The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Economic Review Ä r ^ ^ ^ W M u w ^ ^ I FEDERAL RESERVE BANK OF ATLANTA NOVEMBER 1982 4 BANK SIZE, BANK COSTS COST STRUCTURE A ^ RISK AND MARKET SHARE CHANGES A LARGE BANK ENTRY A PROFITS OR MARKETSHARE? A PAYMENTS SYSTEMS COSTS Economic Review FEDERAL RESERVE BANK O F ATLANTA President: William F. Ford Sr. Vice President and Director of Research: Donald L Koch Vice President and Associate Director of Research: William N. Cox Financial Structure: B. Frank King, Research Officer David D. Whitehead Larry D. Wall National Economics: Robert E. Keleher, Research Officer Stephen O. Morrell Mary S. Rosenbaum Regional Economics: Gene D. Sullivan, Research Officer Charlie Carter William J. Kahley Database Management: Delores W. Steinhauser Payments Research • Veronica M. Bennett Paul F. Metzker Visiting Scholars: James R. Barth George Washington University James T. Bennett George Mason University George J. Benston University of Rochester Gerald P. Dwyer Emory University Robert A. Eisenbeis University of North Carolina John Hekman University of North Carolina Paul M. Horvitz University of Houston Peter Merrill Peter Merrill Associates Communications Officer: Donald E Bedwell Public Information Representative: Duane Kline Editing: Gary W. Tapp Graphics: Susan F. Taylor Eddie W. Lee, Jr. The E c o n o m i c Review seeks to inform the public a b o u t Federal Reserve policies a n d t h e economic environment and, in particular, to narrow the gap b e t w e e n specialists a n d c o n c e r n e d laymen. Views expressed in the Economic Review aren't necessarily t h o s e of this Bank or the Federal Reserve System. Material may be reprinted or abstracted if the Review a n d author are credited. Please provide the Bank's Research Department with a copy of any publication containing reprinted material. Free subscriptions and additional copies are available from the Information Center, Federal Reserve Bank of Atlanta, P.O. Box 1731, Atlanta, G a 3 0 3 0 1 (404/586-8788). Also contact the Information Center t o receive Southeastern E c o n o m i c Insight a free newsletter on economic trends published by t h e Atlanta Fed twice a month. 2 N O V E M B E R 1982, E C O N O M I C REVIEW ••••••••• Economies of Scale in Banking: An Overview 4 Operating Costs in Commercial Banking 6 Do operating costs shrink as a bank increases in size? Are unit banks more expensive to operate than branch banks? Does holding company affiliation affect costs? Some surprising results from a major study. Economies of Scale: A Case Study of the Florida Savings and Loan Industry 22 Bank Size and Risk: A Note on the Evidence 32 A new study examines the cost structure of Florida S&Ls. The evidence suggests some differences in the cost structures of banks and S&Ls. Is there a relationship between a bank's size and its degree of risk? A review of the evidence. Changes in Large Banks' Market Share The Impact of Local Market Entry by Large Bank Holding Companies 41 35 If there are economies of scale in banking, larger banks would be expected to capture larger shares of local markets. A new Atlanta Fed study compares larger banks' performance with smaller banks in the same local markets. An Alternative View of Bank Competition: Profit or Share Maximization What happens to market share, profits, and risk when a large bank holding company enters a local market? Do holding company subsidiaries have a competitive advantage? 48 Future Payments System Technology: Can Small Banks Compete? 58 One theory holds that small banks seek to gain market share, while large banks aim for profits. An Atlanta Fed study tested the theory in southeastern markets. Will the explosion in payments system technology favor larger banks and S&Ls? Not necessarily, especially if smaller institutions are able to share technology and services through networks Statistical Supplement V O L U M E LXVII, N O . 11 68 Economies of Scale i Private institutions and public regulators alike have been erasing boundaries between financial institutions. Some40,000 depository institutions are now able to offer a full range of consumer financial services where there were only 13,000 in March 1980. With the recent enactment of landmark banking legislation, another4,000 new competitors (savings and loan associations) will be able to offer banking services to businesses. Besides these new competitors, nondepository institutions have entered consumer and business markets with an expanding variety of financial instruments and services. How many of these new competitors will prosper? What will the survivors and the financial system look like? The answers to these rather broad questions depend, t o a great extent, on the answer to a narrower question: do larger financial institutions enjoy lower costs than smaller institutions? Evidence on this question will be vital to thinking through the implications of future financial deregulation. For instance, how would interstate deposit-taking operations by banks and S&Ls affect the structure and competitive health of the nation's financial system? Will large institutions "gobble up" smaller ones, or will smaller ones be able to remain competitive? Part of the answer depends on whether the large institutions can produce financial services more cheaply than smaller ones—in the language of economics, whether production of financial services is subject to "economies of scale." This issue of the Economic Review concentrates on present and future competition between larger and smaller financial institutions. With remarkable consistency, the studies reported here suggest that large size does not seem to give a financial institution significant competitive advantages. This implies that, contrary to many predictions, large banks and S&Lsare not likely to 4 drive smaller ones out of business as deregulation progresses. The first study, dealing with operating costs at banking organizations, is presented by George J. Benston, visiting scholar at the Federal Reserve Bank of Atlanta and professor of accounting, economics and finance at the University of Rochester's Graduate School of Management; Gerald A. Hanweck, economist, Board of Governors of the Federal Reserve System, and David B. Humphrey, chief of the Federal Reserve Board's Financial Studies Section. They estimated costs for branch and unit banks (unit banks are individual banks in states that prohibit bank branches) and tested for the impact of bank holding company affiliation on costs. They found that costs per account for banks larger than $50 million in deposits increased as bank size increased, while costs declined with size for banks with less than $25 million in deposits. ^ In the second article, James E. McNulty, assistant vice president and economist, Federal Home Loan Bank of Atlanta, reports on a new study of the cost structure of savings and loan associations. M c N u l t y found decreasing costs forS&Ls with up to about $500 million in deposits but did not fully account for normal methods of expansion. Above the $500 million level, costs increased slightly. Larger S&Ls, then, may capture some economies of scale if they can expand significantly w i t h o u t adding a proportionate number of branches. Since S&Ls specialize in low activity functions such as mortgages and savings accounts, McNulty suggests, these results are not likely to be carried over as S&L operations broaden to include transactions accounts and business and consumer loans. Following McNulty's study, David D. Whitehead, senior financial economist Federal Reserve Bank of Atlanta, and Robert L Schweitzer, assistant N O V E M B E R 1982, E C O N O M I C REVIEW r : An Overview professor of economics, University of Delaware, survey the evidence linking size and risk in banking. Operating cost studies fail to consider the relationship between risk costs and size in financial institutions. Greater risk requires greater compensating returns to shareholders and other providers of funds and thus may put riskier institutions at a competitive disadvantage. In their review of the sparse evidence found in studies of credit risk, interest rate risk, bank capital adequacy, and early warning systems for problem banks, Whitehead and Schweitzer found little evidence that small banks operate with greater risks that cannot be controlled. Moving from studies of relationship between size and costs to the playing out of competition in the real world, B. Frank King, research officer at the Atlanta Fed, asks how well larger banks in local markets have performed versus the smaller ones over the past decade and a half. Citing evidence from several studies, including the Atlanta Fed's new study covering southeastern markets between 1974 and 1981, he concludes that larger banks generally have lost market share to smaller local competitors. The evidence supports that conclusion for any subperiod, any area of the country and any type of market studied over the past 15 years. Carrying the consideration of one-on-one competition a step further, King then looks at the impact of large bank holding companies' acquisition of local banks in three southeastern states. King studied two types of banks—de novo (newly organized) banks and large banks acquired by large holding companies. He compared them with similar independent banks in the same markets. The evidence indicated that after several years the holding company subsidiaries did not generally differ from the independents in size, profitability or risk. If smaller banks operate with different goals than larger ones, the smaller banks may be able FEDERAL RESERVE BANK O F A T L A N T A 5 to survive despite some competitive disadvantages, according to another study. It also found that small banks may contribute a competitive element that would be lost with their demise. David D. Whitehead tested indirectly the theory that small banks attempt to gain market share at the expense of profits, while large banks seek profits even if they must sacrifice market share. He found evidence consistent with his hypothesis that small banks emphasize market share. Smaller banks may survive even if they do not earn quite so much as larger ones. They may also create significant price competition. The final study in this issue takes a forward look at payments system technology and tries to project the impact of this important and rapidly changing segment of financial institutions' operations on their competitiveness. Paul F. Metzker, an economist on the Atlanta Fed's payments research team, foresees substantial and growing economies of scale in the production of payments services. However, he also expects small institutions to capture the benefits of scale by sharing production facilities through networks, service bureaus and franchises. He finds evidence that small banks already are moving toward shared facilities. These studies present considerable new evidence on competition between small and large institutions—and that evidence generally is consistent. Large institutions do not seem to have enjoyed significant competitive advantages over small ones even in the recent past. Further, in the important business of payments services, smaller institutions seem likely to keep up despite broad technological changes. This evidence indicates that interstate banking is not likely to produce rapid consolidation of institutions or to greatly increase productive efficiency or safety for the financial system. Operating Costs in Commercial Banking Evidence indicates that bank operating costs, when adjusted for normal methods of expansion, increase with size in banks with deposits of $ 5 0 million or more. 6 Reasons for Concern with Banks' Operating Costs In most businesses, operating costs are a private matter. In banking, though, these costs receive wider attention. A bank's managers and owners clearly are concerned with costs since they profit if costs are controlled. A bank's customers also are concerned, since banking costs ultimately are passed on to users of the bank's services. In addition, legislators and those charged with bank regulation have a derivative concern; the more efficiently banks are operated, the larger the earnings flows that may improve safety by absorbing losses, the more efficiently the nation's payments system works and the more efficiently savings are channeled into investment. NOVEMBER 1982, E C O N O M I C REVIEW Large versus Small Institutions Legislators and regulators also have a direct concern with banks' operating efficiency, because efficency may influence the way specific regulations are enforced. In particular, if larger banks produce a given level and quality of services at a lower cost than smaller banks—that is, if banking is subject t o economies of scale—then larger banks could offer lower prices or more service than smaller banks. In an unrestricted environment, larger banks might well displace the small banks, reducing competition. If these economies of scale are large enough, the savings of resources might offset anticipated reductions in competition. However, if economies of scale aren't significant, a policy of unrestricted chartering (free entry) would probably not result in failure of new banks due to their inherent inability to compete successfully with larger banks. Instead, failure would be associated more w i t h management skill and specific economic events than with the scale of operation. Form of Organization—Branching and Holding Companies Financial institutions' operating costs also may be related t o their form of organization. Information on this relationship is important for states trying to decide what type of bank branching and bank holding companies they will allow, and for the federal government as it considers various proposals for interstate banking. Branches, independent banks and bank holding company subsidiaries may all have different relationships between size and operating costs per unit of output. Branches, in particular, may be more or less costly to run than similar sized unit banks (banks without branches). Or very small branches may be more efficient than small unit banks, while large branches may be less efficient than similar sized unit banks. And, even if branches were more costly t o operate, they might permit a bank t o grow and take advantage of economies of scale, the savings from which could offset the presumed extra costs of branching. Considering the costs of holding companies' operations is important, since they are an alternative t o branching. on individual products is required; some other questions require information on the bank as a whole. In some cases, the costs and amounts of various services must be separated so that institutions producing different types and quantities of outputs can be compared meaningfully. Estimates of the production cost of a specific service, such as mortgage lending, are desirable in a market analysis for a particular product. For example, in considering competition between commercial banks and savings and loan associations, the possibilities for economies of scale in mortgage lending and time deposits are of primary concern; the costs of many other banking services may not be important. ". . . if banking is subject to economies of scale—then larger banks could offer lower prices or more service than smaller banks." Aggregating the costs of multiple banking services is necessary to determine if a financial institution as a whole is subject to economies of scope. These economies occur when several products (such as time and demand deposits or demand deposits and business loans) are produced together at a lower total cost than when produced separately. If there were significant economies of scope in producing, say, business, consumer and mortgage loans in one operation, specialized institutions such as S&Ls w o u l d be less efficient than full-service commercial banks. The c o m m o n production of transactions in corporate and government securities also could give rise to economies of scope. This has implications for public policy regarding the powers allowed thrift associations and commercial banks as well as for mergers between different types of insti1 tutions. Multiple Products Results of a Current Bank Cost Study Banks produce multiple products. To answer some of the above questions, cost information This study reviews research on bank costs and reports on a recently published study by the 7 FEDERAL RESERVE B A N K O F A T L A N T A authors. Consecutive sections address the technical problems of defining and measuringoperating costs and output, estimating operating efficiency, and obtaining appropriate data for cost analysis. The article then reports the finding of our latest study and discusses them in relation to other recent studies of bank operating costs. Results and implications of the analysis are summarized on page 21. The important objectives of bank cost studies have been to determine: (1) the extent that the scale of bank operations affects bank costs; (2) the effects of branching and holding company organization on costs; (3) the cost effects of banks producing multiple services; and (4) the effects of different types of bank customers on bank costs. A number of fundamental problems are common to all bank cost studies as well as cost studies of other industries. Fortunately, measurement problems are not great using data from the Federal Reserve's Functional Cost Analysis (FCA) program. • The most recent research (using 1975-78 FCA data, a more flexible estimating functional form, and a theoretically supportable index number method of aggregating the various banking outputs) generally confirms many of the results of previous research. New results are summarized as follows: (1) Banks with more than $50 million$75 million in deposits in unit banking states experienced diseconomies of s c a l e that is, average costs increased as the banks increased in size. (2) Banks of all sizes in states that allow branch banking experienced economies of scale with respect to the numbers of deposits and loan accounts. (3) When the mode of e x p a n s i o n increasing the number of accounts or the number of offices for branching banks but only expanding the number of accounts for unit banks was accounted for, banks in both branching and unit states above $50 million-$75 million in deposits experienced diseconomies of scale. (4) Economies of scale appeared to be unchanged over the 1975-78 period. (5) Larger account sizes for both types of banks increased costs less than proportionately. (6) Average costs (unadjusted for certain differences in branch and unit banks) were similar for banks with up to $75 8 million in deposits in branch and unit states; larger banks in branching states had considerably lower average costs than similar banks in unit states. (7) Taking into account the different ways that branch and unit state banks can expand—by increasing offices or average account size, respectively—average costs were similar among banks of similar size in unit and branching states. (8) Bank holding c o m p a n y affiliation was found to have little effect on bank costs. Definition and Measurement of Operating Costs and Output Measuring Operating Costs Public and private policy questions require numbers that meaningfully reflect economic values rather than accounting measurements. Regulators and bankers should be concerned with "opportunity costs," the value of resources given up because of a decision to do one thing rather than another. For example, the cost of making and servicing a mortgage loan is the dollar value of resources that, as a consequence, cannot be used for the next best opportunity. Where outlays are made in cash or in close equivalents (e.g., the present value of promises to pay cash in the future) for such expenses as salaries, supplies and computer services, the accounting numbers reflect economic values well, since the amount of cash used for the stated purpose is no longer available for another. Other costs may not be so well reflected, however. For example, the economic cost of the space occupied by the mortgage loan department is not well measured in accounting statements. These statements include depreciation, an essentially arbitrary allocation of a portion of the original cost of the building and equipment to a given activity over a specific time period. These figures seldom provide valid estimates of the present economic cost of using the building. This cost is measured by the bank's next best opportunity. This would be the amount that it would receive if it rented out the space occupied by the mortgage department or the amount it could save by giving up space occupied by another function that could be moved into the mortgage department's space. Other costs that may be similarly misstated are inventories valued N O V E M B E R 1982, E C O N O M I C REVIEW at acquisition cost and salaries and other tax deductible compensation that are substituted for dividends. Clearly, opportunity costs involved in these situations are very difficult and often impossible t o measure. Even when costs are measured reasonably well, it often is very difficult to assign them to specific types of output with reasonable accuracy. For example, the cost of employing the bank president might be measured correctly if the present value of his pension and other fringe benefits were accounted for along with his salary. But how much of the president's compensation should be charged to demand deposits, business loans, mortgage loans, and the other banking functions? Even if one could keep track of the ^ ^ • ^ • • • • • • • • • • • • • • • • i i i h i . . it is often very difficult to assign [costs] to specific types of output with reasonable accuracy." president's time, there is no reason t o expect a direct relationship between the time spent with a department and the value of the president's services to the bank. Similar problems beset the allocation of other overhead costs.1 Fortunately, the divergence of the numbers that can be obtained (accounting data) from those that one wants (economic values) is relatively small for financial institutions. Salaries, which can be measured by accounting numbers, comprise about 52 percent of commercial banks' operating costs (excluding interest and loan and security losses). Occupancy expense, including building depreciation, represents only 9 percent of those costs.2 The cost of supplies and goods sold, a large part of most other enterprises' costs, is negligible for financial institutions. Thus, except for a relatively small amount of depreciation and the inherent difficulty of assigning costs to specific ' S e e Benston (1982) for a more c o m p l e t e description of the difference b e t w e e n e c o n o m i c values a n d accounting numbers. Federal Reserve, Functional Cost Analysis: 1 9 7 8 Average Banks, p. 20. The percentages are almost the same for other years 1978 figures are used here because data for this year are analyzed further below 2 FEDERAL RESERVE B A N K O F A T L A N T A outputs, the accounting numbers recorded on financial institutions' books provide reasonably good proxy measures of economic values. Further problems arise because not all costs of producing financial services are recorded on the institutions' books; however, these problems are considerably less than those for cost studies of other firms. Two problems specific to financial institutions' costs arise from the possible tradeoffs between revenues and expenses included in operating costs. With respect to lending operations, for instance, higher expenses for monitoring and collecting loans can substitute for higher loan losses. To the extent that this substitution is made, operating expenses will be higher, since actual loan losses are not considered operating costs, but are charged t o reserves. As an alternative, higher loan losses could be compensated for with higher interest and fee income. In this event, not only must loan losses be excluded from operating expense, but higher losses need not even be associated with lower monitoring (operating) expenses. Interest and other payments for deposits present the second problem. These payments are returns to depositors on their investments, as much as dividends and capital gains on mutual funds shares are returns to shareholders. The bank is merely an intermediary, as is a mutual fund. While interest is an important outlay to the bank, it is determined by market forces that reflect alternative investments available to depositors. Thus interest is not an operating expense for purposes of measuring banks' efficiency. However, the amount a bank can pay its depositors in the form of interest is constrained by law. For this and other reasons such as personal income taxes and administrative efficiency, banks may pay customers for their deposits in the form of "free" services, preferred lending arrangements (including lower interest charges) or more convenient banking facilities. "Free" services result in higher operating costs while preferred lending arrangements result in lower revenue. A similar problem may affect the recorded expenses of banks that get services from other banks in exchange for compensating balances. Differences in incurring and recording operating costs may produce misleading conclusions about bank efficiency if the differences are related systematically to the size or organizational form of the institution. Fortunately, this seems not to be the case, at least with respect to the size of banks. For banks with deposits up to $50 9 million, from $50 to $200 million and over $200 million, occupancy expenses average 9.0 percent, 9.7 and 9.7 percent of total operating expenses. 3 For these same bank size groupings, the five-year average amount of loan losses, expressed as percentages of loans outstanding, are .15 percent, .16 percent and .21 percent. The reported gross yields on loans for these banks average 9.8 percent, 9.8 percent and 10.2 percent. 4 Thus the relatively larger loan losses at banks with deposits over $200 million appear t o be compensated for by higher yields. Nor does the understatement of respondent banks' operating expenses (since they pay for some correspondent bank services with compensating balances rather than with fees) appear to distort the cost data with respect to bank size. In t w o previous studies, the opportunity cost of these balances was computed and added to the banks' operating expenses; this adjustment had no significant effect on the estimates of economies of scale.5 Hence, we conclude that bank operating costs, as usually reported, measure economic values reasonably well and, to the extent they do n o t appear not to bias economies of scale estimates. A major problem is the exclusion of the cost of customer inconvenience when there are too few banking offices available due t o legal restrictions on banking. Measurement of Output—What Do Banks Really Produce? Measuring output is a more difficult problem. One's view of what banks produce depends on one's interests. Economists w h o are concerned with economy-wide (macro) issues tend t o view the banks' output as dollars of deposits or loans. Monetary economists see banks as producers of money—demand deposits. 6 Others see banks as producing loans, with demand and time deposits being analogous to raw materials. 7 Thus banking statistics are reported and banks described in dollar terms, as a "million-dollar bank." Customers also often describe the output of a bank in dollar terms, since 3 lbid. Ibid, p. 5. Flannery (1981) and Benston, H a n w e c k a n d H u m p h r e y (1982). Dunham (1982) also made this correction but did not estimate total bank cost output elasticities for t h e bank as a whole. 6 For example, s e e Goldschmidt (1981). ' S t u d i e s that employ this definition (e.g., Greenbaum, 1967) are reviewed 4 5 they hold so many dollars in deposits or want t o borrow a given amount of money. But banks do not incur operating costs directly as a function of the number of dollars of deposits or loans they process. The cost of accepting and collecting a deposited check is affected only slightly by the number of digits that appear around the decimal point. While a $10,000 check involves somewhat more risk to process than a $10 check, the extra cost is not 1,000 times greater. Similarly, a $100,000 loan may involve more careful administration than a $ 10,000 loan, but not by a factor of 10. Indeed, with respect t o operating costs, a bank is best described as a recordkeeping and processing "factory." The key cost-causing output variable is the number and types of pieces of paper and electronic signals processed. Though not proportionately so, the dollars written on the paper processed are also relevant. The risk, should a deposit be mishandled, a check not collected, a loan not repaid, or a security defaulted, is related positively t o the amount of the check, loan or security. The amount of services a bank is willing to give t o its depositors in lieu of direct interest payments (which are legally prohibited or restrained) also is a function of the value of the deposit. Therefore, to be complete, the dollar amount as well as the number of items and accounts processed should be considered operative costcausing factors. Which particular variables are taken t o measure output? It depends on the question asked. For the regulatory and bank management questions posed above, the preferred measure is that which yields answers to such questions as, " w h a t are the operating costs of large compared to small banks, of branches compared to unit banks, automated teller machines compared t o bank offices, of banks compared to thrift associations, and of combinations of financial institutions?" To answer these questions, we must attempt t o relate the costs incurred by various institutions to the same set of cost-causing activities. For these purposes, it is meaningless t o report that a large bank enjoys lower costs per dollar of loans than a small bank if this " e c o n o m y " is achieved because the large bank makes larger loans. Most operating costs are related to the loan, as such, rather than to its a m o u n t Comparing the efficiency of a large and small bank in this manner is like comparing costs per dollar of below. 10 N O V E M B E R 1982, E C O N O M I C REVIEW sales of a wholesaler selling by the case and a retailer selling by the item. Measurements of operating costs per dollar can be useful, however, since the lower cost per dollar loaned may explain why interest rates are lower on larger loans. But unless the size of loans or deposits is expected t o change as a consequence of a regulatory or business decision, the variables used to measure output should refer to the ongoing factors that generate operating costs. Any analysis of the output of financial institutions, then, should be multifaceted, and should include the numbers of deposit items processed and loans made and serviced. Collateral banking services, such as trust and safe deposit boxes, also should be accounted for. If different institutions are compared, differences in the levels and qualities of their outputs should be considered. And, because the costs incurred are functions of local economic conditions and the value of the dollar over time (when data for different time periods are used), the recorded costs should be adjusted for the effect of price-level factors. Problems of Estimating Operating Efficiency Two basic problems beset researchers w h o would analyze operating cost. One is accounting for differences among banks with respect to the amounts of costs recorded and the types of outputs produced. The second is accounting for differences in costs related to unique attributes of specific banks or markets rather than to general attributes, such as size and type of organization. For instance, a bank may have grown large because it was run efficiently, possibly because of the genius of particular individuals. Their compensation for superior abilities would not be included in current operating costs if it was paid for w i t h stock options or if it occurred in the past. The cost of these factors may have been expensed in the past or may have been fortuitously acquired. That such banks currently achieve low operating costs and have grown large does not necessarily mean banking is characterized by economies of scale. Similarly, lower operating costs and, say, small size, could be associated jointly with characteristics of the markets in which banks operate. Thus small towns may have both lower wages and small banks. This relationship FEDERAL RESERVE B A N K O F A T L A N T A "Operating cost is taken to be adequately measured by the amount reported in the bank's earnings statements." is not an indication of economies of scale, since banks in small towns cannot effectively serve customers w h o live in distant large cities. Fortunately, these problems are not serious for studies of financial institutions' costs. The technology and managerial methods used by banks are generally well-known and available. Furthermore, branching and chartering laws restrict banks to given geographic areas for deposit services. Except for large business loans, their markets are constrained by the cost t o customers of dealing with banks far outside their communities and the cost t o banks of acquiring information about and monitoring distant customers. Thus banks tend to be large or small because of the markets in which they operate. The operating cost and output data, therefore, are likely to trace out cost curves from intersections of supply and demand curves rather than tracing out demand curves. The problem of accounting for factors that produce differences among banks in recorded costs and types of output is dealt with in t w o ways. W i t h respect t o cost, one way is to assume, usually implicitly, that these differences either are trivial or are not systematically related t o bank output. Operating cost, then, is taken to be adequately measured by the amount reported in the banks' earnings statements. The bias in this simple "solution" is not likely to be serious, at least with respect t o bank size. However, differences among individual banks should be kept in mind. In situations where differences may be important, the data should be adjusted and checked to see if perceived differences are systematically related to the variables at issue. W i t h respect to output, t w o approaches of measuring comparative bank efficiency have been followed. In one, differences among banks are assumed to be unimportant. Total costs of producing outputs such as deposits or loans 11 are divided by total outstanding amounts of these outputs. This gives costs per dollar of output, which are taken to be a valid measure of operating efficiency. The unit costs are presented in tables (tabular analysis) comparing banks grouped according to a variable of interest (for example, deposit size). The second approach uses the statistical method of multiple regression analysis to account for differences in output in order to isolate the scale effect from other costcausing factors. The advantages, shortcomings, and findings of each type of analysis are presented in the next t w o sections. Data for Cost Analyses All Bank Aggregate Data Since all chartered financial institutions regularly report income, expense and balance sheet data, measuring operating efficiency as total recorded costs divided by total deposits, loans and securities, or assets has considerable advantages. In studies using this method, costs per dollar typically are averaged for banks grouped by deposit size and compared in tables (hence, tabular analysis). This procedure was followed by Alhadeff (1954), who published the first study of bank operating costs, using California data from the years 1938-1950. But this comparison does not account for differences among banks, such as the composition of their earning assets, deposits, type of organization, and location. Horvitz (1963) replicated Alhadeffs study with data from all Federal Reserve member banks for 1949-1960. He subdivided the 1959 data into branch and unit banks and into three groupings according to the ratio of time to total deposits. Other variables were not accounted for. Assuming that these other variables are not important or are unrelated to size, Horvitz found (as did Alhadeff) that "once a bank reaches the relatively small size of $5 million in deposits, additional size does not result in reduced costs to any great extent until a bank reaches the giant size of over $500 million" (p. 37), and that branch banks have uniformly higher costs per dollar of loans and investments than unit banks. But, as the more complete statistical analyses reported below indicate, the assumptions necessary for this exercise are not likely t o be valid. 12 FCA Data One important shortcoming of the data used in these early studies is the lack of detail with respect to specific banking outputs. Fortunately, a rich source of data on the costs of major banking functions has been available since the late 1950s. The Federal Reserve provides banks with a Functional Cost Analysis (FCA) service in which the banks allocate costs and revenue to specific banking functions. These figures are then structured into a standard format that may be compared to other reporting banks. The FCA program is voluntary, with 780 banks responding nationwide in 1978. Among the data allocated to the banking functions are itemized revenue, expense, dollars of deposits and assets (averages of 12 month end amounts), and the number of accounts opened and items processed. Thus, these data provide a wealth of detail that permits the analyst to improve upon potentially misleading tabular analyses. However, even though FCA figures are aggregated by three bank deposit sizes, a comparison of the numbers must rest on several possibly untenable assumptions. One is that the banks' organizational structure is not relevant, since branch and holding company banks are not distinguished from unit banks in the published results. Another is that the location of a bank (e.g., urban or rural) does not affect its costs. Third, the analyst either must remove the allocated overhead expenses or accept the essentially arbitrary allocations made in the FCA data. Perhaps most important for a measure of efficiency is that output can be measured with only one variable for each type of p r o d u c t number of accounts serviced, number of items "We measured output using the average number of deposit and loan accounts serviced, augmented with information on the average balances of the accounts." NOVEMBER 1982, E C O N O M I C REVIEW processed, or dollars of account balances outstanding. The joint effect of items processed, accounts serviced and dollars outstanding cannot be simply measured. In order to assess the joint effect of these variables, economists w h o have used these and other data employ multiple regression analysis. Findings of a Recent Study8 The Model W e recently specified and estimated a bank cost function using multiple regression analysis. We used annual Federal Reserve FCA data for the years 1975 through 1978 involving from 747 to 852 banks, amounting to about 15 percent of the Federal Reserve members. Since this program is voluntary, the banks included are likely t o be more conscious than most of the value of cost control. The banks range up to $1 billion in deposits, since the few banks in the FCA panel larger than this were excluded. Thus, the sample covers the range of U. S. banks except the giants, and it is perhaps overrepresentative of more efficient banks due to the voluntary nature of the FCA program. Operating cost was defined to exclude interest and loan losses—since only the outputs of the banking deposit and loan functions were analyzed, the direct and allocated costs of the functions of safe deposit box, trust, customer computer services and investments were also excluded. The included outputs account for more than 72 percent of total operating costs. W e measured o u t p u t using the average number of deposit and loan accounts serviced, augmented with information on the average balances of the accounts. Since w e wanted to estimate the relationship between total operating cost and output (and other variables), we had to construct an index to represent the multiple products produced by banks. For this purpose we employed the technique suggested by aggregation theory (see Diewert, 1976 and Barnett, 1981) to construct a Divisia multilateral index number of bank output. In effect, w e weighted the number of accounts of each of the five types of output (demand deposits, time and savings deposits, real estate loans, installment loans, and business loans) by their proportionate share in total operating costs using the FCA data. 9 The Divisia Index number of accounts provides a valid measure of the aggregate cost-causing transactions undertaken by banks for their customers. Including the average size of accounts serves as a control for activities that are related primarily to the dollar amounts of the accounts. Differences among the banks in the prices of inputs were accounted for with variables that measured the average annual salaries plus fringe benefits paid per employee and an index of regional office floor space rental costs—the opportunity cost per square foot of bank building space. Two types of organizational differences were recognized. The effect of multibank holding company (MBHC) affiliation on operating cost was captured with two variables, one of which measures its effect on total cost and the other its interaction between branching and M H B C affiliation. The costs of branching also were accounted for in t w o ways. First, the data were analyzed separately for banks in unit banking states and banks in states that permitted branch banking. This procedure avoids forcing the estimated parameters to be equal for branch and unit state banks, a common (but incorrect) assumption of other studies. Second, interactions between the number of offices and output were specified so that the effect of branching was not measured as merely a simple and constant percentage of total cost. W e believe that absence of this more general branching specification seriously mars the findings of previous studies whose effect was t o assume that the cost of adding a branch office was a constant percentage of total operating costs. Thus the cost to a larger bank of adding a branch of a given size w o u l d always be proportionately greater than the cost of adding the same size branch to a smaller bank. 10 Our specification allows the cost of an additional branch, as a percentage of operating costs, to vary across different sized banks. Finally, the functional form fitted, the translog cost function, is designed t o permit increasing "See Benston, Hanweck a n d Humphrey, 1982, for details. For example. Longbrake a n d Haslem (1975. Table 2) estimate the d e m a n d deposit cost of an additional branch at a bank having from t w o to five branches as S 1 2 5 3 while the same sized additional branch at a bank having b e t w e e n twenty-five a n d fifty branches is $3,760 ,0 "Details of this study may be found in Benston, Hanweck and Humphrey 1981 a n d 1982. FEDERAL RESERVE BANK O F A T L A N T A 13 Table 1. Scale Economy (Elasticity) Estimates, Unit and Branch States Banks, 1978 FCA Data Deposit Size Group ($ millions) Operating Cost Number of Accounts A. Unit State Banks $0-10 10-25 25-50 50-75 75-100 100-200 200-300 300-400 400+ Total Sample Including Interest Dollars of Accounts Operating Cost Number of Accounts Excluding Interest Dollars of Accounts .95 1.01 1.07 1.11* 1.13* 1.19* 1.19* 1.24* 1.23* 1.09* .73* .82* .88* .92° .96 1.01 1.06 1.06 1.12 .92° .95 .98 1.00 1.01 1.03 1.05 1.08° 1.06 1.09 1.01 B. Branch State Banks .81* $0-10 .89° 10-25 .93* 25-25 .93* 50-75 .92* 75-100 .94 100-200 .92 200-300 .93 300-400 .92 400+ .92* Total Sample .63* .74* .81* .84* .85* .89* .90 .93 .94 .84 .81 .95* 1.02 1.04* 1.05* 1.06* 1.06* 1.05 1.04 1.04* .96° .95* .96* .98* 1.00 1.02 1.09* 1.13* 1.13* .97* 1.01 1.02 1.02* 1.02* 1.01* 1.01 1.00 .99 .99 1.02* Notes: a. Source - Benston, Hariweck a n d Humphrey 1981 a n d 1982 *(°) Indicates elasticities different from 1.0 (constant costs) at the 0 5 ( 1 0 ) c o n f i d e n c e level in a t w o tail t-test and/or decreasing costs to be measured as output increases, as well as t o provide estimates of economies of scale, of branching, and of account size that are permitted to vary by size of bank. It is not constrained to meet the requirements of a particular production function, such as the Cobb-Douglas function which was almost exclusively used in earlier studies. Three principal shortcomings of this study must be emphasized. One is that the data include only the fraction of U. S. banks that participate in the FCA program. The other shortcomings are that estimates of economies of scope have not been separated from estimates of economies of scale and that scale economies of individual banking functions are not separately estimated. These analyses are in progress. Since economies of scale and average costs per account can and do differ for each bank, derived estimates of economies of scale and 14 other parameters from the regression estimates were based on unit and branch banking state subsamples for nine deposit-size groups. The relevant measures were then calculated for the (geometric) average bank in each group. The following findings may be drawn from our analysis.11 Economies of Scale • Banks in unit b a n k i n g states above $50 million-$75 million in deposits experienced diseconomies of scale w i t h respect t o t h e Divisia n u m b e r of accounts. The smaller banks in these states had scale e c o n o m i e s or constant costs. The elasticities range ' ' T h e numbers presented are for the analysis of 1 9 7 8 data. The analysis of the 1975, 1 9 7 6 and 1977 data yield similar results. See Benston, Hanweck and H u m p h r e y (1982) for details. NOVEMBER 1982, E C O N O M I C REVIEW f r o m .95 t o 1.23. 1 2 (See Table 1, c o l u m n 1.) • Banks in b r a n c h i n g states of all sizes e x p e r i e n c e d e c o n o m i e s of scale w i t h respect to the Divisia n u m b e r of accounts. The elasticities range f r o m .81 for t h e smallest d e p o s i t size group t o .92 for the largest. These elasticities were calculated for each d e p o s i t size group, so t h e fact that larger branch banks have more branches was taken into account. (See Table 1, c o l u m n 1.) The e c o n o m i e s of scale e x p e r i e n c e d by banks in b r a n c h i n g states appear to result from t h e a b i l i t y of branch banks t o o p e r a t e out of many smaller offices, a v o i d i n g the diseconomies associated w i t h a large n u m b e r of accounts per office. To a c c o u n t for d i f f e r e n c e s in the m o d e of branch bank e x p a n s i o n (by a d d i n g branches each w i t h a g i v e n n u m b e r o f accounts) versus that for unit banks (by a d d i n g more accounts at a single o f f i c e ) , w e c a l c u l a t e d a u g m e n t e d elasticities, where the effect on costs of changes in the n u m b e r o f offices as w e l l as the n u m b e r of accounts serviced is a c c o u n t e d for. These a u g m e n t e d elasticities ( n o t shown in Table 1) show that: • Banks in unit states w i t h deposits b e l o w $25 m i l l i o n e n j o y e d e c o n o m i e s of scale. The m i d d l e - s i z e d banks ($75 m i l l i o n t o $300 m i l l i o n in deposits) had significant d i s e c o n o m i e s of scale and t h e others had i n s i g n i f i c a n t scale d i s e c o n o m i e s . ( T h e n u m b e r s are almost t h e same as t h o s e given in Table 1 for t h e u n a u g m e n t e d measure.) • Banks in b r a n c h i n g states w i t h m o r e than $25 m i l l i o n in deposits e x p e r i e n c e d statistically significant d i s e c o n o m i e s of scale (using t h e n u m b e r of accounts as o u t p u t ) . The elasticities range f r o m 1.09 for banks w i t h $25 m i l l i o n t o $50 m i l l i o n in deposit size group t o 1.1 6 f o r t h e largest. Smaller banks experienced approximately constant costs. Thus, w h e n the n o r m a l path of e x p a n s i o n is a c c o u n t e d for, branch and unit state banks '-'An elasticity is the proportional c h a n g e in cost associated with a given proportional c h a n g e in output For example, an elasticity of 9 5 means that a 100 percent increase in output is associated with a 9 5 percent increase in cost. Since cost increases less than proportionately, the banks are said to experience e c o n o m i e s of scale FEDERAL RESERVE BANK O F A T L A N T A e x p e r i e n c e d similar scale e c o n o m y or dise c o n o m y values. O t h e r results i n d i c a t e that: • Economies of scale appear t o be unchanged over the p e r i o d 1975 t h r o u g h 1978, since t h e elasticities c o m p u t e d are either very similar or show unsystematic changes over the years. • W i t h respect t o the average size of accounts, banks in both unit and branching states experienced conside~able economies of scale. Elasticities averaged b e t w e e n .28 and .50 over t h e four years s t u d i e d . The elasticities increase c o n s i d e r a b l y for the larger d e p o s i t size groups of banks. In particular, the smallest banks had almost no a d d i t i o n a l cost associated w i t h larger account sizes, but t h e cost of the largest deposit size banks (particularly t h e u n i t state banks) almost d o u b l e d w h e n t h e i r average a c c o u n t sizes d o u b l e d . This finding indicates that the larger banks, w h i c h have larger loans and d e p o s i t accounts, incur proportionately larger costs, probably arising f r o m m o n i t o r i n g t h e loans and attracting the deposits. Average Cost per Account For many p o l i c y questions, average costs per account are more relevant than economies of scale. It may be t h a t unit banks e x p e r i e n c e d i s e c o n o m i e s of scale but t h e i r costs per account serviced m i g h t be l o w e r than those for branch banks. Average cost p e r a c c o u n t was c a l c u l a t e d in several ways. First, the average ( g e o m e t r i c mean) values of input prices, holding company affiliation, size of account, and n u m b e r of offices of each of t h e nine d e p o s i t size groups were used for the calculation. Thus the calculated average cost p e r a c c o u n t is the average e x p e r i e n c e d for each size group. These are plotted in Chart 1 and show: • Similar average costs for banks in unit and branch states u p t o $75 m i l l i o n in deposits. • Greatly increasing average costs for larger banks in unit states. Costs i n c u r r e d by the largest banks were more than d o u b l e those for the m i d d l e size group. • Higher average costs for larger banks in b r a n c h i n g states that, nevertheless, are considerably b e l o w those for similarly sized banks in unit states. 15 C h a r t 1 . Average Cost per Account, 1978 FCA Data, At Unit and Branch State Banks Cos! Per Account 30 60 90 Deposit Size Groups 120 S millions) A - unit state banks, variables at m e a n s of deposit size g r o u p B - branch state banks, variables at means of deposit size groups. C - u n i t state banks, variables at branch state bank. M e a n s for entire sample, except number of a c c o u n t s a n d offices. "Deposit size group $ 4 0 0 - 1 , 0 0 0 is o m i t t e d since none of the unit state banks had numbers of a c c o u n t s of the magnitude serviced by branch state banks. D - branch state banks, variables at m e a n s for entire sample, except number of a c c o u n t s and offices. therefore, it w o u l d be unrealistic t o assume that higher o u t p u t c o u l d be a c h i e v e d w i t h o u t branch expansion. U n i t banks, t h o u g h , must e x p a n d at a single full-service office. To c o m p a r e average costs of banks in unit states w i t h average costs of banks in branching states, w e s u b s t i t u t e d b r a n c h i n g state cost d e t e r m i n a n t s in t h e unit states' cost equation. W e e n t e r e d averages of n u m b e r of accounts, n u m b e r of offices, a c c o u n t size and h o l d i n g c o m p a n y a f f i l i a t i o n . Average n u m b e r of offices in b r a n c h i n g states c o u l d not be e n t e r e d in the unit state e q u a t i o n ; thus, this cost d e t e r m i n a n t was ignored. W e e m e r g e d f r o m this exercise w i t h an e s t i m a t e of the average costs of unit banks w i t h all of the characteristics of b r a n c h banks but m u l t i ple offices. These calculations ( p l o t t e d in Chart 1) show: • Somewhat higher average costs for larger banks. • Very similar average costs for similar sizes of banks in unit and b r a n c h i n g state banks. These results indicate t h a t w h e n w e take p r o p e r a c c o u n t of the c u r r e n t d i f f e r e n c e b e t w e e n unit and branch state banks, these t w o classes of banks w o u l d experience similar average costs. This reinforces the earlier rep o r t e d similarities w i t h the scale e c o n o m y results w h e n a c c o u n t is taken of t h e d i f f e r e n t expansion m e t h o d s used by unit and branch banks ( d u e t o regulatory constraints). Cost of Branching and Holding Company Affiliation For most p o l i c y questions (for e x a m p l e , mergers and unit versus branch banking), t h e average size of accounts is not relevant. Presumably, bank customers w o u l d d e m a n d loans and hold deposits that met their convenience rather than the bank's. C o n s e q u e n t l y , we e x t e n d e d our analysis t o abstract f r o m this difference among banks, by holding the average size of a c c o u n t constant at t h e mean values for the e n t i r e b r a n c h i n g state bank sample. I n p u t prices and h o l d i n g c o m p a n y a f f i l i a t i o n also w e r e held constant at t h e b r a n c h i n g state sample means; h o w e v e r , t h e n u m b e r of offices operated was not held constant. Branch banks t e n d t o expand by o p e n i n g more offices; W e can also compare estimated operating costs of a branch with those of a similar sized unit bank. Such comparisons are relevant to public choice of unit versus branch banking and of methods of interstate banking. The typical branch serves from 5,000 t o 11,000 accounts. The operating costs of unit banks w i t h the same o u t p u t are $210,000 and $462,000. 13 Branches serving the same number of accounts at smaller branch banking organizations have estimated costs of $213,000 and $483,000, respectively. Similar sized branches 1978 data (as are the following numbers) The a m o u n t s for the other years studied (1975-1977) are similar, though lower due to the effects of inflation over the same period See Benston. Hanweck and Humphrey 1981. Table 6 for the individual year amounts. 16 N O V E M B E R 1982, E C O N O M I C REVIEW at larger banking organizations are estimated to have increasingly higher operating costs (for example, an 11,000-account branch of a bank with $300 million to $400 million in total deposits costs $650,000 to operate.) Thus, small unit banks appear able t o compete successfully with branches, particularly those operated by large banks. W e found the effect on operating costs of bank holding company affiliation to be insignificant. Not only were the coefficients of relevant variables generally statistically insignificant, but calculations of average costs that included or excluded consideration of these variables produced numbers that were almost identical. (Cont. on p. 21) Regression Studies of Bank Operating Costs Before considering the implications of these findings for the policy questions delineated above, we review briefly the previously published regression studies of bank operations costs. These studies may be grouped according to the type of data used, which, in large measure, determined the researchers' definition and measurement of output. The first group used balance sheet and income statement data, with output defined as dollars of assets, while the other used FCA data, with output defined as number of accounts. Output as Dollars of Earning Assets Three of the six regression studies that use dollars of assets as output report economies of scale. (Two of these employed specifications that imparted a strong bias towards this result.) One study (Powers) found some evidence of diseconomies and two others (Schweitzer and Kalish-Gilbert) found strong evidence of U-shaped average cost curves, that is, decreasing costs at lower output levels and increasing costs at higher levels. The use of dollars of earning assets as the measure of output appears to have biased the findings of the studies away from showing higher costs for larger banks and deemphasizes the Ushape of the average cost curves We reach this conclusion in part because when our present study is recast roughly in terms of these previous analyses, the findings are rather similar to theirs. All the studies found branch banking organizations as having higher costs than unit banks. The two studies that included measures of holding company affiliation report conflicting findings. The first two studies that employed multiple regression analysis measured costs as total operating expenses (including interest) divided by year-end total assets (Schweiger and McGee (1961) and Gramley (1962)). This cost measure was regressed on bank size as measured by deposit size groups by Schweiger-McGee and total assets by Gramley. Other variables used to hold bank characteristics constant were the ratios of time to total deposits, various types of earning assets as a percent of total assets, and whether a bank was a unit or branch bank SchweigerMcGee used all member bank data for 1959 and Gramley used a sample of Tenth (Kansas City) District member banks Both studies found evidence of large economies of scale. Schweiger-McGee also reported higher costs per dollar of assets for branch banks (Gramley's sample included only unit banks). Unfortunately, the use of costs per dollar of assets to measure unit costs imparted strong bias toward the finding of economies of scale. The use of dollars of assets as the output measure requires the implicit and incorrect assumption that large and small loans are equally costly per dollar outstanding. Thus banks with larger average loan sizes appear more efficient. Equally important is the fact that neither study included variables that could identify U-shaped cost curves nor made a formal determination that a U-shaped curve was not necessary to properly describe the data they used. Four studies defined output as loans and securities with deposit services being considered an input. Greenbaum (1967), Powers (1969) and Schweitzer (1972), and Kalish and Gilbert (1973) reasoned that the "social" value of these earning assets is equal to the amount people are willing to pay to obtain them. Hence, except for Powers, they weighted each bank's dollars of loans and securities outstanding by the average gross yields on these assets overall banks in the sample to abstract from the effect of local market conditions. Powers used the rates charged at each bank rather than the average rate for all banks because he believed that a loan's social value is measured best by its actual local price. Thus output is equal to gross revenue or a variant thereof. Cost is defined as total operating cost, including interest, less fees for deposit services. 14 Greenbaum (1967) analyzed 1961 data of 745 and 413 insured banks in the Tenth (Kansas City) and Fifth (Richmond) Federal Districts. He found U-shaped average costs in both districts; however, 98 percent of the banks are on the portion of the cost curve that showed economies of scale. The Fifth District sample was analyzed separately for branch and unit banks When thus separated, both groups have downward sloping average costs. The unit banks have lower and more rapidly declining costs than the branch banks, but the unit banks are much smaller. Greenbaum concludes that banks grow larger by branching because large unit banks may be more costly to operate. Powers (1969) analyzed 1962 data from all insured banks in the Seventh (Chicago) Reserve District. He ' " G r e e n b a u m d i d not d e d u c t deposit service fees from costs. FEDERAL RESERVE B A N K O F A T L A N T A 17 stratified his sample of 2,411 banks into 24 subsamples, according to asset size, current operating revenue to total output (which he defined as total revenue), lending output to total output, and branch versus unit. Of these sub-samples, five have some suggestion of a U-shaped cost curve within the range of the data With respect to the branch versus unit dichotomy, Powers finds that the branch banking organizations generally (but far from uniformly) have higher average costs than do unit banks. Schweitzer (1972) analyzed 1964 data from 1,325 Ninth (Minneapolis) District banks. Unlike Greenbaum and Powers, he included the following additional variables: time deposit rate paid, two dummy variables for city location to account for input price differences among banks, and dummy variables measuring the effect of Federal Reserve membership and bank holding company affiliation. The data were sub-categorized into four size of loans and securities groups. Schweitzer reports "scale economies for banks with total assets under $3.5 million, constant returns to scale for banks between $3.5 and $25.0 million, and decreasing returns for larger banks." (p. 265). He also found significant economies associated with large holding company affiliation, except for banks over $25.0 million in assets. Small bank holding company affiliation indicated economies that are not statistically significant. Kalish and Gilbert (1973) analyzed 1968 data from 898 FCA banks, using a method that traced out costs with observations from the most efficient banks The banks were separated into sub-samples of 86 holding company affiliates, 353 branch banks and 460 unit banks. They report U-shaped average cost curves for all three groups, with unit banks having the lowest costs (except for the very largest banks). Branch banks above $1 million in adjusted revenue (or $10 million in loans and investments) show lower costs than holding company subsidiaries and smaller branch banks had higher costs than similar sized holding company affiliates. The four studies that allowed for finding of diseconomies of large scale found them. 15 However, the evidence of eventually upward sloping curves is slight in Greenbaum's study and mixed in Powers' study. The four studies that included measures of branch versus unit banking report higher costs for branch banks. Results on holding company affiliation are conflicting. Using 1964 data Schweitzer found holding company affiliates to have lower costs (Schweitzer, 1964 data); using 1968 data Kalish and Gilbert found them"to have higher costs (Kalish and Gilbert, 1968 data). Before these results can be accepted as meaningful, at least with respect to the banks and time periods analyzed, we must consider the extent to which the more important misspecifications may bias these findings. The three principal problems are (1) ,5 S c h w e i g e r - M c G e e a n d Gramley e m p l o y e d different m e t h o d s which, we' feel, biased their measures of scale (as discussed above). 18 exclusion of variables that account for differences in input prices and types of deposit services, (2) inclusion of interest in operating expenses, and, most important, (3) output defined as dollars of assets outstanding. The exclusion of input price differences in the studies does not appear to be serious, at least with respect to the calculation of elasticities and average costs, where the banks are ordered by size and by type of organization (branch and unit). We calculated these numbers for our samples holding these factor prices constant over the entire sample; the changes in the scale economy estimates and average costs were slight and did not affect any of the conclusions. However, ignoring differences in types of deposits is likely to be serious, since about half of the total operating expenses (excluding interest) are due to deposits. 16 Furthermore, the proportion is higher at the small banks (those with deposits under $50 million): 53 percent compared to 49 percent for banks with $50 to $200 million in deposits and 46 percent for banks with over $200 million in deposits. The smaller banks also have relatively more demand deposit expense (39 percent versus 36 and 35 percent). This misspecif¡cation probably overstates the average costs of smaller banks. Exclusion of interest from operating expenses might understate the costs of large banks, since they generally purchase a greater proportion of theirfunds in the market. We tested this hypothesis by including interest in operating expenses in our data and recalculating the cost functions. The recalculations showed greater economies of scale for all the banks (See Table 1, column 3). With interest included and the Divisia number of accounts as the measure of output, the larger banks (over $25 million in deposits) in unit states had nearly constant costs. The smaller (under $25 million in deposits) branch banks had statistically significant economies of scale, and the medium-sized ($50 million to $300 million in deposits) banks in branching states had small, but statistically significant, diseconomies of scale. Thus, inclusion of interest could account, in part, for the larger economies or lower diseconomies of scale for larger banks found in the other studies. Finally, measuring output with assets (weighted by revenues in most of the studies) may impart a bias for banks that service larger accounts showing them as being more efficient. We tested for this bias by recalculating the elasticities and average costs with dollars of deposits and loans as the output measure instead of the number of accounts and average size of account. These elasticities are considerably lower for all size groups of the unit banks and for all but the larger(up to $200 million to $300 million in deposits) branch banks (see Table 1, column 2). With interest included as part of operating expenses, the middle-sized banks in both unit and branch states show diseconomies of scale, a finding similar to those reported in the studies that ,6 1 9 7 8 Functional Cost Analysis data. N O V E M B E R 1982, E C O N O M I C REVIEW similarly measured cost and output (see Table 1, column 4). Thus using dollars of deposits is equivalent to measuring output by the number of accounts while assuming that the size of accounts is uniform among banks. This assumption biases the elasticity estimates "Measuring output with assets may impart a bias for banks that service larger accounts, showing them as being more efficient." towards greater economies of scale particularly at the large unit banks. This effect also is reflected in the average cost per dollar of deposits and loans (not shown): the cost curve is'less U-shaped. Output as Numbers of Accounts and Average Size of Accounts in a Cobb-Douglas Production Function Benston (1965A) is the first study to use the FCA data, which permitted him to specify output as the average number of accounts serviced and the average size of accounts, and to analyze costs of separate banking functions. He analyzed each of the five functions—demand deposits, time and savings deposits, real estate loans, installment loans, business loans and securities—and overhead costs. Overhead was regressed on total assets rather than the humber and average size of accounts For most of the analyses Benston used a functional form that can show only continously rising, falling, or flat average cost curves. But by including higher powers on the output measure he also tested for the presence of U-shaped cost curves. This study used data from some 80 banks in the Boston Federal Reserve District for 1959, 1960 and 1961 and found economies of scale for all except the business loan function and the overhead costs Slight U-shaped curves were present for installment loans only. The estimated elasticities (averaged over the three years) ranged from .74 (securities) and .87 (demand deposits) to 1.0 (business development and overhead). However, the largest bank included had $55 million in assets(1961 price level), so the data did not cover the full range of bank output. Branch banks were found to have higher operating costs than unit banks that serviced the same number of accounts (Benston 1965B), but because the cost effect of branching was misspecified the validity of this finding is questionable. Bell and Murphy (1968) extended Benston's work considerably. They analyzed data from 210 to 283 banks in the Boston, New York and Philadelphia Federal Reserve Districts that reported FCA data for FEDERAL RESERVE B A N K O F A T L A N T A 1 9 6 3 , 1 9 6 4 and 1965. They did not explicitly test the data for the presence of U-shaped cost curves since they explicitly based their analysis on the CobbDouglas production function. They estimated elasticities (averaged over the three years) ranging from .83 (securities) and .86 (real estate loans) to 1.0 (safe deposits and trust department). They constructed an overall measure of economies of scale by weighting the individual elasticities by the average cost of each function and type of overhead expense. This yielded a total elasticity of .93. Branch banks were found to have higher operating costs than unit banks, but Bell and Murphy calculated that expansion of branch banks did not generate greater average costs, since the higher cost of branching, in their calculations was offset almost exactly by economies from a larger volume of output. Murphy (1972) replicated and further extended this work with a larger sample (967 banks) of large banks (up to $5.5 billion in deposits) nationwide for 1968. The only important difference between the elasticities measured with these data compared to the earlier (1965) results is the change from significant economies of scale in demand deposits to approximately constant costs. Most of the other functions showed only small economies of scale, so that the estimated total elasticity for the 1968 sample is .95. Longbrake and Haslem (1975) also used these data (1968 FCA banks, nationwide), but restricted the analysis to direct demand deposit costs, emphasizing the effect on these costs of organizations structure. They separated the sample of 767 banks into four sub-sets: unit banks and branch banks, each affiliated or not affiliated with holding companies. They found statistically significant economies of scale (elasticities of .93 for unit affiliates and branch non-affiliates and .82 for branch affiliates) for all except the unit bank non-affiliates, which had constant costs (elasticity of 1.03). In general, the branch banks (particularly the larger ones) had higher estimated costs per account or dollar of deposit than did the unit banks, the affiliated unit banks had higher costs than the unaffiliated unit banks, but the affiliated branch banks had lower costs than the unaffiliated branch banks. Mullineaux (1975) also separated a 1970 sample of 196 FCA banks in the Boston, New York and Philadelphia Reserve Districts, into branch (167) and unit (29) banks, to test the hypothesis that the estimated coefficients differ by type of organization. Elasticities were estimated for each of the five principal banking functions and the hypothesis was accepted. For most functions, unit banks had economies of scale and branch banks had diseconomies. As the previous studies found, branch banks appeared to have higher operating expenses than similar unit banks. Benston and Hanweck(1977) reported preliminary findings of a study of nationwide FCA banks for the years 1968 through 1974 (904 banks in 1974). The direct costs of the five principal banking functions were analyzed separately. Unlike the other studies which used the Cobb-Douglas cost function, a quadratic cost function was estimated which permitted estimation 19 of U-shaped cost curves.17 They report finding scale economies for most years only for business loans. Little indication of systematic economies or diseconomies was found for the real estate, installment loan or time deposit functions. The demand deposits function, though, exhibited diseconomies of scale starting at low output levels. The effect of holding company affiliation was tested by separating the sample into affiliated and independent banks. The affiliated banks exhibited somewhat lesser scale economies than the independents. But average costs per account at the affiliates were lower, particularly with respect to demand deposits. For 1974, demand deposits costs averaged 4 percent lower for banks affiliated with a small holding company and 17 percent lower for banks affiliated with a large holding company. Branch and unit banks were not separately identified and this misspecification may have confounded the results somewhat. Finally, a recent study by Dunham (1982) used 1978 FCA data for three Federal Reserve Districts (Boston, New York and Chicago). The Cobb-Douglas form was used and separate estimation was performed for five banking functions—demand deposits, time deposits, commercial and agricultural loans, installment loans and mortgage loans. Operating costs were adjusted for correspondent services paid for with non-interest bearing balances by imputing a cost based on the three-month Treasury bill rate applied to net "due from" balances. Output was measured by the number of accounts for each function. When operating costs are not adjusted for the imputed cost of correspondent services, the demand deposit function exhibited constant costs while the other functions showed significant scale economies. After costs were adjusted, the demand deposit function also exhibited significant economies of scale. In addition, this study also found that branch banks had higher costs than unit banks with the same characteristics. Thus, the results are similar to those of Benston, Bell and Murphy. These studies offer some important insights into costs of individual bank outputs and changes in banks cost structure over time. The analyses indicated economies of scale in the early 1960s for all banking functions and departments except safe deposit boxes and trust. The late 1960s and early 1970s data revealed a change to constant costs for the important demand deposit function, at least for unit banks not affiliated with holding companies. Affiliated unit banks and branch banks were found to have economies of scale in demand deposit operations similar to those computed with earlier data Data for 1970 yielded approximately constant direct costs for branch banks but economies of scale for unit banks. However, one study that used data through 1974 found diseconomies pf scale. This study (Benston-Hanweck) is the only one using data covering a broad range of bank sizes that specified a cost function to test for U-shaped cost curves. Thus, the earlier findings of continuous slight economies of scale might be due to this misspecification or to a change over time in the cost function faced by commercial banks. It also is important to note that none of these studies tested for economies of scale for banks as entire entities. Where overhead costs were analyzed, these costs were regressed on output measured as total assets. Hence, the elasticities of banks that serviced larger average size accounts appear to be biased downward. This problem could have resulted in a finding of economies of scale or constant costs when, in fact, diseconomies of scale were experienced. To test for the possible misspecification effect of not allowing for U-shaped cost curves and of not analyzing the banks as an entire entity, and as a means of separating the effects of sample and specification differences, we reestimated our 1978 findings using the Cobb-Douglas cost function. This functional form yields the following elasticities: branch state banks, .93*, unit state banks 1.07*, and for the entire sample, 1.01 (an asterisk indicates elasticities statistically significantly different from 1.0—constant costs—at the 95 percent confidence level). These are almost the same as the mean elasticities computed with our more complex translog cost function which allows for a U-shaped cost curve. But, as Table 1, column 1 shows, the mean elasticity value masks the finding of lower elasticities for smaller banks and higherelasticities for larger banks. Combining unit and branch banks in a single sample also masks important and statistically significant differences between them. Furthermore, as the augmented measure of branch bank elasticities presented above shows, ignoring the way in which branch banks actually expand understates the elasticities experienced by branch banks when one wishes to determine the scale economy of the entire organization rather than of a single office. " T h e earlier findings of continuous slight e c o n o m i e s of scale might be d u e to this misspecification or to a c h a n g e over time in the cost function faced by c o m m e r c i a l banks." With respect to branch banking, most of the studies reported higher branching costs that appeared to be offset by measured economies of scale. Studies that calculated average costs found them to be higher for branch banks, with larger branching systems having even higher costs per account. Those findings are consistent with our more current study, with one exception. We found branch and unit banks to have very similar costs' per account when differences between them, in terms of their mode of expansion, were accounted for properly. " T h e variables w e r e divided by output to correct for heteroscedasticity 20 NOVEMBER 1982, E C O N O M I C REVIEW Implications of the Studies and Conclusion At this point, only a few implications can be drawn concerning commercial banks' operating costs. To date, w e are unable to weigh the possibilities and magnitudes of economies of scale or the cost savings from producing bank services jointly rather than separately. However, studies using data from 1959 through 1970 provide us with some evidence on banks' costs to produce individual services, such as demand deposits and installment loans. Unfortunately, the functional forms used appear t o have understated the costs for larger banks. Also, their old data may not reflect present cost conditions. Thus w e are unable at this time to go much beyond saying it is unlikely that there are large, if any, economies of scale in producing the most important banking services. (Business loans may be an exception.) costs per account, after we adjust for the higher cost of servicing customers w i t h larger accounts. Small branch and unit banks, though, appear t o be more efficient than larger banks. W e should emphasize that customer-borne costs, such as inconvenience, are not accounted for, nor is the benefit to customers of using collateral services offered by an individual bank. Presumably, the effect of these factors should be reflected in banks' revenues. Considering these caveats, our analysis of the presently available data leads us t o conclude that smaller banks are, at the least, not at an operational disadvantage with respect to large banks. As a result, mergers appear unlikely to result in operating cost savings. Whether or not other benefits or costs flow from a policy of freer entry, or branching, or a more liberal merger policy is beyond the scope of this analysis. —George J. Benston, Gerald A. Hanweck, and David B. Humphrey However, it does appear safe to conclude that branch and unit banks incur about the same Hanweck is an economist Reserve System. Humphrey Studies Section. with the Board of Governors, is chief of the Federal Reserve's Federal Financial REFERENCES 1. Alhadeff, David A M o n o p o l y a n d C o m p e t i t i o n in Banking. Berkeley, California: University of California Press, 1954. 13. Goldschmidt, Amnon " O n the Definition a n d M e a s u r e m e n t of Bank Output." Journal of Banking a n d Finance, 5 (1981), 575-585. 2. Barnett, William A "Divisia Indices." Encyclopedia of Statistical Sciences, New York, New York: Wiley a n d Sons, 1981. 14. Greenbaum, Stuart I. " B a n k i n g Structure a n d Costs: A Statistical S t u d y of the Cost-Output Relationship in Commercial Banking." National Banking Review, 4 (June, 1967), 415-34. 3. Bell, Frederick W „ a n d Neil B. Murphy Costs in C o m m e r c i a l Banking: A Quantitive Analysis of Bank Behavior a n d its Relation to Bank Regulation. Federal Reserve Bank of Boston, Research Report No. 41, 1968. 15. Gramley, Lyle E. A Study of Scale E c o n o m i e s in Banking. Federal Reserve Bank of Kansas City, 1962. 4. Benston, George J. " E c o n o m i e s of Scale a n d Marginal Costs in Banking Operations." T h e National Banking Review, 2 (June 1965A), 507-49. 16. Horvitz, Paul M., " E c o n o m i e s of Scale in Banking," Private Financial Institutions, Englewood Cliffs, N.J.: Prentice-Hall, Inc. 1963, 1-54. 5. Benston, G e o r g e J. " B r a n c h Banking a n d Economies of Scale," T h e J o u r n a l of Finance, XX (May 1965B), 312-31. 17. Kalish, Lionel a n d R. Alton Gilbert "An Analysis of Efficiency of Scale a n d Organizational Form in Commercial Banking." Journal of Industrial Economics, 21 (July 1973), 293-307. 6. Benston, G e o r g e J. " A c c o u n t i n g Numbers a n d Economic Values." T h e Antitrust Bulletin, XXVII (Spring 1982), 161-215. 7 . Benston, G e o r g e J., a n d Gerald A Hanweck "A Summary Report on Bank Holding C o m p a n y Affiliation a n d Economies of Scale." Proceedings of a C o n f e r e n c e o n Bank Structure a n d Competition, Federal Reserve Bank of Chicago, 1977. 8. Benston, George J., Gerald A Hanweck, a n d David B. Humphrey "Scale E c o n o m i e s in Banking: A Restructuring a n d Reassessment." Working Paper, Federal Reserve Board, November 1981. 9. Benston, G e o r g e J., Gerald A H a n w e c k a n d David B. Humphrey "Scale E c o n o m i e s in Banking: A Restructuring a n d Reassessment." J o u r n a l of M o n e y , Credit, a n d Banking, XIII (November 1982, Part I). 10. Diewert, W. Erwin "Exact a n d Superlative Index Numbers." J o u r n a l of Econometrics, 4 (May 1976), 115-145. 11. Durham, Constance " C o m m e r c i a l Bank Costs a n d Correspondent Banking." N e w England E c o n o m i c Review, Federal Reserve Bank of Boston, (Sept e m b e r / O c t o b e r 1981) 22-36. 18. Longbrake, William A a n d John A. Halstern "Productive Efficiency in Commercial Banking: The Effects of Size a n d Legal Form of Organization on t h e Cost of Producing Demand-Deposit Sen/ices."Journal of M o n e y , Credit a n d Banking, 7 (August 1975), 3 1 7 - 3 3 0 . 19. Mullineaux, Donald J. " E c o n o m i e s of Scale of Financial Institutions: A Comment." J o u r n a l of Monetary Economics, 1 (April 1975), 233-240. 20. Murphy, Neil B. "A Re-estimation of the Benston-Bell-Murphy Cost Functions f o r a Larger Sample with GreaterSize a n d Geographical Dispersion." J o u r n a l of Financial a n d Quantitative Analysis 7 (December 1972), 2 0 9 7 - 1 0 5 . 21. Powers, J o h n A " B r a n c h Versus Unit Banking; Bank Output and Cost Economies." Southern E c o n o m i c Journal, 3 6 (October 1969), 153-64. 22. Schweiger, Irving a n d J o h n S. M c G e e "Chicago Banking: The Structure a n d Performance of Banks a n d Related Financial Institutions in Chicago a n d Other A r e a s " Journal of Business, 3 4 (July 1961), 203-366. 23. Schweitzer, A A " E c o n o m i e s of Scale a n d Holding Company Affiliation in Banking." Southern E c o n o m i c Journal, 3 9 (October, 1972), 258-266. 12. Flannery, Mark J. "Correspondent Services a n d Cost Economies in Commercial Banking." Research Paper No. 77, Federal Reserve Bank of Philadelphia, November 1981. 21 FEDERAL RESERVE BANK O F A T L A N T A Economies of Scale: A Case Study of the Florida Savings and Loan Industry A study of Florida S&Ls shows that operating costs decrease as S&Ls increase in size up to $500 million in deposits. H o w does the structure of costs at savings and loan associations compare with that of commercial banks? In view of the trend toward deregulation, the large number of mergers of thrift institutions, proposals for new powers for thrift institutions, and prospects for increased competition and consolidation among all types of financial institutions, this is certainly not an academic question. This article reviews recent studies regarding savings and loan scale economies, discusses the profit implications of S&L consolidation, and then presents some new results based on a study of the S&L industry in Florida. Since the Florida S&L industry ranks second to California nationally in terms of assets, these results should also be reasonably representative of S&L costs in other areas of the country. S&L costs are both different from and similar to commercial bank costs. Operating costs are considerably lower at S&Ls, for instance, because these organizations have specialized in the lowcost functions of mortgage lending and servicing, and the collection and servicing of time deposits. This study indicates that S&Ls experience economies of scale up to about $500 million in assets. Except at small sizes, however, costs savings are not very large as size increases. If S&Ls could expand assets significantly without adding offices, they could capture important economies of scale beyond the $500 million asset level; however, most are not able to do this without expanding their office networks. 22 NOVEMBER 1982, E C O N O M I C REVIEW Measuring Economies of Scale Studies of operating costs typically have measured economies of scale by estimating the elasticity of cost with respect t o output. For example, if assets are used as a measure of bank or S&L output, the elasticity would be calculated as the percent change in cost divided by the percent change in assets. Thus, if a typical bank or S&L with $100 million in assets had operating expenses of $1 million, and one with $200 million in assets had expenses of $1.9 million, the elasticity would be calculated as follows: p e r c e n t change in costs p e r c e n t change in assets (1.9/1.0) - 1 (200/100) - 1 90% = 100% = °-9- If costs increase by less in percentage terms than the increase in asset size, the elasticity is less than one. In such a case average costs would decline as size increases, and economies of scale would exist in this industry. Elasticities that are greater than one indicate that average costs rise as size increases, in economic terms, "diseconomies of scale." Until a few years ago it was generally accepted that economies of scale existed at all types of financial institutions and were significant in most asset size ranges. For example, in a survey of the major studies published in 1972, Benston (4) concluded that, on average, the overall elasticity of cost with respect to output was about 0.93 for commercial banks. Roughly similar results had been found for S&Ls. This led Benston to conclude that banks and S&Ls have approximately the same cost structure. Recent studies, however, have called into question the conventional wisdom about economies of scale at financial institutions. For example, Gilligan, Smirlock and Marshall have concluded that, based on their investigation of the issue, " t h e empirical evidence does not indicate the existence of scale economies in banking, except for relatively low o u t p u t levels...(therefore) a public policy that attempts to increase bank scale through controlling entry or encouraging merger cannot be justified on the basis of cost savings" (11, p. 27). In a recent paper, Benston, Hanweck and Humphrey have also found evidence of diseconomies of scale (increasing expense ratios) for commercial banks at bank office sizes in excess of $25 million of deposits (5). Similar controversy surrounds the existence of economies of scale at credit unions (1 5). Profit Implications To put into perspective the differences between S&L and bank costs, w e should note that if economies of scale exist, the profit implications of changing size are somewhat different for S&Ls than for commercial banks. The typical S&L reported an operating cost ratio (operating costs divided by average assets) of 1.35 percent in 1980, while the typical commercial bank had a ratio approximately twice that amount. 1 This difference is a result of the much greater diversity of functions performed by banks. Demand deposits, consumer lending, and corporate lending involve high turnover and thus require constant activity on the part of a bank's staff just to keep asset and deposit size at a certain level. In contrast, the bulk of S&L assets are long-term mortgage loans. These require servicing but involve little additional operating cost unless they become d e l i n q u e n t In addition, savings deposits (the primary liability item for S&Ls) involve less operating costs per dollar than demand deposits. Table 1 shows a hypothetical situation in which a typical S&L and a typical commercial bank experience growth and enjoy cost savings as a result of economies of scale. Both are assumed t o have the same cost elasticity with respect t o output. Because of its much larger initial operating cost ratio, the commercial bank experiences a pre-tax profit gain of 12 basis points; for the S&L the effect on profits is only half as great. Previous Research Reviewed Earlier studies seem to indicate quite clearly that economies of scale do exist at savings and loan associations. The most recent study to confirm this finding was the Brookings study of the thrift industry, conducted by Andrew Carron (8). As indicated in Table 2, Carron's results (which were based on 1980 data) are reasonably consistent w i t h many earlier findings going back to the late 1960s. These studies suggest, however, that the cost savings from growth will not be particularly large. ' S e e the note to Table 1 for the precise figures a n d the sources of data. 23 FEDERAL RESERVE BANK O F A T L A N T A Table 1 . Hypothetical Effect of a Merger on Profitability Savings and Loan Association Commercial Bank Before Merger Operating Costs (millions) Assets (millions) Cost: Assets Elasticity of Cost With Respect to Output (Assets) $ 1.25 $100.00 1.25% 0.9 $ 2.50 $100.00 2.50% 0.9 After Merger Operating Cost (millions)1 $ 2.38 $ 4.75 Assets (millions) $200.00 $200.00 Cost/Assets 1.19% 2.38% Cost Savings (equals 6 basis points 12 basis points Pretax Profit Gain) 1 A 90% increase in costs, which is the result of the asumed elasticity of 0.9 and the 100% increase in assets. Note: The operating cost ratios are for illustrative purposes only. The actual ratio of operating e x p e n s e to average assets in 1980 was 2.73% for commercial banks a n d 1.35% for S&L's on a national average basis. Source: Calculated for commercial banks from FDIC, Bank Operating Statistics, 1980, a n d the Federal Reserve Bulletin, various issues; for S&L's from FHLBB, C o m b i n e d Financial Statements, 1 9 8 0 a n d miscellaneous news releases. The last column of the table indicates that a doubling of asset size (or other measure of S&L output) is expected to cut the operating cost ratio by less than 25 basis points. The typical estimate, in fact, is substantially less than this— approximately seven to 12 basis points. Cost savings from mergers actually may be even less. In most of these studies, when the relationship between asset size and operating cost was estimated, the number of branches was held constant Mergers, of course, do not conform to such statistical niceties—in a merger, both asset size and the number of offices necessarily increase. The larger number of offices w o u l d increase operating cost, partially offsetting the estimated cost savings shown in Table 2. In their 1969 study, Brigham and Pettit concluded that a merger would be expected t o produce cost savings as a result of economies of scale. They calculated that the total operating cost for a single S&L w i t h assets of $500 million and 9 branches would, in most cases, be about 1 5 to 35 percent less than the cost of operating ten $50 million unit S&Ls (associations without branches). However, in a study done at about the same time, Benston reported results suggesting that a merger would not produce any cost savings. More recently, Henry Cassidy has shown that the equations developed in Atkinson's 1977 study produced results more consistent with Benston's conclusions. He estimated that costs would be about 10 percent higher at an association with five branches than the total for five unit associations, each of which is one-fifth as large. However, the results were mixed, providing some support for the Brigham and Pettit conclusions. Thus, as Cassidy has indicated, the issue of whether a merger will reduce costs has yet to be resolved. In the t w o most recent studies shown in Table 2, a variable representing the number of offices was not held constant in the regression analysis. In other words, the number of branches was allowed to vary with asset size. Both of these studies found evidence of cost savings as size increases, lending support to the idea that a merger reduces cost. Such cost savings should be small, however, for mergers among institutions of roughly equal size. The ratio of operating costs to average assets for the S&L industry was 1.35 percent in 1980 and 1.42 percent in 1981. The estimated cost savings of seven to 12 basis points from doubling asset size, the result suggested by most of the studies reviewed here, would not improve operating expense ratios significantly. Merging a relatively small association into a large association, however, could offer an effective way of dealing with a profit squeeze at the smaller association. For example, estimated operating costs for associations in the $50 million asset range are about 1.4 percent of assets. This drops rapidly to a range of 1.0 percent to 1.1 percent as asset size reaches the $500 million to $1 billion range (12). This would imply a reduction of 30 percent to 40 percent in operating cost at the disappearing institution as a result of the merger. Operating Costs of Florida S&Ls Estimating the extent of economies of scale requires statistical analysis of data on operating cost and other variables at a cross section of associations. Operating cost includes all non- 24 NOVEMBER 1982, E C O N O M I C REVIEW Table 2. Summary of Previous Research Results on Economies of Scale at S&Ls Author, year of publication, and reference number Period studied and sample design Measures of S&L output utilized Elasticity estimate(s) Estimated reduction in the operating expense to assets ratio from a doubling in output' Atkinson, (1979) [1] 1975 (1,878 S&Lsnational sample) Total assets 0,84 to 0.91 7 to 11 basis points Atkinson, (1977) [2] 1974 (1,200 S&Ls7 states2) Total assets 0.86 8 to 9 basis points Benston (1969) [3] 1963-66 (3,159 S&Lsnational sample) (1) Number of loans made; (2) Number of loans serviced; and (3) Number of savings accounts 0.91 to 0.94 5 to 7 basis points Brigham and Pettit (1969) [6] 1962-66 (approximately 450 S&Ls-Chicago, Cleveland, Detroit, Los Angeles SMSAs) Total assets N.A. 12 to 17 basis points Carron (1981) [8] 1980 (approximately 4,000 S&Lsnational sample) Total assets N.A. 12 basis points 4 ' 5 McNulty (1981) [12] 1979 (approximately 360 S&Lssix area samples) Total assets .82 to .98 (median = .92) 10 to 24 basis points 5 Morris (1978) [13] 1976 (187 S&LArizona, California and Nevada) (1) Total loans closed (dollar amount) and (2) Number of savings accounts (1) 0.90 for loans closed (2) 0.98 for number of accounts (1) 6 basis points^ (2) Virtually no cost savings-1 for number of accounts Verbrugge, Shick and Thygerson (1976) [14] 1971-72 (478 S&Lsnational sample) Total assets N.A. 7 basis points 3 'Estimated by the author from the equations a n d / o r other information provided in each study 'Arizona, California, Colorado, Illinois, Kansas, O h i o a n d T e x a s ^Assuming that a doubling of lending activity, or other o u t p u t measure used in the sudy, results in a doubling of a s s e t s " B a s e d on unpublished results d o n e in c o n n e c t i o n with this study, as supplied by the author 5 B a s e d on an a s s u m e d increase in asset size from $ 2 5 0 million to $ 5 0 0 million. interest costs, such as wages and salaries, office occupancy expenses, advertising, fees for professional services, and so forth. This study is based on 1980 data for 117 Florida savings and loan associations. All insured associations file detailed financial reports with the Federal Home Loan FEDERAL RESERVE BANK O F A T L A N T A Bank Board twice a year, and these reports provided most of the required data. The sample excludes associations in existence less than two years. Previous studies have often used national data. However, many factors affecting operating cost 25 vary from area t o area. Associations operating in a similar market environment should provide a more appropriate basis of comparison. The industry is large and diverse in Florida and the range of operating cost ratios experienced by Florida S&Ls is similar to that for the country as a whole. Therefore, the Florida results can be generalized t o apply t o other areas of the country. Elsewhere in this Review, Benston et a/ have criticized the use of assets as a measure of total output for commercial banks. Their argument has to do with a bias introduced by the large variation in average loan and deposit size among commercial banks in different asset size groups. There is also some variation in average loan and deposit size among individual S&Ls, but it is no doubt much less, particularly in a sample restricted to one state. Thus the bias introduced by using assets as the output measure is likely t o be less important in S&L cost studies. On the surface, a simple statistical analysis of data on operating costs and total assets would seem to provide the necessary information. However, many factors besides asset size affect operating cost. These include the number of branches, the extent to which the association is involved in mortgage banking activity, the mix of savings accounts, the percentage of liabilities representing borrowed money, and wage and salary levels in the local market(s) served by the association. Through multiple regression analysis, a mathematical equation was fit to the data on operating cost, asset size, and 14 other variables likely to affect operating costs. This has the effect of holding these "other" factors constant, so that the relationship between size and operating cost can be measured. This S&L cost study is the first to utilize a regression form known as a"translog cost function." Analysts w h o have used the translog form prefer it because of its flexibility, since it does not impose any particular shape on the cost function. This particular functional form has been used increasingly in economies of scale studies during the past six years.2 The appendix contains a detailed description of the regression equation that was estimated, as well as a description of the "other" variables and estimates of their effect on operating cost. 2 O n e of the first studies was by Christensen a n d Greene (10). Another important study is Brown, Caves a n d Christensen (7). This form w a s also used in the two recent commercial bank studies c i t e d earlier (5,11). 26 Chart 1 . Estimated Operating Expense Ratios Florida Savings and Loan Associations, 1980 Percent A= F = 25 branches C = 10 branches G = 3 0 branches H = 3 5 branches D = 15 branches ' I = 4 0 branches J = 4 5 branches K = 5 0 branches E = 2 0 branches 1 branch B = 5 branches I I L Source: Table 3 Statistical Results Table 3 and Chart 1 show how the estimated operating cost ratios change as asset size changes. W e also calculated the normal number of offices for associations in various asset size groups. These results are shown in Table 4, which is based on an equation described in the appendix. Operating cost numbers are omitted from Table 3 for asset-branch combinations outside the normal range. For example, it is possible to estimate the operating cost ratio for an S&L with $2.5 billion in assets and one office. Since no such S&L exists, however, this calculation would not be meaningful, so is not shown in the table. On the surface, the results suggest the existence of substantial economies of scale. For example, an association with assets of $50 million and five offices has an estimated ratio of operating cost to average assets of 1.63 percent (Table 3). The ratio for an association with the same number of offices (five) but an asset size of $500 million would be only 0.94 percent. Similarly, for an association with 20 offices, the ratio goes from 1.24 percent to 0.92 percent as assets increase from $500 million to $2 billion. While these cost savings are substantial, they are not representative of what most associations can achieve. Few association managers can increase their asset size tenfold or even fourfold NOVEMBER 1982, E C O N O M I C REVIEW Table 3. Estimated Operating Expense Ratios Florida Savings and Loan Associations, 1980 Branches Assets $ millions) 25 50 100 200 250 400 500 750 800 1,000 1,250 1,500 1,750 2,000 2,250 2,500 2,750 3,000 1 1.913 1.552 1.235 .965 .888 5 10 15 1.633 1.411 1.196 1.130 .996 .935 1.670 1.494 1.312 1.254 1.132 1.075 .974 .958 .904 1.386 1.332 1.220 1.166 1.069 1.054 1.001 20 1.286 1.235 1.142 1.128 1.077 1.026 .985 .951 .921 25 1.203 1.188 1.139 1.089 1.049 1.015 .986 .960 30 1.144 1.104 1.070 1.041 1.016 .993 35 40 45 50 1.153 1.120 1.091 1.066 1.043 1.023 1.004 1.164 1.136 1.111 1.088 1.068 1.050 1.152 1.130 1.110 1.091 1.191 1.169 1.149 1.130 Source: Estimated from the translog cost model described in the appendix. The "other" factors influencing oeprating cost w e r e held constant at their average level in calculating the ratios. while keeping their number of offices constant, except perhaps over a long period of time. Furthermore, as noted earlier, these cost savings cannot be realized in a merger, which normally involves simply combining t w o branch systems into one. Table 5 provides a better perspective on the cost savings from mergers. The figures represent estimates of the change in the operating cost ratio from selected 50 percent and 100 percent increases in both asset size and the number of offices. For example, if an association with assets of $50 million and five offices were t o merge with another association of the same size, the operating cost ratio is estimated to decline from 1.63 percent to 1.49 percent, a decline of 14 basis points. It can be seen that the cost savings diminish rapidly once asset size reaches a certain level. For example, if t w o $500 million associations with identical branch systems are put together, average costs stay approximately the same. For a merger of two $1 billion institutions, costs actually increase slightly. Thus cost savings from a merger of t w o associations of similar size apparently disappear once the asset size of the resulting institution reaches $500 million. Thus, when branches are allowed t o vary along with assets, the estimated average cost curve for S&Ls is the traditional Ushaped curve. To check the reliability of these results, w e performed an additional test by estimating a model which directly related the operating expense ratio t o asset size and a number of other financial ratios. The results of this "ratio model" follow the general form of the results shown here, and clearly confirm the existence of economies of scale. The ratio model did show lower cost savings at small asset size levels, but costs continued to decline until asset size reached about $1 billion. At this point, both approaches predict little or no further cost savings when asset size and the number of offices are increased proportionately. How to Control Cost I n a study several years ago for the U.S. League of Savings Associations, Verbrugge et a/ (14) 27 FEDERAL RESERVE B A N K O F A T L A N T A 1 Table 4. Estimated Number of Offices for Associations of a Given Asset Size Florida Savings and Loan Associations, 1980 Assets {$ millions) Table 5. Estimated Cost Savings from Selected 50 and 100 Percent Increases in Asset Size and the Number of Branches Estimated Numberof Offices Average Range Assets From 50 100 250 500 750 1,000 1,250 1,500 1,750 2,000 2,250 2,500 5 6 8 12 16 20 23 27 31 35 39 43 1 -13 1 -14 1 -16 4-20 8-24 12-28 23-31 19-35 23-39 27-43 31 - 4 7 35-50 Source:Estimated from the Branch System Size e q u a t i o n described in the appendix. The range is t w o standard errors on each side of the regression line. This interval c o n t a i n s o v e r 9 5 % of the associations under consideration. found the operating cost ratio to be one of the four key financial ratios influencing association profit performance. The data in Table 3 suggest something important about the influence of branch offices on cost: with a given asset size, associations with fewer offices tend to experience much more favorable operating cost ratios. This would indicate that associations (and financial institutions in general) will want to be cautious in expanding their branch systems. Because of differences in their markets, and their aggressiveness in branching, the n u m b e r o f offices varies widely for associations of the same asset size. (Table 4 shows the ranges for the number of offices at which 95 percent of the associations in Florida were operating in 1980.) Associations operating at the lower end of these ranges can achieve operating cost ratios of 0.9 percent to 1.0 percent. However, associations operating at larger asset sizes, but in the middle or upper end of their ranges, experience much higher operating cost ratios. In addition, associations with a total number of offices in the lower end of these ranges can experience operating cost ratios as much as 25 percent lower than associations in the high end of the same range. 28 To No. of Offices From Operating Cost Ratio To From To Change 50 100 5 10 1.63 1.49 -0.14 100 100 200 200 5 10 10 20 1.41 1.31 1.49 1.44 -0.10 -0.05 200 200 400 400 5 10 10 20 1.20 1.13 1.31 1.29 -0.07 -0.02 250 250 500 500 5 10 10 20 1.13 1.08 1.25 1.24 -0.05 -0.01 500 750 10 15 1.08 1.07 -0.01 0.94 0.90 1.08 1.08 1.17 1.19 -0.04 500 500 500 1,000 1,000 1,000 5 10 15 10 20 30 750 750 750 1,500 1,500 1,500 10 15 20 20 30 40 0.97 0.99 1.07 1.10 1.14 1.20 +0.02 +0.03 +0.06 1,000 1,000 1,500 1,500 10 20 15 30 0.90 0.91 1.08 1.10 +0.01 +0.02 1,000 1,000 2,000 2,000 10 20 20 40 0.90 0.92 1.08 1.14 +0.02 +0.06 1,500 1,500 2,250 2,250 20 30 30 45 0.99 1.02 1.10 1.15 +0.03 +0.05 1,500 1,500 1,500 3,000 3,000 3,000 15 20 25 30 40 50 0.91 0.95 0.99 1.05 1.05 1.13 +0.04 +0.06 +0.08 0.00 +0.02 Source: Table 3 One way to control costs is t o limit the number of offices relative t o asset size. The data in Table 3 indicate that this may be an even more effective method of achieving low expense ratios than expanding assets.3 Clearly, the results suggest quite strongly that economies of scale cannot be achieved by aggressive branching, since this would tend t o push up the number of o f f i c e s more than in proportion to asset size. , , • Summary and Conclusion Previous research indicates that economies of scale exist in the savings and loan industry, but 3 Records maintained at t h e F H L B o f Atlanta indicate that, despite the severe profit squeeze, only a few associations in the Southeast have closed any offices within t h e past year. NOVEMBER 1982, E C O N O M I C REVIEW y that the potential cost savings from most types of consolidation are small. This was confirmed in an analysis of 1980 data on 117 savings and loan associations in Florida The estimated cost savings are substantial in movingfrom low asset sizes (for example, $50 million) to the $500 million level but are exhausted after asset size reaches $500 million. This conclusion applies t o proportional increases in asset size and the number of offices. Limiting the number of offices relative t o asset size appears t o be a more effective way of controlling cost than does unrestrained asset growth.^ The author would like to thank David Humphrey, Robert Ott, David Roddy, Philip Webster and James Zabel for helpful comments and discussion, and Kathryn Whitehead for statistical assistance on this paper. —James E. McNulty Assistant Vice President-Economist, Federal Home Loan Bank of Atlanta. REFERENCES 1. Atkinson, J a y F. "Firm Size in the Savings a n d Loan Industry." Invited Research W o r k i n g Paper No. 29, Federal H o m e Loan Bank Board (December 1979). 2 "The Structure of Cost in t h e Savings a n d Loan Industry During 1974." Research Working Paper No. 67, Federal H o m e Loan Bank Board ( M a r c h 1977). 3. Benâton, George J. "Cost of Operations a n d Economies of Scale in Savings a n d Loan A s s o c i a t i o n s " Study of t h e Savings a n d Loan Industry, Federal H o m e Loan Bank Board. Washington: U.S. Gove r n m e n t Printing Office, 1970, 677-761. 4 "Economies of Scale of Financial Institutions." Journal of M o n e y , Credit, a n d Banking (May 1972). 5. Benston, George J., Hanweck, Gerald A a n d Humphrey, David B. "Scale Economies in Banking: A Restructuring a n d R e a s s e s s m e n t " J o u r n a l of M o n e y Credit, a n d Banking (November 1982, forthcoming). 6. Brigham, Eugene F. a n d Pettit, R. Richardson, "Effects of Structure on Performance in the Savings a n d Loan Industry." Study of t h e Savings a n d Loan Industry, Federal H o m e Loan Bank Board. Washington: U.S. Government Printing Office, 1 9 7 0 9 7 1 - 1 2 0 9 . 7. Brown, Randall S., Caves, Douglas W. a n d Christensen, Laurits R. " M o d e l l i n g the Structure of Cost a n d Production for Multiproduct Firms." S o u t h e r n E c o n o m i c J o u r n a l (July 1979). 8. Carron, Andrew S. T h e Plight of t h e Thrift Institutions. Washington: Brookings Institution, 1982. 9. Cassidy, Henry J. "S&L Branching and Operating C o s t s " Research Working Paper No. 75, Federal H o m e Loan Bank Board (March 1978). 10. Christensen, Laurits R. and Greene, Willima H. "Economies of Scale in U.S. Electric Power Generation." J o u r n a l of Political E c o n o m y (August 1976). 11. Gilligan, T h o m a s W. Smirlock, Michael L. a n d Marshall, William J. "Cost Complementarities Scale Economies a n d Natural Monopoly in Banking." Federal Reserve Bank of Chicago. P r o c e e d i n g s of a C o n f e r e n c e o n Bank Structure a n d Competition, 1 9 8 2 (forthcoming; also available a s W o r k i n g Paper No. 82-5, S c h o o l of Business Administration, W a s h i n g t o n University, St. Louis). 12. McNulty, J a m e s E. " E c o n o m i e s of Scale in the S&L Industry: New Evidence a n d Implications for Profitability." Federal H o m e Loan Bank Board J o u r n a l (February 1981). 13. Morris James R. "Economies of Scale at District S & L s " Commentary, Federal H o m e Loan Bank of San Francisco (July 1978). 14. Verbrugge, James A Shick, Richard A a n d Thygerson, K e n n e t h J. "An Analysis of Savings a n d Loan Profit Performance." J o u r n a l of F i n a n c e (December 1976). 15. Wolken, John D. a n d Navratil, Frank J. "Economies of Scale in Credit Unions: Further Evidence." J o u r n a l of F i n a n c e J u n e 1980). 29 FEDERAL RESERVE B A N K O F A T L A N T A APPENDIX Translog M o d e l . This is the first S&L scale e c o n o m y s t u d y t o e m p l o y t h e t r a n s l o g cost model. This m o d e l has b e e n used in industrial sector studies by Christensen a n d G r e e n e (10) a n d Brown, C a v e s a n d C h r i s t e n s e n (7). It recently has b e e n applied to commercial banking by B e n s t o n , H a n w e c k a n d H u m p h r e y (5) a n d Gilligan, S m i r l o c k a n d M a r s h a l l (11). T h e f o r m of t h e e q u a t i o n is: log C = a + b log Q + c (log Q) 2 / 2 + d log B + e log Q • log B + 1 fi log Xj W h e r e C = o p e r a t i n g cost, Q = o u t p u t ( r e p r e s e n t e d by a s s e t s in t h i s study), B = n u m b e r of o f f i c e s a n d r e p r e s e n t s o t h e r f a c t o r s a f f e c t i n g cost. This f u n c t i o n a l f o r m is c o n s i d e r e d by r e s e a r c h e r s in this area to be a substantial improvement over t h e simple logarithmic function which has a constant elasticity, a n d t h u s d o e s not a l l o w t h e a v e r a g e c o s t c u r v e t o t u r n u p w a r d at s o m e point. T h e s i m p l e l o g a r i t h m i c f u n c t i o n " f o r c e s " e c o n o m i e s of s c a l e t o exist at all levels of o u t p u t , if t h e d a t a i n d i c a t e s that t h e y exist o n average. T h e d i s t i n g u i s h i n g f e a t u r e s of a t r a n s l o g f u n c t i o n , in contrast, are t h e (log Q ) 2 t e r m a n d t h e i n t e r a c t i o n t e r m (log Q - l o g B). (The a b o v e f o r m is slightly d i f f e r e n t f r o m a p u r e t r a n s l o g f u n c t i o n , w h i c h would require a separate squared term for each v a r i a b l e i n c l u d e d in an i n t e r a c t i o n t e r m In t h i s c a s e this w o u l d r e q u i r e i n c l u s i o n of a (log B ) 2 t e r m in t h e e q u a t i o n . H o w e v e r , a r e g r e s i o n w h i c h i n c l u d e d this v a r i a b l e p r o d u c e d r e s u l t s w h i c h w e r e not realistic.) R e s u l t s of t h i s t r a n s l o g f u n c t i o n , e s t i m a t e d w i t h d a t a o n 1 1 7 s a v i n g s a n d l o a n a s s o c i a t i o n s in Florida for t h e y e a r 1 9 8 0 , a r e s h o w n in T a b l e A-1. T h e " o t h e r f a c t o r s " a f f e c t i n g o p e r a t i n g c o s t are t h e v a r i a b l e s f o u n d t o b e s i g n i f i c a n t in an earlier s t u d y of o p e r a t i n g c o s t a n d s c a l e e c o n o m i e s by A t k i n s o n (1). All data, e x c e p t f o r t h e local a r e a w a g e rate, c o m e f r o m semia n n u a l r e p o r t s e a c h a s s o c i a t i o n files w i t h t h e F e d e r a l H o m e L o a n B a n k Board. T h e best available proxy f o r inter-area w a g e d i f f e r e n c e s is p r o b a b l y p e r - c a p i t a income, s o c o u n t y or ( w h e n a p p l i c a b l e ) S M S A perc a p i t a i n c o m e w a s u s e d as t h e w a g e variable. T h e p e r f o r m a n c e of t h e e q u a t i o n is impressive, w i t h e i g h t of t h e v a r i a b l e s statistically s i g n i f i c a n t w i t h t h e expected sign. Nonetheless, not too much importance s h o u l d b e p l a c e d o n t h e h i g h R 2 term. T h i s m e r e l y r e f l e c t s t h e fact that l a r g e r a s s o c i a t i o n s have a h i g h e r a b s o l u t e level of total o p e r a t i n g c o s t t h a n s m a l l e r a s s o c i a t i o n s , w h i c h w o u l d b e t r u e r e g a r d l e s s of t h e b e h a v i o r of a v e r a g e cost. N o n e t h e l e s s , it s h o u l d b e n o t e d that t h e c l o s e l y r e l a t e d F-statistic of 6 3 2 . 5 w a s t h e h i g h e s t of a n y e q u a t i o n that w a s tested. In t h e a b o v e m o d e l t h e elasticity of c o s t w i t h r e s p e c t t o o u t p u t c a n b e c a l c u l a t e d as f o l l o w s : T a b l e A - 1 . S u m m a r y of R e g r e s s i o n R e s u l t s Translog Cost Function Variable Coefficient t-Statistic Constant Log (Assets) (Log (Assets))2/21 Log (Total Number of Offices) (Log (Assets)) * (Log (Number of Offices)) Log (Wage Rate) Log (Scheduled Items)2 Log (Other Loans)3 Log (Loans Serviced For Others) Log (Loans Serviced By Others) Log (Borrowed Money) 4 Log (Passbook Savings) Log (Investment in Service Corporations) Stock Or Mutual 5 -7.69645 1.36783 -0.0385634 -1.26834 1.294 2.033 1.004 2.335 0.0733443 2.531 0.242230 -0.00112127 -0.0286108 0.00449619 2.423 0.2298 1.406 2.128 -0.00637744 2.765 0.00235612 0.0749267 0.00841176 0.7661 1.191 1.962 0.106537 2.114 2 R : 0.9876 R2: 0.9861 Durbin-Watson: 1.7018 F-Statistic (13,103): 632.5 ' U s i n g one half of t h e squared term simplifies the calculation of the elasticity. This has no effect on the other results, since t h e percent c h a n g e in the squared term is unaffected. S c h e d u l e d items include foreclosed real estate o w n e d a n d loans that are delinquent or in default, as well as loans to facilitate the sale of foreclosed real estate. 3 Other Loans include c o n s u m e r loans, education loans, loans on savings accounts, h o m e improvement loans a n d mobile h o m e loans. " B o r r o w e d M o n e y includes Federal H o m e Loan Bank Advances a n d other b o r r o w i n g s 5 Stock equals one, zero otherwise. Note: The d e p e n d e n t variable is total operating expense. All variables except the number of offices a n d the d u m m y variable are in dollar amounts. The asset variable is average assets, c o m p u t e d over t h r e e semiannual periods. Thus, t h e elasticity varies w i t h t h e level of o u t p u t a n d t h e n u m b e r of offices, r a t h e r t h a n r e m a i n i n g c o n s t a n t , as it w o u l d in a s i m p l e l o g a r i t h m i c f u n c t i o n . T a b l e A-2 s h o w s t h e e s t i m a t e d elasticity for s e l e c t e d asset sizes a n d n u m b e r of offices. It s h o u l d be n o t e d a g a i n that t h e s e c a l c u l a t i o n s a s s u m e t h a t it is p o s s i b l e t o i n c r e a s e a s s e t s w i t h o u t i n c r e a s i n g t h e n u m b e r of offices. It s h o u l d b e e m p h a s i z e d in t h i s c o n t e x t that t h e n e g a t i v e c o e f f i c i e n t of t h e B ( n u m b e r of offices) t e r m has n o m e a n i n g by itself. T h e e l a s t i c i t y of c o s t w i t h r e s p e c t to t h e n u m b e r of o f f i c e s is: d log C = d + e log Q d log B d log C = b + c log Q + e log B a log Q w h i c h is positive f o r all o u t p u t (asset sizes) levels e x c e p t for t h e very smallest. 30 N O V E M B E R 1982, E C O N O M I C REVIEW a » Table A-2. Estimated Elasticity of Cost with Respect to Output Table A-3. Summary of Regression Results Ratio Model Coefficient Variable Assets ($ millions) Number of Offices 50 50 100 100 250 250 500 500 1,000 1,000 1,500 1,500 2,000 2,000 2,500 2,500 1 10 1 10 5 15 5 20 10 25 20 35 20 40 30 50 Elasticity 0.684 0.853 0.657 0.826 0.740 0.821 0.713 0.815 0.738 0.805 0.773 0.814 0.762 0.812 0.783 0.820 Note: The elasticity has been calculated in the conventional way, which assumes that assets can be increased without an increase in the number of offices. Ratio Model. The results of this model are shown in Table A-3. This model was estimated to serve as a check on the results of the translog model. The fit cannot be compared directly with that of the other equation because the dependent variable is a ratio. However, the fit does compare favorably with ratio models estimated in other studies (e.g. (6), (8), (11), (14)). Nonetheless, there are less statistically significant coefficients than in the translog equation. As noted in the text, the ratios estimated from the ratio equation shown in Table A-3 do confirm the existence of economies of scale; however, economies are less in the under $500 million asset range, but they extend out to $1 billion when the effects of proportional increases in assets and in the number of offices are calculated. Branch System Size Equation. Table A-4 shows the results of a simple regression of the number of offices on asset size, for the same 117 associations. This equation was used to estimate the "average" number of offices for associations of a given asset size. From this we constructed a normal range for the number of offices at each asset size level (Table 4 in the text), so as to avoid reporting estimated cost ratios that would not be representative or meaningful. This normal range was set at two times the standard error of the estimate, on either side of the estimated value. With a standard error of estimate of 4, this led to a range of ± 8 offices on either side of the fitted value. The actual data were then evaluated, and over 95 percent of the observations fell within the confidence interval. Constant Assets (Assets)2 Number of Offices (Assets)*(Number of Offices t-Statistic 2.832 3.049 0.9413 0.1739 0.7927 0.609988 —6.78547(E-10) 1.27676(E-19) 0.00185660 7.14398(E-12) 4.86297(E-5) Wage Rate Ratio: Scheduled Items - 0 . 0 1 0 8 1 9 9 to Assets 0.00479122 Ratio: Loans Serviced For Others to Assets Ratio: Loans Serviced 0.00290794 By Others to Assets 0.00511514 Ratio: Other Loans to Assets 0.00670078 Ratio: Borrowed Money to Assets 0.0173610 Ratio: Passbook Savings to Total Savings -0.0294030 Ratio: Investment in Service Corporations to Assets 0.294797 Stock Or Mutual 2 R : 0.3557 R2: 0.2744 Durbin-Watson: 1.5851 F-Statistic (13,103): 4.374 2.325 0.2026 1.462 1.120 0.2304 1.018 4.433 0.3551 3.597 Note: The d e p e n d e n t variable is the ratio of operating cost to average assets expressed in percentage terms. All other variables are as defined in Table A-l. Table A-4. Summary of Regression Results Branch System Size Equation Variable Coefficient t-Statistic Constant 4.23264 8.330 Assets 1.57135(E-81) 20.420 R2: 0.7838 Durbin-Watson: 1.6535 F-Statistic: (1,115): 416.8 Standard Error of the Regression: 4.275 Note: The d e p e n d e n t variable is the number of offices. 31 FEDERAL RESERVE B A N K O F A T L A N T A Bank Size and Risk: A Note on the Evidence A review of the literature finds little evidence that small banks operate with greater risks than large banks. The relationship between risk and size of financial institutions may play a vital role in the evolution and stability of our financial system. Three possibilities exist: large banks are more risky than small banks, small banks are more risky than large banks, or large and small banks are equally risky. Everything else equal, the greater the risk of variation in income or of failure assumed by banks of various sizes, the greater the cost of capital and other liabilities that they need in order to grow. Therefore, the structure of our financial system (the numberand size distribution of financial institutions) will ultimately depend in part upon the relationship between bank size and risk. This extremely important issue has received little or no direct empirical testing. Most of the evidence on risk and size must be gleaned from side results of empirical studies of other problems such as capital adequacy, interest rate risk, or identification of potential problem banks. There are four main sources of risk for financial institutions. First, credit or default risk, the risk that borrowers will not repay or will not repay on schedule. Second, the risk associated w i t h variability of interest rates, which stems from the fact that financial intermediaries must borrow or at least pay interest on a majority of the funds they lend and may not be able to match maturities of the funds they lend against the funds they acquire. Third, operating risk, a measure of how 32 N O V E M B E R 1982, E C O N O M I C REVIEW well expenses and liquidity are managed. And fourth, the risk associated with fraud or insider abuse by management. Reviewing the empirical literature on each of these risks and their association with bank size indicates that we know far too little about bank size and risk. Credit Risk There is no direct empirical evidence or good indirect evidence showing a systematic relationship between a bank's size and its ability to control its credit risk position. There has not been a string of either large or small bank failures due t o credit risk, but there is evidence that large and small banks each face unique types of credit risk. Small banks may find themselves more geographically limited and hence more dependent on the health of a specific type of industry than d o large banks. Therefore, small bank exposure to unforeseen regional or national economic events may be greater than for large banks, which tend to have greater geographic and industry loan diversity. The vulnerability of small rural banks to problems in the farming industry was highlighted in 1977. Rural banks had liquidity problems because their loan portfolios were heavily directed toward farmers w h o were experiencing hard times resulting in their inability t o repay their loans. The dependence of small banks on their local economy is also demonstrated by the current recession (See 16). These risks, however, may be equalized by those faced by large banks which must keep current on large, diversified portfolios. Recently the problems of diversification have been graphically illustrated to banks w i t h international loans, (some loans to Poland and/or Mexico are being restructured). (See 10, 12). The only empirical study on the attitude of bank managers toward risk looked at the relative riskiness of bank holding company banks (which included almost all large banks) and independent banks (which are typically smaller banks). The study concluded that holding company bank managers take more credit risks (14). Interest Rate Risk Bank interest rate risk depends on the bank's vulnerability t o changes in interest rates. Unless banks are able to perfectly match interest rates and maturities of their liabilities and assets, some interest rate risk will exist. Researchers have found little if any systematic relationship between large and small banks and their interest rate risk exposure (5, 6, 7). Use of financial futures markets is one way to help control interest rate risk. Recent surveys of the use of financial futures markets by banks and savings and loan associations indicate that few small banks presently use these markets (3, 8). As banks increase in size, they are more likely to use the financial futures markets. This may be because small banks have less need t o hedge their portfolios or because they have some inherent disadvantage in hedging (1). Presently, however little quantitative evidence exists that there are differences between large and small banks in interest rate risk exposure. Operating Risk Financial institutions face the risks of poor expense control, poor product design and poor liquidity management that afflict all firms. The extent of these risks depends primarily on how well the institution is managed and on the complexity of the problems it faces. Management problems are likely t o be simpler in smaller organizations. Yet claims have often been made that smaller banks are less able to attract the best managers. In any event, there seems to be no strong tendency for problems associated with poor operating management to be associated with financial institutions of a particular size. Management Risk The inability t o control expenses due t o poor management is a serious problem, but a much more serious problem arises from management dishonesty and greed. These risks are manifest in fraud, forgery, insider transactions and dishonest acts by the bank's staff. Unfortunately, the bank failure and early warning system literature reveals that these risks are the primary cause of most bank failures. Joseph F. Sinkey, Jr. (18) writes, "For over 167 years, the major cause of bank failures, dishonest bank managers, basically has remained the same. The form has varied but the driving force has not changed." These problems are usually the most difficult to protect against. The importance of managerial risk is seen in the record of bank failures since 1960. From 1960 to May 1976,84 insured commercial banks failed. FDIC records indicate that the principal 33 FEDERAL RESERVE BANK O F A T L A N T A causes of failure of 45 of these banks were insider loans and out-of-territory loans (often connected with brokered deposits). Twenty-five more failures were principally caused by embezzlement or other management manipulation. Only 14 failures (17 percent of the total) resulted from bad loans made in the bank's local area (18). Overall Risk Two sets of empirical literature assess overall bank risk. The first is the capital adequacy literature covering the problems of determining how much capital is necessary to buffer banks against potential losses. The second is the literature on early warning systems which would allow regulators t o identify problem banks prior t o the problem. Both sets of studies are extensive, yet the size-risk relationship is seldom addressed. The capital adequacy literature shows that capital ratios are higher at small banks than at large banks. In addition it shows that regulators require higher capital ratios for smaller banks. Therefore, either banks abide by regulation or smaller banks in fact have more risk and compensate for this risk by holding more capital. Studies by Peltzman (15) and Mayne (11) indicate that regulations have little influence on bank capital ratios, while a study by Mingo (14) reaches the opposite conclusion. Wolkowitz (19) finds that small banks are inherently riskier but that they hold more capital to'compensate for that risk. Therefore, when management decisions are made, actual risk appears to balance out for small and large banks. Dince and Fortson (2) also find that bank size and capital adequacy are not related. The "early warning system" literature studies the characteristics of banks which have failed or which have appeared on a regulator problem list (See 4 for a summary). These studies generally controlled for bank size in the sampling process, which implicitly assumes that size is not important One of the early warning studies that did consider size explicitly found no relationship between bank size and vulnerability (9). Another study which did not control for size in its sample selection found that the characteristics which best distinguished problem banks did not include size (17). Conclusion There has been little systematic study of the size-risk relationship for financial institutions. M u c h of the evidence that exists is in the form of incidental evidence from studies of other aspects of these institutions. Although little is known, our review of the literature leads to t w o general conclusions. First, no systematic evidence exists that small banks are at a competitive disadvantage in terms of risk. O n this front, therefore, small banks appear to be on equal footing with large banks. And second, as banking competition becomes more intense, risk borne by banks will become increasingly important. The topic deserves much more attention than it has received. —David D. Whitehead and Robert L Schweitzer Schweitzer is assistant professor of University of economics, Delaware. REFERENCES 1. Dew, J a m e s Kurt. "David a n d Goliath: A Skirmish in the HedgeRows." A m e r i c a n Banker, VoL 147, (September 14, 1982) pp. 4-7. 2. Dince, Robert R. a n d J. C. Fortson. "The Use of Discriminant Analysis to Predict Capital Adequacy of C o m m e r c i a l B a n k s " J o u r n a l of Bank Research Vol. 3 , 1 9 7 2 , pp. 54-62. 3. Drabenstott, Mark a n d A n n e O. McDonley. "The Impact of Financia! Futures on Agricultural B a n k s " E c o n o m i c Review, Federal Reserve Bank of Kansas City, (May 1982), pp. 19-30. 10. Martin, S a r a h "The Secrets of the Polish Memorandum." E u r o m o n e y (August 1981), pp. 9-15. 11. Mayne, Lucille S. "Supervisory Influence on Bank Capital." J o u r n a l of F i n a n c e Vol. 2 7 (June 1971) pp. 6 3 7 - 6 5 1 . 12. " M e x i c o ' s Moratorium Puts Damper on C o m m e r c e (August 27, 1982) p. 5A. Market." J o u r n a l of 13. Mingo, J o h n J. " M a n a g e r i a l Motives, M a r k e t Structure a n d the Performance of Holding C o m p a n y B a n k s " E c o n o m i c Inquiry, Vol. 14 (September 1976) pp. 4 1 1 - 4 2 4 . 4. Eisenbeis, Robert A "Financial Early Warning Systems: Status a n d Future Directions." Issues in B a n k Regulation (Summer 1977), pp. 8-12. 14 . 5. Flannery, Mark J. " M a r k e t Interest Rates a n d Commercial Bank Profitability: An Empirical Investigation." J o u r n a l of F i n a n c e (December 1981), pp. 1 0 8 5 - 1 1 0 1 . 15. Peltzman, Sam. "Captial Investment in C o m m e r c i a l B a n k i n g a n d Its Relation t o Portfolio Regulation." J o u r n a l of Political E c o n o m y (January/February 1970), pp. 1 -26. 6 16. "Recession Has Bankers W a t c h i n g Loan Quality." ABA B a n k i n g Journal, Vol. 7 4 (July 1982). "The Impact of M a r k e t Rates on Small Commercial B a n k s " Rodney L White Center of Financial Research, W o r k i n g Paper No. 10-81, August 1981. 7. Hanweck, Gerald A a n d Thomas E Kilcollin. " B a n k Profitability a n d Interest Rate Risk." Research Papers in Banking a n d Financial Economics, Board of Governors of the Federal Reserve System, July 1981. 8. Koch, Donald L , Delores W. Steinhauser a n d Pamela Whigham. "Financial Futures as a Risk M a n a g e m e n t Tool for Banks a n d SaLs." E c o n o m i c Review, Federal Reserve Bank of Atlanta, Vol. 5 7 (September, 1982), pp. 4-14. 9. Korobow, Leon, David P. Stuhr a n d Daniel Martin, "A Probabilistic Approach to Early Warning of Changes in Bank Financial Condition." Financial Crises: Institutions a n d M a r k e t s in a Fragile Environment, ed. Edward I. Altman a n d Arnold W. Sametz (New http://fraser.stlouisfed.org/ York: J o h n Wiley a n d Sons) 1977. Federal Reserve Bank of St. Louis "Regulatory Influence on B a n k Capital Investment." J o u r n a l of F i n a n c e Vol. 130 (September 1975) pp. 1 1 1 1 - 1 1 2 1 . 17. Sinkey, J o s e p h F„ Jr. "A Multivariate Statistical Analysis of the Characteristics of Problem B a n k s " J o u r n a l of Finance, Vol. 3 0 (1975) pp. 21-36. 18. Sinkey, Joseph F. Jr. "Problem a n d Failed Banks, Bank Examinations, a n d Early Warning Systems: A Summary." F i n a n c i a l Crises: Institutions a n d M a r k e t s in a Fragile E n v i r o n m e n t ed. E d w a r d I. Altman a n d Arnold W. Sametz (New York: J o h n Wiley a n d Sons) 1977. 19. Wolkowitz, Benjamin. " M e a s u r i n g Bank S o u n d n e s s " in Bank Structure a n d C o m p e t i t i o n . Federal Reserve B a n k of Chicago, 1975. Changes in Large Banks' Market Shares From 1974 through 1981, larger banks in the Southeast generally lost market share to smaller local competitors. Evidence strongly supports that conclusion for other areas of the country over the past 15 years. Several factors help determine the competitive position of individual banks. Size, risk assumed, management strategies, and past behavior are thought t o play important roles. Studies of these factors allow us to assemble evidence on their impact t o project how a particular type of bank will perform. An alternative way to study the impact of these factors, particularly size, is t o observe how a certain group of banks performs in a total environment without isolating individual factors. Evidence on bank operating costs and risk indicates that, above a relatively small size, commercial banks do not gain operating efficiency or reduce risk t o any significant degree. Other factors that might give large banks significant competitive advantages, such as economies of scope, the ability to invest in innovations, the ability t o ride out errors, the concern of regulators to keep them from failing, have received less study. Students of banking find it difficult to put together all the evidence to project or explain banks' market performance. Banks, on the other hand, put all of these factors together in their markets every day. Operating costs, risk, regulatory compliance costs, innovation, regulatory attitudes and other factors influence each day's 35 FEDERAL RESERVE BANK O F A T L A N T A operations. One way to determine whether larger banks enjoy significant competitive advantages in local markets is to study their market performance. If larger banks have (and use) advantages over smaller banks, w e would expect t h e m t o gain market share at the expense of smaller banks. Lower costs of operations, risk taking, and the ability t o handle the expense of developing new products or t o ride out errors would allow larger banks to offer lower prices or higher quality service than smaller competitors. If they did so, then customers w o u l d gravitate t o them from their smaller competitors. But studies indicate that personal and business customers are loyal t o their financial institutions. Nevertheless, turnover in both business and consumer markets w o u l d allow banks using price or service advantages to gain business relative t o their competitors over an extended period of time. Evidence on the actual market performance of large and small institutions comes from recent direct studies of the subject and from another group of studies by economists interested in competition in local banking markets. These latter economists have performed several studies of changes in the market shares of larger banks in these markets. While these studies were not conducted specifically t o test the relative performance of larger and smaller banks, the evidence presented in them is relevant. Studies indicate that smaller institutions generally have not been a t a disadvantage relative t o large competitors over the past decade and a half. This seems true in all sizes of markets and in each geographic area studied. Evidence comes from a variety of empirical work already published and is confirmed by new work on banking markets in the Southeast presented here for the first time. • • • • • • • • • • • • ^ • • • • • • • ^ ^ • • • • • • • • • • • • • • • • • • • H H "Studies indicate that smaller institutions generally have not been at a disadvantage relative to large competitors over the past decade and a half." Our new evidence also indicates that, although large banks generally have lost market share in 36 local markets, they have not been losing local market share because outside institutions are taking business away from them rather than from smaller banks. Nor do they seem t o be losing share because they are refraining from using their competitive advantages so as t o keep prices and profits high. Studies of the Performance of Small Institutions Two recent studies of the performance of small financial institutions (11) (12) during 19781980 found no evidence that smaller institutions are not viable competitors. Both studies deal with institutions in Standard Metropolitan Statistical Areas (SMSAs); one covers commercial banks, the other savings and loan associations. The authors of each compare performance of small firms w i t h that of larger ones in the same economic environment. Although smaller banks and S&Ls report somewhat lower returns, they are also clearly less risky. The studies also find that, while there has been no difference in the growth rates of small and large S&Ls, smaller banks have grown faster than larger ones. 1 Studies of Large Banks' Market Shares Another set of relevant studies looks at changes in the market shares of larger banks relative t o their smaller competitors. The most comprehensive of these, covering 213 SMSAs and 233 large non-SMSA counties between 1966 and 1975 (16), recorded changes in market share of the three largest banking organizations in each area. If their share rose, it would offer some indication that large banks enjoy significant advantages. In general, however, their share declined rather than increased. Of the SMSAs, 86 percent recorded a falling share for the largest firms; of the nonSMSA counties, 79 percent recorded a falling share. The average three bank share declined from 75.8 t o 69.3 percent in the SMSAs and from 81.2 t o 78.4 percent in the non-SMSA counties. The study also found that the three largest banks lost most in markets where their 1966 share was greatest. It did not, however, screen out the concurrent influence of other factors on market share change. ' F o r o l d e r s t u d i e s of t h i s issue — w h i c h r e a c h similar c o n c l u s i o n s s e e (1) a n d (8). NOVEMBER 1982, E C O N O M I C REVIEW Another national study of market structure changes covering a smaller sample of markets produced similar results. The Rhoades (9) study covered the 1966-1976 period and included only 152 SMSAs that had not had their boundaries changed during the period and a sample of 129 non-SMSA counties. Over the period and t w o subperiods, more than 80 percent of the SMSA markets showed declining in market shares for the largest three banks. In 71 percent of the county markets, the largest three banks also lost share. Two regional studies - one from the Midwest and one from the Southeast - confirm the findings of the national studies. The most recent of these covers 53 SMSAs and 233 non-SMSA counties with more than three banking organizations in Illinois, Indiana, Iowa, Michigan and Wisconsin (2). This study covers smaller areas and more recent experience (1965-1979) than the national studies summarized above, but its conclusions are quite similar. Of the SMSAs, 85 percent recorded declines in the combined market share of the three largest banks; of the non-SMSA counties, only 53 percent recorded declining concentration. As in the national studies, concentration declined t o a greater extent in areas with higher initial concentration. Another regional study was carried on at the Federal Reserve Bank of Atlanta in 1976 (18) and is the springboard for new empirical evidence presented in the next section of this article. That study covered 98 banking markets in Alabama, Florida and Tennessee during the 1970-1974 period. Its principal purpose was t o determine if market concentration had increased in markets entered by multi-bank holding companies; however, it also presented evidence on the general issue of large bank market performance. A comparison of 1970 and 1974 three-bank concentration in 98 markets w i t h three or more banks in 1970 indicated that the larger banks had lost share in 62 (63 percent) of the markets and only maintained share in 26 more. They had gained market share in only 10 markets. In the 75 markets w i t h five or more banks in 1970—that is, markets in which there are smaller banks available t o compete throughout the period—the largest three banks lost share in 59 (79 percent) and managed only t o maintain shares in seven more. As was the case in other studies, large banks in markets w i t h higher initial concentration were likely to lose a greater market share (7). Table 1 . Concentration Change, Markets with Five Banks or More in June, 1981 State Alabama Florida Georgia Tennessee Total Number of Markets with Decrease Total Number Percent 29 19 65.5 24 29 82.8 69.2 13 9 14 12 85.7 85 64 75.3 with Increase Number Percent 10 34.5 5 17.2 4 30.8 2 14.3 21 24.7 Recent Evidence from the Southeast Our latest study adds both markets and time to evidence from Alabama, Florida and Tennessee. It covers the period from 1974 t o 1981 and includes markets from Georgia as well as the three states covered previously. Eighty-four markets with five banking organizations or more are included. The market areas are those used by the Board of Governors of the Federal Reserve System in decisions on bank holding company acquisitions and bank mergers. They are defined on the basis of study of banking patterns in local areas and are updated on the basis of changing local conditions (1 5 and 17). The study excludes eight markets for which the market definition was changed after 1974. Each of these markets was redefined during the study period t o include an expanded geographic area. This in itself resulted in a decline in the market share held by the largest banks. The markets were excluded t o avoid any bias toward a general conclusion that large bank market shares were declining. The evidence from Sixth District markets is summarized in Table 1 and 2. The three largest banks lost market share in over three quarters of these markets, losing in each state and in markets of all sizes. The average share held by the three largest banks declined from 76.8 percent in 1974 t o 72.2 percent in 1981 —a drop of almost 6 percent. As Table 2 indicates, the three largest banks' share declined by almost 9 percent in Florida markets but by only 1.2 percent in Georgia markets. If changes in bank operations and competition during the 1970s have changed large banks' performance, results of this most recent study 37 FEDERAL RESERVE BANK O F A T L A N T A Table 2. Concentration Change, Markets with Five Banks or More Average Concentration Change 1974-1981 State 1974 1981 In Average Alabama Florida Georgia Tennessee J59 .768 .804 .757 .727 .699 .794 .694 -.032 -.069 -.010 -.063 Percent in Average -4.217 -8.984 -1.244 -8.322 Total .768 .722 -.046 -5.989 should be expected t o differ from previous results. Yet results are quite similar t o previous studies. Explanations for Large Banks' Losses Even if larger banks have lost market share, it does not show conclusively that they have suffered competitive disadvantages. At least t w o alternative explanations are possible. First, outside competitors may be taking more business from larger banks than from smaller ones. Nonlocal and non-bank competitors are not included in measures of local market size and share because data on their local business generally is not available. Consequently, large banks may appear t o lose share to small ones w h e n they are losing to non-local and non-bank competitors. Second, one may argue that large banks refrain from capitalizing on their advantages to charge lower prices, pay more for deposits or provide higher quality services in order to gain higher profits. If this were so, one might find these banks retaining or losing their market share rather than gaining. W e tested for these two alternative explanations and found that neither holds up well. If nonbank and nonlocal competitors have entered markets and taken business away from larger banks w e would expect their entry to have its greatest impact in the most attractive markets. Multivariate tests of the determinants of changes in larger banks' shares indicate that neither market size nor market growth—two indicators of a market's attractiveness—was related t o the decline in large bank shares in southeastern markets. (See the Appendix for an explanation of the tests.) Average Percent Per Market -3.758 -8.686 -1.184 -8.138 5.732 The consistency of the study results also implies that competitors from outside the banking industry and local markets have not differentially affected large bank shares. Markets of all sizes have shown a decline in large banks' share over all time periods since 1965. Yet rapid expansion of nonlocal and nonbank competition has been rather recent. Had this expansion caused large banks to lose more local market share, declining shares would have shown up in later studies. This has not been the case. Evidence does not entirely rule out the possibility that large banks refrained from using their advantages in order t o earn greater profits. However, no supporting evidence has been found in Southeast markets. I n three of the other studies, tabular analysis indicated that the three largest banks lost the highest share in markets where they had held the highest initial share. This type of market performance supports one reason advanced for the ability of smaller banks to compete with larger ones. Large banks may refrain from exploiting some of their competitive advantages if they are able t o earn long-run profits by doing so. Three previous studies of concentration change indicate that small banks gain more ground when large banks hold a greater market share—that is, when they have more incentive to charge higher prices and/or provide less quality. The other studies did not, however, account for other factors that also might have influenced large banks' share. Our study tested a more detailed multivariate model and found no relationship between changes in large banks' share and the level of their share (see Appendix.) Our tests indicate that large banks probably have lost market share in local markets because they have been at a competitive disadvantage of 38 NOVEMBER 1982, E C O N O M I C REVIEW some sort. Neither differential effects of outside and nonbank competitors nor large banks' reluctance t o use advantages seem t o explain their loss of market share. A final element of the multivariate model gives a clue t o the identity of the banks that gained from large bank losses and suggests at least one dimension of large bank disadvantages. Our tests found that the entry of new banks into local markets was closely related to large banks' loss of market share. The introduction of new banks was followed by greater market share loss for large banks. This finding is consistent w i t h results of a studies of de novo entry by Rose and Savage (13 and 14). They found that new banks whether independent or "Our tests found that the entry of new banks into local markets was closely related to large banks' loss of market share." started by bank holding companies made significant contributions t o decreasing concentration of local market deposits in larger banks. That new banks should gain market share seems reasonable for several reasons. Their organizers would not start them nor w o u l d regulators approve them w i t h o u t considerable confidence that they w o u l d attract profitable business, that is, gain market share. In addition, new banks are often organized by investors w h o d o substantial banking themselves and w h o move their business to the new institution. Finally, most new banks in larger markets have opened in suburban areas that grow more quickly than the d o w n t o w n areas that are headquarters of larger banks. An interesting extension of this study w o u l d be an examination of the market shares of smaller banks that existed in local markets at the beginning of our study period. Did they also lose share to new banks or did they also gain share at the expense of larger banks? Studies reported here consistently indicate that in the recent past smaller banks have performed at better than par with larger ones in local markets. The smaller banks have been about as profitable (when profits are adjusted for risk) and have generally gained market share. Attributing this phenomenon to mismeasurement of market share and noncompetitive behavior of large banks does not seem t o fit. It seems more likely that larger banks have been at a competitive disadvantage in relation to smaller banks in some basic product lines. —B. Frank King REFERENCES 1. Darnell, J e r o m e C. a n d H o w a r d Keen, Jr. "Small B a n k Survival: Is the Wolf at t h e Door?" Business Review Federal Reserve Bank of Philadelphia (November 1974) pp. 16-23. 10. Rhoades, S t e p h e n A. "Structure a n d Performance Studies in Banking: A S u m m a r y a n d Evaluation." Staff Economic Studies, No. 92, Board of Governors of the Federal Reserve System, 1977. 2. Erdevig, Eleanor. "District Trends in Banking Concentration." E c o n o m i c Perspectives, Federal Reserve B a n k of Chicago, Vol. 5 (March/April 1981) pp. 6-12. 11. Rhoades, S t e p h e n A a n d Donald T. Savage. " C a n Small B a n k s Compete?" T h e Bankers M a g a z i n e (January/February 1981) pp. 5965. 3. Farrar, D . E a n d R.R. Glauber. "Multi-collinearity in Regression Analysis: The Problem Revisited." R e v i e w of E c o n o m i c s a n d Statistics, Vol. 4 9 (February 1967), pp. 92-107. 12. Rhoades, S t e p h e n A a n d Donald T. Savage. "The Performance of Small versus Large Savings and Loan Associations: Can the Small Associations Survive?" T h e B a n k e r s M a g a z i n e (forthcoming). 4. Heggestad, Arnold A. a n d S t e p h e n A. Rhoades. "An Analysis of C h a n g e s in Bank M a r k e t Structure." Atlantic E c o n o m i c Journal, Fall 1976, pp. 64-69. 13. Rose, J o h n T. a n d D o n a l d T. Savage. " B a n k Entry a n d Market Share Redistribution." S e p t e m b e r 1 9 8 2 (mimeo). 5. Hooks, Donald L. a n d T e r r e n c e F. Martell. " M u l t i b a n k Holding Company Acquisitions a n d Local Market Structure: An Analysis of Pooled Cross Section a n d Time Series D a t a " Research Paper 81 0 1 0 , Federal Reserve Bank of S t Louis, 1981. 14. Rose, J o h n T. a n d D o n a l d T. Savage. " B a n k H o l d i n g C o m p a n y Entry and Banking Market Déconcentration." Journal of Bank Research, Vol. 13 ( S u m m e r 1982), pp. 9 6 - 1 0 0 . 15. Schweitzer, Paul R. "The Definition of B a n k i n g Markets." Banking Law Journal, Vol. 9 0 (September 1973). 6. King, B. Frank. " C h a n g e s in Seller C o n c e n t r a t i o n in Banking Markets." Working Paper, Federal Reserve Bank of Atlanta (March 1977). 16. Talley, S a m u e l H. " R e c e n t Trends in Local B a n k i n g M a r k e t Structure." Staff Economic Studies, No. 8 9 Board of Governors of t h e Federal Reserve System. 7. King, B. F r a n k "Entry, Exit, a n d M a r k e t Structure C h a n g e in Banking." W o r k i n g Paper, Federal Reserve Bank of Atlanta (March 1979). 17. Whitehead, David D. "Relevant Geographic Banking Markets. How S h o u l d They Be Defined?" E c o n o m i c Review Federal Reserve Bank of Atlanta (January/February 1980), pp. 20-29. 8. Kohn, Ernest. T h e F u t u r e of S m a l l Banks. Albany, N.Y.: N e w York State B a n k i n g Department, 1966. 18. Whitehead, David D. a n d B. Frank King. " M u l t i b a n k Holding C o m p a n i e s a n d Local Market Concentration." Monthly Review, Federal Reserve Bank of Atlanta (April 1976) pp. 34-43. 9. Rhoades, Stephen A "Geographic Expansion of Banks a n d Changes in Banking Structure." Staff Economic Studies No. 102, Federal Reserve Board. 39 FEDERAL RESERVE B A N K O F A T L A N T A APPENDIX To provide some evidence on the relationship of market power of the largest banks to changes in their shares, we developed and tested a model of concentration change. It follows the development in Working Papers previously published by the Federal Reserve Bank of Atlanta (6 and 7). As explained in the text, large bank forebearance might be thought to result from use of market power even though large banks were more efficient. One would expect such forebearance to be more likely the larger the large banks' market share. Thus one feature of the model is market share of the market's three largest banks measured in the beginning year, 1974. Market growth may also influence large banks' share in two ways. First, by providing new business opportunities and attracting migration, growth brings in bank customers not previously attached to a local bank. Second, growth may attract new bank competitors that apparently take market share from large banks. (See 6) Market growth has two dimensions in this case: the percentage change in market size and the absolute change in size. In order to capture both aspects, variables for percentage change in market deposits and for total market deposits were used. effect, but some studies find multi-bank holding companies both increasing and decreasing concentration of business in larger banks. A variable to measure the change in the number of multibank holding companies operating in each market was included in the model. These variables were regressed against the percentage change in concentration in the local markets used in this study. The table below gives the results of this multiple linear regression. Only one factor was closely related to changes in the market shares of the three largest banks. That was the change in the number of organizations competing in the market. Greater increases in the number of competitors—greater entry—were associated with greater declines in large bank shares. Market growth, size, concentration and bank holding company activity were not significantly related to concentration change. 1 The equation explained almost 30 percent of concentration change, and the relationship was highly significant. This level of explanation is quite satisfactory in view of the slowness with which changing market conditions appear to be felt in market We would also expect large banks' market share to be influenced over time by changes in the number of firms competing in the market. Additional firms would be expected to enter only if they could take market share from larger firms; exiting firms would give larger firms an opportunity to increase their share. Bank holding company activity has also been discussed as an influence on concentration (See (5) and the next article in this Review, for summary of the evidence.) Evidence of holding company effects is mixe d A majority of studies find no bank holding company ' W e c o n s i d e r e d the possibility that c h a n g e in t h e n u m b e r of c o m p e t i t o r s a n d market g r o w t h might be closely related, c a u s i n g problems related to multicollinearity. P e r c e n t a g e market growth a n d market size explained less than 7 percent of the c h a n g e in the number of competitors in m a r k e t s in this sample. In addition exclusion of the market growth variable from the model h a d only minor influence on other coefficients. Both facts indicate minimal multicollinearity. See (3). Appendix Table Concentration Change Model: Regression Results Variable Deposit growth 1974-81 (percent, annual average) Deposits 1974 (billion $) Three organization concentration, 1974 Change in number of competing organizations, 1974-1981 (percent) Change in number of multibank holding companies represented, 1974-1981 Constant Regression Coefficient Standard Error -.213 .182 -1.169 -1.435 1.039 -1.380 184 6.661 .027 -184 —5.009a .036 -245 1.039 .360 -.882 5.273 -.167 Dependent variable: Change in three-organization concentration, 1 9 7 4 - 1 9 8 1 (percent) "R2 — . 2 5 1 a a D i f f e r s f r o m 0 at .01 level. 40 N O V E M B E R 1 9 8 2 , E C O N O M I C REVIEW The Impact of Local Market Entry by Large Bank Holding Companies Five years after selected large bank holding companies entered new local markets, their subsidiaries showed no significant advantages over comparable independent banks. In fact, after seven years, the independents had gained market share on the larger holding company banks. If there are advantages conferred by size in banking, they may not be limited t o large individual institutions. Subsidiaries of larger multi-institution organizations may have competitive advantages over smaller independent organizations. Three states in the Sixth Federal Reserve District have large multibank holding companies that acquired banks during the 1970s. If membership in large organizations conferred cost or product advantages on an acquired bank, its market share would grow and/or its returns w o u l d increase relative t o competitors. W e w o u l d expect banks acquired by the largest organizations in District states to gain market share or have higher profits than similar independent banks. This article reports on a study that compared the performance of t w o groups of Sixth District banks acquired by large bank holding companies with the performance of their independent counterparts. W e looked at a group of banks started c/e novo (not acquired by merger or acquisition) by the four largest bank holding companies in Alabama, Florida and Tennessee between 1972 and 1975 and a group of larger banks—the largest or second largest in their markets—acquired by the four largest bank holding companies in the same states during those years. W e compared the de novo banks' market share and profit performance w i t h that of a control group of independent de novo banks chartered in their markets during the same period. W e compared the larger banks' performance with that of larger independent banks in their markets. Both groups of banks were followed for several years after holding company acquisition—the small banks for five years and the large ones for seven. Our study indicates that large bank holding companies did not offer significant advantages to the banks they started or acquired, even several years after acquisition. De novo banks started by holding companies did enjoy higher assets, deposits and rates of returns than independents after five years of operation, but this seems to be the result of special circumstances in which the holding companies merged their smaller de novo banks into other subsidiaries. Larger banks ". . . large bank holding companies did not offer significant advantages to the banks that they started or acquired, even several years after acquisition." acquired by large holding companies lost market share to independent competitors through seven years after acquisition. They also lost ground in rates of return and in risk. Our results cover a longer period after acquisition than d o most studies of banks acquired by holding companies, but the results seem quite consistent with other studies. Evidence From Other Studies Previous studies indicate little advantage for subsidiaries of multibank holding companies as a whole. Four groups of studies are relevant. A fairly extensive volume of literature compares the performance of banks acquired by multibank companies w i t h independent banks of similar size in the same markets. (See Currey (2) for a detailed summary.) The studies are consistent in their major conclusions that bank holding company acquisition results in some changes in acquired banks' asset portfolios and increases in their operating expenses and income. The net effect is that these changes produced no significant changes in return on equity of acquired banks relative t o independents. Acquired banks, however, have been leveraged t o a greater extent. Studies that look at growth of acquired and independent banks find no difference between the two. Another set of studies culminates in the survey of bank costs detailed in this Review. These studies generally conclude that subsidiaries of multibank holding companies enjoy no cost advantages over banks that are not holding company subsidiaries. A third set of studies analyzes changes in concentration in markets entered by bank holding companies. (A highly concentrated market is one where a small number of firms hold a large share of the market.) These studies attempt to determine whether banking business in markets entered by multibank holding companies becomes more concentrated in a few firms after entry (4, 6, 7, 8, 10, 11, 17, 18, 21). A majority of these studies conclude that bank holding company entry has no impact on the concentration of deposits in the entered market. There are, however, contradictory studies concluding that concentration is either increased or decreased by holding company entry. A study of concentration change in 228 SMSAs by Heggestad and Rhoades (4) concludes that concentration became greater in markets entered by multibank companies. O n the other hand, three studies of Alabama markets by Hooks and Martell (6, 7, 8) and a case study of Colorado markets by Schweitzer and Greene (17) find declining concentration after entry by multibank companies. Whitehead's study of the impact of large bank holding companies on local market performance concludes that large companies tend to influence market prices if not market structure (20). The mixed results of these studies should not be surprising. Bank holding companies enter markets in many ways likely t o impact concentration differently. If acquisition confers advantages, w e would expect the market share held by 42 N O V E M B E R 1982, E C O N O M I C REVIEW large banks to increase if a large bank is acquired but to shrink if a small or de novo bank is acquired. Thus the manner of entry would be crucial to the impact of bank holding company acquisition. Yet only t w o of the studies considered that determinant. A more relevant set of studies looks at market performance of individual banks acquired by multibank holding companies. These studies follow the acquired banks for several years. In a study of 71 banks acquired between 1965 and 1970, Goldberg (3) found no significant change in market shares. His results were mirrored in Burke's broader study of 227 banks acquired between 1962 and 1970 (1). Burke reported some subtle tendency for larger banks to lose market share and smaller ones t o gain share. In a series of studies, Rose and Savage examined the performance of bank holding company subsidiaries after acquisition. These studies emphasize market share changes. Rose and Savage found that relatively large banks acquired by holding companies not otherwise represented in their market lost share and that small ones gained share (12). De novo banks acquired by bank holding companies performed no differently from independent de novo banks in aspects other than market share (15). I ndependents had a large edge in market share in less concentrated markets but holding company banks had a slight edge in highly concentrated markets (13 and 14). One case study that addresses this question found minimal impact from large New York City banks' entry into upper N e w York state after branching and bank holding restrictions were removed. The large New York City banks' market penetration was modest and their competitors' performance did not suffer (9). In a study of banks acquired by t w o Florida companies, Hoffman found no significant increase (or decrease) in the market shares of acquired banks relative to a control group of independent banks in their market (5). These studies find little evidence of holding company impact on market share. The studies are, however, inadequate in one way or another. Only Hoffman's study includes a control group, and it covers a relatively short period and only two companies. The studies by Goldberg and Burke use no control group of independent banks t o isolate bank holding effects. Despite their limitations, the studies reviewed here cast serious doubt on the proposition that multibank holding companies confer enough advantages on their subsidiaries to a l l o w t h e m t o make substantial inroads on independent competitors. The studies are neither conclusive nor— with the exception of the studies by Rose and Savage—without fault; their weaknesses in today's world relate to their treatment of all multibank companies as the same, their lack of coverage of recent years and their lack of control groups. If holding company size is important, only large companies may confer advantages. Recent innovations may have increased large companies' ability to help their subsidiaries. W i t h o u t a control group of independent banks, w e do not know whether bank holding companies or some other factor accounts for acquired banks' performance. Our study looks only at banks acquired by the largest multibank organizations in their states. It follows these banks over most of a decade to the present, and provides a control group of independent banks against which to test the holding company subsidiaries. New Evidence From the Southeast Large bank holding companies entered many markets in Alabama, Florida and Tennessee during the early 1970s and have competed in these markets since then. To test whether acquistion by these large organizations conferred advantages that allowed acquired banks t o gain at the expense of independent banks, we selected t w o extreme groups of banks acquired by the four largest bank holding companies in each state and paired them w i t h similar independent banks in their markets. W e then traced three major elements of performance from the acquisition during the 1972-1975 period to recent dates. To capture crucial aspects of holding company influence, we studied de novo acquisitions and acquisitions of the largest or second largest banks in the relevant markets. De novo acquisition is in many ways the purest type of holding company acquisition. The acquired bank begins life as a subsidiary. All future performance is under holding company influence and there is no past t o boost or drag down the bank. The acquiring company has no one t o blame (or congratulate) but itself for the bank's performance. 43 FEDERAL RESERVE B A N K O F A T L A N T A Acquisition of one of a market's largest banks may leave a holding company w i t h a residue of past management's brilliance or mistakes, but it also gives the acquiring company the potential to exercise market power. Thus this type of acquisition w o u l d seem likely t o confer advantages on organizations that already had operated successfully in a local market During 1972-1975, the four largest bank holding companies in each of the states of Alabama, Florida and Tennessee acquired 26 de novo banks for which w e could find matches of independent de novo banks in the same local market. 1 All but t w o of these pairs were in Florida. During the same period, companies in Alabama, Florida and Tennessee acquired 13 banks that were the largest or second largest banks in their markets and could be matched with an independent in their market that was also of that rank.These t w o sets of pairs were studied. W e measured three aspects of performance. To get an overall indication of relative performance, we studied market share differences. If one type of institution possessed advantages , over another, then w e would expect it to widen the gap between its market share and that of the other type of institution. W e tested differences between acquired and independent de novo banks. W e followed each pair of banks up through 1981 or until one of the pair changed its status by being acquired, merged or divested. Since each de novo pair started from scratch, we tested for significant differences in assets, deposits, rates of return on assets and equity, capital t o total assets and capital t o risk assets one, three and five years after acquisition. Each larger bank started with its own established market share and earnings and risk ratios from the time of acquisition through one, three, five and seven years. W e tested for differences in market share after acquisition; that is, whether gaps in performance widened or narrowed. Profits were analyzed because market share might be gained by a bank willing to sacrifice returns by pricing lower to attract customers. Models of this type of behavior have been •In o r d e r to be a m a t c h , t h e i n d e p e n d e n t must have b e g u n o p e r a t i o n no more t h a n a year b e f o r e o r after t h e h o l d i n g c o m p a n y b a n k a n d h a d t o b e l o c a t e d in t h e s a m e local market. 44 Table 1 . Pairs of Banks Remaining After Acquisition Years After Acquisition One Three Five Seven De novo Banks Larger Banks 26 25 14 7 13 13 12 12 developed to analyze the N O W account experience in New England and the behavior of southeastern banks (19). The situations tested in this study do not closely parallel the New England experience; however, systematic differences in profitability over time may well cast doubts upon the long-term viability of a group of institutions. W e tested differences in both return on assets and return on equity. Finally, profitability difference may be related to the risk taken by institutions. Consequently, we tested for risk differences among institutions by looking at capital t o assets and capital t o risk assets ratios. Our samples began w i t h 26 pairs of de novo banks and 13 pairs of larger banks. Over time, mergers of sample banks removed pairs from the sample. Florida—a unit banking state before 1975—authorized countywide branching in 1975 and statewide branching by merger in 1980. Large bank holding companies in Florida chose to merge their unit banks into larger multioffice institutions after 1975, removing some of their de novo banks and one large bank from our samples. The number of remaining pairs is shown in Table 1. Our test followed de novo banks for five years after acquisition and larger banks for seven years. New Banks As the first t w o panels of Table 2 indicate, holding company and independent de novo banks performed much the same during the first three years after the holding company acquisition. Differences between the groups are small enough in any case that one cannot say they were not accounted for by chance. Holding company subsidiaries were somewhat larger after one year but somewhat smaller than independents after three years. N O V E M B E R 1982, E C O N O M I C REVIEW T a b l e 2 . P e r f o r m a n c e of D e N o v o B a n k s Table 3. Large Bank Performance Performance Measure Variable BHC Mean Independent Mean Difference 5.63 4.60 5.55 4.32 .08 .28 -.58 -.63 .05 -2.57 -2.24 -.33 29.36 30.95 -1.59 132.57 129.43 3.14 End of Third Year After Charter Year (n=25) Assets (millions S) Deposits (millions $) Return on Assets (percent) Return on Equity (percent) Equity to Assets (percent) Equity to Risk Assets (percent) 11.92 10.55 12.97 10.92 -1.05 -.37 -.37 -.25 -.12 -3.00 -1.77 -1.23 11.06 16.75 -5.69 16.33 22.90 -6.57 End of Fifth Year After Charter Year Assets (millions $) Deposits (millions $) Return on Assets (percent) Return on Equity (percent) Equity to Assets (percent) Equity to Risk Assets (percent) Independent Mean Difference* YearO End of Charter Year (n=26) Assets (millions $) Deposits (millions $) Return on Assets (percent) Return on Equity (percent) Equity to Assets (percent) Equity to Risk Assets (percent) BHC Mean 21.78 19.52 13.9 12.42 7.88° .142 -.221 .363 1.434 -6.924 8.360 b 8.37 9.98 -1.615 12.77 15.18 -2.405 7.09® a Differs from 0 at .005 level (two tailed test) b Differs from 0 at .025 level (two tailed test) "-Differs from 0 at .05 level (two tailed test) The independent de novo banks had lower operating losses and took less risk through three years. The differences are not statistically significant, however. By the fifth year after holding company acquisition, the subsidiaries appear to have established a considerable advantage over independent banks. Deposits and assets of the holding company de novos are significantly larger than those of independents. Holding company banks have become profitable while paired independents are still losing. These results however, are not indicative of bank holding company advantages. During the same period between the third and fifth year after acquisition, Florida's holding companies chose to merge nine of the Assets (millions $) Market Share (%) Return on Assets Return on Equity Equity to Assets Equity to Risk Assets 41.80 30.70 1.07 14.45 7.43 9.42 46.56 35.38 1.05 13.75 7.67 9.71 -4.76 -4.68 .02 .70 -.24 -.29 Year 7 Assets (millions $) Market Share (%) Return on Assets Return on Equity Equity to Assets Equity to Risk Assets 62.32 29.74 1.02 11.64 8.81 10.14 82.85 40.01 1.20 13.24 9.11 10.92 -20.53 -10.27 -.18 -1.60 -.30 -.78 *BHC mean minus independent mean. sample's de novo banks into other subsidiaries. (Three of the independent banks were also merged into other banks. One was part of a pair with a merged holding company bank.) Banks that the holding company merged were significantly smallerand less profitable than those they did not merge. The fifth year results thus indicate a choice by the holding companies to eliminate smaller, less profitable banks rather than demonstrating competitive advantages by the remaining banks. In o r d e r t o see if this selection by bank holding companies influenced results of our tests in the fifth year, we tested third year differences in only the pairs remaining after five years. These tests showed that third year results for the remaining pairs were quite similar to fifth year results. Overall, there is little indication that de novo banks chartered by large bank holding companies possess significant advantages over independent de novo banks. In the three years before holding company mergers eliminated their smaller, less profitable subsidiaries from the sample, there was no difference in assets, deposits, profitability or risk that could not be accounted for by chance. Large Banks Large banks acquired by bank holding companies lost ground to their paired independent banks in the seven years covered in this study. The changing relative position of bank holding company subsidiaries is shown in Table 3. At year 45 FEDERAL RESERVE BANK O F ATLANTA Table 4. Relative Performance of Larger Bank Holding Company Acquisitions Mean Difference* Variable End of Acquisition Year n = 1 3 Market Share Return on Assets Return on Equity Equity to Assets -.377 .075 .047 .530 c End of Third Year After Acqusition n = 1 3 Equity to Risk Assets Market Share Return on Assets Return on Equity Equity to Assets Equity to Risk Assets .340 —2.248d -.124 -2.614d .920 3.304 b End of Fifth Year After Acquisition n = 1 2 Market Share Return on Assets Return on Equity Equity to Assets Equity to Risk Assets -1.909d -.215 -2.247 -.0255 -.935 End of Seventh Year After Acquisition n = 1 2 Market Share Return on Assets Return on Equity Equity to Assets Equity to Risk Assets -4.970* -.222 -2.291 -.0811 -1.043 »Changes in bank holding c o m p a n y data minus c h a n g e s in independents' data, stated in percentage p o i n t s a D i f f e r s f r o m 0 at .005 level (two-tailed test) ^Differs from 0 at .01 level (two-tailed test) c D i f f e r s from 0 at .025 level (two-tailed test) d D i f f e r s from 0 at .05 level (two-tailed test) end before acquisition—when all banks in the sample were independent—banks that remained independents had a 4.68 percent market share advantage on the banks acquired by bank holding companies. The independents were less profitable and somewhat less risky. By the seventh year after acquisition, the independent banks had gained market share while the holding company subsidiaries had lost share. The independents enjoyed a 10.27 percent market share advantage, reported greater returns on assets and equity and remained less risky. Larger independents gained on the larger banks acquired by bank holding companies several years after acquisition. As Table 4 indicates, the independents and holding company subsidiaries had not, with one exception, changed in significantly different ways during the first year after acquisition. Duringthe first yearthe holding company banks raised their equity-to-assets ratio significantly. During the three years after acquisition the independents' relative increases in market share and return on assets were statistically significant. Bank holding company subsidiaries increased equity-to-risk assets significantly more than independents. In the first five years after acquisition, the gap between independents' market share and that of holding company banks widened t o a statistically significant extent. The gap also widened in the same direction in each of the other variables but not to a statistically significant extent. After seven years the independents had established statistically significant gains in market share only. Their gains in return on assets and equity and equity-to-assets and risk assets were greater than those of the subsidiary banks but not significantly so. On their face, these results indicate that independents gained market share t o a significantly greater extent than bank holding company subsidiaries in the same markets without sacrificing either returns or safety. The position of the holding company as a resource for its subsidiaries makes statements about returns and risk somewhat ambiguous, however. Through various devices, holding companies may reduce profits in their subsidiaries by charging various flows to the parent company to tax deductible expenses rather than dividends. At the same time parent companies may bear risks for subsidiaries. Thus, relatively lower reported returns and risk for bank holding company subsidiaries may indicate a downward bias in the reporting of returns and risk. N o such downward bias exists in reporting holding company banks' market shares. Summary and Implications This study indicates that acquisitions by large bank holding companies of de novo or larger banks have not helped these banks increase their market share relative to similar independent banks over a period of several years.2 2 A l t h o u g h this s t u d y c o n c e n t r a t e s on large h o l d i n g c o m p a n i e s , t h e evid e n c e may apply to all c o m p a n i e s . Rose a n d S a v a g e (12) p r e s e n t e v i d e n c e that h o l d i n g c o m p a n y s i z e has little impact on p e r f o r m a n c e of large companies de novo banks. 46 NOVEMBER 1982, E C O N O M I C REVIEW "These larger (bank holding) organizations do not seem to be in a position to drive independent banks out of business." • • ^ • H H M B B H B I i ^ H H M H B B H B H B M B H B H H i Indeed, the larger banks that remained independent fared better than those acquired by holding companies, at least in market share. Results indicate also that large independents did not gain share at the expense of higher profits or lower risk. These results are consistent with most other evidence on the advantages of multibank holding companies. This study improves on the others by observing acquired banks over a longer period—up t o seven years—and by providing a specific control group, studying market share, and bringing the experience to the present all at the same time. Its samples, however, are relatively small and confined t o three states. Additional work should be done. If one accepts the findings of this study and most similar studies and projects them into the future, one finds implications on several fronts. On-site results seem consistent w i t h cost study results, which indicate no bank holding company advantages. These larger organizations d o not seem t o be in a position t o drive independent banks out of business. Their entry w o u l d seem unlikely to have the dire effects predicted by some opponents of bank holding company expansion at both state and national levels. In addition, their seeming inability t o raise returns on assets and equity at acquired banks casts doubts on their capability to muster resources needed to acquire large numbers of smaller —B. Frank King REFERENCES 1. Burke, James. " B a n k Holding C o m p a n y Behavior a n d Structural Change." J o u r n a l of Bank Research, Vol. 9 (Spring 1978), pp. 4351. 11. Rose, J o h n T. " B a n k i n g Holding C o m p a n y Affiliation a n d Market Share Performance." J o u r n a l of M o n e t a r y Economics, Vol. 9 (January 1982), pp. 110-119. 2. Curry Timothy J. "The Performance of Bank Holding Companies" T h e Bank H o l d i n g C o m p a n y M o v e m e n t t o 1 9 7 8 . A Compendium. Washington, Board of Governors of the Federal Reserve System, 1978, pp. 95-120. 12. Rose, John T. and Donald T. Savage. "Bank Holding Company Performance a n d Holding C o m p a n y Size." August 1 9 8 2 (mimeo). 3. Goldberg, L a w r e n c e G. " B a n k Holding C o m p a n y Acquisitions a n d Their Impact on Market S h a r e a " J o u r n a l of M o n e y , C r e d i t a n d Banking, Vol. 3 (February 1976), pp. 127-30. 4. Heggestad, Arnold a n d S t e p h e n A Rhoades "An Analysis of C h a n g e s in Bank Market Structure." Atlantic E c o n o m i c J o u r n a l Vol. 4 (Fall 1976), pp. 64-69. 5. Hoffman, Stuart G. "The Impact of Holding Company Affiliation on Bank P e r f o r m a n c e : A C a s e S t u d y of T w o F l o r i d a M u l t i b a n k Holding Companies." Research Paper, Federal Reserve Bank of Atlanta (January 1976). 6. Hooks, D o n a l d L. a n d Terrence F. Martell, "Effects of Multibank H o l d i n g C o m p a n i e s on Local M a r k e t Concentration." J o u r n a l of t h e M i d w e s t F i n a n c e Association (September 1979), pp. 5770. 13. Rose, J o h n T. a n d Donald T. Savage. " B a n k Holding Company de novo Entry a n d Market S h a r e Accumulation." T h e Antitrust Bulletin, Vol. 2 6 (Winter 1981), pp. 7 5 3 - 7 6 7 . 14. Rose, J o h n T. a n d Donald T. Savage. " B a n k Holding C o m p a n y de novo Entry, Bank Performance, a n d Holding C o m p a n y Size." August 1982 (mimeo). 15. Rose, J o h n T. a n d D o n a l d T. Savage. " B a n k H o l d i n g C o m p a n y Performance: B a n k H o l d i n g C o m p a n i e s Versus I n d e p e n d e n t Banks." July 1 9 8 2 (mimeo). 16. Schull, Bernard, "Multiple-Office Banking a n d t h e Structure of Banking Markets: The New York a n d Virginia Experience," Conf e r e n c e on Bank Structure a n d Competition, Federal Reserve Bank of Chicago, 1972. 17. Schweitzer, Paul a n d J o s h u a Green. "Greeley in Perspective." Staff E c o n o m i c Studies # 9 1 Board of Governors of the Federal Reserve System, 1977. 7. Hooks, D o n a l d L. a n d Terrence F. Martell. "The Impact of Multibank Holding Company Acquisitions on Local Market Structure." Working Paper, University of Alabama (August 1980). 18. Ware, R o b e r t F. " B a n k C o n c e n t r a t i o n in Ohio." E c o n o m i c C o m m e n t a r y , Federal Reserve B a n k of Cleveland (November 1975). 8. Hooks, D o n a l d L a n d T e r r e n c e F. Martell. " M u l t i b a n k Holding C o m p a n y Acquisitions a n d Local M a r k e t Structure: An Analysis of Pooled Cross Section a n d Time Series D a t a " Research Paper 810 1 0 Federal Reserve Bank of St. Louis, 1981. 19. Whitehead, David D. "An Alternative View of Bank Competition: Profit or Market Share Objectives?" E c o n o m i c Review, Federal Reserve Bank of Atlanta, Vol. 6 7 (November 1982). 9. Kunreuther, J u d i t h Berry. " B a n k i n g Structure i n New York State: Progress a n d Prospects." M o n t h l y Review, Federal Reserve Bank of New York (April 1976), pp. 107-115. 10. Rhoades, S t e p h e n A "The Impact of Foothold Acquisitions on Bank Market Structure." Antitrust Bulletin, Vol. 2 2 (Spring 1977), pp. 119-28. 20. W h i t e h e a d , D a v i d D. " H o l d i n g C o m p a n y Power a n d M a r k e t Performance: A New Index of Concentration." W o r k i n g Paper, Federal Reserve Bank of Atlanta, D e c e m b e r 1977. 21. Whitehead, David D. a n d a Frank King. " M u l t i b a n k H o l d i n g C o m p a n i e s a n d Local Market Concentration." M o n t h l y Review, Federal Reserve Bank of Atlanta (April 1976), pp. 34-43. 47 FEDERAL RESERVE B A N K O F A T L A N T A An Alternative View of Bank Competition: Profit or Share Maximization A study of 5 9 0 banks over eight years finds consistent evidence that small banks seek to increase their market share even at the expense of profits. Large banks, on the other hand, apparently aim to maximize both profits and market share. Market forces w i t h i n the financial services industry are driving regulators and legislators to consider further deregulation. Both product and market restraints on financial institutions are being reconsidered seriously in light of new communication technology, product innovation, and cooperative competitive agreements among what traditionally were considered noncompeting financial institutions. As d e r e g u l a t i o n removes the artificial restraints t o product and market d e v e l o p m e n t , consolidations among financial institutions appear inevitable. 1 Although consolidations will occur, the degree of consolidation is questionable because there will always be a ' T o the e x t e n t that r e g u l a t i o n has r e s t r a i n e d t h e o p t i m a l size of f i n a n c i a l institutions, d e r e g u l a t i o n will lead to m e r g e r s or c o n s o l i d a t i o n a m o n g f i n a n c i a l institutions. As d e r e g u l a t i o n occurs, a larger n u m b e r of f i n a n c i a l i n s t i t u t i o n s will be able to offer similar f i n a n c i a l services. This will t e n d to i n c r e a s e t h e n u m b e r of c o m p e t i t o r s a n d r e d u c e t h e level of m a r k e t c o n c e n t r a t i o n , m a k i n g it easier for similar i n s t i t u t i o n s t o merge. place for relatively small financial organizations w h i c h are specialized and highly efficient. O n one hand, consolidations may offer certain benefits, such as those associated w i t h economies of scale and the consumer's ability t o obtain many financial services from a single institution. O n the other hand, consolidations may involve certain costs t o society such as the potential loss in c o m p e t i t i v e market performance from t h e removal of some small and potentially innovative competitors. Given the likelihood of deregulation and the probability that many small competitors will be eliminated through consolidations, it is essential t o learn w h e t h e r c o m p e t i t i o n w o u l d be decreased if the n u m b e r of small competitors were reduced. This study addresses the question by analyzing the c o m p e t i t i v e performance of relatively large and relatively small banks in the Sixth Federal Reserve District. Conventional antitrust analysis is based on a traditional theoretical m o d e l that assumes a 48 N O V E M B E R 1982, E C O N O M I C REVIEW firm's primary goal is t o maximize profits. Given profit maximization as the firm's objective, it is then possible t o d e p i c t t h e firm's conduct in various market settings. The firm's conduct or behavior t h e n determines its performance (that is, prices, o u t p u t , profit or rate of return). It follows that the interaction a m o n g firms in the same market, all a t t e m p t i n g t o maximize profits individually, will determine the market's competitive performance. D e p i c t i n g all firms as profit maximizers then allows for a rather simple analysis of markets. Knowledge of the markef s structure (number and size distribution of firms in the market) and the presumed conduct of firms in their efforts t o maximize profits t h e n allows us t o project the market's competitive performance. Markets w i t h high concentration ( t w o or three firms controlling a relatively large p o r t i o n of the market) are presumed t o be less c o m p e t i t i v e (higher prices and lower levels of o u t p u t ) than those w i t h low levels of concentration. This rather simple model forms the basis for the w e l l - w o r n analytical tool that uses a market's structure t o project its performance. retail and commercial accounts. Second, these competitive actions are almost certainly directed at new customers, those just moving into a new area or seeking t o establish a banking relationship. It is then primarily the growth in banking consumers for w h i c h bankers are overtly c o m p e t i n g — n o t the entire customer base of a market. This leads to a t h i r d and perhaps more important observation: small banks may emphasize deposit growth t o a greater extent than do their larger competitors. Both large and small banks overtly c o m p e t e for the same set of customers, but small banks have more t o gain by attracting new customers than d o large . . if small banks are indeed the most likely to stimulate competition within a market, then the loss of small banks may weaken these markets' competitive performance." Casual Observation Yet the competitive interaction among banks in market settings raises questions about t h e premise that all banks are profit maximizers. Commercial banks are d e p a r t m e n t stores of financial services. The average consumer seems t o develop a close relationship w i t h his banker because access t o future financial services may d e p e n d on it. This leads t o customer loyalty perhaps unparalleled in other industries. O n c e people decide w h i c h bank t o deal with, it is extremely difficult t o persuade t h e m t o change. Consumer loyalty and mutuality of benefits appear t o be at the root of this t y p e of behavior. As a consequence bankers are less likely t o c o m p e t e for each o t h e r s " current customers than for u n c o m m i t t e d customers. This is probably more t h e case for a small retail account than for large retail or commercial accounts. C o m p e t i t i o n for these latter types of accounts is probably more personal and individualized than c o m p e t i t i o n for t h e small retail or small commercial account. Two observations, then, seem w o r t h w h i l e . First, most o v e r t measures of c o m p e t i t i o n within banking markets in fact measure the intensity of c o m p e t i t i o n for relatively small banks. Numerous studies on economies of scale in banking show that significant reductions in average costs are experienced only up t o approximately $50 t o $75 million in deposits. 2 Past this point, as a bank expands its deposits it experiences relatively constant average costs. Therefore, by emphasizing deposit growth, t h e small bank may lower its average costs and thus increase its profit potentials. It follows that a small bank may undertake overt competitive actions ( b o t h price and nonprice) in order t o attract proportionately more deposits than its larger c o m p e t i t i o n . Only after t h e small bank has o b t a i n e d a size sufficient t o realize available economies of scale (lowest average costs) w o u l d it turn its attention t o profits. N o w if w e f o l l o w this line of reasoning further, w e may hypothesize that relatively large banks and smaller banks in t h e same markets may have different objectives. The 2 See G e o r g e J. Benston, Gerald A. Hanweck and David B. Humphrey, " O p e r a t i n g Costs in Commercial Banking," this Review 49 FEDERAL RESERVE B A N K O F A T L A N T A large bank may find it less desirable t o compete overtly for market share than t o simply take advantage of its customer base and maximize profits subject t o some m i n i m u m market share constraint. O n the other hand, the smaller bank finds it advantageous t o expand its customer base in order t o achieve sufficient size t o take advantage of scale economies. Therefore, the small bank finds it advantageous t o undertake overt c o m p e t i t i v e action t o increase its deposit base. To t h e e x t e n t t h a t p r o f i t m a x i m i z a t i o n behavior and share maximization behavior are inconsistent w e should be able t o devise an empirical test of the hypothesis. This hypothesis is important. If large and small banks in fact have different objectives, and if the small bank is most likely t o undertake overt price and nonprice stimuli seeking t o expand its market share, t h e n the loss of many small banks may adversely affect the c o m p e t i t i v e performance of banking markets. In other words, if small banks are indeed the most likely t o stimulate c o m p e t i t i o n w i t h i n a market, t h e n the loss of small banks from deregulation and consolidations may weaken these financial markets' competitive performance. Theoretical Rationale Thinking of a firm as something other than a profit maximizer is by no means pathbreaking. 3 In t h e late 1950s, W. J. Baumol asserted: " I a m prepared t o generalize f r o m these observations a n d assert that t h e t y p i c a l oligopolist's o b j e c t i v e can usefully be characterized, a p p r o x i m a t e l y , as sales m a x i m i z a t i o n subject t o a m i n i m u m profit constraint. Doubtless this premise over-specifies a rather vague set of a t t i t u d e s b u t I b e l i e v e it is n o t t o o far f r o m t h e truth. So long as profits are high e n o u g h t o keep stockholders satisfied a n d c o n t r i b u t e a d e q u a t e l y t o the financing of c o m p a n y growth, m a n a g e m e n t w i l l b e n d its efforts t o t h e a u g m e n t a t i o n of sales revenues rather t h a n t o f u r t h e r increase profits." 4 Baumol's generalization was based on his v i e w of how large firms actually behaved. The management of business firms seems t o be obsessed w i t h sales growth. In attempts t o impress directors, attract stockholders or simply impress market observers, management consis3 F. M a c h l u p , " T h e o r i e s o t t h e Firm: Marginatist, Behavioral, Managerial," A m e r i c a n E c o n o m i c Review, M a r c h 1 9 6 7 , pp. 1-33. • W i l l i a m J. Baumol, B u s i n e s s B e h a v i o r , V a l u e a n d G r o w t h , ( N e w York: Macmillan, 1959), pp. 4 9 - 5 0 . 50 tently emphasized sales growth. This led Baumol t o hypothesize that the primary objective of the management of larger corporations is sales maximization subject t o some minimum level of profits. This laid the foundation for a series of empirical studies attempting to verify what came to be known as the "sales maximization hypothesis." A number of these studies attempted to test the phpothesis, but most found little support for it.5 One exception, a study by Robert J. Saunders that used a cross section of commercial banks from the Fourth Federal Reserve District, reported: "This observed profit-depressing, high growth-oriented behavior of some commercial banks w o u l d be expected in a s i t u a t i o n w h e r e the sales m a x i m i z a t i o n hypothesis is t r u e . . ." 6 M o r e importantly, however, Saunders f o u n d that some commercial banks seemed t o display profit maximizing behavior while others pursued policies consistent w i t h sales maximization. Issues concerning the proper behaviorial model for t h e firm clearly are far from settled as a n u m b e r of recent articles on expense preference behavior reveal. 7 The predictability of the relationship between a market's structure ( n u m b e r and size distribut i o n of firms) and its c o m p e t i t i v e performance (i.e. level of prices, profits, and o u t p u t ) largely depends on the objectives of firms in that market. The structure—performance relationship in banking has proved statistically significant but quantatively weak. In other words, t h e level of market concentration (a structural measure of how much of the market is controlled by the largest'firms) matters, but only very large changes in market concentration are associated w i t h very small changes in price, profits or the other performance indicators. O n e probable reason for f i n d i n g that this relationship quantitatively weak is that all firms in a market do not operate w i t h the same objective. Numerous studies testing various explanations (for example, the expense preference hypothesis, or the 5 S e e f o r e x a m p l e , William C. Pardridge, " S a l e s o r Profit M a x i m i z a t i o n in M a n a g e m e n t Capitialism," W e s t e r n E c o n o m i c J o u r n a l , Spring, 1 9 6 4 ; M a r s h a l l Hall, " S a l e s E x a m i n a t i o n " J o u r n a l of I n d u s t r i a l E c o n o m i c s , April, 1 9 7 7 ; a n d Bevars D. M a b r y a n d David L. Siders, " A n E m p i r i c a l Test of the Sales Maximization Hypothesis," S o u t h e r n E c o n o m i c J o u r n a l , January, 1 9 6 7 . • R o b e r t J. S a u n d e r s , " T h e S a l e s M a x i m i z a t i o n H y p o t h e s i s a n d the Behavior of C o m m e r c i a l Banks," Mississippi Valley J o u r n a l of Business a n d E c o n o m i c s , Vol 6, Fall 1 9 7 0 . 7 S e e S t e p h e n A. R h o a d e s , " A S u m m a r y a n d Evaluation of S t r u c t u r e P e r f o r m a n c e S t u d i e s in B a n k i n g : A n Update," W o r k i n g Paper, Staff of t h e B o a r d of G o v e r n o r s of t h e F e d e r a l R e s e r v e System, 1 9 8 2 . N O V E M B E R 1982, E C O N O M I C REVIEW limit price hypothesis, or Hick's q u i e t life hypothesis or the linked oligopoly hypothesis) bear out the fact that the behaviorial element especially of the banking firms is a complex animal. 8 A t t e m p t i n g t o understand a c o m p l e x relationship often entails separating the components and analyzing t h e m independently. Since it is q u i t e probable that not all firms operate w i t h the same objective, a basic distinguishing characteristic should separate firms into groups sharing the same objective. Perhaps the simplest characteristic is t h e relative size of firms in their respective markets. For those markets chosen in the Sixth Federal Reserve District, banks w i t h less than 3 percent market shares are consistantly smaller than the deposit size necessary t o take advantage of scale economies. Banks w i t h more than 15 percent of the market's deposits are consistently largerthan the m i n i m u m size necessary t o take advantage of scale economies. Therefore, this study attempts t o gather empirical evidence on the possibility that relatively large banking firms' behavior (those w i t h fifteen percent or more of market deposit) differs significantly from that of smaller firms (those w i t h three percent or less of market deposits) w i t h respect t o t h e i r p r o f i t and m a r k e t share maximizing behavior. Hypothesis O u r basic hypothesis is that relatively large banks a t t e m p t t o maximize profit subject t o some minimal market share constraint w h i l e smaller banks a t t e m p t t o maximize market share subject to some minimal profit constraint. This hypothesis asserts that, in general, large banks are willing t o sacrifice market share t o take advantage of pricing or nonprice policies allowing the bank to maximize profits. Therefore, we w o u l d expect that higher levels of market profit would be associated with declining market share for large banks. Conversely, small banks are in general attempting t o acquire market share, since increased market share equates to increased market power. It is only after the small " S e e for e x a m p l e , F r a n k l i n R. Edwards, " M a n a g e r i a l O b j e c t i v e s in R e g u l a t e d I n d u s t r i e s : E x p e n s e - P r e f e r e n c e B e h a v i o r in B a n k i n g , " J o u r n a l of Political E c o n o m y , Vol. 85, F e b r u a r y - D e c e m b e r 1 9 7 7 o r T i m o t h y H. H a n n o n , " E x p e n s e - P r e f e r e n c e B e h a v i o r in B a n k i n g : A R e e x a m i n a t i o n , " J o u r n a l o f Political E c o n o m y , Vol. 8 7 , FebruaryD e c e m b e r , 1 9 7 9 or J a m e s A. V e r b r u g g e a n d J o h n S. Zahera, " E x p e n s e P r e f e r e n c e B e h a v i o r in t h e S a v i n g s a n d Loan Industry," J o u r n a l of Money, Credit and Banking, November, 1981. Figure 1. Graphic Depiction of the Hypothesized Relationship \ Relatively large banks maximize profits at the expense of share. Profits \ \ \ \ \ \ \ \ \ \ % Change Market Share Controlled 0 Relatively small banks maximize growth in the market share at the expense of market profits. + % Change Market Share Controlled bank has acquired some level of market power, or relative size, that its objective would change t o maximizing profits. Therefore, we would expect lower levels of profit to be associated with increasing market share for small banks. This w o u l d be expected if small banks have the same or higher but declining average cost as their larger competitors. It w o u l d not be expected only in the case w h e r e the average cost of the larger firm is higher than that of the smaller firm, in other words, where diseconomies of scale are encountered. Graphically, Figure I describes the relationship expected if the t w o sets of banks, large and small, in fact perform as if they were a t t e m p t i n g t o maximize profits or market shares, respectively. Conceptually, the relationship b e t w e e n profits and growth or decay in market share need not be continuous, but the general relationship described by Figure I w o u l d need t o be f o u n d t o support the hypothesis. Sample To test the hypothesis we selected banking markets in the Sixth Federal Reserve District that contained five or more banks in 1969. 9 The criterion of five or more banks assures the inclusion of an ample n u m b e r o f both large and small banks. Because of the large n u m b e r of " T h e g e o g r a p h i c m a r k e t s used w e r e t h o s e d e f i n e d by t h e Federal Reserve B a n k of A t l a n t a a n d a c t u a l l y u s e d in t h e analysis of h o l d i n g e o m p a n y a c q u i s i t i o n s or bank mergers. 51 FEDERAL RESERVE BANK O F A T L A N T A banking markets in the Sixth District containing just t w o or three banks, inclusion of markets w i t h less than five competitors w o u l d have artificially w e i g h t e d t h e sample in the direction of relatively large banks. In total, 54 markets were selected w h i c h contained 590 banks in 1969. Income and Call Report items were studied for each of these 590 banks for the period 1969 through 1977. Although this period includes a relatively severe recession, it also includes more normal periods of the business cycle. Overall w e believe the period is fairly representative. Market shares and other market related calculations i n c l u d e d new entrants in each of the markets; however, only banks existing in 1969 were used as observation points. These banks were segmented according t o relative size, and the profit performance and share experience of each of these banks was tracked for the nine year period. market shares of 3 percent or less, i.e. our small bank segment. This gives us t w o groups of banks, one relatively large the other relatively small. The profit or share growth objectives of banks in the mid-range (those with more than 3 percent of the market but less than 15) are more likely t o differ among organizations than are the objectives of either the banks in the large or small bank groups. Therefore, an analysis of the relationship between bank profits and market share growth for banks in the large group relative to those in the small group should provide some indication of whether or not the t w o groups of banks perform as if they have significantly different objectives in conformity with the hypothesis. The model used to gather empirical evidence took the following form: tr ¡J = f(Msjj, Abjj, ( D D / T D ) j j , Rjj, ( C + l / T Q j j , E Hcjj where: 7r ¡j = Model and Test Given that this is an exploratory study, a direct test of the hypothesis proved too expensive to undertake. Statistical techniques to derive the appropriate share and profit constraints require tremendous amounts of computer time. 10 Therefore, as a first approximation of the hypothesized relationship a test for differences in the actual profit and share performance of relatively large and relatively small banks was devised. This test is simply intended to establish whether or not relatively large and relatively small banks display the hypothesized patterns w i t h respect t o their profits and market share growth. Since the exact point at which a bank would opt t o maximize profits instead of market share was of little interest in this exploratory study, we simply split our sample into t w o segments based on the relative size of the bank within its relevant market. O n e segment included all banks w i t h beginning (1969) market shares of 15 percent or more, i.e. our large bank segment. The second segment included all banks with beginning ' " I d e a l l y , to test t h e h y p o t h e s i s in t h e f o r m in w h i c h it is p r e s e n t e d w e w o u l d n e e d a m o d e l in w h i c h a b a n k ' s m a r k e t s h a r e a n d its profits are d e t e r m i n e d s i m u l t a n e o u s l y . In a d d i t i o n , d e t e r m i n a t i o n of t h e m a r k e t share g r o w t h w o u l d r e q u i r e a profit c o n s t r a i n t a n d d e t e r m i n a t i o n of m a r k e t p r o f i t s w o u l d require a profit constraint. M a x i m u m l i k e l i h o o d e s t i m a t i o n may be u s e d to i d e n t i f y t h e p r o p e r profit a n d s h a r e c o n s t r a i n t s , however, this p r o c e d u r e p r o v e d t o o e x p e n s i v e f o r t h e purpose at hand. Therefore, g i v e n t h a t this is an e x p l o r a t o r y s t u d y , an indirect t e s t w a s d e v i s e d in o r d e r to e x p l o r e t h e r e l e v a n c e of t h e h y p o t h e s i s w i t h o u t t h e e l a b o r a t e d i r e c t test. 52 Profits = ( N e t Income/Total Assets of the ith bank in the j t h market.) Msjj = Change in the market share b e t w e e n 1969 and 1977 (Market Share 977 - Market Share 1969) Abjj = Absolute size (Total Deposits in millions of dollars)) of the ith bank in the jth market in 1969. ( D D / T D ) j j = Average ratio of the ith bank's d e m a n d to total deposits from 1969 t o 1977. Rjj = a risk measure w h i c h is the bank's average loanto-asset ratio over the period relative t o the market's loan-to-asset ratio. 11 ( C + l / T L ) j j = the proportion of commercial and industrial loans to total loans held by the ith bank in the jth market averaged for 9 years. 12 E = the n u m b e r of new banks entering the market during t h e 1969 to 1977 period. Hj = The average of the jth market's Herfindahl concentration ratios over the 9 year period. He = a d u m m y variable w h i c h indicates w h e t h e r or not the ith bank is a subsidiary of a bank holding company. Rij = V T Ali (f '2 < c + ™ = * ' ' r M ä-/N)9 (?5r)/9 N O V E M B E R 1982, E C O N O M I C REVIEW Table. 1 Empirical Results on Relationship Between Profits and Share Growth for Large and Small Banks (With Bank Profits the Dependent Variable) Classification of B a n k S i z e A Intercept Ms;; < .03 market share Small Bank .008826 -.075266 > . 1 5 market share Large Bank .012025 .024150 (2-45) < .03 or > . 1 5 market share Large and Small Subset .010708 All Banks .011252 x xx xxx A = = = = (1.66) Aby He R2 .009440 (1-41) x .000001 .07270 2.9903 -.002370 (0.49) -.004326 (0.50) -.000017 (1.42) x .19560 3.3348 .001607 (2.06) -.006579 (2.66) .000102 .04841 2.6962 379 (0.00) .000001 (0.63) XX XXX .000848 (1.44) x -.007490 (3.83) -.000521 (0.13) -.000001 (1.11) .04372 3.8012 590 (DD/TD)¡j R¡j (C+l/TUjj .000483 (2.80) .001872 (1.59) .000413 (0.31) -009956 (3.22) xxx -.000013 (0.61) .000280 (0.34) .001700 (1.25) x .000008 (0.38) .001043 (1.35) XX XXX .014973 (1.39) .008887 (1.08) -.000002 (0.09) .001314 (1.78) (0.88) Cases 275 A 104 A XXX Significant at the .10 level. Significant at the .05 level - one tail t Significant at the .01 level F significant at the .01 level for all equations. Given the hypothesis, w e are interested in the relationship between the relative size of a bank and its profit and market share performance during the period. Specifically the expected relationship is: Change in Firm Market Share Profits Small > < Large < > Therefore, w e would expect the relationship between both small and large bank profits and changes in their market shares t o be negative. Small banks w o u l d show lower profits as they gained market share and large banks w o u l d show higher profits as they lost market share. Since the signs of both share coefficients will not indicate whether or not the banks actually lost market share as profits increased (as w e w o u l d expect case for large banks) or gained market share at the expense of profit (as we w o u l d expect for small banks), a scatter diagram was used t o differentiate the t w o events. The expected signs on the remaining independent variables were as follows: profits are expected t o expand with absolute size of the bank, (the larger the absolute size of the bank the more probable is profit maximizing behavior); profits are expected t o be positively associated with the banks' demand t o total deposit ratio, (this ratio proxies the bank's cost of funds; therefore, the higher the ratio the lower the cost of funds, thus increasing the potential for higher profits); profits and risk are assumed t o have a positive association, (as risk increases w e expect higher returns); the proportion of the bank's loans allocated to commercial and industrial borrowers in a portfolio proxy which should be negatively associated w i t h profits (commercial and industrial borrowers normally obtained preferred rates); the entry variable is expected to be positively associated w i t h bank profits, the higher the level of bank profits the more attractive the market for new entrants; and, He or the holding company dummy variable is expected to be positively associated with bank profits (either because of deep pocket assumptions or potential economies associated with holding company organizational structure). The empirical results of estimating the model appear in Table I. As the table indicates, the model was run three times, once including only banks with 1969 market shares 15 percent or greater, once including only those banks w i t h 1969 market shares of 3 percent or less, once including banks with either 3 percent or less of the market or 15 percent or more in 1969 and once for all banks. The empirical results were mixed with respect to the hypothesis tested. The F test on all three 53 FEDERAL RESERVE BANK O F ATLANTA equations was significant at the .01 level. Each of the independent variables with significant (t) tests showed the expected conclusion except in the large bank category where both the change in market share and holding company affiliation variables showed signs opposite those expected. The relationship between holding company affiliation and profits for large banks was negative and significant at the .10 level. This would indicate that, at least for banks which are large relative to their markets, holding company affiliation is negatively associated w i t h profits. This is an interesting finding but outside the scope of this particular study. The magnitude of the coefficient (.000016), however, is so small that although it is statistically significant, it is quantitatively very weak (a little over 1/10 of a percent). The major unexpected occurrence was the relationship between profits and change in market share in the large bank category. As hypothesized, the relationship was expected to be negative, indicatingthat the greater the loss in market share the higher the profits. Instead, this relationship proved positive for large banks and significant at the .01 level. This surprisingly positive relationship indicates that, within the large bank category, the larger the loss in market share the lower the profits. To restate, the smaller the loss in market share or the larger the increase in market share the higher the profits. This relationship suggests that, at least for large banks in our sample, profit maximization and share maximization are not inconsistent objectives. In order to get a visual picture of the relationship between changes in market share and profits, w e ran scatter diagrams for both large and small banks. W i t h the large bank category, scatters were run on all banks w i t h 10 percent or more of market deposits, banks w i t h 1 5 percent or more and banks w i t h 20 percent or more of market deposits. The three groups all showed a significant positive relationship between change in market share and profits. Interestingly, as the criterion of bank size for each group increased—from a 10 percent market share, t o 15 percent and then t o 20 percent—the R2 increased from .08 to .19 and the (t) values increased from 7.2 to 9.5. Those findings indicate that the relationship between changes in market share and profits became more predictable as the bank's market share increased in size. 54 Most importantly, however, two-thirds of all banks holding market shares of .10 or more in 1969 lost market share during the observation period. Three-fourths of the banks in the other large bank categories actually lost market share— those w i t h market shares of .15 percent and greater, and those with 20 percent and more. In the large bank category, then, those banks that gave up smaller market shares during the observation period actually performed better in terms of profits than those that gave up more. In addition, larger banks that increased market share t e n d e d to enjoy higher profits than those showing declining market share. Therefore, it appears from this data that share maximizing behavior is not inconsistent w i t h profit maximization for large banks. For relatively small banks—those holding .03 percent or less of market deposits in 1969— there appears t o be a tradeoff between share growth and profit growth. The hypothesized relationship between profits and change in market share for small banks is negative, and, as expected, the regression analysis shows a statistically significant negative relationship between change in market share and profits for this group. In addition, the scatter diagram of this relationship indicated that 82 percent of the small banks increased market share, contrasting sharply w i t h the 75 percent of large banks showing decreases. The small banks' actual profit and share performance is consistent with the hypothesis that relatively small banks tended t o emphasize share growth, accepting relatively small profits as they gained larger market shares during the period. Figure 2 compares the hypothesized relationship between profits and changes in market shares with the empirical findings. The empirical relationship shows that the profit and market share performance of relatively small banks conforms t o the hypothesis that profits and change in market share are inversely related. This same relationship was hypothesized for relatively large banks, but the empirical results show a positive relationship between the relatively large banks' profits and change in market share. In other words, as market share increases, profits increase. Therefore, small banks face a trade off between growth in market share and growth in profits which the large banks don't face. The large banks appear t o increase profits by minimizing the market share loss or gaining market share. NOVEMBER 1982, E C O N O M I C REVIEW Figure 2. Hypothesis Compared to Empirical Relationship of Large and Small Bank Profit Versus Market Share Experience Hypothetical Table 2. Mean Level of Profits and Standard Deviation of Profits for Relatively Small Banks, Relatively Large Banks and All Banks. Empirical Finding -1 Standard Deviation \ \ Large Banks \ \ Profits (Return on Assets) Small Banks Profits (Return on Assets) Large % C h a n g e in Market Share Controlled Banks Small Banks % C h a n g e in Market Share Controlled Table 2 shows the mean level of profits for small banks, large banks, and all banks represented in the sample as well as the standard deviation of profits within each of these groups. Banks with relatively large market share tend to show higher profits than either small banks or the average of all banks. In addition, the large banks show less variation in profits than d o small banks. It appears that large banks do in fact exercise their market power which results in slightly higher profit, approximately 7 percent greater return on assets than small banks. Although small banks tend to show lower profits than their larger counterparts, small banks tend to gain market share (82 showing positive growth in market share) while maintaining profits just slightly lower than the mean return on assets associated with the large banks. The mean levels of profits for large and small banks are not significantly different. One conclusion w e can draw is that small banks appear t o be at no disadvantage in competing with their larger counterparts. The large banks appear to place more emphasis on profits than do small banks, but they also evidently attempt t o minimize share loss or gain share in concert with their profit objectives. Thus the fact that small banks are capable of gaining market share given these circumstances without giving up much of their relative profitability indicates they can compete with their larger counterparts. Small Banks Large Banks All Banks .0046 .0066 .0051 M e a n II + S t a n d a r d Deviation .0103 .0110 .0105 .0160 .0154 .0159 An analysis of the three equations presented in Table 1 indicates not only that the relationship between profits and changes in market share is significantly different for large and small banks but also that the model's ability t o predict profit is reduced substantially w h e n all banks are taken together (i.e. equation 3). Grouping the banks by relative share of the market increases our ability to predict market performance. This may be taken as another indication that the market behavior of large and small banks is significantly different. It also argues strongly for more research into the behavior of banks based on their relative market size. In order to test the consistency of these findings with regard to market share performance we reorganized the variables in our model, allowing profits t o become an independent variable and change in market share to become the dependent variable. Table III shows the results of rerunning the three equations. The results of running the three equations a second time t o evaluate market share are entirely consistent with the analysis of profit equation results. For relatively small banks profits were negatively associated w i t h change in market share and statistically significant Relatively large banks showed a statistically significant positive relationship between profits and change in market share. Relatively small banks tend to trade profits for increased market share and relatively large banks experience increased profits with increased market share. In addition the R2 for both the relatively small bank sample and the large bank sample are higher than the R2 for all banks taken together. This w o u l d again tend t o indicate that small and large bank behavior is in fact different and that each group's behavior with respect to growth in market share is more predictable than all banks taken together. 55 FEDERAL RESERVE B A N K O F A T L A N T A Table 3. Empirical Results on Relationship Between Profits and Share Growth for Large and Small Banks (With Change in Market Share the Dependent Variable) Classification of Bank Size A Intercept Profits Abjj Small Bank .03 market share or less -.0004 -.1338 (1.66) .0010 (4.22) Large Bank .15 market share or greater -.0308 XX 2.4554 (2.53) -.0000 (.08) Significant Significant Significant Significant (C+l/TUjj ii .0071 (5.12) .0020 (.46) -.0099 (1-22) .0050 .2156 (1.04) at at at at the the the the .10 .05 .01 .01 E .0002 (2.35) R2 F Cases -.0000 (.17) .149 6.7 275 -.0002 (1.57) -233 4.2 104 -.0000 (-89) 1.06 9.8 590 He XXX XXX XXX .0042 (-37) -.0531 (1.16) -.0034 (2.11) XX -.0005 (5.48) XXX = = = = -.0006 (-35) R XXX All Banks x xx xxx F (DD/TD)|jj -.0095 (2.58) XXX .0025 (1.00) -.0174 (1-70) XX -.0004 (2.23) XX level level - o n e tail t level level for all equations. ) A final note of interest, the entry variable is statistically significant in each of the three equations. Entry, however, is positively associated with growth in small bank market share and negatively associated w i t h larger bank growth in market share. This, again, indicates that small banks compete effectively against their larger counterparts. Conclusion In testing the hypothesis that large banks tend to be profit maximizers and that small banks tend t o be share maximizers, w e come t o the following conclusions. First, the tests were structured as an indirect test of the hypothesis, but the results indicate a statistically significant difference in the profit and change in market share relationship between large and small banks. A statistical test of significance (Chow test) indicates the coefficients in the equation for large banks are significantly different from the coefficients in the small bank equation. This is true for both sets of equations, the set predicting profits as well as the set predicting change in market share. This gives further evidence that the profit performance relative t o market share performance is in fact different for large and small banks. Our major finding is that the profit and market share performance of large banks is significantly different from that of small banks. Small banks 56 t e n d t o experience a tradeoff between increases in their market share and profits, increasing market share at the expense of profits. Since the vast majority of small banks increased market share during the period, it appears they have a higher desire for share growth than do the larger banks, most of which gave up market share. O n the other hand, our finding indicates that the relatively large banks do not experience a trade off between profits and growth in market share. To the contrary, it appears that relatively large banks may simultaneously maximize profits and market share. Given this difference and hence the probable difference in objectives of large and small banks, it is not hard to understand why the literature is full of studies that show such weak relationships between market structure and performance. "Small banks tend to . . . increase market share at the expense of profits. . .[while] large banks may simultaneously maximize profits and market share." N O V E M B E R 1982, E C O N O M I C REVIEW Small banks show a higher propensity to acquire market share. Assuming that their actual performance represents an actual objective and not simply a mathematical necessity, small banks attempting to increase market share add to the competitiveness of the marketplace. But because large banks may undertake profit and share maximization behavior simultaneously and because the profit differential between large and small banks is so small, the competitive interaction of relatively small banks may retard the ability of large banks to acquire market share and hence profits. To this extent the consolidation of small banks into larger organizations may be detrimental to competition. On the other hand, the fact that relatively large organizations may simultaneously maximize share and profits indicates that few organizations in a market may be necessary to provide adequate levels of competition. Along the same line, because the differential in return on assets to relatively large and small banks is so slight, relatively small banks appear to be at little competitive disadvantage in relation to large banks. It follows that small banks may then feel little pressure to consolidate with larger organizations to compete effectively. By simply reducing their aggressiveness and attempting to maintain share, they can improve their return on assets.13 [ ' 3 The finding that large relatively large banks appear to simultaneously maximize share a n d profits is consistent with either increasing returns to scale or constant returns to scale in banking. The relatively small profits differential b e t w e e n relatively large a n d relatively small banks w o u l d argue strongly for constant returns in banking. This seems to be consistent with most of the empirical work on e c o n o m i e s of scale in banking. —David D. Whitehead and Jan Lujytes *The authors would like to express their thanks to Robert A Eisenbeis, Wachovia Professor of Banking, University of North Carolina at Chapel Hill, for his many helpful comments and suggestions. REFERENCES Ballensperger, Ernst "Alternative Approaches to the Theory of The Banking Firm" J o u r n a l of M o n e t a r y Economics, vol. 6, no. 1 , 1 9 8 0 , pp. 1-37. Baumal, William J. "The Theory of Expansion of the Firm." T h e A m e r i c a n E c o n o m i c Review, vol. 52, 1962, pp. 1 0 7 8 - 1 0 8 7 . Business Behavior, V a l u e a n d Growth, New York: Harcourt Brace a n d World, Inc. Beighley, Prescott H. a n d Alan S. McCall. " M a r k e t Power a n d Structure a n d C o m m e r c i a l Bank Installment Lending." J o u r n a l of M o n e y , Credit, and Banking, November, 1975, pp. 4 4 9 - 4 6 7 . Benston, G e o r g e J., Gerald H a n w e c k a n d David B. Humphrey. "Scale Economies in Banking: A Restructuring a n d Reassessment." Research Paper, Board of Governors of the Federal Reserve System, Washington, D.C., October, 1 9 8 0 (Revised November, 1981). Edwards, Franklin R. a n d Arnold A. Heggestad. "Uncertainty, M a r k e t Structure, a n d Performance: The Galbraith-Caves Hypothesis a n d Managerial Motives in Banking." Quarterly J o u r n a l of E c o n o m i c s , vol. 87, no. 3, August, 1973, pp. 4 5 5 - 4 7 3 . Glassman, C y n t h i a A. a n d S t e p h e n A R h o a d e a " O w n e r vs. M a n a g e r Control Effects on Bank Performance." T h e Review of E c o n o m i c s a n d Statistics, vol. 62, 1980, pp. 2 6 3 - 2 7 0 . Hall, Marshall. Sales Revenue Maximization: An Empirical Examination. Hannon, Timothy H. "Expense-Preference Behavior in Banking: A Reexamination." J o u r n a l of Political E c o n o m y , vol. 87, no. 4 , 1 9 7 9 , pp. 891-895. Klein, Michael A "A Theory of the B a n k i n g Firm," J o u r n a l of M o n e y , Credit a n d B a n k i n g , vol. 3, May, 1971, pp. 205-218. Gale, Bradley, T. "Market S h a r e a n d Rate of Return." Review of E c o n o m i c s a n d Statistics, N o v e m b e r 1972, pp. 412-423. Mabry, Bevars, D. a n d David L Siders. "An Empirical Test of the Sales Maximization Hypotheses," Southern E c o n o m i c Journal, January 1967, pp.367-377. McCall, Alan S. a n d Manfred O. Peterson. "A Critical Level of C o m m e r c i a l Bank Concentration." J o u r n a l of Banking a n d Finance, no. 4, 1980, pp. 353-369. Mullineaux, Donald J. " E c o n o m i e s of Scale a n d Organization Efficiency in Banking: A Profit Function Approach." T h e J o u r n a l of F i n a n c e , vol. 33, no. 1, March, 1978, pp. 259-280. Pardridge, William D. "Sales or Profit Maximization in M a n a g e m e n t Capitalism." W e s t e r n E c o n o m i c Journal, Spring, 1964, pp. 134-141. Rhoades, S t e p h e n A "Structure Performance Studies in Banking: A Summary a n d Evaluation." Staff E c o n o m i c Studies No. 92, Board of Governors of the Federal Reserve System, Washington, D.C., 1977. Rhoades, S t e p h e n A a n d Paul Schweitzer. " F o o t h o l d Acquisitions a n d Bank Market Structure." Staff Economic Studies No. 98, Board of Governors of the Federal Reserve System, Washington, D C., 1978. Saunders, Robert J. " O n the Interpretation of M o d e l s Explaining Cross Sectional Differences A m o n g Commercial Banks." J o u r n a l of Financial a n d Quantitative Analysis, March, 1969, pp. 25-35. Saunders, Robert J. "The Sales Maximization Hypothesis a n d the Behavior of C o m m e r c i a l Banks." Mississippi Valley J o u r n a l of Business a n d Economics, vol. 6, no. 1, Fall, 1970, pp. 21-32. Sealey, C. W., Jr. a n d J a m e s T. Lindley. "Inputs, Outputs, a n d a Theory of Production a n d Cost at Depository Financial Institutions." T h e J o u r n a l of Finance, vol. 32, no. 4, September, 1977, pp. 1 2 5 1 - 1 2 6 6 . 57 FEDERAL RESERVE BANK O F A T L A N T A v * • ' Future Payments System Technology: Can Small Financial Institutions Compete? In payments system technology where there are significant economies of scale, small financial institutions may not necessarily suffer relative to large institutions. Shared networks and services promise to make sophisticated technology available to small institutions. According t o many of the studies cited elsewhere in this issue, the difference between unit operating costs at large and small financial institutions has been insignificant in the recent past Even though small financial institutions appeared to have slightly lower unit costs for some payments system services, large institutions appeared t o enjoy a slight unit cost advantage in others. Taken together, the respective "comparative advantages" seem to even out, making both large and small financial institutions formidable competitors. Do these trends remain in effect? Will the historical trend of unit operating cost "parity" continue through the rest of the decade? This article addresses these questions and develops a model for the most-likely source of payments system services offered by small financial institutions.* * This article's focus is not on large financial institutions versus small institutions in all facets of their operations; instead, it c o n c e n t r a t e s on payments system-related products and services offered by large a n d small financial institutions. It explores the impact of future technology on the traditional unit cost parity in payments systems among financial institutions of all sizes. Generally speaking, this article refers to small financial institutions as commercial banks with less than $1 50 million in assets, savings, a n d loan institutions with less t h e n $ 2 5 0 million in assets, a n d credit unions. Large financial institutions generally consist of commercial banks and savings a n d loan associations with assets in excess of the small financial institutions limits. This article assumes that references to small banks in their payments system behavior are also generally applicable t o the behavior of other small financial institutions. It assumes a similar relationship for large commercial banks a n d their large counterparts in other s e g m e n t s of the industry. The term "technology" here refers to computer hardware a n d software, terminal devices, and electronic c o m m u n i c a t i o n networks. Finally, the article d o e s not attempt to c o m p a r e large a n d small financial institutions on a point a n d counterpoint basis The emphasis is on w h e t h e r payments system technology will be a friend or foe of small financial institutions in the 1980s. 58 N O V E M B E R 1 9 8 2 , E C O N O M I C REVIEW The Situation Today Other articles in this issue suggest that small banks enjoyed a slight edge in their payments system unit costs during the 1970s. What kinds of unit cost differences exist today? Data in the Federal Reserve System's Functional Cost Analysis (FCA) show that similar trends continue. Recently released data for 1981, covering 614 commercial banks nationwide, show that: Interest-bearing checking accounts were most profitable at the smaller banks (deposits of $50 million or less), earning them $11.22 monthly per account.... Medium-sized banks (deposits of $50 million t o $200 million) earned $8.35 each month per account... The largest banks (more than $200 million in deposits) earned $5.77 per month on the accounts.... The cost of administering a N O W account grew in all size groups over the previous year. Smaller banks again had the lowest administrative expenses. The accounts cost the smallest banks $6.06 monthly per account t o handle. The medium banks had expenses of $6.69 per month per account. And it cost the largest banks $8.59 per account. 1 A review of other FCA data shows that the lowest-cost mantle varies between large and small financial institutions in the other payment systems. Time and savings deposits functions show variation too. Thus, when performance in all of these payments and deposit generating activities is netted out, the traditional overall parity remains generally intact. The historical evidence implies that, so far anyway, payments system technology has not given either large or small financial institutions an insurmountable advantage in the delivery of such products as checking accounts, savings accounts, credit cards, and ATMs. Technological change during the 1970s and early 1980s did not appear to alter materially the unit cost parity between large and small financial institutions in payments system services. The Status Quo Is Under Attack During the rest of the 1980s, though, both the payments system (as now structured) and the 'Carson, Teresa. " I n t e r e s t B e a r i n g A c c o u n t s Gain, a n d So D o e s Their Profitability," A m e r i c a n B a n k e r ( N e w York), A u g u s t 19, 1 9 8 2 , p. 3 FEDERAL RESERVE BANK O F A T L A N T A 59 traditional unit cost parity among financial institutions are expected t o be altered in several respects. First, the Federal Reserve System has begun to charge for its services and will charge even more as attempts intensify to price all Fed float. Second, as Regulation Q is phased out, interest expenses on all kinds of deposits should escalate. Third, interstate banking is progressing on a de facto basis and is likely to become a legal activity before the decade is over. Fourth, savings and loan associations, mutual savings banks and credit unions are now providing such products as checking accounts, credit cards, ATMs and debit cards. Fifth, new players, offering payment services of one sort or another, are entering the a r e n a Sears, Merrill Lynch, and American Express to name a few. Sixth, backroom operations in the paper-based payment systems are labor and energy intensive. Costs in paper-based systems are expected to continue spiraling. When itemized, these trends may sound foreboding to bankers. Yet such lists usually cite payments system technology as the salvation of financial institutions from these threatening trends. The general thrust of the technology argument is that computer and communications advances offer lower cost electronic alternatives t o the paper-based payment systems and their gloomy future. Even more exciting, the argument contends that new, innovative alternatives to today's narrow, " m e too" product lines can be designed and marketed electronically, at lower cost and with higher profit margins. Explicit or implicit in many of these upbeat, technology-inspired forecasts is a suggestion that the future success of financial institutions will not only require use of this technology, but also use of large, networked payment systems. In other words, large financial institutions will be in the best position to reap the technological windfall and gain a unit cost superiority. Many forecasts of this genre imply that electronic banking concepts are an important weapon in efforts to improve the nation's payments system. Such scenarios envision a necessary and desirable shakeout period as small, inefficient financial institutions are forced to merge or fold. The following prediction is typical of those anticipating much greater concentration in the industry: The financial services industry of the future is unlikely to consist of the same familiar types of institutions that we know today. It is not ordained that all or most commercial banks, savings and loans, mutual savings banks, credit unions, insurance companies, investment bankers and others of the old forms of financial institutions will survive. It is more likely that many of them will not. 2 According t o the shakeout theory, only large organizations have the capital and technical expertise t o support large, complex electronic networks. One disagreement among proponents of this theory is the rate at which the transition will occur. In any case, the projected result for small financial institutions is an inevitable d e a t h fast as with an electric chair or slow as with a Chinese water torture. Such forecasts seem logical, tidy, neat and clean. But another scenario, just as plausible, is 2 Kaufman, George, Larry M o t e a n d Harvey Rosenblum, "Product Lines in Geographic M a r k e t s " Bankers Monthly Magazine. Volume 9 4 (May 15,1982), p.22. This article, as do others that it is representative of d o e s not suggest that all small financial institutions are expected to survive. By contrast, there is another school of thought, which the divisibility theory supports, that believes that a very large number of small financial institutions will survive—if they elect to d o so a n d are well managed. 3 A review of the history of the retail grocery industry provides another example of the divisibiity theory. In t h e early years of this century, separate retail outlets typically sold separate lines of groceries, e.g., meat markets sold meat, dry g o o d s grocers sold canned/packaged food products, and produce markets sold vegetables a n d fruits. Many of t h e s e outlets w e r e proprietorships a n d often the o w n e r lived on premise. T h e n in the 1 9 3 0 s w h e n the supermarket chain c o n c e p t s began evolving, many industry forecasters predicted a shakeout of all proprietorships and specialized s t o r e s The demise was e x p e c t e d to occur b e c a u s e small retailers w e r e not e x p e c t e d to achieve the economies of scale available to the supermarket chains with their centralized buying, warehousing a n d distribution systems. 60 emerging to challenge the shakeout theory. The concepts embodied in the alternative theory are not new or unique. They have been evident in many other industries. 3 The alternative is based on the ease with which electronic products can be produced in large quantities for several smaller customers on a pooled or shared basis. The divisibility theory, as it will be called here, suggests that the foreseeable technological trends are not the private preserve of an elite large few. Instead, the divisibility features of the new technology could offer even more unit cost competition between all sizes of financial institutions. More importantly, the new technology could lead to greater service differentiation and market segmentation than is now possible. Likely Technological and Unit Cost Trends During the 1980s To visualize fully the impact of payments system technology on financial institutions during the rest of this decade, let's review some of the general technological and cost trends expected during the 1980s. The review provides a preface for a more specific discussion of how technology can impact the payments system services of small financial institutions. Inexpensive, powerful, computer hardware. Computing power will be less expensive and more accessible to households, small businesses, and small financial institutions. Whether we look at trends in the price-performance of large mainframe computers, minicomputers, or microcomputers, the cost of computing per calculation is declining. This article, for example, is being written with a microcomputer and word-processing These forecasts ignored the easy divisibility of these centralized functions into lots usable by several smaller retailers on a cooperative basis. Wholesalers a n d cooperatives w e r e established to support i n d e p e n d e n t supermarkets as well as corner stores or " m o m a n d pop" operativon. The divisibility theory enabled "independents" to achieve chain store economies of scale without the massive capital investment of a chain. The pooled resources of all t h e wholesaler's users/cooperatives's m e m b e r s were sufficient to c o m p e t e with the large chain's economies of scale. Granted the era d i d see the decline of the corner store concept in most areas of the country. Ironically another mutation, in the 1 9 6 0 s and 1970s, revived the corner store concept in a slightly different format and in a chain structure. The m u t a t i o n is t h e convenience store w h e r e premium prices are c h a r g e d for access to many supermarket items at locations or during hours w h e n easy access to a supermarket is not possible. Today, the retail grocery industry is a mixture of chain supermarkets, chain convenience stores, independent s u p e r m a r k e t s independent convenience stores, a n d specialty outlets such as meat martkets and produce m a r k e t s Their c o e x i s t e n c e demonstrates how an easily divisible product, food, can be provided by organizations with many different organizational structures, with either similar shelf prices or with extra convenience at a premium price. N O V E M B E R 1982, E C O N O M I C REVIEW "The divisibility theory . . . suggests that the foreseeable technological trends are not the preserve of an elite few." package that together cost less than $2,500. The memory capacity of this microcomputer equals the combined memory capacity of the IBM 1401 computers in four check reader-sorters, costing thousands of dollars per year in the early 1970s, at the Federal Reserve Bank of Atlanta. Price-performance improvements are being accompanied by miniaturization. The four readersorters of 1970 required a large room w i t h special electrical wiring, humidity control and temperature control. Today, memories of comparable size and power are available in desk-top and smaller computers requiring none of the special environmental support of earlier years. These trends during the 1980s should produce an exponential growth in the sales of cheap memory and data storage devices. As a result, computing power will represent an exception t o the resource scarcity now being faced with energy, strategic minerals, and even water. The surfeit of computing power will provide electronic payments with a significant price-performance advantage over more-labor intensive payment systems. Inexpensive, user-friendly, computer programs. Canned, easy-to-use software for general management, as well as for payments products and services, will be readily available at reasonable prices. Much current literature on bank operations points out that programming costs are escalating rapidly and possibly threatening the economies that arise from automation and electronic payments. These concerns are based on a traditional reliance on the in-house development of software. The rationale for in-house software development has been that a particular financial institution is unique and cannot possibly use software written for another financial institution. Unfortunately for those a d h e r i n g t o t h e in-house philosophy, some small financial institutions are living proof that several organizations can use the same software and even share a large computer system. Instead of owning the entire rock, so t o speak, small financial institutions are finding ways to o w n just the pieces of a larger rock that they need in their organization. Thus the idea that software development costs are rising and will continue t o rise needs to be tempered with the realization that an individual financial institution may actually reduce software expenses during the 1980s by moving from proprietary, in-house software t o canned or shared software concepts. Inexpensive electronic communications. Advances in communications technology will lead to less expensive movement of information between locations. As energy and other air or surface transportation costs continue to rise, electronic transmission of information will become a greater bargain. Wide-scale networking. Intelligent termials, costing under $200 a unit, should facilitate electronic access of households and small businesses t o a w i d e array of products and services. Such terminals, linked t o inexpensive computer hardware, user-friendly software, and inexpensive electronic communications networks, will create an electronic network with inestimable impact. Among the products and services readily adaptable t o such a network are those often referred to as " h o m e banking" services. New concepts of "branch" banking. The cost advantages of electronic banking will lead to the demise of extensive networks of brick-and-mortar, full-service banking facilities. Many facilities will be replaced with a wider distribution of ATMs, home terminals, and point-of-sale (POS) systems. More sophisticated households, workers, and small businesses. Over time, the nation's population is being exposed t o a more electronic world. Today's children, are growing up in a more computerized world. The work force and household bases of 1990 will be more receptive t o price-competitive electronic alternatives for existing services, in general, and payments services in particular. Tomorrow's small businessman and consumer will be more knowledgeable and sophisticated in selecting and using credit, payment, and investment vehicles. Idle, non-earning balances left with financial institutions will continue their movement into easily accessible accounts/investment vehicles. Widespread use of such vehicles 61 FEDERAL RESERVE B A N K O F A T L A N T A should virtually eliminate demand deposits as a low-cost source of funds for financial institutions. Accordingly, financial institutions can be expected to rely less on spread income and more on fee income. Finally, the trend throughout American industry and trade will continue to be away from blue collar jobs toward technical and white collar occupations. In the financial industry, the trend translates to less reliance on clerical work forces and more emphasis on technically skilled professional staffs that are comfortable working with a more computerized environment. These technological trends will drastically change the unit cost structure for financial institutions. Donald C. Long, a finance industry consultant with IBM Corporation, recently summarized the anticipated impact of these technological trends on the unit cost of several payments system products. The following unit cost trends are excerpted from his recent article in the journal of Bank Retailing: Branch Automation. In the traditional office, an average transaction accepted by a teller over the counter without online capability at the teller station cost61 cents in 1981. At a conservative 6 percent c o m p o u n d rate of growth, the cost will be 82 cents in 1986. Providing full function on-line capability at the teller w i n d o w could reduce that cost by approximately 21 percent to 48 cents.... Automated Teller Machines (ATMs). While the figure is not directly comparable to the above because of differences in transaction mix, the cost per transaction of an ATM doing 6,000 transactions per month exclusive of inquiries is approximately 28 cents . . . .A major additional consideration is that the very high component of technology involved in this service delivery mechanism can significantly reduce the rate of cost growth. Based on this , I estimate that the same volume and transaction mix on the ATM in 1986 will average 30 cents per transaction — Point-of-Sale. At the merchant pointof-sale, the current cost of a check accepted for goods or services is approximately 39 cents of merchant cost (57 cents general merchandise retailerand 33 cents supermarket skewed to reflect the majority of such checks in supermarkets) and 9 cents net bank cost (13 cents handling 62 and processing less 4 cents deposit charges per item). This POS financial transaction cost could be reduced to about 31 cents (27 cents merchant cost and net 4 cents bank cost) through the use of an on-line debit card system providing both electronic authorization and data capture.... The Automated Clearinghouse....A recent study sponsored by the Bank Administration Institute estimated the cost to the bank of an EFT deposit at 7 cents versus 24 cents for an over-the-counter teller deposit and 59 cents for a bank-bymail d e p o s i t Home Banking....The net cost t o the payor, biller, and bank of a payment made by mail exceeds 65 cents. Providing a " M o d e l T" capability, such as telephone bill payment, can reduce that net cost to less than 30 cents. 4 Clearly, payments system technology is positioned to play a strong role in the unit costs of financial services and products as the decade unfolds. Yet as formidable as these technological trends and the unit cost projections may seem, they still skirt the original question: What does technology hold in store for unit costs at small financial institutions? Impact of Payments Technology On Small Financial Institutions' Unit Costs Small financial institutions probably will not generate sufficient volumes in electronic payment systems to develop proprietary, completely in-house hardware, software, and communications functions. But, just as obviously, the divisibility theory permits shared efforts that can achieve, collectively, the break-even volumes to be pricecompetitive with larger financial institutions. 5 j Let's explore some of the implications of the divisibility theory further. First, regulators have tended to require small banks to maintain higher capital ratios than large banks. One often-stated reason is that small " L o n g , D o n a l d G. "The B u s i n e s s Case for E l e c t r o n i c Banking," J o u r n a l of Retail B a n k i n g , V o l u m e 4 (June 1982), pp. 19-20. 5 P l e a s e s e e A p p e n d i x A, T r a n s a c t i o n V o l u m e s a n d Unit Costs, f o r a more t h o r o u g h d e s c r i p t i o n of t h e c o n c e p t s involved in e c o n o m i e s of scale for an e l e c t r o n i c p a y m e n t p r o d u c t / s e r v i c e a n d h o w t h e divisibility theory makes t h e s e e c o n o m i e s available to small f i n a n c i a l institutions. N O V E M B E R 1982, E C O N O M I C REVIEW ) ) Table 1. Primary Source of Computer Processing Percent of banks by asset size in millions of dollars Processing source Under$100 1974 On-premise operations Holding co. arrangement Correspondent bank Joint venture Facilities management Servicing by non-bank Other 1977 $100-$500 $500 and above 1980 1974 1977 1980 1974 1977 1980 15% 24% 62% 58% 45% 77% 82% 75% 10 12 12 12 17 19 17 13 18 58 5 52 4 41 2 7 6 7 5 11 4 1 0 0 1 1 1 8 5 6 10 6 9 5 3 4 8 0 10 0 13 2 3 0 9 0 10 2 0 0 1 0 0 1 9% Source: " O p e r a t i o n s / A u t o m a t i o n Scoresheet: c o m p a r e your bank with its peers," ABA Banking Journal, vol. 74 (Feb. 1982), p. 98. financial institutions experience more risk because of their size. However, one risk not usually discussed is the relative automation investment of large versus small financial institutions. Table 1 shows the primary sources of computer processing for small, medium and large banks as determined from responses to the American Bankers Association's Operations and Automation Surveys in 1974, 1977 and 1980. Large banks typically use in-house equipment and much inhouse software. Small banks rely much more frequently on third-party sources. Conversations with vendor representatives indicate that, with few exceptions, small banks use canned software even if they have in-house processing capability. Exhibit 1 (chart and table) shows the much larger percentage increase in expenditures for hardware and software that large banks anticipate in comparision w i t h small banks. At a time when banks are experiencing new competitive pressures and squeezed profit margins, it appears that a significantly greater proportion of large bank expense and investment resources will be tied up in automation efforts. In contrast, small banks are buying what they need, in the quantity they need, when they need it. The latter approach would seem to carry less risk. FEDERAL RESERVE B A N K O F A T L A N T A Exhibit 1 . Projected Hardware/Software Expenditures per Bank - 1982 % 1 0 0 F | | software Computer Equipment 50 25 0l 1 1 1 Under $ 1 0 0 S100-S500 1 * $ 5 0 0 a n d above Asset Size (millions of dollars) Average Hardware/Software Expenditures per Bank Asset Size (millions of dollars) Under$100 $100-$500 $ 5 0 0 and Above C o m p u t e r Equipment 1980-actual 1981-budgeted 1982-anticipated Software 1980-actual 1981-budgeted 1982-anticipated ($000s) 39 48 50 ($000s) 198 222 242 ($000s) 1.978 2,229 2,482 11 11 13 24 33 36 180 258 298 Source: Operations/Automation Scoresheet: C o m p a r e your bank with its peers," ABA Banking Journal, Vol. 74 (Feb. 1 982), p. 96. 63 Why? A large bank is likely t o be heavily committed and heavily invested in a software/hardware system that becomes prematurely obsolete from the rapid innovation spawned by a highly competitive marketplace. As a result, it will feel management pressure either to continue offering an obsolete product or service until the initial investment has been recouped or t o drop the incompletely depreciated product taking a heavy loss on the initial investment. O n the other hand, a small bank, using third-party suppliers, can "A large bank is likely to be heavily committed and heavily invested in a software/hardware system that becomes prematurely obsolete. . . change products or services as the market dictates with much less internal financial impact. Divisibility, in a sense, decreases small bank automation risk and increases small bank ability t o respond to a rapidly changing marketplace. Second, there are several indications that the economies of scale in payment systems now require volumes larger than are achievable even at large financial institutions. For example, during the past year several large financial institutions have turned their Visa and/or Mastercard backroom operations over to a third party. Almarin Phillips, Professor of Economics, Law and Public Policy at the University of Pennsylvania, recently predicted: Economies of scale in funds management, distribution, and in the electronic and mechanical aspects of clearing mean that many institutions will be unable independently to enter new markets with new products, or indeed, competitively t o maintain their old services in t h e i r o l d markets....In many respects, the sharing of product offerings, with compatible hardware and software and general customer recognition is highly procompetitive. It keeps in the marketplace numbers of firms that would otherwise disappear; it also makes entry easier if 64 procompetitive, non-exclusionary participations are allowed. 6 Perhaps it is the large financial institutions that will become more like their small brethren with respect to automation resources. Perhaps it is the small institution's approach to harnessing automation resources that kept its unit costs so competitive during the last decade. If so, the small financial institution is really the automation model for the 1980s, not the large financial institution. If so, the divisibility theory actually has a broader application than once visualized. Third, networking opens up markets for smaller financial institutions that were not profitably serviceable in the traditional payments system environment. William S. Anderson, chairman of NCR Corporation, observes that: A score of regional EFT networks in the United States—each capable of providing retail banking services over a broad geographic area—have now been organized. It is only a matter of time until full-scale national networks of this type, as well as international networks will be in operation. W h e n that occurs, any bank regardless of location or size will at least theoretically have the entire world as its market. 7 I n this respect, technology opens new doors of opportunity for small financial institutions. Surely it would be naive to believe that small institutions can provide mass-market products nationally or internationally. But it does not seem naive at all to think that small financial institutions could use this technological windfall in certain well-defined market segments. A good example of how new technology can be accessed by large and small institutions is the sweep and invest feature now being marketed by several large and small institutions in conjunction with their checking accounts for large balance customers. The size of the financial institution does not appear to be a deterrent to offeringthis feature. In fact, the sweep and invest feature offered by many large and small institutions is being obtained from the same third-party source. Fourth, the literature in support of the shakeout theory seems to ignore a major question: Where •Phillips, Alamarin. " F i n a n c i a l I n s t i t u t i o n s in a R e v o l u t i o n a r y Era," J o u r n a l of C r e d i t U n i o n M a n a g e m e n t a n d E c o n o m i c s , V o l u m e 2 (Spring 1982), p. 17. ' A n d e r s o n , William S. " E l e c t r o n i c F u n d s T r a n s f e r is R e a c h i n g the Pointof-Sale," A m e r i c a n B a n k e r ( N e w York), J u l y 2 8 , 1982, p. 58. N O V E M B E R 1982, E C O N O M I C REVIEW will large financial institutions obtain the financial wherewithal to go on a merger binge? Raoul D. Edwards, senior editor, United States Banker, suggests that a source is not available. Therefore, he observes: What has happened is a new and less certain economic pattern, marked increasingly by shrinking earnings margins, escalating labor costs, increasing pressures on the bottom line—and a consequent reluctance to make the kind of massive commitment implicit in major acquisitions. Some deals will still get made; the joining of a good sized pair of banks as equals, where no dilution of value occurs, for example. But no one in today's environment wants to go out and buy up a bunch of smaller institutions at bookand-a-half o r t w o - t i m e book, especially if their o w n stock is selling under book. 8 According to Edwards, a major shakeout of small financial institutions is unlikely. What is likely is that technology will be the tool with which all financial institutions will attempt to offer products and services equal to or better than those in the past, but at a lower price and, hopefully, a better profit margin. The same article that presented Edwards' scenario touched on a point that bears amplification here as a potential advantage for small financial institutions in the competition of the 1980s: This (trend) means that the vendor of products will increasingly find his market opening up at the bottom. Moreover, the range of services he can sell will broaden.... The vendor who recognizes this and prepares for the new market is going to find it large and growing; the banker who understands this will find his options will be far greater, far more useful, than ever before. 9 If vendors behave in this manner, small financial institutions will find more, not fewer, potential sources of supply for their hardware, software, technical, product, and service support in the 1980s than they can find now. If we assume that more vendors mean more price competition, "Edwards, R a o u l D. " T h e V e n d o r a n d C h a n g i n g Worlds," U n i t e d S t a t e s Banker, V o l u m e 9 3 ( M a y 1982), p. 6 8 9 lbid„ pp. 6 8 a n d 71. then small financial institutions could find more of a "buyer's market" when shopping for suppliers than will large financial institutions deeply engrossed in more complex, longer lead time efforts to develop proprietary products and services. If so, intense price-performance competition among vendors will help cap potential cost increases for small institutions. If so, the divisibility theory provides them yet another benefit. Fifth, m i n i c o m p u t e r s — a n d i>ow m i c r o c o m puters—constitute new technological weapons for small businesses and small financial institutions. Their full benefit has not been tapped yet by small institutions. Admittedly, their impact transcends the payments system functions of financial institutions. Until the advent of microcomputers, only large businesses and financial institutions could afford automated planning and control. Robert H. Long, president of Long, Inc., and editor of Microbanker, recently wrote in The Southern Banker that: The microcomputer is one of the first pieces of technology that places a potent competitive weapon in the hands of community bankers—a piece of technology that can help them competitively serve the more profitable markets and reposition themselves for success in a deregulated environment. Large banks will use them the same way, but the effect will not be as dramatic. 10 Stated another way, large financial institutions achieved the major planning and control software on their large computer systems during the last decade. Thus, for large banks, microcomputers will esentially serve as "fine tuning" devices. But microcomputers provide the first reasonable method for small financial institutions to achieve the management and operations improvement economies available from scientific planningand control processes. The " q u a n t u m leap," so to speak, could help small financial institutions improve their efforts to control expenses and to make better product/service decisions. In the process, the comparative advantages of sophisticated management processes at large financial institutions will trickle down to small ' " L o n g , R o b e r t H. " W e l c o m e M i c r o T e c h n o l o g y , " T h e S o u t h e r n B a n k e r , V o l u m e 1 5 8 (July 1982), p. 26. 65 FEDERAL RESERVE B A N K O F A T L A N T A "Small financial institutions could even enjoy a unit cost advantage over large institutions during the 1980s." institutions. As a result, large financial institutions would lose one of their significant comparative advantages in unit cost control. The ability of small financial institutions t o compete would be enhanced as this aspect of the divisibility theory materializes. The small financial institution's approach is also more flexible and more capital conservative, no small considerations in the volatile, unpredictable marketplace of the 1980s. On the basis of the qualitative information we ? have reviewed, it appears that unit cost parity is possible for small financial institutions in the > 1980s. More significantly, evidence suggests that , small financial institutions could even enjoy a unit cost advantage over large institutions during the 1980s. If the number of small financial institutions contracts during the 1980s, the contraction should not be caused by a lack of state-of-the-art tech- s nology at competitive prices for small institutions. ! Other causes will need to be sought if such a contraction occurs. » Will Unit Cost Parity Be Sustained in the 1980s? Unfortunately, our research uncovered little quantitative data that address the question. M u c h qualitative information was found t o structure a logical argument for or against continued unit cost parity. Technology, as represented in the divisibility theory, appears to provide small financial institutions with an enhanced ability t o compete. The preceding discussion suggests that technology in the 1980s will trickle down t o small financial institutions bringing many of the beneficial management control processes and systems once the province of large financial institutions w i t h large computer installations. The divisibility theory also suggests that the sourcing model for financial institutions during the 1980s will not be the in-house proprietary route so often favored by large financial institutions during the 1970s. Instead, the most cost-effective alternative appears to be the third-party, shared systems concepts practiced so effectively by small financial institutions during these years. The Bottom Line Future payments system technology appears n unlikely t o prevent small financial institutions from competing effectively during the 1980s. In fact there are several indications that the benefits of the divisibility theory are so attractive that } small institution sourcing arrangements will become 1 the norm with many large financial institutions as well. The available quantitative evidence, albeit ' sparse, indicates that future payments system technology should not hinder the competitiveness of small institutions. But any elation over this conclusion needs strong qualification. Technology will not save a * financial institution—large or small—that is not • well managed. Harnessing newtechnology is just $ one of the many responsibilities that management must oversee. Payments system technology is merely a tool by which an organization can achieve results in its payments activities. Without sound management, technological advances will not achieve the cost benefits potentially available l t o all financial institutions.^ —Paul F. Metzker 66 N O V E M B E R 1982, E C O N O M I C REVIEW APPENDIX A TRANSACTION VOLUMES AND UNIT COSTS The following discussion is excerpted from an article by David A Walker, associate professor, Georgetown University, titled "Electronic Funds Transfer Issues and Experience: What Credit Unions Should Consider," and published in the Journal of Credit Union Management and Economics, Volume 2 (Spring 1982), pp. 5-6. Besides the absolute costs of developing EFT systems, the per unit costs are a major concern. The average cost per transaction (total costs divided by total transactions) over the long run is used to judge whether or not there are economies of scale associated with an economic activity. If average costs decline as output (transactions) increase, economies of scale exist and the marginal cost per unit (additional cost for an additional transaction) is below the average cost in this transaction range. Two economic approaches have been applied to developing cost models forfinancial services. One approach assumes that average costs decline, or remain constant or increase throughout the relevant output range; and curve is selected, from among ones shaped like AA', BB', or CC', respectively, in Figure 1. The relevant marginal cost relationships for these three cases are aa', bb', or cc', respectively. The second, broader approach is to assume that average costs decline at low output levels (0 to Q*) and then begin to rise beyond some higher output level (Q); in this approach, instead of selecting one of the three possibilities from Figure 1, all three are presumed to represent parts of the relationship as shown in Figure 2. The empirical evidence on the average and marginal costs per EFT transaction is very meager.... It appears that average costs per transactions decline as cash dispenser and automated teller transactions increase (as shown in Figure 2). Lower average costs are expected to be observed until approximately 45,000 transactions per month (Q* in Figure 2) occur at a particular retail terminal. The preceding discussion shows how economies of scale can be measured in an EFT system. Walker's example pertains to a single cash dispenser or ATM. The same basic rationale applies to a payment product, overall, offered by a financial institution. More importantly, using Figure 2 in Walker's discussion, it is possible to show how the divisibility theory benefits small financial institutions A small financial institution quite possibly can only generate a volume of transactions in the range of 0 to less than Q*. If so, the small institution is not operating in the portion of the cost curves that allows its unit costs to be most competitive with larger competitors. But, if several small financial institutions use a thirdparty source, their combined transaction volume will more than likely allow them to operate in the Q* to 0 range of the cost curves. In this range, each of the small financial institutions will achieve unit costs that are competitive with larger institutions. Figure 2 Costs Average Marginal Marginal Average Figure 1 O O Q* Q Output Q* Q Output O t o Q* average costs may decrease as ouptut increases (economies ol scale); Q* t o Q average c o s t s m a y remain c o n s t a n t as o u t p u t increases (constant returns to scale); above Ö average costs may increase as output increases (diseconomies ol scale). 67 FEDERAL RESERVE BANK O F A T L A N T A 777771 f T f i r l FINANCE SEPT 1982 AUG 1982 SEPT 1981 ANN. ANN. % CHG. SEPT 1982 AUG 1982 SEPT 1981 534,675 10,649 91,577 433,178 JUL 501,678 15,865 534,363 10,519 91,963 432,781 JUN 503,964 16,753 511,245 6,713 94,014 410,260 JUL 507,767 17,135 + 5 + 59 - 3 + 6 78,877 1,704 11,536 65,809 JUL 69,736 2,981 78,882 1,696 11,558 65,727 JUN 69,968 3,117 74,933 1,025 11,765 62,031 JUL 74,120 3,608 + 5 + 66 4,530 91 544 3,926 JUL 3,957 47 4,517 89 546 3,908 JUN 3,946 78 4,339 53 595 3,710 JUL 4,010 94 4 72 9 b 47,661 1,158 7,689 38,810 JUL 41,191 2,345 47,681 1,155 7,693 38,783 JUN 41,364 2,519 45,369 718 7,821 36,648 JUL 45,173 3,041 9,869 193 1,176 8,598 JUL 8,996 167 9,898 190 1,190 8,599 JUN 9,062 171 9,559 107 1,221 8,251 JUL 9,476 140 7,898 110 1,215 6,591 JUL 7,332 281 7,893 111 1,223 6,585 JUN 7,331 229 7,215 62 1,193 5,973 JUL 7,047 222 2,466 52 232 2,197 JUL 2,177 20 2,436 53 228 2,168 JUN 2,180 20 2,387 26 236 2,132 JUL 2,206 34 6,453 100 680 5,687 JUL 6,083 121 6,457 98 678 5,684 JUN 6,085 100 6,064 59 699 5,317 JUL 6,209 77 CHG. $ millions Commercial Bank Deposits Demand NOW Savings Time Credit Union Deposits Share D r a f t s Savings & Time 11 1 31 0 18 30 39 27 Savings & Loans Total Deposits NOW Savings Time 11 2 33 1 17 31 24 27 Savings & Loans Total Deposits NOW Savings Time + 8 + 1 + 32 Savings & Loans Total Deposits NOW Savings Time 1,170,291 1,163,103 1,051,230 282,892 286,095 287,174 59,257 58,171 45,362 149,974 150,116 150,193 704,138 695,861 594,949 48,930 49,479 37,581 3,153 3,324 2,268 41,962 42,093 33,090 Commercial Bank Deposits Demand NOW Savings Time Credit Union Deposits Share D r a f t s Savings & Time 124,831 32,456 7,649 14,656 72,434 4,617 302 3,896 124,861 33,180 7,536 14,673 72,160 4,634 330 3,880 112,286 33,025 5,746 14,762 61,659 3,534 244 3,061 Commercial Bank Deposits 13,886 3,300 672 1,557 8,767 837 59 677 13,953 3,439 653 1,557 8,715 815 64 654 12,903 3,279 509 1,564 7,844 551 48 494 40,620 11,298 3,313 6,176 20,349 2,083 165 1,640 40,642 11,605 3,268 6,180 20,293 2,116 183 1,647 37,034 11,875 2,504 6,332 17,142 1,602 139 1,253 + 10 + + + + 5 32 2 19 30 19 31 17,375 5,772 1,101 1,637 9,656 855 29 775 17,400 5,911 1,089 1,635 9,623 853 32 773 15,150 5,757 835 1,584 7,912 689 19 659 + + + + + + + + 15 0 32 3 22 24 53 18 Savings & Loans Total Deposits NOW Savings Time 22,859 5,758 1,071 2,428 13,989 123 9 116 22,758 5,876 1,036 2,448 13,888 126 10 116 20,299 5,855 776 2,395 11,754 95 6 88 + 13 + + + + + + Savings & Loans Total Deposits NOW Savings Time 10,375 2,296 555 736 6,999 N.A. N.A. N.A. 10,372 2,205 571 733 6,993 N.A. N.A. N.A. 9,330 2,239 426 733 6,142 N.A. N.A. N.A. + 11 + 3 + 30 + 0 + 14 19,716 4,032 937 2,122 12,674 719 40 19,736 4,144 919 2,120 12,648 724 41 690 17,570 4,020 696 2,154 10,865 597 32 567 + 12 + 0 + 35 NOW Commercial Bank Deposits NOW Credit Union Deposits Savings & Time GEORGIA Commercial Bank Deposits NOW Credit Union Deposits Savings & Time I.OTTURI ANA Commercial Bank Deposits NOW Credit Union Deposits Share D r a f t s Savings & Time Commercial Bank Deposits NOW Credit Union Deposits TENNESSEE Commercial Bank Deposits NOW Credit Union Deposits Share D r a f t s + + + + + + Mortgages Outstanding Mortgage Commitments - + + + + - + - - 0 12 52 23 37 2 38 1 19 29 50 32 Mortgages Outstanding Mortgage Commitments Mortgages Outstanding Mortgage Commitments Savings ic Loans Total Deposits NOW Savings Time Mortgages Outstanding Mortgage Commitments Mortgages Outstanding Mortgage C o m m i t m e n t s Mortgages Outstanding Mortgage Commitments Savings & Loans Total Deposits NOW Savings Time Mortgages Outstanding Mortgage Commitments - + + + + 1 17 20 25 21 Savings k Loans Total Deposits NOW Savings Time Mortgages Outstanding Mortgage Commitments - 1 7 _ - 2 + 6 - 6 I - 17 - 1 50 - - i> + 61 - + - 2 6 9 23 3 + 8U 4 4 - - b 19 y + TI + 2 + 10 4 + ri I* ) 3 +100 2 3 - - 1 41 6 + 69 + 3 V - 2 - + bV Notes: All deposit data are e x t r a c t e d from the Federal Reserve Report of Transaction Accounts, other Deposits and Vault Cash (FR2900), \ and are reported for t h e average of the week ending the 1st Wednesday of the month. This data, reported by institutions with r ", over $15 million in deposits as of December 31, 1979, represents 95% of deposits in the six s t a t e area. The major differences between f this report and the "call report" are size, the t r e a t m e n t of interbank deposits, and t h e t r e a t m e n t of f l o a t . The data generated from ~ j the Report of Transaction Accounts is for banks over $15 million in deposits as of December 31, 1979. The total deposit data generate^ f r o m t h e Report of Transaction Accounts eliminates interbank deposits by reporting the net of deposits "due to" and "due f r o m " other . depository institutions. The Report of Transaction Accounts s u b t r a c t s cash in process of collection from demand deposits, while the calif report does not. Savings and loan mortgage data a r e f r o m the Federal Home Loan Bank Board Selected Balance Sheet Data. The Southeast data represent the t o t a l of the six s t a t e s . Subcategories were chosen on a selective basis and do not add to t o t a l . N.A. = fewer than four institutions reporting. FRASER Digitized for http://fraser.stlouisfed.org/ 68 Federal Reserve Bank of St. Louis FEDERAL RESERVE BANK O F A T L A N T A EMPLOYMENT ANN. % CHG. AUG 1982 JUL 1982 AUG 1981 Civilian Labor Force - thous. Total Employed - thous. Total Unemployed - thous. Unemployment R a t e - % SA Insured Unemployment - thous. Insured Unempl. R a t e - % Mfg. Avg. Wkly. Hours Mfg. Avg. Wkly. Earn. - $ 111,887 101,177 10,710 9.8 112,526 99,732 11,036 9.8 110,099 102,152 7,947 7.3 39.0 332 39.0 333 39.9 320 Civilian Labor Force - thous. Total Employed - thous. Total Unemployed - thous. Unemployment R a t e - % SA Insured Unemployment - thous. Insured Unempl. Rate - % Mfg. Avg. Wkly. Hours Mfg. Avg. Wkly. Earn. - $ AL Civilian Labor Force - thous. Total Employed - thous. Total Unemployed - thous. Unemployment R a t e - % SA Insured Unemployment - thous. Insured Unempl. Rate - % Mfg. Avg. Wkly. Hours Mfg. Avg. Wkly. Earn. - $ 14,329 12,936 1,392 9.7 14,329 12,902 1,428 9.4 13,901 12,844 1,056 7.6 39.1 288 38.8 285 40.4 278 - 3 + 4 1,690 1,450 240 14.1 1,704 1,454 250 13.6 1,658 1,483 175 10.5 + 2 39.3 284 38.8 40.1 282 282 + 1 Civilian Labor Force - thous. Total Employed - thous. Total Unemployed - thous. Unemployment R a t e - % SA Insured Unemployment - thous. Insured Unempl. R a t e - % Mfg. Avg. Wkly. Hours Mfg. Avg. Wkly. Earn. - $ 4,865 4,486 379 7.7 4,854 4,489 365 7.3 4,603 4,294 308 6.4 + 6 38.6 274 38.3 271 40.4 267 Civilian Labor Force - thous. Total Employed - thous. Total Unemployed - thous. Unemployment R a t e - % SA Insured Unemployment - thous. Insured Unempl. Rate - % Mfg. Avg. Wkly. Hours Mfg. Avg. Wkly. Earn. - $ 2,694 2,494 2,692 2,473 219 7.6 2,610 2,452 158 5.7 + 3 39.2 265 38.6 262 40.4 257 - 3 + 3 Civilian Labor Force - thous. Total Employed - thous. Total Unemployed - thous. Unemployment R a t e - % SA Insured Unemployment - thous. Insured Unempl. Rate - % Mfg. Avg. Wkly. Hours Mfg. Avg. Wkly. Earn. - $ 1,907 1,700 207 1,872 1,719 152 8.5 + 2 - 1 11.0 1,901 1,685 216 11.1 39.4 374 39.6 375 41.8 358 Civilian Labor Force - thous. Total Employed - thous. Total Unemployed - thous. Unemployment R a t e - % SA Insured Unemployment - thous. Insured Unempl. R a t e - % Mfg. Avg. Wkly. Hours Mfg. Avg. Wkly. Earn. - $ 1,052 920 132 12.7 1,064 933 131 11.2 1,048 963 85 8.2 + 0 39.0 251 38.3 244 39.5 237 - 1 + 6 Civilian Labor Force - thous. Total Employed - thous. Total Unemployed - thous. Unemployment R a t e - % SA Insured Unemployment - thous. Insured Unempl. Rate - % Mfg. Avg. Wkly. Hours Mfg. Avg. Wkly. Earn. - $ 2,120 2,111 1,933 178 8.4 + 0 - 2 234 11.4 2,115 1,868 247 11.2 39.3 283 39.2 278 39.9 269 nyär 200 7.2 + 2 - 1 +35 - 2 + 4 + 3 + 1 +32 - 2 +37 - 2 + 4 +23 + 3 + 2 +27 ANN. % AUG 1981 JUL 1982 AUG 1982 CHG. Nonfarm Employment- thous. Manufacturing Construction Trade Government Services Fin., Ins., & Real Est. Trans. Com. <5c Pub. Util. 89,195 18,793 4,167 20,547 14,902 19,191 5,429 5,048 89,362 18,725 4,149 20,598 15,082 19,209 5,422 5,051 91,087 20,370 4,431 20,664 15,097 18,771 5,374 5,180 Nonfarm Employment- thous. Manufacturing Construction Trade Government Services Fin., Ins., & Real Est. Trans. Com. & Pub. Util. 11,264 2,154 674 2,669 2,039 2,234 641 697 11,300 2,144 678 2,679 2,063 2,241 643 698 11,386 2,313 746 2,651 2,045 2,142 635 700 Nonfarm Employment- thous. Manufacturing Construction Trade Government Services Fin., Ins., & Real Est. Trans. Com. & Pub. Util. 1,319 334 63 272 291 213 60 71 1,323 333 63 272 295 214 60 71 1,347 366 67 273 281 211 60 7 2 + + Nonfarm Employment- thous. Manufacturing Construction Trade Government Services Fin., Ins., & Real Est. Trans. Com. & Pub. Util. 3,699 443 255 1,004 572 906 278 231 3,721 440 258 1,011 586 908 279 230 3,660 466 290 973 568 853 272 227 + 1 - 5 -12 + 3 + 1 +6 + 2 + 2 Nonfarm Employment- thous. Manufacturing Construction Trade Government Services Fin., Ins., & Real Est. Trans. Com. & Pub. Util. 2,150 495 100 497 424 368 117 142 2,149 492 100 497 425 368 117 143 2,180 523 104 503 421 362 115 145 - 1 - 5 - 4 - 1 + 1 + 2 +2 - 2 Nonfarm Employment- thous. Manufacturing Construction Trade Government Services Fin., Ins., & Real Est. Trans. Com. & Pub. Util. 1,609 197 133 370 305 297 76 131 1,613 198 134 369 307 296 77 132 1,635 216 159 365 310 284 76 129 - 2 - 9 -16 +1 - 2 + 5 0 + 2 Nonfarm Employment- thous. Manufacturing Construction Trade Government Services Fin., Ins., & Real Est. Trans. Com. & Pub. Util. 783 206 40 163 170 118 33 40 790 207 40 163 173 121 33 40 813 224 44 165 175 118 33 41 4 8 9 1 3 0 0 - 2 Nonfarm Employment- thous. Manufacturing Construction Trade Government Services Fin., Ins., & Real Est. Trans. Com. <5c Pub. Util. 1,704 479 83 363 277 332 77 82 1,704 474 83 367 277 334 77 82 1,751 518 82 372 290 314 79 86 - 3 - 8 +1 - 2 - 4 + 6 -3 - 5 - 2 - 8 - 6 - 1 - 1 + 2 + 1 - 3 - 1 - 7 -10 + 1 - 0 + 4 +1 - 0 2 9 6 0 4 1 0 - 1 A Notes: 1,886 +36 - 6 + 4 - 4 +55 +31 - 2 + 5 - All labor force d a t a are from Bureau of Labor Statistics reports supplied by s t a t e agencies. Only the unemployment rate data are seasonally adjusted. The Southeast data represent the t o t a l of t h e six s t a t e s . The annual percent change calculation is based on the most recent data over prior year. http://fraser.stlouisfed.org/ NOVEMBER 1982, E C O N O M I C REVIEW Federal Reserve Bank of St. Louis 69 CONSTRUCTION ANN % CHG AUG 1982 JUL 1982 AUG 1981 Nonresidential Building Permits Total Nonresidential Industrial Bldgs. Offices Stores Hospitals Schools $ Mil. 47,160 5,498 13,392 5,458 1,694 861 48,090 5,780 13,884 5,602 1,701 849 52,478 8,394 13,464 6,693 1,425 706 10 35 1 - 18 + 19 + 22 Residential Building Permits Value - $ Mil. Residential Permits - Thous. Single-family units Multi-family units T o t a l Building Permits Value - $ Mil. SOUTHEAST Nonresidential Building Permits Total Nonresidential Industrial Bldgs. Offices Stores Hospitals Schools $ Mil. 6,275 737 1,334 1,035 212 94 6,489 763 1,378 1,054 272 95 7,359 897 1,342 1,054 260 75 15 18 1 2 - 18 + 25 Residential Building Permits Value - $ Mil. Residential Permits - Thous. Single-family units Multi-family units Total Building Permits Value - $ Mil. ALABAMA Nonresidential Building Permits Total Nonresidential Industrial Bldgs. Offices Stores Hospitals Schools $ Mil. 387 78 55 66 21 8 398 78 54 67 21 8 434 46 62 71 24 5 11 7 - 13 + 60 Residential Building P e r m i t s Value - $ Mil. Residential Permits - Thous. Single-family units Multi-family units Total Building P e r m i t s Value - $ Mil. FLORIDA Nonresidential Building Permits Total Nonresidential Industrial Bldgs. Offices Stores Hospitals Schools $ MiL 3,154 362 624 555 97 18 3,269 381 639 563 157 20 4,189 481 582 593 125 26 25 - 25 + 7 6 - 22 - 31 Residential Building P e r m i t s Value - $ MiL Residential Permits - Thous. Single-family units Multi-family units Total Building Permits Value - $ MiL GEORGIA Nonresidential Building Permits Total Nonresidential Industrial Bldgs. Offices Stores Hospitals Schools $ Mil. 1,020 160 240 103 26 35 1,045 156 247 104 27 34 1,015 177 250 116 21 14 + 0 10 4 - 11 + 24 + 150 Residential Building Permits Value - $ MiL Residential Permits - Thous. Single-family units Multi-family units Total Building P e r m i t s Value - $ Mil. LOUISIANA Nonresidential Building Permits Total Nonresidential Industrial Bldgs. Offices Stores Hospitals Schools $ Mil. 884 88 265 162 15 25 905 91 263 168 21 25 937 108 305 116 73 21 6 - 19 - 13 + 40 - 79 + 19 Residential Building P e r m i t s Value - $ MiL Residential Permits - Thous. Single-family units Multi-family units T o t a l Building P e r m i t s Value - $ MiL MISSISSIPPI Nonresidential Building Permits Total Nonresidential Industrial Bldgs. Offices Stores Hospitals Schools $ MiL 170 13 42 38 4 1 174 15 42 39 4 2 177 20 37 37 8 1 4 35 + 14 + 3 - 50 0 Residential Building Permits Value - $ MiL Residential Permits - Thous. Single-family units Multi-family units Total Building P e r m i t s Value - $ MiL TENNESSEE Nonresidential Building Permits Total Nonresidential Industrial Bldgs. Offices Stores Hospitals Schools $ Mil. 660 35 107 110 40 7 698 41 133 114 33 6 607 65 105 120 9 8 + AUG 1982 JUL 1982 AUG 1981 ANN % CHG 12-month Cumulative Rate - - - - 11 + 70 - - 35,018 34,772 47,399 - 26 463.5 395.1 461.5 392.6 681.1 482.4 - 32 - 18 82,178 82,862 99,877 - 18 6,432 6,467 9,890 - 35 93.9 81.3 92.7 83.9 148.5 126.6 - 37 - 36 12,707 12,956 17,259 - 26 221 239 394 - 44 3.9 4.0 4.0 5.2 7.6 8.0 - 49 - 50 607 6 37 827 - 27 3,993 4,062 6,841 - 42 50.0 50.5 49.6 52.6 89.8 89.4 - 44 - 44 7,147 7,332 11,030 - 35 1,118 1,077 1,202 21.4 10.7 20.9 10.0 25.2 9.8 2,137 2,123 2,217 - 7 - 15 + 9 - 4 _____ - - - 9 - 46 + 2 8 +344 - 13 Residential Building Permits Value - $ MiL Residential Permits - Thous. Single-family units Multi-family units Total Building P e r m i t s Value - $ MiL 580 579 681 - 15 9.2 8.5 9.2 8.5 11.4 9.1 - 19 - 7 1,463 1,483 1,618 - 10 150 142 232 - 35 3.1 2.0 2.8 1.9 4.5 3.7 - 31 - 46 321 315 409 - 22 371 368 541 - 31 6.3 5.6 6.2 5.7 10.0 6.6 - 37 - 15 1,031 1,066 1,158 - 11 _ Data supplied by the U. S. Bureau of the Census, Housing Units Authorized By Building P e r m i t s and Public Contracts, C-40. Nonresidential data excludes the cost of construction for publicly owned buildings. The southeast data represent the t o t a l of the six states. The annual percent change calculation is based on the most recent month over prior year. Publication of F. W. Dodge construction c o n t r a c t s has been discontinued. http://fraser.stlouisfed.org/ Federal Reserve Bank of St. Louis 70 FEDERAL RESERVE B A N K O F A T L A N T A GENERAL Personal Income-? bil. SAAR (Dates: 2Q, IQ, 2Q) Taxable Sales - $ bil. Plane Pass. Arrivals (thous.) JUL Petroleum Prod, (thous. bis.) Consumer Price Index 1967=100 Kilowatt Hours - mils. (MAY) SOUTHEAST Personal Income-$ bil. SAAR (Dates: 2Q, 1Q, 2Q) Taxable Sales - $ bil. Plane Pass. Arrivals (thous.) JUL Petroleum Prod, (thous. bis.) Consumer Price Index 1967=100 Kilowatt Hours - mils. (MAY) ALABAMA Personal Income-? bil. SAAR (Dates: 2Q, 1Q, 2Q) Taxable Sales - $ bil. (JUN) Plane Pass. Arrivals (thous.) JUL Petroleum Prod, (thous. bis.) Consumer Price Index 1967=100 Kilowatt Hours - mils. (MAY) FLORIDA Personal Income-? bil. SAAR (Dates: 2Q, 1Q, 2Q) Taxable Sales - ? bil. Plane Pass. Arrivals (thous.) JUL Petroleum Prod, (thous. bis.) Consumer Price Index - Miami Nov. 1977 = 100 Kilowatt Hours - mils. (MAY) GEORGIA Personal Income-? bil. SAAR (Dates: 2Q, 1Q, 2Q) Taxable Sales-? bil. (1Q-4Q-1Q) Plane Pass. Arrivals (thous.) JUL Petroleum Prod, (thous. bis.) Consumer Price Index - Atlanta 1967 = 100 Kilowatt Hours - mils. (MAY) LOUISIANA Personal Income-? bil. SAAR (Dates: 2Q, 1Q, 2Q) Taxable Sales - ? bil. Plane Pass. Arrivals (thous.) JUL Petroleum Prod, (thous. bis.) Consumer Price Index 1967 = 100 Kilowatt Hours - mils. (MAY) Personal Income-$ bil. SAAR (Dates: 2Q, IQ, 2Q) Taxable Sales - $ bil. Plane Pass. Arrivals (thous.) JUL Petroleum Prod, (thous. bis.) Consumer Price Index 1967 = 100 Kilowatt Hours - mils. (MAY) Personal Income-$ bil. SAAR (Dates: 2Q, 1Q, 2Q) Taxable Sales - $ bil. Plane Pass. Arrivals (thous.) JUL Petroleum Prod, (thous. bis.) Consumer Price Index 1967 = 100 Kilowatt Hours - mils. (MAY) ANN. % CHG. SEP 1982 AUG 1982 SEP 1981 2,541.5 N.A. N.A. 8,684.3 2,518.6 N.A. N.A. 8,669.1 2,370.9 N.A. N.A. 8,640.2 + 1 293.2 158.6 292.2 167.4 279.3 160.6 + 5 - 1 301.8 N.A. 4,353.0 1,386.5 297.0 N.A. 4,192.5 1,387.5 280.5 N.A. 4,292.5 1,421.3 + 8 N.A. 24.9 N.A. 25.4 N.A. 26.0 - 4 33.6 20.9 107.8 57.0 33.0 20.9 112.9 56.5 31.7 20.2 127.1 60.5 + 6 + 3 -15 - 6 N.A. 3.4 N.A. 3.5 N.A. 3.7 - 8 111.3 66.7 2,277.5 73.0 SEP 156.1 7.0 108.7 66.6 2,056.9 75.0 JUL 155.1 6.9 102.1 65.3 1,961.3 97.4 SEP 150.2 7.0 + 9 + 2 + 16 -25 52.5 34.5 1,504.1 N.A. AUG 295.6 3.7 51.8 34.3 1,564.1 N.A. JUN 291.1 3.9 49.2 32.1 1,750.8 N.A. AUG 276.1 3.8 + 7 + 8 -14 43.7 N.A. 273.5 1,164.0 43.0 N.A. 259.4 1,164.0 40.4 N.A. 258.0 1,168.0 + 8 N.A. 4.3 N.A. 4.3 N.A. 4.4 19.7 N.A. 32.8 92.5 18.9 N.A. 33.0 92.0 18.5 N.A. 39.1 95.4 N.A. 1^6 N.A. U> N.A. 1J5 41.0 25.5 157.3 N.A. 41.5 25.4 166.2 N.A. 38.6 23.5 156.2 N.A. N.A. 4.9 N.A. 5.2 N.A. 5.5 + 7 + 1 - 2 + 4 0 + 7 - 3 + 6 - 0 - 2 + 6 -16 - 3 + 6 + 9 + 1 -11 SEP 1982 SEP 1981 AUG (R) 1982 ANN. % CHG. Agriculture Prices Rec'd by Farmers 136 Index (1977=100) Broiler Placements (thous.) 78,072 60.00 Calf Prices (? per cwt.) 27.1 Broiler Prices (« per lb.) 5.28 Soybean Prices (? per bu.) 209 Broiler Feed Cost (? per ton) 133 80,612 61.90 26.3 5.59 215 133 77,721 61.40 26.3 6.21 222 + 2 + 0 - 2 + 3 -15 - 6 Agriculture Prices Rec'd by Farmers 120 Index (1977=100) Broiler Placements (thous.) 30,677 55.58 Calf Prices (? per cwt.) 26.6 Broiler Prices ($ per lb.) 5.43 Soybean Prices (? per bu.) 204 Broiler Feed Cost (? per ton) 120 31,843 57.84 25.6 5.83 213 125 30,723 56.77 24.9 6.34 219 - 4 - 0 - 2 + 7 -14 - 7 Agriculture Farm Cash Receipts - ? mil. 903 (Dates: JUN, JUN) Broiler Placements (thous.) 9,478 Calf Prices (? per cwt.) 56.50 25.0 Broiler Prices (<t per lb.) 5.54 Soybean Prices (? per bu.) Broiler Feed Cost (? per ton) 205 9,938 57.20 24.5 5.73 210 904 9,770 55.70 24.0 6.09 235 - 0 - 3 + 1 + 4 - 9 -13 Agriculture Farm Cash Receipts - ? mil. (Dates: JUN, JUN) 2,905 Broiler Placements (thous.) 1,795 59.70 Calf Prices (? per cwt.) Broiler Prices (<t per lb.) 27.0 5.54 Soybean Prices (? per bu.) 210 Broiler Feed Cost (? per ton) 1,839 61.10 25.0 5.73 220 2,588 1,800 59.90 25.0 6.09 230 +12 - 0 - 0 + 8 - 9 - 9 Agriculture Farm Cash Receipts - ? mil. 1,270 (Dates: JUN, JUN) Broiler P l a c e m e n t s (thous.) 12,281 49.10 Calf Prices (? per cwt.) 26.5 Broiler Prices ($ per lb.) 5.50 Soybean Prices (? per bu.) 200 Broiler Feed Cost (? per ton) 12,423 54.20 25.0 6.25 215 1,265 12,312 52.50 24.5 6.46 210 + 0 - 0 - 6 + 8 -15 - 5 Agriculture Farm Cash Receipts - ? mil. 561 (Dates: JUN, JUN) N.A. Broiler Placements (thous.) 58.50 Calf Prices (? per cwt.) 27.5 Broiler Prices (<t per lb.) 5.51 Soybean Prices (? per bu.) Broiler Feed Cost (? per ton) 250 N.A. 60.70 27.5 5.84 250 595 N.A. 58.60 26.5 6.57 245 - 6 Agriculture Farm Cash Receipts - $ mil. (Dates: JUN, JUN) 839 Broiler P l a c e m e n t s (thous.) 5,927 Calf Prices ($ per cwt.) 57.50 Broiler Prices (<t per lb.) 29.0 Soybean Prices (? per bu.) 5.31 Broiler Feed Cost ($ per ton) 200 805 5,574 58.70 26.5 6.30 205 + 4 5,973 58.10 28.0 5.83 205 Agriculture Farm Cash Receipts - $ mil. 713 (Dates: JUN, JUN) 1,217 Broiler Placements (thous.) 51.90 Calf Prices ($ per cwt.) 25.5 Broiler Prices (<t per lb.) 5.36 Soybean Prices ($ per bu.) Broiler Feed Cost ($ per ton) 176 1,326 55.90 25.5 5.63 181 647 1,266 54.80 25.0 6.28 195 + 10 - 4 - 5 + 2 -15 -10 - - 0 + 4 -16 + 2 + 6 - -16 - 2 Notes: Personal Income data supplied by U. S. Department of Commerce. Taxable Sales are reported as a 12-month cumulative total. Plane Passenger Arrivals are collected from 26 airports. Petroleum Production data supplied by U. S. Bureau of Mines. Consumer Price Index data supplied by Bureau of Labor Statistics. Agriculture data supplied by U. S. Department of Agriculture. Farm Cash Receipts data are reported as cumulative for the calendar year through the month shown. Broiler placements are an average weekly r a t e . The Southeast data represent the t o t a l of the six s t a t e s . N.A. = not available. The annual percent change calculation is based on most recent data over prior year. revised. Digitized Rfor= FRASER http://fraser.stlouisfed.org/ NOVEMBER 1982, E C O N O M I C Federal Reserve Bank of St. Louis REVIEW 2 + 9 71 Federal Reserve Bank of Atlanta P.O. Box 1731 Atlanta, Georgia 30301 Bulk Rate U.S. Postage Address Correction Requested Atlanta, Ga. P e r m i t 292 Lß LIBRARY FEDERAL RESERVE BANK PHILADELPHIA PR PAID IRIDI