At the processing site, which must meet certain security standards, labor employs a variety of capital goods (mainly high-speed sorters and computers) to sort the checks and keep track of the settle ment operation. Monetary Control Act 6 These firms include savings and loans, credit unions, and foreign banks. 7 Unusual local economic factors accounted for much of the check volume growth in Columbus. FI GURE 2 GLS Estimates of Ray Average Cost Cents per u n it o f aggregate outp ut a. Millions of items per quarter. SOURCE: Author’s calculations. another way. Humphrey (1981) estimates scale economies in check processing at 36 Federal Reserve Banks and branches for the 1974-76 period. He concludes that 78 percent of checks deposited with the Fed during that time were processed at offices with significant scale dis economies, a finding he attributes to a lack of market competition. The increase in competition after pricing be gan led to greater cost-control incentives and im proved resource allocation. By 1982, constant (rather than decreasing) returns to scale were the mle in Federal Reserve check-processing opera tions (see Humphrey [1980,19851 ). II. Frontier Estimation and Unit Cost Decomposition The cost function, C (y,w,z), for a firm simply yields the minimum cost of producing any specified level of outputs (y) given technologi cal constraints, input prices (w), and environ mental effects (z). Foreshadowing our results somewhat, figure 2 plots the generalized least squares (GLS) estimates of the ray average cost function for check processing using the Sys tem’s averages for output mix, input prices, and site-specific characteristics.8 The curve in dicates the lowest ray average .cost that can be achieved for a given level of output, provided the site is operated efficiently.9 The concept of a frontier is quite natural in the context of a cost function. Even allowing for random events that may lead to temporar ily lower or higher unit costs, we would expect most offices to operate on or above the cost function. In the context of this theoretical con struct, there are many ways for things to go wrong and only one way to get them exactly right. Thus, observed costs will tend to be above the corresponding ray average cost curve. The cost function is a particularly useful con cept because many characteristics of the techno logical constraints facing the firm can be derived from it. For example, from figure 2 we can see that for low levels of output, check-processing services face scale economies— that is, ray aver age costs fall as both outputs are increased pro portionally. For the average mix of outputs, the advantages of running a larger operation are al most exhausted after about 105 million aggregate items per quarter. Increasing the level of output from about 76 million items per quarter (the mean value from 1983 to 1990) to the level re quired for scale efficiency (holding the output mix constant) lowers ray average costs only 2.6 percent. Once these levels of output are reached, we will see that cost efficiency (the ratio of the cost on the frontier to observed cost) becomes a more important consideration. ■ 8 Ray average cost is defined as C (ky, w, M z) / X = C (y, w, z) / £ y t . /= 1 Although the denominator appears to be arbitrarily summing over the various outputs, since the output mix is held constant for this calculation, the rather arbitrary output aggregator function imposes no additional re strictions. ■ 9 Holding the mix of outputs constant by increasing them proportion ally is extremely restrictive. In the results section, I demonstrate how the scale-efficient level of output depends crucially on the mix of outputs. Given this demonstration of the usefulness of cost functions, there is one problem— these functions must be estimated from data generated by sites in operation. A number of empirical techniques have been developed to estimate frontier cost functions. Generally, they can be divided into two classes: 1) estimators based on econometric techniques, such as maximum likeli hood estimation and panel data estimation, or 2) estimators based on linear programming tech niques, such as data envelopment analysis.10 In this paper, I report only estimates derived from the GLS approach.11 Econometric Techniques Broadly speaking, econometric techniques em ploy a specific (although flexible), functional form for the cost function and impose some ad ditional assumptions about the statistical prop erties of the inefficiency terms. As a category, these techniques assume a compound error term that comprises both cost inefficiencies and statistical noise. Within the category, the tech niques differ in the assumptions used to de compose this error term to obtain estimates of cost efficiency. All of the econometric techniques impose an explicit functional form for the cost function. The translog functional form is employed be cause it is a second-order approximation to any cost function about a point of approximation (here, the sample mean). Essentially, this means that it can model many different possible relation ships among outputs, inputs, and environmental factors, depending on its parameter values. The translog cost function can be written as M (1) ln C (, = p0+ £ $ J n y nut L m= 1 1990 + X ^ j ° j t + U i + Vit’ j = 1984 where y is a vector of M outputs, w is a vector of K input prices, z is a vector of L environ mental variables, D is a set of T -1 dummy variables (one for every year except the first), u ( u > 0 ) measures cost inefficiency, and v represents statistical noise. Estimation of this function involves finding the values of the parameters that best fit the observed data given the imposed assumptions. Equation (1) is estimated, along with the corre sponding equations for input shares, imposing the usual mathematical restrictions of symme try and linear homogeneity in input prices. The symmetry constraints come from assuming that the cost function is twice differentiable, so that d Wjd Wj diV jdw i ' and d 2C _ dy^yj d 2C d y ^ y ,' This forces 8 k[ = 8 lk and (3 kl = (3 lk, for every k and /. Linear homogeneity in input prices, t ■C(y, w) = t ■w), stems from defining the cost function as yielding the minimum cost of producing a given output level when faced with a particular set of input prices. Propor tional changes in input prices affect only the cost level, not the cost-minimizing input bun dle. This property imposes constraints on all parameters related to the \nwkit's-. m=l M +1/2X m= (3 ) M X I Y „= 1 k PmMVnuM’/U and 1 1=1 K I + X i k lnwkit k= 1 M 9 - = I k S t( = 0 , V / , m . k K +X I Qmklnym:t[nWkit m= 1 k=1 K + 1 /2 X k=l K X 1=1 b k !l n W kUl n W IU ■ 10 A more detailed description of the techniques employed in this paper can be found in Bauer and Hancock (1993). For a thorough treat ment of these two classes of techniques, see Greene (1993) and Ali and Seiford (1993). ■ 11 Bauer and Hancock (1993) report estimates using a variety of econometric and linear programming techniques. Here, I choose to con centrate on one set of results in order to provide a sharper focus. 29 The use of longitudinal data often allows us to avoid assuming a specific distribution for the inefficiency terms. Repeated observations over time identified site-specific, time-invariant inefficiencies.12 For the GLS technique, the inefficiency terms are calculated by using the average of the residu als by site, a ( . The most efficient site in the sam ple is taken to be the best estimate of where the cost frontier lies and is thus assumed to be fully efficient. The inefficiency of the i th site is meas ured by the proportionate increase in predicted costs over the predicted costs of the most effi cient site. An index bounded by zero (costs are inclined, but no output is produced) and one (a site on the cost frontier) can be calculated as r/\ W A exp (min a (5) In [ ( C / y u ) / ( C / y , ) ] T = 1/ I n [C(yit,wit, z it) t£ t= i exp (Uj+ v it)/y ut) T N - 1 / 77 V ] T ] T I n [C(yit,wit,zit) T= 1 i = l exp (u i + vi() / y li(] . Equation (5) can be rearranged to (6) In [(C / y () /( C / y ) ] = { ut- u 1 A - a / ). m= 1 The GLS technique mns an iterative, seemingly unrelated regression (ITSUR) on the system of cost and K -1 input share equations using panel data. One of the share equations, which are de rived using Shephard’s lemma, must be dropped in order to avoid singularity of the system.13 How ever, since the estimates are obtained using ITSUR, the numerical estimates are the same no matter which one is dropped. M M + 1 / 2 XX P m / lnyu[nymi n j / ) m= 1 1=1 + Inyu - In V, X yk(lnw k i~ [nwk) k= i Unit Cost Decomposition For the moment, assume that only one output is produced. In this case, unit cost is just C/y, where C is observed cost and y is observed output. If w e wanted to compare one site to the average of all sites, we could do so by tak ing the ratio of that site’s unit costs to the over all average. This would readily tell us whether a site’s costs were above or below average, but we would not know why. Using the definition of the cost function and the error specification developed for the GLS estimation technique, we can rewrite the ratio of a site’s average unit costs to the overall aver age unit costs as follows in order to derive a more informative set of measures.14 12 See Schmidt and Sickles (1984) for further explanation. Berger (1993) contains some possible extensions. K K + 1/2 X k=i M X dk/(lnw kilnw n - ln w klnw ,) 1=1 K +1 X X emk([nyn,:lnu;ki- ln^ m= 1 k= 1 ln ^ ) m —1 where the expressions in braces can be defined as effects resulting from differing cost efficiencies, outputs, input prices, the interaction of outputs and input prices, and environmental effects.15 ■ ■ 13 For more details on the treatment of the share equations, see Bauer, Ferrier, and Lovell (1987). ■ 14 Although any two observations could be chosen to compare http://fraser.stlouisfed.org/ unit costs, comparing the sample mean for the ithsite to the overall sam Federal Reserve Bank of St. Louis ple mean causes the term involving statistical noise, v, to drop out. ■ 15 I derive the decomposition for the general case when there are M outputs and arbitrarily use the first output (forward items) as the denominator in the construction of unit costs. Empirically, the resulting measure of unit cost is highly correlated with the Federal Reserve's measure because forward proc essing appears to account for more than 80 percent of the costs of processing services, but has the advantage that this specification can be exactly decom posed into the various effects described below. While these are logarithmic differences, as long as the numerical values are close to zero, they can be roughly interpreted as the percentage difference in costs stemming from these vari ous effects.16 Clearly, unit costs provide a useful measure of a site’s relative ability to produce a given level of output at the lowest possible cost, be cause it summarizes the overall effect of a variety of cost factors. Once the trouble and expense of collecting the data have been incurred, the unit cost measures are easy to calculate. O n the other hand, the cost-function approach imposes greater structure and requires more effort to calculate, but it also provides a much more de tailed set of information. Now one must explicitly consider the com plications posed by the presence of multiple outputs. The Federal Reserve constructs unit cost measures for each of its services and then weights them by cost shares to obtain an over all measure of performance across service lines. A potential problem is that the accounting rules employed to allocate the costs of joint inputs (those used to produce more than one service, like computer systems) may not accurately re flect the flow of services from these inputs to the various services. This will cause the calcu lated unit cost measures to be biased up or down, depending on whether the service in question receives more or less of its share of costs associated with the joint inputs. In the case of some joint inputs, there may be no simple accounting mle that could accurately al locate their costs because of nonlinear techno logical relationships among the various outputs and inputs. Rather than relying on arbitrary accounting rules, the cost-function approach allows the data (combined with the imposed assumptions) to allocate costs to the various outputs by find ing the parameters that best fit the cost model. Marginal costs for each of the outputs can then be readily calculated by differentiating the esti mated cost function. For pricing and output decisions, marginal costs should be more rele vant than unit costs. III. Data Construction Quarterly data for the 1983-90 period were col lected on total costs, check volume, input prices, and environmental variables for 47 Federal Re serve check-processing sites.17 The primary data source was annual functional cost accounting re ports, which are prepared by the Federal Re serve via its Planning and Control System to monitor costs and improve resource allocation within the System. These data were supple mented by other cost and revenue figures, infor mation from occasional Federal Reserve surveys, price index data from the Commerce Depart ment’s Bureau of Economic Analysis and the La bor Department’s Bureau of Labor Statistics, and pricing data from industry sources. Production costs for forward items, return items, and adjustments were included in total costs, but certain overhead expenses, such as special District projects, were excluded. The two measures of output were the total number of forward items and return items processed at each site. Reflecting the earlier discussion of the vast array of products offered by the vari ous offices, this measure is at best an approxi mation. Some of the environmental variables discussed below attempt to adjust for the differ ent product mixes across offices. Inputs to the check-processing function fall into the catego ries of buildings, materials, transit, and labor. Labor expenditures— salaries, retirement, and other benefits— accounted for 47.1 percent of total costs in 1990:IVQ. Buildings’ total cost share was only 5.6 per cent in 1990:IVQ, in part because the interest expenses associated with the acquisition of build ings are not represented in the cost-accounting framework (these are included in the PSAF rather than in direct and indirect costs). Expenditures for materials (office equip ment and supplies, printing and duplicating, data processing, computers, and check readersorters) accounted for 29.8 percent of total costs in 1990:IVQ. Transit expenditures— the expenses associated with data and other com munications, shipping, and travel— made up just over 17.5 percent. Environmental variables, which control for a variety of site-specific characteristics, include the item-pass ratio, the number of endpoints, the machine error rate, and the type of machine used. The item-pass ratio, defined as the aver age number of times a check must pass through a reader-sorter, is a measure of the exogenous check-sort pattern and has been found in pre vious studies to influence costs significantly. The number of endpoints is the number of locations ■ 16 For the exact percentage difference, one must take the antilog minus one. ■ 17 For complete details, see Bauer and Hancock (1993). The New York check-processing operation was omitted because it was closed in 1988. m FI GURE 3 M1 Locus Millions of return items per quarter 4 .0 3 .5 3 .0 2 .5 2.0 1 .5 1.0 0 .5 0.0 0 20 40 60 80 100 120 140 160 180 200 220 Millions of forward items per quarter SOURCE: Author’s calculations. to which checks must be sorted and delivered. The machine enor rate is the number of incom ing enors per 100,000 checks at each office and is largely a matter of poor MICR (magnetic ink character recognition) encoding. The last environmental variable indicates whether the site used IBM or Unisys machines and allows for differences in maintenance expenses, fail ure rates, and downtime. IV. Empirical Results Scale Efficiency A scale-efficient office operates at the output level at which ray average costs are minimized for its output mix or, equivalently, at the out put level at which cost elasticity equals one (that is, a 1 percent increase in output would cause costs to rise by 1 percent).18 Conversely, a scale-inefficient office operates at an output level larger or smaller than the scale-efficient level. If the office processes less than the scale-efficient volume, the cost elasticity is less than one (a 1 percent increase in output would raise costs by less than that amount), meaning that the office could achieve lower unit costs by boosting out put. Alternatively, a scale-inefficient office that processes more than the scale-efficient volume has a cost elasticity greater than one, and unit costs can be lowered by reducing output. Thus, ■ 18 Cost elasticity is defined as 9 lnC(ky, w,z)/dlnXlx = 1 . This turns out to be identical to the sum of the cost elasticities with re spect to each output. estimates of cost elasticities yield direct esti mates of scale efficiency. When multiple outputs are produced, the mix of outputs must also be considered when exam ining scale efficiency. The M locus (see figure 3) is defined as the set of all outputs with unitary cost elasticities.19 For any level of forward items, the M locus reveals the conesponding level of return items required to achieve scale efficiency. A site operating below (above) the M locus ex periences scale economies (diseconomies). The estimated M locus indicates that a site process ing a large number of return items relative to for ward items reaches scale efficiency at a lower level of forward items. It may not be possible for every office to achieve scale efficiency, despite the best efforts of managers. The volume of checks and return items processed at an office depends on the size of the market and the prices charged. The eco nomic size of managers’ payments markets is out side their control, and although managers may have some authority over prices, their need to re cover costs may prevent them from setting a price low enough to attract a scale-efficient vol ume of output. In short, even the best-run office will be scale inefficient if it is in a market too small to achieve scale efficiency. Figure 2 demonstrated that the ray average cost curve for check and return processing was U-shaped (meaning that at low levels of aggre gate output, ray average cost falls as outputs are increased proportionally, but that scale econo■ 19 The M locus in figure 3 is drawn with the input prices equal to their values at the sample mean. Unfortunately, the estimated M locus be comes increasingly speculative as it moves away from the output ratio found at the sample mean. mies are exhausted at some point, so further in creases result in higher ray average costs). In table 1, we present estimates of cost elasticities using site-specific characteristics for 1990JVQ. Most sites are fairly close to achieving scale effi ciency— given their individual output mixes, in put prices, and environmental variables. Even so, these estimates suggest that the average of fice could lower its costs about 12 percent if it could generate scale-efficient volumes. Full scale efficiency would require the average office to increase its scale of operations significantly. However, as revealed by the ray average cost function in figure 2, most of the gains occur be fore 76 million items per quarter are processed. Although some smaller offices appear to be operating in the output range where further scale economies could be exploited in the future, addi tional float costs that are not incorporated in our model may make this infeasible. As items from more-distant banks are processed, additional shipping costs and delays will be incuned that may outweigh the associated cost savings. Estimates of marginal cost, or the incremental cost of processing one more item, can provide additional information for pricing. One of the beneficial outcomes of competitive markets is that competition forces prices to be set equal to marginal costs. In other words, the price that con sumers pay for a good or service equals a firm’s incremental cost of producing it. If the Federal Reserve set its actual prices for forward and re turn items in 1990:IVQ to equal estimated mar ginal costs, those prices would have averaged $0,009 and $0,643 per item, respectively (see ta ble 1). In practice, prices are based on account ing data, and the Federal Reserve’s calculated unit costs for forward and return items averaged $0.0135 and $0,159, respectively. Neither unit costs nor marginal costs can be directly used for pricing because they fail to ac count for several characteristics (such as the time the checks are submitted for processing), yet to the extent that pricing is based on unit costs, the estimated marginal costs imply that forward items could be regarded as overpriced, whereas return items could be regarded as underpriced. Even though the Federal Reserve sets prices to re cover costs, econometric estimates indicate that the accounting data appear to assign too much of the costs to forward items and too little to re turn items. Market conditions are consistent with econometric estimates, since there are no entry barriers into either market, yet the Federal Re serve faces little competition for return items, and there are many private-sector competitors for forward items. Clearly, this is an issue that requires further study. Cost Efficiency Cost inefficiencies appear to raise costs more than does scale inefficiency. If all offices could be operated on the cost frontier, costs could be lowered by about 23.5 percent. Table 2 compares GLS estimates of the cost efficiencies calculated using the multiproduct cost function employed here with the single-product esti mates reported in Bauer and Hancock (1993). Overall, the results change remarkably little when return items are treated as a separate out put: O n average, estimated cost efficiency rises only 3 5 percent. However, one site with a rela tively large number of return items (FR27) saw its estimated efficiency increase by 16.3 per centage points. Many of the top-ranked offices were in the same Federal Reserve District, indicating that management differences may be important. Aside from superior managerial skill, estimated cost efficiency could vary across sites because some Districts may focus on cost performance while others may stress customer service, which is largely uncontrolled for in this study. For ex ample, one District may specify precisely how checks must be submitted and refuse to accept them otherwise, while another may accept checks in any form but charge higher fees for packages that require more attention. The for mer District will appear to be more efficient than the latter, other things being equal, because it re ceives the checks exactly as it wants them. How ever, the latter District receives a higher fee by providing a service desired by its customers. Unit Cost Decomposition The average unit cost measure for each of the 47 offices over the 1983-90 period relative to the overall sample mean is presented in table 3, along with estimates of each of the component effects. Unit costs vary substantially across of fices, from -0.388 to 0.309, or from about a third below to a third above the overall average. The largest single component appears to be the costefficiency effect, with a conelation between it and unit cost of more than 80 percent. TABLE 1 Estimates of Marginal Costs and Cost Elasticities, 1990MVQ Cost Elasticities Office FR1 FR2 FR3 FR4 FR5 FR6 FR7 FR8 FR9 FR10 FR11 FR12 FR13 FR14 FR15 FR16 FR17 FR18 FR19 FR20 FR21 FR22 FR23 FR24 FR25 FR26 FR27 FR28 FR29 FR30 FR31 FR32 FR33 FR34 FR35 FR36 FR37 FR38 FR39 FR40 FR41 FR42 FR43 FR44 FR45 FR46 FR47 Average Marginal Costs Return Items Overall Forward Items Return Items 0.381 0.400 0.416 0.392 0.464 0.483 0.405 0.359 0.444 0.418 0.469 0.450 0.391 0.382 0.412 0.429 0.396 0.533 0.489 0.439 0.436 0.381 0.407 0.439 • 0.509 0.492 0.375 0.493 0.502 0.393 0.452 0.371 0.427 0.413 0.453 0.473 0.366 0.485 0.473 0.415 0.480 0.448 0.403 0.442 0.450 0.421 0.414 0.568 0.568 0.439 0.565 0.333 0.240 0.501 0.670 0.436 0.443 0.424 0.388 0.590 0.629 0.506 0.421 0.526 0.219 0.162 0.556 0.495 0.548 0.519 0.500 0.198 0.303 0.708 0.402 0.362 0.519 0.130 0.528 0.400 0.470 0.430 0.341 0.604 0.382 0.312 0.463 0.378 0.467 0.576 0.445 0.491 0.378 0.447 0.949 0.968 0.855 0.958 0.797 0.723 0.906 1.029 0.880 0.860 0.893 0.838 0.982 1.011 0.918 0.850 0.922 0.752 0.651 0.996 0.932 0.929 0.926 0.939 0.706 0.795 1.083 0.896 0.864 0.912 0.582 0.899 0.827 0.882 0.883 0.814 0.971 0.867 0.784 0.878 0.859 0.915 0.979 0.887 0.941 0.799 0.861 0.007 0.010 0.007 0.011 0.010 0.009 0.007 0.009 0.009 0.009 0.012 0.007 0.010 0.011 0.009 0.008 0.008 0.015 0.010 0.011 0.009 0.006 0.012 0.012 0.006 0.008 0.011 0.011 0.013 0.005 0.006 0.009 0.006 0.006 0.010 0.010 0.010 0.013 0.008 0.009 0.010 0.009 0.009 0.007 0.013 0.006 0.009 0.738 0.797 0.655 1.011 0.618 0.520 0.657 0.865 0.584 0.817 0.525 0.486 0.907 0.902 0.700 0.631 0.767 0.433 0.561 0.466 0.473 0.662 1.077 0.673 0.296 0.331 0.653 0.346 0.389 0.531 0.601 1.258 0.570 0.531 0.547 0.534 1.131 0.496 0.480 0.805 0.457 0.481 0.631 0.475 0.601 0.671 0.895 0.433 0.446 0.880 0.009 0.643 Forward Items SOURCE: Author’s calculations. The output and input price effects can also exert a significant influence on some offices’ unit costs, but the correlations with unit costs are much lower. In fact, the correlation be tween unit costs and the input price effect is negative, hinting that some input quality may vary across sites and that higher-priced inputs may be more productive. By construction, in put prices and estimates of cost efficiency are uncorrelated, so in this case, the unit cost measures have revealed an issue that requires further study. The environmental effects tend to be mini mal across all sites except FR25, which serves an unusually small number of endpoints. The interactive effect is slight for all offices, with the largest estimated effect shifting relative unit costs only about 6.8 percent. Productivity Growth Including year dummies in the cost function al lows us to estimate whether it shifts down (or up) over time as a result of changes in technol ogy or in the regulatory environment. Estimates of a technical change index are presented in figure 4. For 1983, the index equals 100; for later years, it rises or falls depending on the be havior of the estimated cost function. As of 1990, costs had risen about 8.7 percent. Most of the upward shift that occurred in 1989 appears to have stemmed from transitory costs related to the implementation of regulations designed to post checks more quickly, since costs fell sharply in 1990. Measured productivity growth in check proc essing has been anemic for two main reasons: 1) some of the cost savings have been plowed back into producing a higher-quality product (such as expedited funds availability), and 2) even though prices for computer equipment and other office machinery have fallen precipi tously over the last 10 years, the price of high speed check reader-sorters has remained roughly unchanged in real terms. Apparently, the limit of how quickly paper checks can be read and sorted has nearly been reached, and further advances will have to await the in creased use of electronics in collecting checks. Bal TABLE 2 GLS Cost Efficiency Estimates, 1983-90 Average Single Product Multiproduct Office GLS Rank GLS Rank FR1 FR2 FR3 FR4 FR5 FR6 FR7 FR8 FR9 FRIO FR11 FR12 FR13 FR14 FR15 FR16 FR17 FR18 FR19 FR20 FR21 FR22 FR23 FR24 FR25 FR26 FR27 FR28 FR29 FR30 FR31 FR32 FR33 FR34 FR35 FR36 FR37 FR38 FR39 FR40 FR41 FR42 FR43 FR44 FR45 FR46 FR47 0.864 0.535 0.997 0.587 0.625 0.669 0.634 0.633 0.656 0.629 0.581 0.765 0.717 0.713 0.715 0.714 0.754 0.647 0.645 0.707 0.683 0.919 0.530 0.612 0.802 0.880 0.627 0.738 0.615 0.969 0.693 0.630 0.939 1.000 0.711 0.660 0.574 0.610 0.792 0.557 0.747 0.742 0.668 0.843 0.610 0.630 0.689 7 46 2 42 37 25 31 32 28 35 43 11 16 19 17 18 12 29 30 21 24 5 47 39 9 6 36 15 38 3 22 34 4 1 20 27 44 40 10 45 13 14 26 8 41 33 23 0.884 0.604 0.998 0.615 0.708 0.687 0.689 0.696 0.673 0.632 0.668 0.861 0.736 0.718 0.737 0.721 0.728 0.645 0.679 0.770 0.694 0.928 0.585 0.661 0.837 0.927 0.790 0.798 0.665 0.961 0.720 0.636 0.939 1.000 0.756 0.685 0.656 0.644 0.815 0.567 0.820 0.721 0.745 0.914 0.659 0.685 0.638 8 45 2 44 25 29 28 26 33 43 34 9 19 24 18 22 20 39 32 15 27 5 46 36 10 6 14 13 35 3 23 42 4 1 16 31 38 40 12 47 11 21 17 7 37 30 41 Average 0.708 SOURCE: Author’s calculations. 0.742 Change in Efficiency 0.020 0.069 0.002 0.028 0.083 0.018 0.056 0.063 0.017 0.003 0.086 0.095 0.019 0.005 0.022 0.007 -0.026 -0.002 0.034 0.063 0.011 0.009 0.055 0.048 0.035 0.047 0.163 0.060 0.050 -0.008 0.027 0.006 0.000 0.000 0.045 0.025 0.082 0.034 0.023 0.011 0.074 -0.021 0.078 0.071 0.049 0.055 -0.051 0.035 Chang in Ran 1 -1 0 2 -12 4 -3 -6 5 8 -9 -2 3 5 1 4 8 10 2 -6 3 0 -1 -3 1 0 -22 -2 -3 0 1 8 0 0 -4 4 -6 0 2 2 -2 7 -9 -1 -4 -3 18 Unit Cost Decomposition3 Logarithm ic Differences from Sample Means, 1983-90 Office Unit Cost ($) FR1 FR2 FR3 FR4 FR5 FR6 FR7 FR8 FR9 FRIO FR11 FR12 FR13 FR14 FR15 FR16 FR17 FR18 FR19 FR20 FR21 FR22 FR23 FR24 FR25 FR26 FR27 FR28 FR29 FR30 FR31 FR32 FR33 FR34 FR35 FR36 FR37 FR38 FR39 FR40 FR41 FR42 FR43 FR44 FR45 FR46 FR47 0.015 0.021 0.013 0.023 0.018 0.018 0.016 0.022 0.019 0.020 0.020 0.014 0.018 0.018 0.017 0.015 0.018 0.024 0.021 0.018 0.016 0.012 0.022 0.022 0.014 0.016 0.020 0.020 0.024 0.012 0.019 0.020 0.014 0.013 0.018 0.021 0.021 0.021 0.016 0.019 0.017 0.017 0.019 0.013 0.023 0.016 0.017 U nit Cost -0.179 0.176 -0.349 0.242 0.022 0.028 -0.087 0.212 0.047 0.133 0.131 -0.221 0.023 0.029 -0.036 -0.145 -0.017 0.304 0.182 -0.007 -0.112 -0.376 0.193 0.219 -0.217 -0.124 0.120 0.134 0.309 -0.388 0.042 0.122 -0.269 -0.352 0.025 0.155 0.162 0.141 -0.086 0.038 -0.051 -0.036 0.046 -0.288 0.259 -0.104 -0.049 a. Office mean relative to overall sample mean. SOURCE: Author’s calculations. Cost Efficiency -0.185 0.195 -0.307 0.178 0.037 0.067 0.064 0.054 0.087 0.151 0.096 -0.159 -0.002 0.022 -0.003 0.019 0.009 0.130 0.078 -0.047 0.057 -0.234 0.227 0.106 -0.130 -0.232 -0.073 -0.083 0.100 -0.269 0.019 0.145 -0.245 -0.308 -0.029 0.070 0.113 0.132 -0.103 0.258 -0.111 0.018 -0.014 -0.218 0.109 0.069 0.142 Total O u tp u t -0.054 0.001 -0.110 -0.163 -0.136 0.010 -0.133 -0.122 0.018 -0.119 0.174 -0.038 -0.148 0.053 -0.125 -0.085 -0.155 0.278 0.054 0.149 0.005 -0.191 -0.081 0.188 0.134 0.149 0.357 0.356 0.375 -0.193 -0.027 -0.288 -0.165 -0.179 0.117 0.105 -0.093 0.029 0.017 -0.126 0.102 0.187 0.152 0.001 0.133 -0.252 -0.162 Direct In p u t Price 0.017 0.061 0.038 0.225 0.213 0.012 0.068 0.211 -0.046 0.086 -0.044 0.083 0.223 -0.167 -0.029 -0.084 0.073 -0.159 0.076 -0.183 -0.156 0.078 0.113 -0.167 0.404 -0.136 -0.233 -0.270 -0.275 0.065 0.024 0.219 0.101 0.124 -0.189 -0.151 0.034 0.076 -0.004 0.031 -0.058 -0.338 -0.150 -0.005 -0.130 0.217 0.101 Interactive Effect -0.006 0.003 0.005 0.008 -0.022 -0.006 0.000 0.019 0.006 0.002 0.004 -0.002 0.001 -0.023 0.001 0.000 0.000 0.030 -0.029 0.004 0.002 0.002 -0.001 -0.002 -0.068 0.020 -0.047 0.020 0.028 0.002 -0.003 0.012 0.003 0.001 0.008 0.016 0.001 -0.004 0.004 0.001 0.008 0.019 -0.018 0.003 0.002 -0.007 0.003 Environm ental Effect 0.048 -0.084 0.024 -0.005 -0.070 -0.054 -0.086 0.050 -0.018 0.013 -0.099 -0.106 -0.051 0.143 0.119 0.005 0.055 0.026 0.002 0.070 -0.020 -0.031 -0.066 0.094 -0.556 0.075 0.117 0.111 0.080 0.008 0.029 0.034 0.038 0.010 0.119 0.115 0.107 -0.092 0.001 -0.126 0.008 0.077 0.076 -0.069 0.144 -0.132 -0.133 fia FIGURE 4 Technical Change Index Index, 1983 = 100 140 120 100 80 60 40 20 0 1983 1984 1985 1986 1987 1988 1989 1990 SOURCE: Author’s calculations. V. Conclusions and Prospects for the Future This study finds scale economies to be sufficiently large to enable most offices to proportionally in crease their forward and return volumes, yet still lower their ray average costs by roughly 12 per cent. Costs appear to increase much more rapidly as more return items are processed than as more forward items are processed. Although there ap pear to be opportunities for most offices to im prove their performance by further exploiting scale economies, costs could be lowered even more (up to 23-5 percent overall) if all offices could operate closer to the cost frontier. It is necessary to keep in mind three impor tant caveats. First, some offices may be located in areas where it may not be possible to expand output enough to achieve scale efficiency. Sec ond, the cost-efficiency measure is relative to the most efficient office observed in the sample. Third, although I use the concise term ’‘cost effi ciency,” the concept is more fully described as “once factors included in the cost function are controlled for, there remain unexplained cost dif ferences across processing sites.” Every effort has been made to control for the factors that affect the costs of check-processing offices, but no one can hope to account for every factor that might significantly affect costs. Future research will ex tend the analysis by trying to control for product quality in a more detailed way.20 The multiproduct cost-efficiency estimates for the 47 offices covered here are highly conelated with earlier single-product estimates presented in Bauer and Hancock (1993). O n average, cost effi ciency rose only 3.8 percentage points when returns were treated as a separate output. How ever, one office that processed an atypically high level of returns had its cost-efficiency in dex increase by 16.3 percentage points. The overall level of cost efficiency is roughly the same as that found for private financial institu tions, using similar estimation techniques.21 In the single-product setting, unit cost meas ures provide an easily calculated overall indicator of relative office performance. Unfortunately, they do not reveal the sources of superior or inferior performance. In the multiproduct set ting, unit cost measures could be biased if the costs of joint inputs are misallocated across services. The cost-function approach over comes both of these drawbacks, but requires imposing a number of potentially restrictive as sumptions. The decomposition of unit costs re veals that, for this sample, cost efficiency tends to be the largest single component, but consid erable office-specific variation results from the other components. Only the interaction effect between output levels and input prices is con sistently small in magnitude for all offices. ■ 20 Product quality can affect cost efficiency measures because it is expensive to provide higher quality. If output is not adjusted for product quality, sites providing lower-quality output will, other things being equal, appear to be more cost efficient. ■ 21 For example, see Bauer, Berger, and Humphrey (1993), Ferrier and Lovell (1990), Fried, Lovell, and Vanden Eeckaut (1993), and Mester (1993), to name just a few. While these studies examine the cost efficiency of producing outputs other than check-processing services, their estimated effi ciency levels suggest that the Federal Reserve pursues its behavioral goal about as well as private financial institutions pursue theirs. The net effect of technological and regulatory changes seems to have shifted the multiproduct cost frontier up slightly over time, a finding that supports the prevailing view that much greater use of new technologies, such as check trunca tion and imaging, will be required to achieve sig nificant technical change in check processing. This finding is also consistent with earlier work by Bauer and Hancock (1993). In the coming years, check processing at the Federal Reserve will face a number of new challenges, since volume is likely to rise less rapidly, and may even fall. One cause is merg ers and acquisitions in the financial service sec tor, which have resulted in more on-us items that can be cleared internally. Other causes in clude bilateral agreements among banks to swap checks directly, the emergence of private nation wide check processors, same-day settlement, and technological advances such as electronic check presentment and the shift to electronic payments. The introduction of pricing, the evolution of technology, and the consolidation of the banking industry during the past few years have led to many changes in the check-processing market. Moreover, increased competition between bank ers and nonbank providers of financial services, along with more competition between checks and other payment media, indicates that more changes will follow. In the future, market forces will largely determine the number and location of check-processing sites across the country. Re search studies can contribute to a more complete understanding of developments in this dynamic payment service. References Ali, A., and L. Seiford. “The Mathematical Pro gramming Approach to Efficiency Analysis,” in H.O. Fried, C.A.K. Lovell, and S.S. Schmidt, eds., The Measurement o f Produc tive Efficiency: Techniques a n d Applications. Oxford: Oxford University Press, 1993Bauer, P. W., A. N. Berger, and D. B. Hum phrey. “Efficiency and Productivity Growth in U.S. Banking,” in H.O. Fried, C.A.K. Lovell, and S.S. Schmidt, eds., The Measure ment o f Productive Efficiency: Techniques arid Applications. Oxford: Oxford University Press, 1993Bauer, P. W., G. Ferrier, and C. A. K. Lovell. “A Technique for Estimating a Cost System That Allows for Inefficiency,” Federal Re serve Bank of Cleveland, Working Paper 8704, May 1987. Bauer, P. W., and D. Hancock. “The Efficiency of the Federal Reserve in Providing Check Proc essing Services,” Journal o f Banking a n d Fi nance, vol. 17 (April 1993), pp. 287-311. Berger, A. N. “Distribution-Free Estimates of Ef ficiency in the U.S. Banking Industry and Tests of the Standard Distributional Assump tions,” Jo u rn a l o f Productivity Analysis, vol. 4, no. 3 (September 1993), pp. 261-92. Board of Governors of the Federal Reserve Sys tem. The Federal Reserve System: Purposes a n d Functions. Washington, D.C.: Board of Governors, 1984. Brunner, Karl, and Allan H. Meltzer. “The Uses of Money: Money in the Theory of an Ex change Economy,” Am erican Ecoomic Re view, vol. 61, no. 5 (December 1971), pp. 784-805. Ferrier, Gary D., and C. A. K. Lovell. “Measuring Cost Efficiency in Banking: Econometric and Linear Programming Evidence,” Jo u rn a l o f Econometrics, vol. 46, nos. 1/2 (October/ November 1990), pp. 229-45. Fried, H. O., C. A. K. Lovell, and P. Vanden Eeckaut. “Evaluating the Performance of U.S. Credit Unions,” Jo u rn a l o f Banking a n d Finance, vol. 17 (April 1993), pp. 251-65. Garbade, Kenneth D., and William L. Silber. “The Payment System and Domestic Exchange Rates: Technological versus Institutional Change,” Journal o fMonetary Economics, vol. 5, no. 1 (January 1979), pp. 1-22. Greene, W. H. “The Econometric Approach to Efficiency Analysis,” in H.O. Fried, C.A.K. Lovell, and S.S. Schmidt, eds., The Measure ment o f Productive Efficiency: Techniques a n d Applications. Oxford: Oxford University Press, 1993. Humphrey, David B. “Costs, Scale Economies, Competition, and Product Mix in the U.S. Payments Mechanism,” Board of Governors of the Federal Reserve System, Research Pa pers in Banking and Financial Economics, No. 37, May 1980. ________ . “Economies to Scale in Federal Re serve Check Processing Operations,” Jo u r n a l o f Econometrics, vol. 15, no. 1 (January 1981), pp. 155-73. ________ . “Resource Use in Federal Reserve Check and ACH Operations after Pricing,” Jo u rn a l o f B ank Research, vol. 16, no. 1 (Spring 1985), pp. 45-53Mester, L. J. “Efficiency in the Savings and Loan Industry,” Jo u rn a l o f B anking a n d Fi nance, vol. 17 (April 1993), pp. 267-86. Niehans, Jiirg. “Money and Barter in General Equilibrium with Transaction Costs,” Am eri can Economic Review, vol. 6l, no. 5 (De cember 1971), pp. 773-83. Schmidt, P., and R. Sickles. “Production Fron tiers and Panel Data,” Jo u rn a l o f Business a n d Economic Statistics, vol. 2 (1984), pp. 367-74. Third Quarter Working Papers Current Working Papers of the Cleveland Federal Reserve Bank are listed in each quarterly issue of the Economic Review. Copies of specific papers may be re quested by completing and mailing the attached form below. Single copies of individual pa pers will be sent free of charge to those who request them. A mailing list service for personal subscribers, however, is not available. ■ 9305 Generational Accounting in Norway: Is the Nation Overconsuming Its Petroleum Wealth? ■ 9306 The Evolving Legal Framework for Financial Services Institutional subscribers, such as libraries and other organiza tions, will be placed on a mail ing list upon request and will automatically receive Working Papers as they are published. by Walker F. Todd by Alan J. Auerbach, Jagadeesh Gokhale, Laurence J. Kotlikoff, and Erling Steigum, Jr. Please complete and detach the form below and mail to: Research Department Federal Reserve Bank of Cleveland P.O. Box 6387 Cleveland, Ohio 44101 Check item(s) requested Please send the following Working Paper(s): □ 9305 □ 9306 Send to: Please print Name Address City State Zip Economic Review ■ 1992 Quarter 3 H Comparing Central Banks’ Rulebooks by E.J. Stevens Generational Accounts and Lifetime Tax Rates, 1900-1991 by Alan J. Auerbach, Jagadeesh Gokhale, and Laurence J. Kotlikoff Forbearance, Subordinated Debt, and the Cost of Capital for Insured Depository Institutions by William P. Osterberg and James B. Thomson Has the Long-Run Velocity of M2 Shifted? Evidence from the P* Model by Jeffrey J. Hallman and Richard G. Anderson An Introduction to the International Implications of U.S. Fiscal Policy by Owen F. Humpage ■ 1993 Quarter 1 Examining the Microfoundations of Market Incentives for Asset-Backed Lending by Charles T. Carlstrom and Katherine A. Samolyk 1992 Quarter 4 White- and Blue-Collar Jobs in the Recent Recession and Recovery: W ho’s Singing the Blues? by Erica L. Groshen and Donald R. Williams Assessing the Impact of Income Tax, Social Security Tax, and Health Care Spending Cuts on U.S. Saving Rates by Alan J. Auerbach, Jagadeesh Gokhale, and Laurence J. Kotlikoff History of and Rationales for the Reconstruction Finance Corporation by Walker F. Todd H 1993 Quarter 2 Using Bracket Creep to Raise Revenue: A Bad Idea Whose Time Has Passed by David Altig and Charles T. Carlstrom . Cyclical Movements of the Labor Input and Its Implicit Real Wage by Finn E. Kydland and Edward C. Prescott Money and Interest Rates under a Reserves Operating Target by Robert B. Avery and Myron L. Kwast