,1, .,,- . ...\...-. nun-Juan. nununv‘ IIHIUWHMllllHIHIHIIIHllllHHIIIIHHZIWIWIHIHI A} l O 949 (o ‘0 31293005612134 Tim LIBRARY Michigan State University This is to certify that the dissertation entitled A Microeconomic Model of the Banking Firm and the Pricing of New Automobile Loans at Commercial Banks presented by Mark Arlington White I has been accepted towards fulfillment of the requirements for Ph . D . degree in Business Administration 6; ‘ ; Major professor D ; MSU Lt an Affirmative Action/Equal Opp tttttt ' ty Institution 0-12771 MSU RETURNING MATERIALS: Place in book drop to LIBRARIES remove this checkout from , your record. FINES will be charged if book is returned after the date stamped below. " if: H 4 its , Qiti.§ ifiéfi fl 9’?“ car. (151333 ,. A MICROECONOMIC MODEL OF THE BANKING FIRM AND THE PRICING OF NEW AUTOMOBILE LOANS AT COMMERCIAL BANKS BY Mark Arlington White A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Finance and Insurance 1989 msux ABSTRACT A MICROECONOMIC MODEL OF THE BANKING FIRM AND THE PRICING OF NEW AUTOMOBILE LOANS AT COMMERCIAL BANKS BY Mark Arlington White Consumer debt accounts for a large and growing portion of the financial liabilities of the United states. Many observers see the increased willingness of households to incur debt as a sign of weakness and the imminent collapse of the financial system. The growth in non-residential debt financing, or instalment credit, is viewed as a particular worry because much of it is used to fund investment in nondurable goods. At the same time, commercial banks are eager to expand their presence in the consumer lending arena to offset decreased demand for commercial and international loans. These factors contribute to the growing interest in consumer loan pricing at commercial banks. This dissertation develops a microeconomic model of the banking firm in the theory-of—the-firm tradition of Klein (1971). It extends Slovin and Sushka’s (1983) work on the pricing of commercial loans by introducing an alternative outlet for bank funds (consumer loans) and subjecting the resultant predictions of the theoretical model to empirical verification. There are three primary differences between the work at hand and previous studies: 1) It introduces a ii Mark Arlington White more comprehensive model of bank loan portfolio behavior, 2) The implications of the model are tested using the appropriate microeconomic data and 3) It is the first study to investigate the behavior of consumer loan rates vis-a-vis commercial loan and competitors’ rates over time. This paper provides tentative evidence that the automobile loan market is highly competitive with respect to automobile loan rates; these findings should be pursued with a longer and more comprehensive data series. iii ACKNOWLEDGEMENTS I am grateful to my dissertation chairman, Professor Alan Grunewald, the members of my dissertation committee, Professor Richard Simonds and Professor Robert Rasche, and Professor John O’Donnell for their guidance and advice throughout this undertaking. Don Mann of the Michigan Department of Commerce and Gary Meyers of Meyers and Associates graciously allowed access to data collected by their respective organizations. I thank Manjit Virdi for his assistance with the computer analyses and wish to acknowledge the support and helpful comments of my comrades—in-arms, Maurice Tse, Rick Osborne and Marcus Chung. Finally, I am deeply indebted to my parents, Harry and Nancy White, and my beautiful bride—to-be, Susie Reid, for their encouragement, patience and love. iv TABLE OF CONTENTS I. Introduction ... ......... . ................................ 1 Problem Identification ................................. 4 Research Objectives..... ............................... 5 Contributions and Significance..... ............ . ....... 5 II. Literature Review ....................................... 7 Consumer Credit Markets ............... .. ..... . ......... 7 Types and Characteristics of Consumer Credit ...... 9 History of Automotive Finance .................... 21 Regulation of Consumer Lending ................... 34 Consumer Credit Knowledge ........................ 38 Models of the Banking Firm ............................ 41 Why Do Banks Exist? .............................. 42 Portfolio Management Models ...................... 46 Theory-of—the-Firm Models ........................ 48 Credit Rationing Models .......................... 49 Determinants of Bank Loan Rates ....................... 50 Cost Factors ..................................... 52 Competitive Factors .............................. S4 Empirical Studies ................................ 57 Determinants of Bank Deposit Rates .................... 62 III. A Model of the Consumer Loan Rate ..................... 65 Assumptions and Objective Function .................... 66 First Order Conditions ................................ 69 Second Order Conditions ............................... 70 Solution and Predictions .............................. 70 IV. Methodology ............................................ 72 Sources of Data ........... . ........................... 72 Consumer Loan and Deposit Rate Data .............. 72 Security Rates ................................... 76 Other Variables .................................. 78 Econometric Analysis .................................. 78 V. Results and Discussion .................................. 81 Hypotheses ............................................ 81 Regression Results... ................................. 83 Automobile Loan Rates ............................ 83 Deposit Rates .................................... 97 Discussion ........................................... 103 Is the Model Wrong? ............................. 104 Is the Data Suspicious? ......................... 105 Is the World Different from Theory? ............. 106 VI. Summary and Conclusions ............................... 108 The Theoretical Model ................................ 109 Empirical Results .................................... 110 Implications ......................................... 113 Areas for Future Research ............................ 115 vi Appendix A: A Mathematical Model of the Banking Firm ...... 116 Appendix Bl: Characteristics of Financial Institutions ....125 Appendix 82: Financial Institutions Ranked by Size ........ 129 Appendix C1: Auto Loan Rates (Descriptive Statistics) ..... 130 Appendix C2: Deposit Rates (Descriptive Statistics) ....... 131 References ................................................ 132 vii Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table Table l. 2. 3. 4. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. LIST OF TABLES 1986 Credit Market Borrowing By Sectors ............ 9 Outstanding Consumer Installment Credit ........... 14 Consumer Debt by Family Characteristics ........... 18 Federal Consumer Credit Legislation ............... 35 Development of the Theory of the Banking Firm ..... 45 Loan Portfolios of Insured Commercial Banks ....... 50 Costs of Installment Lending in 1985 .............. 52 Functional Cost Analysis of Bank Lending .......... 53 Empirical Models of Bank Loan Rates ............... 58 Correlation Matrix of Security Rates ............. 77 Hypothesis Testing ............................... 82 Automobile Loan Rate Regressions ................. 84 Signed Coefficients of Loan Rate Regressions ..... 88 Automobile Loan Rates vs. Competitors’ Rates ..... 91 Institutions Classified by Market Structure ...... 94 Intraregional Variation of Auto Loan Rates ....... 96 Intra-Size Class Variation of Loan Rates ......... 97 Deposit Rate Regressions ......................... 98 Signed Coefficients of Deposit Regressions ...... 102 Summary of Empirical Testing Results ............ 111 viii Figure Figure Figure Figure Figure Figure Figure wNH LIST OF FIGURES U.S. Credit Market Debt Outstanding ............... 8 . Major Types of Consumer Installment Credit ....... 11 . Major Holders of Consumer Installment Credit ..... 16 Outstanding Household Debt ....................... 20 Average New-Auto Finance Rates, 1924—1962 ........ 25 GMAC's Six Percent Plan .......................... 27 The Effects of Binding Usury Ceilings ............ 35 ix I. IIHPEKMDUKEPICN! Commercial banks have historically concentrated their lending activities in three areas: business loans, international loans and consumer loans. Consumer lending has grown more attractive in recent years as developments in the commercial and international loan markets have curtailed lending opportunities. At the same time, deregulation of the financial services industry has increased the uncertainty of profits derived from the extension of consumer credit. These and other factors have contributed to a renewed interest in consumer lending and the determinants of consumer loan rates. Commercial loan demand has fallen off for two primary reasons. First, large corporations always have had the alternative to finance their short-term needs by issuing commercial paper. The explosive growth in money market mutual funds has increased the demand for these securities leading to 1) More attractive terms for the established market participants and 2) The issuance of paper by less well-established or more risky firms. Second, the growth of the junk bond market has increased the access of smaller firms to the capital markets and decreased their reliance on bank financing. International lending by commercial banks has followed a similar course. The weak U.S. dollar and unattractive levels of perceived default risk have limited banks’ foreign lending opportunities. Before the Depository Institutions Deregulation and Monetary Control Act of 1980 and the Garn—St. Germain Depository Institutions Act of 1982, non-bank financial institutions were relatively restricted in the scope of their lending activities, while commercial banks were similarly restricted in their investment opportunities. A wave of unprecedented high interest rates and the widespread availability of computer technologies, coupled with a strong pro-business Presidential administration led to the collapse of many barriers to entry throughout the industry. In addition to their traditional competitors among the other depository institutions (savings and loans, credit unions, and mutual savings banks), banks must now defend their loan markets against intrusion from brokerage houses, automotive finance companies, and retailers. Each of these institutions brings a particular strength to the consumer credit market, and it is not yet clear who the victors will be. If the retailing industry’s recent experience may be taken as an example, one might anticipate a few large and growing chains of discount financial centers, in which price is the main competitive weapon, e.g. Wal-Mart and K—Mart. Specialty financial intermediaries might also prosper in well-defined niches, competing primarily on quality and service. On a macroeconomic level, the availability of consumer credit is an important factor in determining the level of consumer spending. Recovery from the 1973-75 and 1980-81 recessions was speeded by the ability of consumers to defer payment of goods purchased. On the other hand, the burgeoning debt held by consumers has significant implications for governmental and monetary policy-making. Population demographics are such that record numbers of younger married couples are entering the economy. Traditionally, these households have been the primary users of consumer credit and these same individuals appear less reticent about taking on higher debt levels than earlier generations have been. Consumer loan rates are of special interest for one last reason. Since biblical times, usury regulations, or limitations on the rate of interest one may charge on loans, have been a part of many cultures’ moral and legal codes. The recent bout of high interest rates and the consequent effects of many state usury laws has generated some debate regarding the desirability of interest rate ceilings and their benefit to the consumer. The importance of consumer credit in the American economy and commercial banks' willingness and ability to provide credit at a fair price will have an important impact on our financial markets. An article in Fortune on Citicorp, the nation’s largest commercial bank, notes: “For the past five years consumer revenues have been growing at 30% a year, three times as fast as the wholesale banking side, which lends to corporations and governments. If the trend continues, virtually all Citicorp’s revenues will come from consumer banking within 15 years. Retail banking m is rapidly becoming the only job that matters m ” (Norton, 1987). PROBLEM IDENTIFICATION Despite the increased interest in consumer lending shown by commercial banks and the importance of consumer credit in determining consumer spending patterns, little theoretical work has been done in the area of consumer loan pricing. Most of the literature on the theory of the banking firm fails to differentiate between the different types of loans present in a bank's portfolio, although significant differences in size, risk, and market characteristics clearly exist. In fact, very few studies directly address loan pricing at all. Slovin and Sushka (1983) derived a model of the commercial loan rate which was applied (incorrectly) to the behavior of consumer loan rates in a recent working paper by Sullivan and Fain (1984). Other empirical investigations of bank loan rates have generally concentrated on geographic and competitive differences with mixed conclusions. The lack of a substantive theoretical base presents difficulties for bank managements wishing to maximize the value of their firms. The recent wave of bank deregulation has led to the entry of many new competitors in the financial services arena. Now more than ever, it is imperative that a bank have a thorough understanding of the characteristics underlying the supply and demand of its various product offerings. As banks’ monopoly power decreases, lower margins of error in the pricing of bank services will be allowed. RESEARCH OBJECTIVES The objective of this dissertation is to develop a microeconomic model of the banking firm with explicit attention given to the pricing of consumer loans. The model will be tested with microeconomic cross-sectional time—series data from depository financial institutions in six metropolitan areas of the United States. An important goal of this research is to identify the influence of competitors’ offerings on consumer loan rates charged by commercial banks; this information may be useful in the formulation of strategic policies by bank managements. CONTRIBUTIONS AND SIGNIFICANCE The results of this research will make several theoretical and practical contributions. As previously noted, no formal model of the consumer loan rate has been proposed in the academic literature. Moreover, the microeconomic model proposed in this dissertation will be tested with micro-, rather than macro-, economic data, which has NOT been the case in other examinations of loan rate behavior (Slovin and Sushka, 1983; Sullivan and Fain, 1984; Lee, 1985). The development and testing of this model should add to our understanding of the the behavior of consumer loan rates at depository financial institutions. Bank managements should find the results of this dissertation useful in establishing a framework for consumer loan pricing. High loan rates may affect shareholder wealth by discouraging profitable loans to good customers. Low loan rates may affect shareholder wealth through lost profit opportunities. Automobile finance companies may gain insight into their competitors’ behavior by a reading of this paper. Although the model is intended as a general approach to consumer lending, its testing is limited to the rates charged on automobile loans, which comprise the largest category of installment lending in the United States. The recent use of incentive financing plans to bolster automobile sales suggests that the automobile finance companies may have a keen interest in commercial banks’ reactions to their pricing policies. Finally, consumers may benefit from the results presented in this dissertation if they are in a position to shop for lower-cost financing. Alternative sources of consumer credit generally increase market efficiency, and knowledge of the loan pricing decision may increase one’s ability to negotiate or choose better loan terms. II. ILITHNRAJHDKE IUTVIIH' This review is developed in sections. First a general overview of consumer credit markets is presented, with especial attention to the history of automobile finance, regulation of consumer lending, and consumer credit knowledge. Second, the academic literature on the theory of the banking firm is reviewed and characterized. Third, previous work on the determinants of bank loan and deposit rates is discussed. CONSUMER CREDIT MARKET S The Board of Governors of the Federal Reserve defines consumer credit as including all short— and intermediate-term credit extended through regular business channels to finance the purchase of commodities and services for personal consumption, or to refinance debts incurred for such purposes (Chapman, 1967). A significant portion of the nation’s outstanding credit market debt is held in this form. In 1985, consumers were responsible for $1066 billion of nonmortgage debt, or about 33 percent of the total debt held by individuals (Figure 1). siwso EGEfifis 36 E...§=E-§z :8 $.20 3.5356 300 3333.33 033an Sgtl EOEEQ—aéoz as, ...... c.30Eo._._< c033 mama c053 3N8 853 Dan; c023 033 33.5330 38 vague: uaeouo :3 3.6 are: 595 BR ¢o==a vwwoo . H 9.25.: Households borrowed $291 billion in 1986, making them the largest single sector in net credit market borrowings (Table 1). Approximately one-fourth of these funds were raised in consumer credit markets. Table 1. 1986 Net Credit Market Borrowing Bx Sectors1 SECTOR .AMOUNT PERCENT Households $291.2 billion 27.2 % Nonfinancial Business 262.2 billion 24.5 US Government 214.3 billion 20.0 Government Agencies 171.1 billion 16.0 State/Local Governments 60.0 billion 5.6 TYPES AND CHARACTERISTICS OF CONSUMER CREDIT Non-Installment Credit Consumer credit is classified as installment (multiple repayment periods) or non—installment (single repayment period). There are three major types of non-installment credit: charge accounts, service credit, and single-payment loans. Charge Accounts Unpaid balances owed to retailers for the purchase of various goods comprise this category of non- installment credit market debt. Some travel and entertainment cards debts, e.g. American Express, are 1Board of Governors of the Federal Reserve System, Flow of Funds Accounts (March 1987) 10 included in this category. This type of credit is declining in popularity as VISA and MasterCharge accounts gain ever- wider usage. Service Ckedit These are accounts owed to service professionals and institutions, such as doctors, lawyers and utility companies. Singlefipayment Loans Also called ‘bullet loans’, this category refers to loans which are repaid in a single payment. Rates are usually low because most of these loans are heavily collateralized. Security and policy loans on one’s life insurance are examples of this type of credit. Installment Credit Installment credit is the more important form of consumer credit and is the focus of this dissertation. Eighty percent of consumer credit is in this form. There are four major categories of installment debt: Automobile Credit, Revolving Credit, Mobile Home Credit and Other. The volume of outstanding consumer installment credit has increased tremendously in recent years (Figure 2). ll CONSUMER INSTALLMENT CREDIT MAJOR TYPES SEASONALLY ADJUSTED, END OF QUARTER RATIO SCALE, BILLIONS OF DOLLARS 400 l 200 100 60 cums 4O REVOLVING ITTII 20 AUTOMOBILE PAPER 10 MOBILE HOME 1 ll I. LiLLiillLJlilLiiliLiiLiLiUiLJiiiLJulLl 1950 1955 1960 1965 1970 1975 1980 1985 1990 Figure 2. Major Types of Consumer Installment Credit2 Automobile credit This is the largest category of consumer installment credit, with over $240 billion outstanding. The importance of the automobile in American culture is well-documented; most purchases are made with borrowed funds. In addition to the volume of extensions and outstandings, the Federal Reserve reports the average interest rate, maturity, loan size, and loan-to-value ratio for new and used cars in the MOnthly Bulletin. Automobile 2Board of Governors of the Federal Reserve System, Historical Chart Book (1986) 12 loan rates will be used to test the consumer loan rate model developed later in this dissertation. Revolving Credit Credit card debts comprise the majority of revolving consumer credit. Nearly one trillion of these instruments are in estimated use worldwide (Rose, 1986). Revolving credit refers to the existence of a prearranged line of credit where the timing of both extensions and repayments are at the option of the creditor. Another form of revolving credit is the overdraft loan available at many depository institutions. nubile acme Credit This type of credit facilitates the purchase of mobile homes by consumers. Other The final category of consumer installment credit encompasses loans made for a variety of purposes. Household expenses, debt consolidation, vacations, and purchases of durable goods other than automobiles are included in this lending miscellany. Installment Credit Terms In an installment loan contract, the issuer agrees to lend a specified sum for a specified period of time at a specified rate of interest. The loan itself is to be repaid according to a payment schedule, which determines the amount and timing of these payments. Each of these terms affects the amount of credit lenders are willing to supply and the amount demanded by consumers. From the lender’s perspective, increasing loan size decreases per-dollar loan costs, as l3 origination costs are generally fixed. Increasing the term of the loan exposes the lender to greater default risk and to greater interest rate risk if the loan rate is fixed. The response of borrowers to installment credit terms will be reviewed in a later section. Lenders appear to Specialize in the types and terms of loans offered. Interest rates on loans obtained from finance companies usually are higher than those on loans obtained from commercial banks. This difference can be explained in part by lower individual loan sizes and the greater risk of borrowers at finance companies (Boczar, 1975, 1978). A trend towards longer maturities has developed in recent years, particularly in automotive lending. Higher automobile prices and borrowers’ desires for low monthly payments have contributed to this pattern (Luckett, 1986). Suppliers of Consumer Credit Consumer debt obligations are held by a number of financial institutions. Commercial banks, finance companies, credit unions and thrift institutions (mutual savings banks and savings and loans) are the major suppliers of installment credit. Retailers and gasoline companies are net suppliers of revolving and charge card credit. The relative amounts and types of installment credit supplied by these intermediaries is in Table 2. 14 Table 2. Outstandin Consumer Installment Credit COMMERCIAL FINANCE CREDIT GASOLINE BANKS COMP ANI ES UNIONS THRI FTS RETAILERS COS . TOTAL Automobile $118.1 5 98.9 $ 47.1 $ 18.8 - - $282.8 Revolving 109.6 - 8.4 15.5 $ 37.7 $ 3.7 174.9 Mobile Home 8.9 7.4 - 9.5 - - 25.9 Other 65.5 37.5 30.0 24.4 6.0 - 163.4 —- _ _ u _ -—- -_ Total $302.0 $143.8 $ 85.6 $ 68.2 $ 443.6 s 3.7 $646.9 Source: Federal Reserve Bulletin, (November 1988), Table 1.55. Figures in millions of dollars. Commercial Banks Commercial banks make short and intermediate-term loans to individuals for the purchase of automobiles, home improvements, education and a variety of other goods and services. At the end of April 1987, banks held $263 billion, or about 45% of total installment credit outstanding, making them the largest direct participants in the consumer credit markets. Banks provide indirect consumer financing as well through loans and loan commitments to finance companies and retailers. Finance companies Finance companies extend credit to consumers in the form of personal loans and automobile paper. Formerly, a distinction was made between companies which made direct cash loans to consumers (personal finance companies) and companies which purchased installment sales contracts from dealers or retailers (sales finance companies). Presently, a more useful dichotomy might be made between the captive finance companies and independent finance companies. Captive finance companies such as General Motors Acceptance 15 Corporation (GMAC), IBM Credit Corporation and Sears Roebuck Acceptance Corporation provide financing to facilitate the sale of their parent company’s products. Independent finance companies specialize in the provision of short-term cash loans to higher risk individuals. C.I.T. Group Holdings, Associates Corporation of North America and Household Financial Services Corporation are among the largest independent finance companies in the United States (Cacace, 1987). Credit Uhions Credit unions specialize in the granting of credit to persons sharing a common bond. Members utilize these funds to purchase automobiles, homes and to fund nondurable items like education and vacations. They are the most rapidly-growing market participants. Thrift Institutions The recently-deregulated thrift institutions (savings and loans, mutual savings banks) are a growing presence in the increasingly-competitive consumer credit arena offering automobile and personal loans to individuals in more-or-less restricted geographical areas. Consumer installment credit experienced rapid and widespread growth in the post-World War II years. The total outstanding rose from $4.4 billion in 1946 to $577.8 billion in 1986, an annual rate of increase of 13.3 percent per year. Population growth, rising incomes, and the increased availability and widespread use of health and unemployment insurance have been cited as important factors in this growth 16 (Chapman, 1967). Figure 3 presents information on the growth of consumer debt held by financial intermediaries. CONSUMER INSTALLMENT CREDIT MAJOR HOLDERS SEASONALLY ADJUSTED, END or QUARTER RATIO SCALE, BILLIONS OF DOLLARS / 200 '- COMMERCIAL BANKS I’,‘~____,r’ 100 - If,_—/” 400 60 40 — 20 - RETAIL OUTLETS 10 _. FINANCE .. COMPANIES CREDIT UNIONS 6 4 2 1 .6 4 'lllLllllLJlllLllllLllllLllllLJlllLJlll ' 1950 1955 1960 1965 1970 1975 1980 1985 1990 Figure 3. Major Holders of Consumer Credit3 Consumer’ Credit Demand The demand for consumer credit has been investigated using annual surveys of consumer attitudes towards various loan characteristics and repayment policies. From 1947 through 1970 these were conducted by the Survey Research 3Board of Governors of the Federal Reserve System, Historical Chart Book (1986) 17 Center at the University of Michigan (Katona, Mandell and Schmiedeskamp, 1970). In the late 19703, balance-sheet data were collected as part of a consumer credit sponsored by the Federal Reserve Board (Durkin and Elliehausen, 1977). The 1983 Survey of Consumer Finances is the most recent comprehensive study generally available (Avery et al.; 1984a, 1984b). Each of these surveys used personal interviews to determine the distribution of assets and liabilities among families with various characteristics (Table 3). 18 Table 3. Consumer Debt by Family Characteristics PERCENT OF MEAN MEDIAN FAMILIES (dollars) (dollars) Family Income (dollars) Less than 5000 33 2834 677 5000 - 7499 40 1919 573 7500 - 9999 48 4152 1006 10000 - 14999 54 3452 1451 15000 19999 66 4295 1639 20000 24999 72 4149 2336 25000 29999 72 4632 2929 30000 39999 77 5138 3594 40000 49999 80 7079 4365 50000 or more 75 12,772 5529 Age of Family Bead (years) Under 25 64 3584 2263 25 - 34 77 4781 2265 3S - 44 79 6673 3030 4S - 54 71 5780 3152 55 - 64 57 6325 1700 65 - 74 31 3567 943 75 and over 15 1117 308 Race of Family Bead Caucasian 63 5577 2503 Nonwhite and Hispanic 60 4573 1330 Life-Cycle State of Family Bead Under 45 years Unmarried, no children 64 4864 1900 Married, no children 86 4877 2949 Married, with children 83 5922 3076 45 years and over Head in labor force 66 6403 2949 Head retired 27 2967 677 All ages Unmarried, with children 65 4433 1135 All Families 62 5400 2382 Source: Avery et al., (1984b) Up to an income of $20,000 to $24,999, the greater the income, the larger the proportion of families that owe consumer debt. Above that level of income the proportion remains relatively stable, although as incomes rise the mean and median amounts of consumer debt outstanding increase. Outstanding debt increases with the age of the family head 19 until 44, when it begins to decline. Married couples are more likely to use consumer credit than unmarried persons. Peterson and Peterson (1981) investigated the relationship between borrower characteristics, downpayments and defaults. They conclude that the default rate on new car loans decreases as downpayments rise and that occupations are significantly related to default rates. It appears that workers most vulnerable to layoffs are more likely to default. The most useful predictor of default risk, however, is age. Younger borrowers (< 30 years) are more than three times more likely to default than older borrowers (> 30 years). Borrower segmentation of the consumer loan market was analyzed in a paper by Johnson and Sullivan (1981). They found that consumers were aware of differences in rates between banks and finance companies and that finance companies were patronized by choice, rather than necessity. They report, “m very little of the market segmentation of banks and finance company customers appears to be a supply-side phenomenon. Many - perhaps three— fourths - of the customers of finance companies were sufficiently creditworthy to warrant loans from commercial banks. Instead, most of the market segmentation apparently stemmed from consumers demands.” Consumer' Debt Burdens The record levels of household indebtedness have given rise to concerns ranging from economic recession through slowed future spending to a nationwide depression arising out 20 of borrower defaults (Kaufman, 1986; Malabre, 1987). Although the “debt burden”, or ratio of consumer installment credit to disposable income has increased steadily since World War II, it has taken a sharp upward turn in recent months (Figure 4). HOUSEHOLD DEBT OUTSTANDING PERCENTAGE OF DISPOSABLE PERSONAL INCOME AMOUNT OUTSTANDING: END OF YEAR, 1950-51; SEASONALLY ADJUSTED, END OF QUARTER, 1952- RATIO SCALE 100 30 L. TOTAL LIABILITIES 6O - F— HOME MORTGAGES ‘,,r-‘__~_.v"’ WM 30 r- CONSUMER 20 INSTALLMENT CREDIT 15 -* __ ’9’ ALL OTHER 10 fl 8 I-— 6 1,1 l LJ_l l ll_l II 1 LJ ill] l_ll 1 11.1 ll,l LJ [ill L_ll ll 1950 1955 1960 1965 1970 1975 1980 1985 1990 Figure 4. Outstanding Household Debt4 4Board of Governors of the Federal Reserve System, Historical Chart Book (1986) 21 Factors contributing to credit growth include the relaxation of binding usury ceilings, greater usage of adjustable-rate financing and increased lending involvement by savings and loans. Other influences, such as the growing use of credit for convenience purposes, the lengthening of maturities on new consumer loans, and changing demographics towards the 25-34 and 35-44 year old age categories (traditionally heavy users of consumer credit) inflate outstanding credit without suggesting any seriously negative impact. Moreover, when outstanding debt (a stock) is measured as a percent of total assets (a stock), rather than as a percent of income (a flow), the spectre of financial collapse appears less imminent (Luckett and August, 1985; Pearce, 1985). HISTORY OF AUTOMOTIVE FINANCE A review of the history of automotive finance is of interest for three reasons. First, automobile credit is the single largest component of consumer installment credit, which is the item of primary interest in this dissertation. Second, the development of automobile credit markets are responsible in large part for the market structure we observe today. Third, this study uses automobile loan data to test its model of the consumer loan rate. The history of automobile finance is necessarily tied to the invention and development of the automobile market. Henry Ford’s 1909 landmark decision to concentrate production 22 on the affordable Model T is credited as the cornerstone of the American automobile industry. Installment selling of cars was rare before WOrld War I, although by the mid- twenties nearly three-quarters of all cars were sold in this manner. Michelman (1970) writes, “Not only was the old American morality concerning buying on time about to crumble in the face of such a convenience, but generations of young couples were handed a suitable locale for various degrees of misconduct.” Shay’s (1964) article reviews the history of automobile finance through 1962. He identified four stages through which automobile financing had passed to that point in time. The Experimental, Period (1915-1922) Seligman (1927) has suggested that consumers were financing automobile purchases through Morris Plan Banks as early as 1910. By 1913, L. F. Weaver, a San Francisco automobile dealer, had established an organization to finance the installment sales of his cars (Phelps, 1952). The Guaranty Securities Company of Toledo was formed in 1915 to facilitate the sale of Willys-Overland cars but soon expanded to provide credit for the purchase of competing makes. In 1916 the Commercial Credit Company and the Commercial Investment Trust (C.I.T.) Corporation were established; these firms continue in existence today. At the close of 1917 nearly forty firms had devoted themselves to the financing of automobile purchases (Phelps, 1952). 23 In the early days of the automobile, most purchasers either paid cash or obtained credit from a sales finance company with whom the dealer had a ‘preferred’ relationship. In 1919, General Motors established the General Motors Acceptance Corporation (GMAC) to better service the credit needs of its automobile customers. That same year, the Industrial Finance Corporation (IFC) signed an exclusive agreement with the Studebaker Corporation to finance the sales of its cars. The involvement of auto manufacturers in automobile sales financing was to have an important impact on the development of this new market. A variety of methods developed for the setting and quoting of finance charges, due in part to the different backgrounds of the respective credit companies. Interest charges, service charges and insurance fees were manipulated to produce an often-confusing panoply of rates. The recession of 1920-21 sharply restricted automobile financing as sales dropped and interest rates rose. Shay (1964) estimated annual interest rates over the period 1919-22 at approximately 14 percent. The Stage of Manufacturer Participation (1922-1935) Following the recession, agreements between automobile manufacturers and large sales finance companies emerged providing mutually-advantageous subsidies. Dealers began to share in finance charge income, although reserve agreements (wherein manufacturers withheld a portion of the purchase 24 price of the automobile paper as a contingency against losses) encouraged them to maintain the quality of their loans. This subsidy system encouraged credit purchases and held rates at lower levels than would otherwise have been the case. The number of independent companies grew as well - Seligman (1927) reports nearly 1600 in operation as of 1925! By 1932, increases in reserves increased the rates charged purchasers as loan losses mounted during the Depression. Chrysler Corporation signed a financing agreement with Commercial Credit Corporation in 1926. The Hudson Motor Car Company affiliated itself with C.I.T. during this period as well. In 1928, Ford Motor Company formed the Universal Credit Corporation to finance the sale of its automobiles, which it Operated until its sale to C.I.T. Corporation in 1933. Finance charges fluctuated rather extensively over the period 1922-35 (Figure 5). 25 —a—a.—a—a 0"me _a h New-auto finance rate (four-large companies) -a_a_a ‘NQ I I 3 if I to l C dNQ‘U’IO’VG Short-term business loan rate, NYC C I llllllllLlll llllllll llll lLll [III III] 1924 '30 '35 '40 '45 '50 '55 '60 '62 Figure 5. Average New-Auto Finance rates, 1924-19625 Prior to 1930, banks were not significantly involved in consumer lending. A generally-conservative banking profession had strong predjudices against lending to consumers for the purpose of purchasing household goods or luxuries. Business loan demand was more than adequate for the placement of deposit funds. Operational problems, perceptions of high risk and an unwillingness to try something new and different prevented banks from immediately embracing these new opportunities. A number of observers SShay (1964), p. 213. 26 viewed consumer lending as morally reprehensible. Phelps (1952) reports, “The use of credit, and particularly the installment type, by consumers was characterized as ‘an economic sin,’ as ‘enervating the character because it leads straight to serfdom,’ as setting ‘utterly false standards of living,’ causing judgement to become ‘hopelessly distorted,’ and tending to ‘break down credit morale.’ It was attacked as ‘marking the breakdown of traditional habits of thrift,’ as tending to ‘weaken the moral fiber of the Nation,’ and as dangerous to the economy of the United States. It was accused of ‘breaking down character and resistance to temptations, to extravagance, and to living beyond one’s means, breeding dishonesty,’ causing ‘many young people to get their first experience of being deadbeats through yielding to temptations that are placed before them,’ and ‘creating a new type of criminal or causing professional deadbeats to shift to this new and highly lucrative opportunity.” By the mid-19303, a more tolerant attitude towards consumer borrowing had developed. Changing mores and weak business loan demand led banks to enter the consumer loan market during the Great Depression. The Period of Anti-Trust Activity (1935-1952) An abundance of funds and active commercial bank participation brought increased competition to the growing automobile credit market, causing a decline in loan rates. Federal anti-trust authorities became concerned over the long-term direction of the industry, and many states passed legislation designed to protect dealer and consumer interests. The first recorded incentive financing plan debuted in 1935. Known as the “Six Percent Plan”, it was introduced by GMAC and quickly copied by the other large 27 sales finance companies. Although created as a competitive weapon, it also sought to increase the public’s knowledge of finance rates. Figure 6 presents a facsimile of an advertisement of that era. GMAC General Motors Acceptance Corporation Reduces Time Payment Costs on New Cars with a New 6% Plan Simple as A, B, C A — Take your Unpaid Balance B - Add Cost of Insurance C* - Multiply by 6% - 12 month’s plan (One-half of one per cent per month for periods more or less than 12 months). That’s your whole financing cost. No extras. No service fees. No other charges. *In some states a small legal documentary is required. Figure 6. GMAC's Six Percent Plan6 The Federal Trade Comission (FTC) charged GM with misleading and deceptive advertising in 1936, arguing purchasers might misinterpret the offering as a 6 percent annual interest rate on the unpaid balance. General Motors and its subsidiaries were ordered to cease and desist in 1939 (Plummer and Young, 1940). The Department of Justice instituted anti-trust proceedings against General Motors, Ford and Chrysler in 1937, leading the latter two firms to divest their finance subsidiaries through consent decrees in 1939. GM contested 6Shay (1964), p. 214. 28 its indictment, and after a lengthy court battle, was finally allowed to maintain ownership of GMAC in 1952. It was prohibited from extending factory preferential treatment to dealers utilizing GMAC services, however. During the war, new car production decreased as production efforts were shifted to war materiel and armaments. Government controls on interest rates (Regulation W) and relatively high disposable incomes maintained rates at generally low levels. The IEra of Inter-Agency' Competition, (1952-present) By 1952, commercial banks held a 42% share of the auto credit market, up from 28% in 1939. They became the largest holder of automotive credit in 1958. Ford Motor Credit Company began operations in 1959 after GM’s consent decree was extended to the other automakers. Consumers have benefited from rate decreases engendered by increased competition. Shay (1964) writes, “The growth of competing credit agencies, particularly commercial banks and credit unions, has tended to replace the four large sales finance companies as low-cost credit sources in the new- auto installment credit market m the major result of such competition has been the secular decline in the cost to consumers of obtaining larger amounts of credit for progressively longer periods of time.” Outstanding automobile credit grew with automobile sales during the late 1970s, contracting somewhat during the early 19803 and rapidly expanding during 1984 and 1985. Commercial banks in particular cut back on consumer lending in the 1980- 29 82 period, reducing their share of the automobile credit market to 45 percent from 58 percent. Manufacturers stepped in to fill the void, offering below-market rates in 1981 on certain slow-selling models. Over the three-year period ending in 1982, finance companies had increased their automobile credit holding by more than 80 percent. The strong resurgence of the American economy beginning in 1983 has had a dramatic impact on new car sales and automobile financing. Banks flush with funds began expanding once more into consumer lending, attracted by ‘sticky’ rates that dropped more slowly than general market rates. In late 1982 the captive finance companies briefly attempted to stimulate sales with reduced-rate or incentive financing programs, while at the same time reducing their investment in outstanding automobile receivables. Commercial banks and the newly-deregulated savings and loans eagerly stepped in the fill unmet consumer needs. In 1985, the automotive finance companies entered the market once more with an aggressive round of financing incentives that remains with us today. Banks, savings and loans and credit unions are concerned about the future competitive structure of the industry. Recent Developments in Automobile Financing Luckett (1986) identified three topics of especial importance in his review of recent developments in automobile finance. 30 Leasing' Leasing is becoming more attractive to consumers as car prices increase and the interest expense deduction is eliminated by the Tax Reform Act of 1986. It offers an alternative to customers wishing to obtain the use of an automobile without the risk of ownership. The initial outlay on a lease agreement is generally smaller that that required for purchase; payments may be greater or less than those of the purchase option. A survey Sponsored by the Consumer Bankers Association (1986) reports the industry is growing more competitive and that profit margins at commercial banks are being squeezed. Note that the same Act which sparked consumer interest in leases also eroded lender profits by removing the investment tax credit and accelerated depreciation benefits. Consumer pressure to disclose leasing terms may impact banks’ willingness to supply funds to this market in the future. Securitisation Until recently, the issuance of asset-backed securities was largely confined to the repackaging and sale of mortgage loans. In 1970, the Government National Mortgage Association (GNMA) developed the Ginnie Mae pass-through as a method of selling participation in single—family FHA and VA mortgages. With over $2 trillion outstanding, mortgage loans account for the majority of loans securitized. In early 1985, Salomon Brothers underwrote a $10 million debt issue secured by auto loans on behalf of Lloyd Andersen, a large West Coast finance company. Salomon titled these 31 instruments “CARS”, for Certificates of Automotive Receivables. Shortly thereafter, Marine Midland Bank and Valley National Bank of Phoenix issued similar packages. GMAC became the first captive finance company to issue securities backed by auto loans in December 1985 with a $525 million issue. Since that time, GMAC has dominated the market. Its most recent issue was a record-breaking $4 billion. Chrysler Corporation and Nissan Motors have sold similar securities. Issuers of asset-backed securities are attracted by opportunities to improve capital ratios and rid themselves of unwanted assets. Investors appreciate the relatively high yields and apparently low risk levels. Investor risk on auto-backed securities appears moderate due to historically low delinquency rates and low inclinations of consumers to pre-pay their loans when rates decline. In addition, most issues have limited repayment guarantees attached. As the market expands, however, it is likely that lesser-quality loans will be pledged, increasing the risk to investors. Standard and Poor’s and Moody’s have not yet issued definitive ratings for these instruments. The development of a secondary market in consumer receivables could lead to lower rates for consumers as lenders compete more aggressively for auto loans to repackage and sell. The general decline in mortgage rates sparked by the movement of mortgage bankers into the once savings and 32 loan-dominated mortgage market can be cited as an example of this phenomenon (Monroe, 1985). Incentive Financing Reduced rate financing (subvention financing) by the automotive finance companies has significantly impacted the structure of automobile credit markets. Periodic offerings of cut—rate financing were first used in 1982 to boost sales of cars left over from the previous model year. Initially, these programs were limited in SCOpe and Offered rates not Significantly less than those available at banks and other depository institutions. More recently, incentive rates have dipped deeply below market rates. Prior to its aquisition by Chrysler Corporation, American Motors offered 0.0% financing on new cars. Through subvention financing, the auto finance companies have established control over the new automobile finance market (Consumer Bankers Association, 1987). The National Credit Union Administration filed a complaint with the Federal Trade Commission in 1986 charging the finance companies with antitrust and consumer protection violations, although those allegations were recently found to be without merit (Luipo, 1987). The promotion of finance charges as a reason to purchase a particular type of automobile at a particular time has had some negative effects for the automakers as well. First, price is the easiest variable in the marketing mix for competitors to duplicate (Kotler, 1986). Second, limited-duration promotions may lead consumers to postpone purchases until more favorable terms 33 are available. Third, the monies necessary to fund below- market loans by GMAC and the other captives are transferred from the parent operating corporation. The consolidated impact on the consumer of higher car prices financed at lower costs has not yet been determined. The ability to sell joint products such as credit life insurance and extended protection plans suggest it is unlikely that the auto finance companies will quickly relinquish their gains in market share. The National Automobile Dealer’s Association (NADA) estimates that finance and insurance income has come to account for approximately 15 percent of the combined new and used vehicle departments’ gross profits (Lukasiak, 1987). The impact of automobile incentive financing plans on commercial bank loan portfolios and loan rates has not yet been established. Respondents to an annual financing survey sponsored by the Consumer Banker’s Association reported a 17 percent increase from 1985 to 1986 with average loan size increasing slightly from $7140 to $7260 (Consumer Banker’s Association, 1987). Loan delinquency data do not support the contention that banks are lending to more risky borrowers, however. The major impact of subvention financing on commercial banks appears to be in a lengthening of the average loan portfolios. Most of the incentives offered by the captive finance companies apply to relatively Short-term loans, i.e. 24- and 36-months. Opportunities exist for loans 34 of longer maturities or for models not covered by incentives, e.g. foreign makes. REGULATION OF CONSUMER LENDING Most consumer legislation was passed in response to the rapid growth of consumer credit in the 19603 and 19703. The Consumer Credit Protection Act of 1968, or Truth-in-Lending, was a major step in ensuring that consumers received accurate information about credit costs in order to facilitate comparison shopping. Discrimination in the granting of consumer credit was a concern of the civil rights movement of the mid-19603 and led to the passage of legislation prohibiting these practices. The Fair Housing Act of 1968 and the Equal Credit Opportunity Act of 1974 forbid discrimination based on sex, age, race, marital status, color, religion and national origin. Abuses in the extension, collection and reporting of consumer credit are addressed by a variety of more specific federal laws. Table 4 chronicles the major events in consumer credit legislation. Table 4. YEAR 1968 1968 1970 1974 1974 1975 1975 1976 1977 1977 1978 1978 LEGISLATION Truth-in-Lending Act [Regulation 2] Fair Housing Act Fair Credit Reporting Act Real Estate Settlement Procedures Act (RESPA) [Regulation X] Equal Credit Opportunity Act [Regulation E] Home Mortgage Disclosure Act [Regulation C] Federal Trade Commission Improvement Act Consumer Leasing Act [Regulation M] Fair Debt Collection Practices Act Community Reinvestment Act (CRAl Right to Financial Privacy Act Electronic Funds Transfer Act [Regulation E] 35 Federal Consumer Credit Legislation DESCRIPTION Requires uniform disclosure of credit charges and established method of reporting charges as the ‘annual percentage rate’ Prohibits discrimination in any part of a credit transaction involving housing Set requirements for consumer credit reporting agencies and users of credit information to prevent innaccurate or inappropriate information disclosure Requires lenders to inform potential homebuyers in writing of settlement charges to prohibit kickbacks and limits use of escrow accounts Prohibits discrimination in personal and commercial credit transactions Requires depository institutions to disclose home loan information: intended to prevent redlining’ Establishes procedures for investigating consumer complaints against financial institutions Requires meaningful, accurate disclosures of the terms of personal property leases and established procedures for resolving disputes over the terminal liability Designed to eliminate abusive and deceptive debt collection practices Intended to encourage depository institutions to help meet the credit needs of their community Requires a Federal agency to obtain either the customer's authorization, subpoena or search warrant to obtain access to financial records Focuses on the unsolicited issuance of access devices, liability for unauthorized use, error resolution and transfer documentation Adapted from Spong (1985) and Rose (1986) The Bankruptcy Reform Act of 1978 impacted consumer credit markets during the early 19803 by subsituting 36 relatively lenient Federal bankruptcy conditions for more onerous state regulations. Personal bankruptcies soared, causing lenders to set higher standards for loan approvals. The Bankruptcy Amendment and Federal Judgeship Act of 1984 addressed these problems and provided for improved monitoring systems and judicial relief in cases of abuse. Usury Laws Prohibitions on the maximum rate of interest that may be charged on a loan have a long history (Tauesch, 1942). They originated from the belief that unsophisticated borrowers required protection from unscrupulous lenders. During the early part of this century, loan sharks were a particular concern in the not-well-developed personal loan market, and many states adopted ‘small loan laws’ designed to prevent usurious rates of interest (Michelman, 1970). Today, most states have ceilings on the rate of interest that may be charged for certain types of consumer loans. In Michigan, the interest rate on new automobile loans may not exceed 16.5 percent. When rate ceilings are non-binding, they generate little interest. During the late 19703 and early 19803 when interest rates rose to record levels, usury ceilings prevented lenders from earning normal profits and led to a decrease in the amount of credit supplied. This can be shown in a loanable funds framework (Figure 7). 37 E: Supply .2 a a: r' '5 0) .§ 'U‘ c _ Demand “\ cls 0' Cb Quantity of Funds Supplied (3) Figure 7. The Effects of Binding Usury Ceilings7 This situation illustrates the negative aspects of usury legislation. First, they affect the efficiency of financial markets. Lenders may require larger downpayments, shorter maturities or up-front processing fees in order to obtain required returns. Second, binding usury ceilings result in credit rationing. High risk borrowers are passed over in favor of less risky applicants. Ironically, this has the potential of harming the very individuals the laws were intended to protect, as the less well-informed or low-income consumers are frequently higher risks. Empirical studies of usury ceilings support the notion that loan supplies decline, noninterest compensation is increased, and borrowing by 7Adapted from Vandenbrink (1982) 38 higher-risk individuals is restricted (Goudzewaard, 1968; Greer, 1974, 1975; Shay, 1970, 1975). Competition in consumer loan markets has been partially explained by restrictive rate ceilings (Sullivan, 1984). Binding interest rates reduce the intensity of competition between banks and finance companies for customers of similar risks. Recent Legislation The passage of the Depository Institutions Deregulation and Monetary Control Act (DIDMCA) brought about a number of developments in the consumer credit markets. Title IV of this Act allowed Federally chartered savings and loans to invest in consumer loans and issue credit cards. Title V preempted state usury ceilings on residential mortgage loans and Title VI revised the Truth-in-Lending Act to make it easier for lenders to comply with disclosure requirements. Title III of the Garn-St. Germain Depository Institutions Act of 1982 increased the permissible percentage of consumer loans at Federal thrift institutions from 20 to 30 percent of total assets. CONSUMER. CREDIT KNOWLEDGE Consumer knowledge and sensitivity to credit terms is a matter of substantial historical interest. Early work by Due (1955) established that American families were relatively unaware of carrying charges or interest rates and 39 characterized consumer credit markets as imperfectly competitive. Families were found to be less concerned with credit costs than with their ability to meet monthly payments. Juster and Shay (1964) confirmed Due’s conclusions in a comprehensive survey of consumer borrowers. They reported that although consumers appear to have “institutional” knowledge of credit terms, i.e. some kinds of credit cost more than others, this did not guarantee they would seek the lowest cost alternative. Congress passed the Consumer Credit Protection Act, or Truth—in-Lending in July 1969. Its purpose is found in Section 102, “The informed use of credit results from an awareness of the cost thereof by consumers. It is the purpose of this title to assure a meaningful disclosure of terms so that the consumer will be able to compare more readily the various credit terms available to him and avoid the uninformed use of credit.” The efficacy of Truth-in-Lending has been controversial. Mandell (1971, 1973) reported that although consumers who borrowed after Truth-in-Lending were more aware of the true rate of interest than consumers who borrowed before, there were still significant gaps in their knowledge. Education and income were the key determinants of consumer credit knowledge and understanding. Other reports from the mid- 19703 substantiated and extended Mandell’s findings (Day and Brandt, 1974; Parker and Shay, 1974; Brandt, Day and Deutscher, 1975; Durkin, 1975; Shay and Schober, 1975b; Kinsey and McAlister, 1981). 40 By 1981, Durkin estimated that 54 percent of consumers could be classified as aware of annual percentage rates, compared with 14 percent before Truth-in—Lending’s implementation. Awareness of interest rate charges was still lowest among less-educated and low-income groups, although awareness of the dollar amount of finance charges was greater for this population as compared to persons with more education and income. It is apparent from these studies that Truth-in-Lending’s main purpose has been accomplished. Significantly more consumers are aware of credit terms than before the laws passage. Although total awareness has not yet been achieved, it is perhaps not necessary for market efficiency. A relatively small number of well-informed borrowers may suffice to ‘police’ the market and prevent gross abuses. Recent work in this area has focused on consumers’ choice of loan contract terms. Peterson and Black (1982) analyzed the results from surveys of 3,572 consumers in Purdue University’s 1979 Consumer Financial Survey. The most frequently cited factors in their choice of credit sources were interest rates, finance charges and monthly payments. Slightly over twenty percent of the consumers who had obtained credit in the past year actively shopped around for the best terms, mostly at commercial banks. According to the authors, “m differences in non-rate payment terms appeared to have little influence on the selection of a borrowing source.” 41 Sullivan (1985) evaluated consumers’ choice of adjustable versus fixed-rate automobile loans. Consumers attracted to adjustable-rate loans have higher credit risk and higher monthly payment burdens than those choosing the fixed-payment option, suggesting that lenders who offer adjustable-rate loans may be substituting default risk for interest rate risk in their loan portfolios. In Worden and Sullivan’s (1987) recent study of consumer credit shopping, consumers were asked to state the most important factor on a credit contract. Almost 40 percent mentioned the interest rate, while another 40 percent mentioned the monthly payment size. They found that consumers’ propensity to shop for credit 1) Increases with education, income and liquid asset holdings, 2) Is less for consumers with previous borrowing relationships, 3) Is less for consumers with past credit rejections, 4) Is higher for young, single persons with no children, and 5) I3 lower for residents of rural areas and states with restrictive interest rate ceilings. MODELS or TEE BANKING FIRM The theory of the banking firm is still in its developmental stages. Although a substantial literature exists, “m this literature, however, is still unsettled and rather heterogeneous. There exist a number of rival models and approaches which have not yet been forged together to form a coherent, unified and generally accepted theory of 42 bank behaviour” (Baltensperger, 1980). Some of this dissonance arises from conflicting notions on what a bank should be Optimizing, while a further source of confusion stems from the multiple functions served by banking firms. Before reviewing the attempts which have been made in this area, it is useful to identify the main functions of financial intermediaries. WHY DO BANKS ExIST? Financial intermediaries contribute to the efficient functioning of a financial system through a variety of products and services. Commercial banks Specialize in a subset of these activities. Gurley and Shaw (1960) and Goldsmith (1969) have discussed the basic reasons for bank existence. Asset Transformation Banks perform a useful function by facilitating the transformation of small, highly liquid assets (bank deposits) into larger denomination, less liquid ones (loans and investments). Kane and Buser (1979) argue that divisibility problems favor the use of financial institutions. Risk Transformation Banks also provide a valuable service in the form of risk transformation. Deposit interest is paid from the earnings of a bank’s loan and investment portfolio. 43 Diversification among a variety of investments reduces the risk of default and unrealized expectations. Klein (1973) suggests that financial firms offer a risk-return combination that dominates households' constrained opportunity set. Brokerage Banks enjoy cost and informational advantages in the evaluation of credit market applicants allowing them to profitably intermediate between ‘sellers’ (borrowers) and ‘buyers’ (depositors). Benston and Smith (1976) argue that banks’ role in minimizing transaction costs justifies economic profits earned from the exploitation of depositors’ constrained opportunity sets. The role of imperfect information in financial intermediation is a subject of substantial recent interest (Leland and Pyle, 1977; Campbell and Kracaw, 1980). Transactions Services The importance of commercial banks in administering the payments mechanism is well-established (Laidler, 1977). Demand deposits, wire transfers and credit cards are the primary means of making payment in the United States. In fact, maintaining a stable payments system is one of the major reasons for the proliferation of bank regulations (Spong, 1985). 44 Miscellaneous Services Commercial banks provide safekeeping services for business and consumer wealth, aided by Federal deposit insurance, which significantly lessens the probability of bank default. Banks often act in a fiduciary capacity and supply trust services to their customers. Convenience functions, e.g. foreign exchange, loan guarantees, securities brokerage and insurance are also performed by banks. Two fundamental approaches have been used to model the banking firm. The portfolio management approach assumes banks behave as risk-averse investors and allocates resources among asset categories to maximize expected returns for a given level of risk. The chief alternative to this approach views banks as microeconomic profit-maximizing firms intent on increasing short-term profits or net worth. A related literature in economics attempts to capture the effects of unmet loan demand (credit rationing) and its impact on the monetary system. Table 5 summarizes the major contributions of the first two approaches to the theory of the banking firm. 45 Table 5. Developgent of the Theogy of the Banking Firm AUTHOR YEAR SOURCE DESCRIPTION Shull 1963 Banking 8 A monopoly model that presents commercial Mon Studies banks as multiple-product price- discriminating firms. Kane and 1965 Qrtly J Econ A.portfolio-management model with deposit Malkiel variability. Pesek 1970 Can J Econ A real-resource model focusing upon the role and behavior of banks and the financial system. Klein 1971 JMCB A monopoly model that determines bank size and portfolio structure. Pyle 1971 JF A model based upon modern portfolio theory with hedging and quantity-setting behavior. Hart & Jaffee 1974 Rev Econ Stat. A portfolio-theory approach in which the intermediary is a quantity-setter facing random deposit rates. Towey 1974 JMCB A real-resource model in the traditional theory-of~the—firm framework. Pringle 1974 JMCB An imperfect-markets and risk-aversion model focusing upon capital decisions in banks. Edwards 1977 JPE An expense-preference model applied to banking. Sealey 1977 JF A neoclassical model based upon production & Lindley and cost theory. Fame 1980 JME A general equilibrium look at banking in the theory of finance. Baltensperger 1980 JME A survey of alternative approaches to the theory of the banking firm with emphasis on ‘partial’ vs. ‘complete’ models. Sealey 1980 JP A model of the banking firm with analysis of deposit rate-setting, risk aversion, market conditions and resource costs. O’Hara 1983 JF A dynamic theory of the banking firm incorporating the bank’s roles as an intermediary, managed firm and regulated enterprise. Prisman, 1986 JME A stochastic programming model of the Slovin 8 Sushka banking firm incorporating uncertainty and a liquidity constraint. Modified from Sinkey (1983). JMCB - Journal of Money, Credit and Banking, JF - Journal of Finance, JPE - Journal of Political Economy, JME - Journal of Monetary Economics 46 The following sections briefly discuss the three major approaches which have been taken in modelling the banking firm. A more detailed review may be found in Santomero (1984). PORTFOLIO MANAGEMENT MODELS The chief advantage of the portfolio management models is their explicit treatment of uncertainty. They have been used most extensively to solve problems in bank reserve and liquidity management. Edgeworth's research in the late 18003 serves as the basis for much of this work (Santomero, 1984). Under this framework bank managements struggle to maintain adequate stocks of liquidity in the face of stochastic deposit demand while maximizing return on earning assets. Reserve requirements act to constrain bank behavior. Different solutions to this problem are obtained depending on the assumptions of market structure (Orr and Mellon, 1961; Poole, 1968; Hester and Pierce, 1975). Parkin (1970), Pyle (1971) and Hart and Jaffee (1974) applied modern portfolio theory techniques to the problem of asset choice. Price-taking in loan and deposit markets leads to optimal portfolio choice in the quantity of various earning asset categories. In Pyle's model an intermediary must choose between three securities: a risk-free asset, risky loans and risky deposits. Intermediation, i.e. the transformation of deposits into loans is found to occur only with the existence of positive risk premia. Hart and Jaffee 47 verified Tobin’s (1958) separation theorem for financial intermediaries constrained by institutional and market factors without assuming that intermediaries can borrow or lend at the risk—free rate. In a recent paper O’Hara (1983) assumed the novel objective function of maximizing the utility of a bank manager’s share of bank profits. She used stochastic dynamic programming to solve a multiperiod model incorporating uncertain loan and deposit rates. Although perfect competition is still assumed in both asset and deposit markets, O’Hara makes a significant contribution through her attempt to incorporate risk, capital adequacy and resource costs within a general model. Portfolio management models have been criticized for their assumptions of perfect competition, failure to incorporate real resource costs and ‘stand-alone’ nature. Baltensperger (1980) refers to them as ‘partial’ rather than ‘complete’ models, since they typically treat but a portion of the balance sheet. A serious shortcoming of the portfolio approach is its general failure to include the effects of deposit uncertainty. In most models banks are assumed to be quantity-setting price-takers with respect to deposit supplies, while the reverse situation is a more likely description of the actual state of affairs (Sinkey, 1983). Sealey and Lindley (1977) note that a major flaw in portfolio models of the banking firm is their neglect of the effects of firm size. 48 THEORY-OF-THE-FIRM “ODILS Banks generally possess significant monopoly power owing to informational advantages, the local nature of their major product, loans, and restrictive entry and exit regulations (Havrilesky, 1985). Shull (1963) and Klein (1971) were among the first to develop this notion as an explanation of bank behavior. Shull applied Clemens’ multiple-product price- discriminating model to the analysis of bank competition. In Klein’s model a bank may choose among three assets: cash reserves, government securities and loans. Liabilities consist of demand deposits, time deposits and bank capital, although no special role is assigned to the latter. Asset markets are characterized by imperfect competition for loans and price-taking for government securities. The bank is a price-setter in unconstrained deposit markets. Klein obtained the surprising result that a bank’s optimal loan policy was independent of the rate paid on its deposits. Critics of Klein’s model targeted key assumptions of 1) Risk neutrality, 2) Imperfectly competitive deposit markets, 3) Single-period planning horizon and 4) Neglect of real resource costs (Pringle, 1973; Sealey and Lindley, 1977; Sealey, 1980; Baltensperger, 1980; Slovin and Sushka, 1983; O’Hara, 1983; Prisman, Slovin and Sushka, 1986). Mest of these writers have developed models of their own designed to “correct” these deficiencies. Klein’s model viewed deposit interest as the sole cost factor in the determination of bank profit. In the 49 traditional theory of the firm, production costs are an important factor in determining firm output and pricing. Applications of this approach to financial intermediaries have been made by Pesek (1970), Towey (1974), Sealey and Lindley (1977) and Sealey (1980). These models have a particular advantage in explaining firm size, although the related empirical literature on bank economies of scale does not seem to have been fully developed. Sealey’s (1980) model makes an important contribution in its simultaneous incorporation of production costs, risk, deposit rate- setting, and liquidity effects. Baltensperger (1980) strongly advocates the theory-of-the-firm approach to bank behavior, alluding to its ability to include joint determinations of asset structure, liability structure and firm scale. CREDIT RATIONING MODELS The traditional credit-rationing literature identified and offered solutions to the problem of legal or social constraints on the availability of credit. If lenders are able to discriminate between borrowers of different risk classes, they may be in a position to allocate funds in a manner which, while profitable to an individual firm, is inconsistent with increased efficiency of the financial system. Freimer and Gordon (1965), Kane and Malkiel (1965) and Jaffee and Modigliani (1969) offer explanations for this behavior. A more thorough exposition of this problem is 50 provided in Jaffee (1971); Baltensperger (1978) and Blackwell and Santomero (1982) offer critiques. DETERMINANTS Ol' BANK LOAN RATES The majority of a bank’s income is derived from its loan portfolio. Commercial bank lending falls into four major classes: Commercial and Industrial, Real Estate, Consumer and Agricultural. The remainder is comprised of loans to foreign governments, state and local governments, and other financial institutions. Table 6 provides information on the loan portfolio of insured commercial banks. Table 6. Loan Portfolios of Insured Commercial Banks LOAN TYPE 1975 1980 1983 Commercial/Industrial $176 35 % $283 34 % $404 36 t Real Estate 136 27 263 31 328 29 Consumer 107 21 182 22 218 19 Financial Institutions 39 8 47 6 77 7 Agricultural 20 4 32 4 39 3 Other 24 S 33 4 60 5 Source: FDIC Annual Reports, Table 109. All dollar figures are in billions. A bank must make several decisions regarding its loan portfolio. First, the total size of the portfolio as a percentage of bank assets must be decided upon. This is primarily a function of loan demand and market structure. Second, choices must be made regarding the loan mix. This 51 decision is affected by the institution's expertise and operating environment. Third, the average maturity of the loan portfolio must be determined. Average deposit and liability maturities are of crucial importance in this decision. After a bank has established its overall portfolio structure goals, emphasis shifts to a consideration of loan pricing. (These events are not temporally independent; it’s quite possible, for instance, for a bank to first establish its pricing policies and then adjust its portfolio accordingly. The theoretical model developed in this dissertation assumes this to be the case). Many factors are involved in the pricing of consumer loans, and there is no set of variables which is universally agreed-upon as determining loan rates. However, most influences on the consumer loan rate may be classified as originating from 1) Cost, 2) Competition, 3) Risk and 4) Regulation. Of these four influences, only cost and competitive factors will be addressed in this dissertation. If it is assumed that banks practice perfect credit scoring, then loans are granted only to customers who repay their loans in the agreed-upon fashion. Default risk and collection risk are thus nonexistent. Alternatively, loan losses may be treated as a cost of lending to be subtracted from the rate of interest charged. In any case, loans are assumed to be priced correctly for their risk characteristics. 52 Despite the significant current interest in the effects of regulation on consumer lending, this aspect of bank loan rate determination will not be covered in depth. A number of recent books discuss the probable impacts of the DIDMC and Garn-St. Germain Acts of 1980 and 1982 on the financial services industry (Fraser and Kolari, 1985; Cooper and Fraser, 1986). The markets are still in too great a state of flux to achieve concrete conclusions. COST FACTORS The most comprehensive source of cost data is found in the Functional Cost Analysis prepared annually by the Federal Reserve. Information on the cost of installment lending at commercial banks during 1985 is shown in Table 7. Table 7. Costs of Installment Lendixg in 1985 DEPOSITS Less THAN BTWN $50 AND OVER $200 $50 MILLION $200 MILLION MILLION Origination cost $81 . 20 $84 . 41 $85 . 67 (per loan) Collection cost $ 5. 47 $ 5. 69 $ 5. 67 (per payment) Loss rate $27.43 $18.45 $13.27 (S-yr. avg. , based on avg. loan size) Average loan size $3,424 $3,852 $4,439 Percent accepted 80 . 74% 81 . 71% 72 . 84% Source: Board of Governors of the Federal Reserve System, Functional Cost Analysis (1985) 53 The administrative and operating expenses of lending vary significantly with the type of loan activity. Consumer lending functions, e.g. credit card and installment loans, are the most expensive to administer. Consumer loan rates must be higher than commercial or real estate loan rates to compensate the bank for increased administrative and default costs. Table 8 compares the costs of various loan categories. Table 8. Functional Cost Analysis of Bank Lending TYPE OF LOAN CREDIT CARD INSTALLMENT COMMERCIAL REAL ESTATE SECURITIES Gross Yield 23.65 % 15.21 % 12.44 % 11.51 % 11.20 % - Cost of Funds (7.95) (8.01) (7-93) (8.05) (7.97) Interest Spread 15.70 7.20 4.51 3.45 3.23 - Oper. Expense (11.64) (3.64) (1.82) (1.01) (0.16) - Loan Losses (2.12) (0.58) (1.05) (0.15) (0.00) Net Return 1.94 t 2.98 t 1.64 i 2.29 8 3.07 % Source: Board of Governors of the Federal Reserve System, Functional Cost Analysis (1983) Economies of Scale There is a rich literature on scale economies in banking. Bell and Murphy’s (1968) seminal monograph and the works of George Benston (1965a, 1965b, 1972) are of particular interest for their broad scope and attention to detail. The study of bank economies of scale is important to at least three groups of people. Regulators are interested 54 in estimating the cost consequences of merger decisions. Bank managements may use comparative cost data to improve firm efficiency. Financial economists rely on empirical estimates of cost functions in the development of theoretical models. Most early studies assumed a Cobb-Douglas production function and measured output as the number of accounts or loans serviced over a given period of time. In general, significant economies of scale were found for most banking functions, including installment lending. A recent paper by Benston, Hanweck and Humphrey (1982) criticizes the early work for its failure to address the optimal scale of bank operations either due to lack of data or the assumed form of the production function. They propose a more general translog function in which bank output is measured by a Divisia index. Their main conclusions are: 1) Average operating costs for both unit and branch state banks are U- shaped or upward-sloping for three different measures of bank output (including total costs or deposits), 2) An optimal size of bank office exists from $10 million to $25 million in deposits and 3) Both branch and unit state banks experience significant operating cost scale diseconomies on a “plant” level. COMPETITIVE FACTORS Some of the most widely discussed issues in American banking over the past twenty years revolve around the topics 55 of bank structure and competition. Regulators believe bank structure (the number and distribution of banks operating within a particular area) affects the degree of competition within a market, which in turn affects loan rates, deposit rates and the price of bank services. One goal of regulation is to determine the optimal bank structure leading to lower prices for bank outputs without the undue sacrifice of market diversity and/or bank failures. A voluminous literature has arisen to provide information on the effects of bank mergers on competition and cost structures. Several literature surveys have been published (Benston, 1973: Rhoades, 1977, 1982), the most recent of which is by Gilbert (1984). Regulation of market structure is effected by legislation governing the entry and exit of banking firms from a particular market, limitations on bank merger activity and branching restrictions. Structure-Conduct-Performance Hypothesis The traditional method of inquiry into bank competition begins with the structure-conduct-performance (S-C-P) hypothesis drawn from the industrial organization literature. According to this hypothesis, a high concentration of sellers fosters collusion among oligopolistic competitors and increases firm performance. The S-C-P hypothesis has received strong support from studies of the behavior of manufacturing firms, but has not been unanimously accepted as a description of the commercial banking industry. Gilbert 56 (1984) reviews 56 studies of bank structure and competition and finds support for the hypothesis in only 27 of them. Smirlock (1985) argues that this is because there is no relationship between concentration and profitability, but rather between market share and profitability. He follows Demsetz (1973) in this latter point, claiming increased market share results from superior efficiency rather than market power. Consumer credit markets have long been recognized as imperfectly competitive, that is, markets in which supply and demand are not equated solely on the basis of price (Yntema, 1938; Phelps, 1944). According to a recent textbook author, “the simple truth, however, is that the market for consumer loans is not nearly as competitive as for commercial loans. Banks have successfully found individuals to be full-service customers of single institutions” (Koch, 1987). Hancock (1986) tested commercial banks’ price-taking behavior in loan and deposit markets and concluded that “competitive markets are the exception rather than the rule.” Recent work by Smirlock and Brown (1986) challenge this assertion of bank monopoly power. These authors support Smirlock’s earlier contention that increased market concentration is the result of superior efficiencies, and present evidence that leading firms act as price-setters while secondary firms act as price-takers. 57 EMPIRICAL STUDIES A limited number of studies have addressed the empirical behavior of bank loan rates over time. Consumer loan rate behavior has been examined only once, in Sullivan and Fain's (1984) working paper. Most other studies have used commercial loan rates or the prime rate in their investigations. The next few paragraphs review the empirical results from four of the more relevant attempts. Slovin and Sushka (1983) used aggregate data from the Federal Reserve to test their theoretical model of the bank loan rate. For the period 1952-1980, current commercial paper and mortgage rates were found to be significant in explaining the behavior of the aggregate rate charged on commercial loans on a quarterly basis. A one-quarter lagged commercial paper rate was significant, although a two-quarter lag was not. Additional regressions led them to conclude “ the commercial loan rate is primarily a function of Open market interest rates and that under normal conditions the setting of the loan rate is dichotomized from conditions in deposit markets.” Table 9 presents the simplest of Slovin and Sushka's regression results. 58 Table 9. Empirical Models of Bank Loan Rates AUTHOR MODEL PERIOD Slovin and RCL - -.31 + .26RCP + .44RCP-1 - .06RCP-2 + .56RM 1952-1980 Sushka (1983) (-.94) (4.39) (5.61) (-.96) (6.73) (quarterly) Goldberg (1984) RCL - .51 + .41RCD + .57RCD-1 - .12RCD-2 + ... 1972-1981 (.97) (7.19) (9.41) (3.36) (monthly) Lee (1985) RCL - -3.4 + .45FF + .39FF-1 + .43RP + .O3IV + ... 1960-1983 (-2.96) (13.07) (11.46) (2.46) (1.77) (quarterly) Sullivan and RAL - 3.71 + .17RTB.1 + .19RTB-2 + .53RM 1972-1983 Fain (1984) (10.02) (1.89) (2.71) (4.08) (quarterly) *Student t-statistics in parentheses. RCL - commercial loan rate, RCD - CD rate, RCP - commercial paper rate, RM - mortgage rate, FF - Federal funds rate, RP - risk premium on AAA corporate bonds, IV - average nonfarm inventory investment, RAL - auto loan rate, RTB - Treasury bond rate. Subscripts indicate lagged variables. Two subsequent tests of Slovin and Sushka’s model have been conducted. Lee (1985) reexamined their results using the Federal funds rate to proxy both open market rates and signalling effects. He also incorporated business loan demand, interest rate expectations, risk, and asset/liability relationships in his empirical model. Unlike Slovin and Sushka (who also examined these factors), Lee found significant relationships for each of these variables, suggesting that commercial loan rates are significantly influenced by factors other than open market interest rates. Lee’s sample period was both shorter and more recent than Slovin and Sushka's; according to his research, the commercial loan rate has grown more responsive to the Federal funds rate and the market risk premium. Sullivan and Fain (1984) used Slovin and Sushka's model to investigate the behavior of consumer loan rates. Their 59 work represents the sole application of time series techniques to consumer loan rate behavior. They regressed aggregate automobile loan and personal loan rates against the rates on constant-maturity Treasury bonds and mortgages. They found lagged quarterly Treasury bond rates and contemporaneous mortgage rates useful in explaining the movements of automobile loan rates. Their best results are shown in Table 9. Goldberg (1984) has argued that the prime rate (the rate charged by commercial banks on loans to their most creditworthy customers) is principally determined by the average cost of a bank’s currently- and previously-issued liabilities. He regressed monthly averages of the prime rate charged by large money center banks (as reported in the Federal Reserve Bulletin) against contemporaneous and lagged values of the average rate paid on large certificates of deposit for the period 1972-1981. For the period as a whole, both current and lagged CD rates are useful in describing the behavior of commercial loan rates (Table 9). Over shorter sub-periods, however, only lagged values are significant. Goldberg interprets these results as evidence the prime rate is “... stable relative to money market rates but responsive to ‘fundamental’ changes in the bank’s cost of purchased liabilities.” In summary, the recent empirical literature is divided on exactly what factors are important in determining the behavior of loan rates offered by commercial banks. Slovin 60 and Sushka's model hinges on the substitutability of open market securities for a bank’s loan portfolio. Lee sacrificed theoretical rigor for empirical results, and concluded that many factors (albeit with different weights) explain the movement of commercial loan rates. Sullivan and Fain found long-term lagged rates useful in explaining the behavior or automobile loan rates. Goldberg’s model, based on bank funding costs, finds commercial loan rates to be functions of lagged but not contemporaneous market rates. Automobile Loan. Rates Previous sections have discussed general issues surrounding the pricing of consumer loans at commercial banks. The final paragraphs of this review will summarize the relatively few studies that have directly addressed automobile lending by commercial banks. Early research by Shay (1963, 1964) laid the cornerstone for work in this area. His careful analysis of the historical determinants of new automobile financing led others to more in-depth examinations. Greer and Shay (1975) and Greer and Nagata (1975) developed an elaborate macroeconomic model of the new automobile credit market for the National Commission on Consumer Finance in the early 19703. Stafford and Dunkelberg (1969) used data gathered from the 1966 Survey of Consumer Finances to characterize borrowers for new and used cars. Finance charges paid by 61 borrowers decreased with age and education but increased with disposable income. Credit charges decreased with loan size and increased with loan maturity. Regional variations were found in loan rates, with banks in the southern United States charging the highest rates. Heggestad and Mingo (1976) included automobile loan rates as a performance measure in their study of competition in commercial banking. They reported a significant increase (79 basis points) in new car loan rates from the least concentrated to the most concentrated markets in their study, suggesting that commercial banks at that time wielded measurable market power. A more thorough investigation of commercial bank auto loan rates was conducted by Peterson and Ginsberg (1981). Using data from approximately 320 banks in all but one state, they examined the effects of government regulations, geographic location, market structure and bank size on 36- month direct loan rates. They found rates to be reduced by rate ceilings and creditor restrictions on the collection of delinquent debts. While smaller banks were found to charge lower loan rates than larger banks, rates offered by rural banks were higher than those of their urban counterparts. Regional geographic differences were found to exist, although they were not consistent with those reported by Shay (1963) or Stafford and Dunkelberg (1969). In Sullivan and Fain’s (1984) working paper the ‘most common’ automobile loan rate was determined to follow 62 movements in home mortgage rates and constant—maturity Treasury securities over the period 1972-1983. They reported that automobile loan rates appear to lag Treasury bill rates by several quarters while mortgage rates move concurrently in the same direction as auto rates. Regressions using a ‘bank willingness index’ to measure the impact of restrictive rate ceilings on auto loan rates were not significant. Sullivan and Fain based their empirical work on Slovin and Sushka’s (1983) theoretical model of the commercial loan rate. Like the latter authors, they used macroeconomic data to test microeconomic phenomenona. Thus, their results are open to further interpretation. DETERMINANTS OF BANK DEPOSIT RATES A number of different approaches have been taken in attempts to describe the behavior of bank deposit rates. This section discusses the major theoretical contributions in this area and reviews the supporting empirical evidence. The theory-of-the-firm approach exemplified by Klein (1971) underlies much of the the theoretical work on the determination of deposit rates at banking firms. Klein’s model assumed banks to be price-setters in both loan and deposit markets and showed that deposit rates were unrelated to the returns on a bank’s loan portfolio. Rather, deposit rates were found to be functions of the (exogenously- determined) rate on government securities. Klein and Murphy (1971) made further refinements to this model and applied it 63 specifically to the pricing of demand and time deposits, with similar conclusions. Baltensperger (1972) and Sealey and Lindley (1977) incorporated production costs into the bank’s profit function and found optimality conditions in which marginal revenue did not always equal marginal cost. The latter authors’ work appears to rest on their peculiar view of bank output, which they defined as “ m total services provided to debtor institutions m ” and measured by the total dollar volume of earning assets. The tendancy for changes in changes in demand deposit rates to lag behind time deposit rate changes was explained by Flannery (1982) using a quasi-fixed production function. He noted that deposit rates have specific investment costs associated with them and that banks might be more likely to pay “excessive” rates in the short run to preserve valuable customer relationships. The trade-off between explicit and implicit payments of interest on deposit accounts has been analyzed by Barro and Santomero (1972), Santomero (1979) and Mitchell (1979). Each of these authors notes that binding interest rate ceilings encourage banks to recompense depositors with nonmonetary returns. Check clearing at prices below cost is the perhaps the most important example. Mitchell (1979) developed a theoretical model in which an increase in explicit interest rates actually reduces the fees charged on check clearing; his results assume that savings accounts are not subject to 64 binding ceilings and that consumers view checking and savings accounts as close subsitutes. The relationship of deposit rates to market interest rates has not received extensive treatment in the theoretical literature, most probably due to the existence of Regulation Q ceilings up until 1982. Weber (1966) demonstrated that deposit rates tend to be sluggish whenever the rates on an institution’s earning assets adjust slowly to market innovations. Goldfeld and Jaffee (1970) and Stigum (1976) elaborated on this observation and noted that when mortgage rates decline, deposit rates will not fall below some minimum feasible rate; however, when rates increase, the deposit rate may stay constant or even fall. The most recent theoretical paper of note in this area is Sealey (1980). He develops a portfolio model of the banking firm incorporating market conditions, cost considerations, and deposit rate-setting behavior. Sealey’s work is important because it demonstrates how risk, resource costs and liquidity constraints interact in the determining optimal bank behavior. III . A MODEL OF THE CONSUMER LOAN RATE This dissertation develops a theoretical model of the banking firm emphasizing the pricing of consumer loans and empirically tests it with monthly loan and deposit data gathered from individual banks across the United States. Myron Slovin and Marie Sushka’s 1983 article in The Jburnal of Finance entitled, “A Model of the Commercial Loan Rate,” provided the foundation for this endeavor. In Slovin and Sushka’s paper, banks are assumed to maximize short-term profits under conditions of monopolistic competition. Banks may hold assets in either of two forms: open-market securities and “loans." Three kinds of deposits - demand deposits, time deposits and negotiable certificates of deposit (CDs) - constitute claims on bank assets. Each deposit type is subject to its own reserve requirements. As demand deposits pay no interest and savings account rates are constrained by regulatory influences, the bank’s decision variables are the CD rate, Ito: and the loan rate, rL, which Slovin and Sushka construe to be the rate on commercial loans (hence the title of their paper). Slovin and Sushka expend some effort in deriving their model under conditions of both nonbinding and binding liquidity constraints. While their results are interesting, it is not clear that banks actually experience liquidity crises of the sort imagined by Slovin and Sushka, or that banks would continue to maximize profits if such situations 65 66 arose. Further, characterizing the return on a bank’s entire loan portfolio with a single number (the the rate charged on commercial loans) raises a number of interesting questions regarding the behavior of individual loan types. As discussed in Chapter II, Slovin and Sushka test their model with aggregate quarterly interest rate data from the Federal Reserve. My work improves on Slovin and Sushka’s theoretical model in three ways. First, I allow for two kinds of loans in bank portfolios: business loans and consumer loans. Second, I explicitly consider the influence of competitors’ rates in determining the rates paid on consumer loans. Third, I provide solutions for the optimal determination of deposit rates, which are not presented in Slovin and Sushka’s original work. Most importantly, however, my model provides the only theoretical explanation for the behavior of consumer loan rates yet to appear in the literature. ASSUMPTIONS AND OBJECTIVE FUNCTION The bank’s assets are assumed to consist of two types of loans (business and consumer), securities, and required reserves. Deposits consitute the sole claim on bank assets. Following Sealey (1980) and Slovin and Sushka (1983), bank capital is omitted. In a review of Sealey’s model, Sinkey (1983) notes, “m if it disturbs you that the model ignores bank capital, think of the banking firm in hlS framework as a going concern. Bank capital becomes important mainly in crisis situations. Moreover, m the 67 adequacy of a bank’s capital is determined primarily by its profitability and liquidity.” The bank’s balance sheet takes the form B + C + s + qD = D (1) where B = business loan volume C = consumer loan volume S = securities qD = required reserves D = deposits The bank is assumed to maximize the single-period profit function 1t=rCC+rBB+rs[(l-q)D—C—B]-rDD " Xc(C) " XB(B) " XD(D) (2) r8 = business loan rate rc = consumer loan rate rs = marketable securities rate rD = deposit rate q = percentage of deposits held for reserve and liquidity requirements XB(B) = cost function of originating and servicing the business loan portfolio XC(C) = cost function of originating and servicing the consumer loan portfolio cost function of supplying deposit services XD(D) 68 Business loan rates are exogenous and set by competitive forces beyond an individual bank’s immediate control. Consumer loan rates are determined in an imperfect market and are a function of bank preferences and competitors’ offerings. S: From the balance sheet constraint, (1 — q)D — B - C and can be substituted into the profit equation. The following assumptions are made regarding the behavior of certain variables and functions: C 83 era as arc 8C arc 8C arx an arD C(rc,rx) B(rB,rC) D(rD) (consumer loan volume is a function of the loan rates offered and competitors’ rates) (business loan volume is a function of the business loan rate and the rate paid on consumer loans) (demand for deposit balances is a function of deposit rates) (increases in business loan rates increase business loan supply) (increases in consumer rates decrease business loan supply) (decreases in consumer rates increase consumer loan demand) (increases in competitors’ rates increase consumer loan demand) (increases in deposit rates increase deposit demand) 69 X'C(C) > 0 (marginal costs increase as loan and deposit volumes increase) X'B(B) > 0 200(1)) > X"C(C) > O (marginal costs are non-diminishing) mus) mm) F IRST ORDER COND I T IONS The bank’s decision variables are the consumer loan rate and the deposit rate. The profit-maximizing levels of consumer and business loans follow from their choices, as do deposit and securities volumes. Following Klein (1971), Tobin (1982), and Slovin and Sushka (1983), the bank is assumed to be risk-neutral with respect to profit maximization. Differentiating the objective function with respect to the choice variables gives two first-order conditions: an 22. - -hg) ls( - -§x_8)+c=o (3) 37:: 3:. (rc rs ac + 3:. r3 r5 as 81: 3D 3X0 _ = _ 1 — — — — - D = O (4) Bro Bro (r5( q) r0 3D 70 SECOND ORDER CONDITIONS The second-order conditions (5) and (6) must be negative for the bank to obtain a profit maximum. 2 2 erg arg 6C aré C 2 2 arg rD 8D arD SOLUTION AND PRED I CTIONS Since the loan and deposit functions are only generally specified, it is not possible to achieve a closed-form solution to this model. However, testable hypotheses of the model’s behavior may be derived from the comparative statics of the first-order conditions.1 Equations (7) and (8) show the response of consumer loan rates and deposit rates to changes in the exogenous variables r3, rB and rx. a_c ac ) * = dr + -——-d (7) drc K “ €9—rc aarc)}drs + 83:} r8 {arx rx drs (8) dr6= L{-ar—1:(l - (I) where K and L reflect constraints imposed by the second-order conditions. Without explicit specifying the supply, demand and cost functions, the magnitudes of the coefficients cannot 1See Appendix A for further explanation. 71 be determined. Their signs, however, provide information on the predicted direction of loan and deposit rate movements. If the second-order conditions are met, the coefficients may be unambiguously signed as follows: arc arc arc -—— > O -—— < O -—— > 0 9 81's 31's arx ( ) arD 81]) Dr.) -— > 0 -—— = 0 -- = 0 10 31's Bra arX ( ) The model predicts that the interest rate charged on consumer loans increases as interest rates on securities rise, reflecting increased opportunity costs on the consumer loan portfolio. Consumer loan rates should decrease as the prime, or business loan rate increases, as resources are directed towards more profitable business loans. Financial institutions should increase consumer loan rates in response to competitor’s offerings. Increases in security rates should increase the rate paid on deposits at financial institutions, while business loan and competitor’s rates should have no effect. These predicted relationships will be examined empirically in Chapters IV and V. IV. METHODOLOGY The solution to the theoretical model implies a bank’s consumer loan rate should increase as security rates and competitor’s rates increase and decrease with increases in the business loan rate. Its deposit rate should increase with increasing security rates and not respond to changes in the business loan rate or competitor’s rates. This section outlines the methodology used to test the theoretical model’s predictions and presents the results of these tests. SOURCES OF DATA Empirical data for testing the theoretical model were obtained from a number of sources. Because the inputs to the theoretical model were specified only in general terms, reasonable effort was expended to find a data set which was representative and which added positive marginal informativeness. The quality and comparability of the data set was also an important consideration. CONSUMER LOAN AND DEPOSIT RATE DATA The primary data source for this dissertation was obtained from Gary Meyers and Associates, a Chicago consulting firm. This company publishes a weekly report, The Meyers Report, of interest rates and terms on consumer financial products at over 200 depository institutions 72 73 throughout the country. The firm surveys each of its reporting institutions by telephone each week on Monday and Tuesday. Usually, the institution’s contact person is telephoned by the same Meyers employee week after week. Information is collected on the current rate structure and terms on consumer mortgages, installment loans and deposit accounts. An attempt is made to gather data on “standard” products, e.g. 48-month fixed-rate new automobile loans. Where this is not possible, data is collected for the closest approximation to this standard. The data is organized and collated on Wednesday and Thursday and the report is mailed to subscribers on Friday. Care is taken to verify the data before is printed: employees are required to cross-check each other’s work. The Meyers data is particularly valuable because it reports the rates and terms on individual products at individual banks. Previous empirical studies on the determinants of bank loan rates used composite figures generated from hundreds of reporting banks. It is quite possible for microeconomic relationships to differ from their macroeconomic aggregates. Unfortunately, comparative microeconomic analyses tend to yield less clear-cut results. It is not the intention of this dissertation to claim that the theoretical model applies to ALL depository institutions, however, it should provide insight into the behaviors of particular subsets. 74 The.Meyers Report offers a unique data set for research, but its shortcomings should also be recognized. First, the data set exists only on paper. The data is originally entered into a microcomputer spreadsheet, but after it is printed, new data is written over the old and the printed hardcopy is the only remaining evidence of its existence. Second, although the firm has been collecting and reporting data in more-or-less the same way since 1984, its portfolio of reporting financial institutions has changed over time, severely restricting the number and length of usable time series. Third, not all institutions within a given market are represented. Some banks do not choose to report their loan and deposit rates. A related, and hopefully very minor problem, is the the possibility that the institution might report inaccurate or misleading information. Finally, Meyers’ hardcopy archives are incomplete. Certain issues are missing pages and/or no longer obtainable. This creates gaps in the data set which may or may not be significant. Consumer loan information is collected for New Car Loans, Unsecured Personal Installment Loans, Home Improvement/Equity Loans, and New RV/Marine Loans. Not all of the reporting institutions offer all of these loans all of the time, and thus certain time series are nonexistent or interrupted. Automobile loan rates provide the most consistent and complete set of consumer loan rates for empirical testing. They are the largest component of consumer credit, have the most closely—comparable terms and 75 have the most homogenous purpose and risk characteristics of the available loan categories. The majority of these rates were quoted for 48-month, fixed-rate new automobile loans. Four kinds of deposit rates were reported: Money Market Accounts, NOW Accounts, IRAs and Certificates of Deposit with maturities from three months to ten years. Again, the most consistent and comprehensive series, Money Market Account rates, was chosen as a proxy for deposit rates. The theoretical model is tested with a monthly data set. The Meyers Report is published weekly, but two conditions led to the choice of monthly periods. First, automobile loan rates do not vary all that much. Generally, they are changed only through the action of a loan committee, which may meet on a biweekly or monthly basis. A weekly series might obscure more fundamental responses. Second, the problem of missing issues in the data set would perhaps be worsened by weekly observations. The data for this study was taken from issues of the Meyers Report published between July 1986 and July 1988. In all but two cases, data came from the first week of each month. In the remaining cases, data was taken from the second week of each month. Loan rate data was manually entered into a computer database; deposit data was scanned with an Optical character reader and further manipulated by computer. Spot checks were conducted in both cases to verify the accuracy of the database. Forty-five institutions from New York City, Chicago, California, Boston, Detroit and Cincinnati were included for 76 a total of 25 weeks x 45 banks x 2 interest rates = 2250 observations. Appendix B provides additional information on the institutions used in this study and descriptive statistics for the consumer loan and deposit rate series. SECURITY RATES The theoretical model developed earlier presupposes a bank holds “securities” in its asset portfolio for invesment and liquidity reasons. Including the rates on all possible securities in an empirical regression equation is not a good idea, as it creates multicollinearity problems and may hide even more fundamental results. Thus, a single, commonly-held security should be chosen as a proxy for ‘SEC,’ the security rate. Four security rates were taken from the Federal Reserve Bulletin and considered as possible proxies for the security rate. Three of these (Treasury bills, federal funds and certificates of deposit) were money market rates and one (3— year constant maturity Treasury bonds) was a capital market rate. Their correlation with each other and the average prime rate is shown in Table 10. 77 Table 10. Correlation Matrix of Security Rates SECURITY I'I'R CD TBL TBD BUS Federal Funds (FFR): 1.000 Certificates of Deposit (CD): 0.703 1.000 U.S. Treasury Bills (TBL): 0.721 0.779 1.000 U.S. Treasury Bonds (TBD): 0.648 0.953 0.777 1.000 Prime Rate (BUS): 0.673 0.935 0.798 0.924 1.000 Monthly observations corresponding to the first week of each month from July 1986 through July 1988. FFR - federal funds rate, CD- secondary market rates on 3-month certificates of deposit, TBL - secondary market rates on 3-month U.S. Treasury bills, TBD - rate on 3-year constant maturity U.S. Treasury bonds, BUS - average rate charged by banks on short-term business loans. Data from the Federal Reserve Bulletin. Much preliminary analysis was conducted to determine the appropriate proxy rate. Federal funds and certificate of deposit rates were dismissed fairly early on, citing the volatility of the former and the aggregate nature of the latter. Treasury bills and three-year constant maturity Treasury bonds were considered in a number of forms, including lags of up to four weeks prior. Treasury bonds provided slightly better results in both loan and deposit rate regressions. The loan rate results may arise from the similar maturities between the bonds and the automobile loans. The results of the deposit rate regressions were more surprising - money market account rates are better explained by Treasury bond, rather than Treasury bill, rates! Thus, three-year constant maturity Treasury bond rates were chosen to proxy the ‘SEC’ rate in the empirical regressions. 78 OTHER. VARIABLES The Federal Reserve Board reports the prime rate charged by banks on short-term business as a monthly average. This figure was used to proxy the rate on a bank’s business loan portfolio. Competitors’ rates were proxied by an arithmetic average of the rates charged by all other banks in the regional sample for a given date. Differences due to geographical location and bank size were investigated through the use of dummy variables. ECONOMETRIC ANAL! S I S In the previous chapter, a theoretical model of the banking firm was developed in which rates on consumer loans and customer deposits were hypothesized to be functions of security rates, business loan rates and competitors' rates. In testing this model, it is important to keep in mind the model’s assumptions and the nature of the available data. This section briefly describes the econometric techniques used to test the theoretical model. Essentially, two types of tests will be conducted: F-tests, which measure the degree of explanatory power of the regression equation, and t-tests, which measure the explanatory power of a particular regression coefficient. The classical linear regression (CLR) model using the ordinary least squares (OLS) estimator is, “m probably the most popular estimator among researchers doing empirical work” (Kennedy, 1985). For many econometric problems, this 79 model has very desirable properties and is computationally efficient. In the CLR, a set of independent variables (regressors) is used to describe the behavior of a dependent variable. The OLS estimator obtains values for the intercept and regression coefficients by minimizing the sum of squared residual disturbances. The applicability of this technique depends in large part on the nature of the data; the CLR model makes five assumptions about the way in which the data is generated. Assumptions of the CLR Model As its name implies, the classical linear regression model assumes the dependent variable can be expressed as a linear function of a set of independent variables plus an error term. The coefficients of this function are assumed to be constants. Specification errors arise when these assumptions are not met. Perhaps the wrong independent variables are included in the regression, or the correct ones are left out. Maybe the regression relationship is nonlinear. Or, the coefficients may be non-stationary over the sample period. The second assumption of the CLR model is that the expected value of the disturbance term equals zero. If it does not, the regression may have a biased intercept. Third, the disturbance terms are assumed to have the same variance across time (homoskedasticity) and to be uncorrelated with 80 one another. Heteroskedasticity and autocorrelation are problems associated with the violation of this assumption. The fourth assumption is that the observations on the independent variables can be considered fixed in repeated samples. If they are not, autocorrelation, errors in variables, or simultaneous equation estimation problems may occur. The final assumption of the CLR model is that no linear relationships exist between the independent variables and the number of independent variables is less than the number of observations. Violating this assumption causes problems of multicollinearity. If its assumptions are met, the ordinary least squares estimator has a number of desirable properties. Is is optimal on the least squares, highest R?, unbiasedness and asymptotic criteria considered most important by econometricians (Kennedy, 1985). However, different methods of estimation may be appropriate when these assumptions are not met. V. RESULTS AND DISCUSSION This chapter describes the hypotheses arising from the theoretical model and the results of the empirical testing procedures. It concludes with discussion and interpretation of the results. HYPOTEESES Initial hypothesis testing was conducted using classical linear regression techiques to estimate the parameters of the equations AUTit = do 4» alssct + azBUSt + a3AUTXt + at (11) MMAit = 80 + Bissct + (323051; + (33AUTxt + at (12) where AUTit = bank i’s automobile loan rate at time t MMAit = bank i’s money market account rate at time t SECt = security rate at time t BUSt = business loan rate at time t AUTXt = competitors’ rate at time t As explained earlier, the general nature of the loan and deposit rate functions underlying the solution to the theoretical model only predicts the signs of the coefficients, and not their magnitudes. Eight hypotheses associated with the model’s predictions (Table 11). 81 82 Table 11. Hypothesis Testing NULL HYPOTHESIS THEORETICAL MODEL PREDICTS . . . Test of Regression Equation REJECT null hypothesis; 1. HO: a1=a2=a3=0 rC is a function of rs, r3, and rx Test of Regression Coefficients 3r 2. Ho: ad 5 O REJECT null hypothesis; 5:2 > 0 5 3r 0 REJECT null hypothesis; 5;9 B (A) :1: O Q N IV .3r_c>o REJECT null hypothesis; arx Mb :1: 0 Q m IA 0 Test of Regression Equation [32 = (33 = o ACCEPT null hypothesis; 5. Ho: rD is NOT a function of rB and rx Test of Regression Coefficients . aro 6. HO: B1 S 0 REJECT null hypotheSIS; 5;— > 0 s arp [32 = o ACCEPT null hypothesis; arB II C) 7. HO: any 8. Ho: [33 — 0 ACCEPT null hypothesis; 8?- = O X 83 REGRESSION RESULTS Ordinary least-squares and first-order autoregressive procedures were run on the data set described earlier. The SAS statistical software package running on Michigan State University’s IBM 3090 mainframe computer was used for these analyses. Only the ‘best’ results corresponding to Equations 11 and 12 are reported.1 AUTOMOBILE LOAN RATES Table 12 shows the results of the regression model specified in Equation 11. Data from Bank 28 was excluded because its automobile loan rate did not vary over the sample period. 1For instance, each of the regression equations was estimated using levels as well as first differences; the latter was generally not useful in explaining the behavior of automobile loan or deposit rates. 84 Table 12. Automobile Loan Rate Regressions AUT - a0 + a1(rso) + a2(BUS) + a3(AUTX) + 81 Bank. a0 91, a2 a3 SE R2 F-rat. DW 1 6.05 -0.58 0.88 0.09 0.041 0.50 8 69 1.74 n-24 (1.00) (0.14) (0.19) (0.10) 2 4.46 0.18 -0.48 0.84 0.127 0.47 7 63 2.20 n=23 (1.80) (0.26) (0.47) (0.25) 3 -8.70 1.94 -3.64 3.23 0.241 0.84 41.76 1.49 n=24 (2.82) (0.33) (0.53) (0.37) 4 -1.90 0.20 -0.41 1.33 0.013 0.97 19.38 1.52 n-24 (0.62) (0.08) (0.11) (0.07) 5 -8.38 -0.21 1.51 0.91 0.255 0.80 31.71 1.37 n=24 (2.90) (0.34) (0.47) (0.29) 6 5.26 -0.37 0.88 0.04 0.027 0.73 21 45 1.84 n-24 (0.81) (0.11) (0.16) (0.08) 7 -3.20 0.15 -0.12 1.31 0 168 0.73 21.39 1.05 n=24 (2.23) (0.27) (0.40) (0.25) 8 7.38 -0.54 0.44 0.38 0.115 0 20 2.93 1.13 n=24 (1.66) (0.23) (0.33) (0.18) 9 —2.11 0.27 -1.09 1.81 0 085 0.80 27.20 1.58 n=21 (2.17) (0.28) (0.44) (0.26) 10 1.28 -0.37 -0.23 1.18 0.127 0.67 16.88 1.40 n=25 (2.49) (0.28) (0.41) (0.28) 11 8.56 0.05 0.68 -0.24 0.061 0.76 26.44 1.19 n=25 (1.60) (0.20) (0.27) (0 16) 12 5.74 0.21 -0.15 0.50 0.057 0 28 4 18 1.10 n=25 (1.58) (0.19) (0.27) (0.17) 13 -1.76 -0.96 1.72 0.49 0.133 0.64 14.98 1.54 n=25 (2.66) (0.28) (0.38) (0.26) 14 1.38 0.63 -1.32 1.42 0.091 0.63 14.46 1.19 n=25 (2.10) (0.25) (0.37) (0.25) 15 1.83 0.49 -0.56 0.95 0.205 0 19 2.82 1.54 n=25 (3.19) (0.38) (0.53) (0.35) 16 1.51 -0.23 .43 0.64 0.232 0.61 n=25 (5.41) (0.42) (0.64) (0.50) 17 -0.21 -0.16 0.18 1.02 0.043 0.77 28.50 2.28 n=25 (1.51) (0.15) (0.20) (0.13) 18 8.18 0.40 -0.37 0.30 0.086 0 10 1 76 0.93 n=21 (2.87) (0.22) (0.31) (0.28) 19 0.03 -0.29 0.41 0.86 0.101 0.31 3 99 1.14 n=21 (3.50) (0.23) (0.32) (0.33) Table 12. (contd.). AUT - a0 + a1(rso) + a2(BUS) + a3(AUTX) + 81 Bank 010 01 _az_ 013 SE R2 F-rat . ow 20 1.51 0.09 -0.30 1.04 0.118 0.53 9.92 2.09 n=25 (2.49) (0.26) (0.35) (0.22) 21 6.77 -0.27 -0.09 0.58 0.126 0.44 6.60 1.19 n-22 (2.41) (0.26) (0.36) (0.22) 22 0.94 -0.72 1.26 0.49 0.086 0.61 13.30 1.33 n=25 (2.09) (0.21) (0.28) (0.17) 23 -4.22 0.55 -0.42 1.30 0.107 0.61 13.64 1.75 n=25 (2.47) (0.25) (0.33) (0.21) 24 2.29 0.45 -0.15 0.63 0.317 0.14 2.32 1.35 n-25 (4.32) (0.41) (0.55) (0.37) 25 4.46 0.06 0.25 0.42 0.227 0.10 1.90 2.79 n=25 (3.54) (0.34) (0.46) (0.30) 26 -7.34 -1.05 1.07 1.52 0.239 0.59 12.37 1.38 n=25 (4.15) (0.33) (0.46) (0.36) 27 1.11 -0.63 0.86 0.57 0.142 0.35 5.32 1.32 n-25 (2.90) (0.26) (0.36) (0.24) 29 0.31 -0.23 0.34 0.90 0.041 0.70 20.03 2.67 n=25 (1.57) (0.14) (0.20) (0.14) 30 1.31 0.73 -1.61 1.52 0.328 0.40 6.43 1.53 n=25 (4.45) (0.42) (0.60) (0.43) 31 -7.19 -0.23 0.66 1.14 0.039 0.86 47.28 1.60 n=23 (1.71) (0.17) (0.23) (0.16) 32 12.81 -0.44 -0.10 0.23 0.055 0.78 27.67 1.70 n=23 (2.05) (0.21) (0.28) (0.19) 33 -0.45 -0.39 0.78 0.59 0.047 0.70 18.25 1.06 n=23 (1.90) (0.19) (0.26) (0.17) 34 4.97 0.37 -1.17 1.03 0.127 0.67 16.04 1.19 n=23 (3.11) (0.32) (0.42) (0.28) 35 -8.68 0.08 0.38 1.28 0.087 0.83 n=23 (3.65) (0.30) (0.43) (0.32) 36 -5.31 0.53 -0.60 1.40 0.038 0.81 32.51 2.01 n=23 (1.70) (0.17) (0.23) (0.15) 37 -2.70 0.90 -1.41 1.53 0.109 0.64 13.88 1.45 n=23 (2.88) (0.29) (0.39) (0.26) 38 7.80 -0.42 .80 -0.11 0.030 0.77 n=23 (2.03) (0.17) (0.25) (0.18) 39 1.51 0.53 -1.19 1.27 0.097 0.62 9.27 1.44 n'16 (2.11) (0.27) (0.46) (0.27) Table 12. (contdrl. AUT - a0 + 01(180) + 02(805) + a3(AUTX) + ei Bank 010 011 “L 013 SE R2 F-rat. ow 40 5.52 -1.13 2.23 -0.32 1.192 0.05 1.35 1.53 n-22 (4.06) (0.73) (1.11) (0.37) 41 7.27 -0.69 0.97 0.10 0.068 0.48 7.68 1.77 n=23 (1.20) (0.19) (0.28) (0.10) 42 1.86 0.39 -0.32 0.66 0.205 0.42 6.34 2.08 n=23 (2.19) (0.36) (0.55) (0.22) 43 -0.69 -0.45 0.92 0.57 0.046 0.86 45.87 1.50 n=23 (1.04) (0.17) (0.26) (0.10) 44 0.11 -0.27 0.89 0.43 0.024 0.92 82.36 1.65 n=23 (0.75) (0.12) (0.19) (0.08) 45 2.07 -0.45 1.15 0.19 0.375 0.28 3.85 1.31 n=23 (2.97) (0.48) (0.75) (0.30) Regressions computed using ordinary least squares (OLS) unless F-ratio and DW are not shown, when first-order autoregression (ARl) techniques were used. SE = standard error of the regression, R2 is adjusted except in the ARI regressions, F-rat.- F-statistic of the regression, DW - Durbin-Watson statistic. Hypothesis 1: Consumer Loan Rates React to Innovations in Security Rates, Business Loan Rates and Competitors’ Rates The theoretical model postulates that changes in the consumer loan rate are associated with changes in open-market security rates, the business loan rate, and competitors’ rates. For this to be true, the coefficients in Equation 11 must be non-zero. The F-statistics shown in Table 12 show this generally to be the case. Only seven banks (Marine Midland, First Chicago, Northern Trust, First Interstate, First National Bank of Monterey, Bank of Boston and Bank of New England) have F-ratios less than the required critical value,and cannot reject the null hypothesis with a 95% level are of confidence. The F-statistics by themselves, however, 87 not enough to support the theoretical model. Hypotheses 2, 3 and 4 must also be rejected for its confirmation. Hypothesis 2: Consumer Loan Rates Increase with Increases in Security Rates The significance of individual regression coefficients is determined with t-tests. If an individual t-statistic (calculated by dividing the coefficient in question by its standard error) is greater than a predetermined critical value, the null hypothesis must be rejected. A one-tailed test is appropriate for testing coefficient signs. Table 13 shows the results of t-tests on the coefficients in each of the regression equations shown in Table 12. The results are applicable at the 95% confidence interval. 88 Table 13. Signed Coefficients of Loan Rate Regressions AUT - a0 + 0110181)) + 0.2(BUS) + 013 (AUTX) + 81 (11 012 a3 a0 BANK NAME: + O + 0 23 Wells Fargo Bank + 0 0 + 18 First Interstate Bank + - + 0 14 La Salle National Bank + - + 0 30 Union Warren Savings Bank + - + 0 37 National Bank of Detroit + - + O 39 Ameritrust + - + - 3 Chase Manhattan + - + - 4 Chemical Bank + - + - 36 Manufacturers National Bank 0 + + 0 29 State Street Bank 0 + + - 5 Citibank 0 + + - 31 Comerica 0 + 0 + 11 Exchange National Bank 0 + 0 0 40 Bank One 0 0 + + 2 Bowery Savings 0 0 + + 12 First Chicago 0 O + + 21 Home Federal Savings and Loan 0 0 + 0 7 Manufacturer's Hannover 0 O + 0 10 Continental Illinois 0 0 + 0 15 Northern Trust Bank 0 0 + 0 17 Bank of America 0 0 + 0 19 First National Bank - Monterey 0 0 + 0 20 Great Western Savings 0 0 + 0 42 Fifth Third Bank 0 o + - 35 Great Lakes Savings Bank 0 0 0 0 16 Talman Home Savings and Loan 0 0 0 0 24 Bank of Boston 0 0 0 0 25 Bank of New England 0 0 0 0 45 Society Bank 0 - + 0 9 Ben Franklin Savings and Loan 0 - + 0 34 First Federal Savings Bank - + + 0 13 Harris Trust and Savings Bank - + + 0 22 Sears Savings Bank - + + 0 27 First Mutual of Boston - + + 0 33 First of America - + + 0 43 First National Bank - + + 0 44 Provident Bank - + + - 26 Boston Five Cents Savings Bank - + 0 + 1 Anchor Savings ~ + 0 + 6 First American Bank - + 0 + 38 Standard Federal - + 0 + 41 Central Trust - O + + 8 Marine Midland Bank - 0 0 + 32 Empire of America Entries represent coefficients significantly different from zero at a 95% confidence level. ‘+’ indicates the coefficient is significantly greater than zero, ‘-’ indicates the coefficient is significantly less than zero, and ‘0’ indicates the coefficient does not differ significantly from zero. 89 It is generally not possible to reject Hypothesis 2 at the 95% level of significance, implying that consumer loan rates are not very sensitive to changes in the security rate (as proxied by the interest rate on 3-year constant maturity Treasury bonds). Of the banks which show a positive relationship, most are strong regional institutions and two (Chase Manhattan and Chemical Bank) must be considered ‘money-center’ banks. Hypothesis 3: Consumer Loan Rates Decrease with Increases in the Business Loan Rate As shown in Table 13, Hypothesis 3 is not well- supported. For most banks, the rate offered on new automobile loans is not negatively related to the prime rate. Hypothesis 4: Consumer Loan Rates Increase with Increases in Competitors’ Rates With few exceptions, Hypothesis 4 can be rejected at a 95% confidence level, indicating interbank competition is very important in setting consumer loan rates. These results support the prediction of the theoretical model and suggest the new automobile loan market is imperfectly competitive, a claim which runs counter to some published research but which supports observations made by practioners. The strong influence of competitors’ rates suggests that they alone might be useful in explaining the behavior of automobile loan rates over time. Table 14 presents the results of regression analyses performed using just an 90 intercept and competitors’ rates charged on consumer loans as explanatory variables. 91 Table 14. Automobile Loan Rates vs. Competitors’ Rates AUT - a0 + “1(AUTX) + 81 Bank 010 011 SE R2 F-rat. ow 1 8.30 0.15 0.077 0.10 2.38 1.16 n-24 (1.17) (0.10) 2 4.13 0.64 0.122 0.52 22.61 1.86 n=23 (1.55) (0.13) 3 -10.49 2.01 0.405 0.76 n=24 (5.57) (0.49) 4 -2.41 1.21 0.020 0.95 24.43 1.38 n-24 (0.68) (0.06) 5 -3.29 1.43 0.374 0.73 n=24 (4.81) (0.43) 6 7.74 0.21 0.059 0.45 n=24 (1.46) (0.12) 7 -3.84 1.38 0.156 0.76 68.65 1.20 n=24 (1.91) (0.17) 8 7.88 0.31 0.112 0.29 n=24 (2.27) (0.20) 9 -1.26 1.09 0.093 0.80 n=21 (3.83) (0.35) 10 2.85 0.61 0.114 0.73 n=25 (4.07) (0.37) 11 12.16 -0.01 0.046 0.83 n=25 (2.47) (0.23) 12 4.47 0.65 0.042 0.52 n=25 (1.97) (0.19) 13 4.16 0.60 0.105 0.74 n=25 (3.81) (0.35) 14 1.83 0.80 0.054 0.80 n=25 (2.84) (0.26) 15 3.61 0.70 0.205 0.22 6.45 1.30 n=25 (2.90) (0.27) 16 3.74 0.60 0.204 0.63 n-25 (5.16) (0.48) 17 -0.62 1.08 0.041 0.79 87.15 2.33 n=25 (1.30) (0.12) 18 9.47 0.17 0.043 0.60 n=21 (3.28) (0.30) 19 3.23 0.68 0.073 0.55 n=21 (4.18) (0.37) Table 14. (contd.). AUT - so + a1(AUTX) + 81 Bank (10 (11 SE R2 F-rat . DW 20 -0.03 1.02 0.120 0.54 26.97 1.82 n=25 (2.20) (0.20) 21 4.35 0.56 0.101 0.59 n-22 (2.91) (0.26) 22 4.95 0.58 0.101 0.58 n=25 (3.20) (0.28) 23 “0.15 1.00 0.123 0.59 n=25 (3.17) (0.28) 24 8.32 0.28 0.287 0.29 n=25 (4.96) (0.44) 25 7.66 0.36 0.247 0.06 1.55 2.39 n=25 (3.30) (0.29) 26 3.07 0.70 0.142 0.78 n=25 (3.86) (0.34) 27 1.54 0.73 0.168 0.27 8.40 1.24 n=25 (2.93) (0.25) 29 0.61 0.97 0.043 0.70 54.56 2.41 n=25 (1.49) (0.13) 30 5.12 0.50 0.332 0.45 n=25 (5.73) (0.50) 31 -6.69 1.41 0.053 0.83 n=23 (2.74) (0.23) 32 6.24 0.43 0.110 0.61 n=23 (4.23) (0.35) 33 0.28 0.82 0.067 0.62 n=23 (2.45) (0.20) 34 3.19 0.61 0.109 0.74 n-23 (4.34) (0.36) 35 —5.64 1.34 0.091 0.80 n=23 (3.97) (0.32) 36 -2.40 1.08 0.054 0.74 61.22 1.41 n=23 (1.67) (0.14) 37 '1.59 1.02 0.153 0.54 n=23 (3.74) (0.31) 38 7.71 0.19 0.022 0.81 n=23 (2.07) (0.17) 39 3.02 0.70 0.074 0.75 n=16 (3.16) (0.24) 93 Table 14. (contd.). AUT - a0 + a1(AUTX) + 81 Bank (10 011 58 R2 F-rat. DW 40 10.08 0.11 1.217 0.12 n-22 (lTU (02% 41 7.97 0.27 0.103 0.24 6.67 1.48 n-23 (ram (040) 42 2.35 0.65 0.214 0.42 15.39 1.73 n=23 (22m (01m 43 -O.14 0.85 0.076 0.78 73.56 1.50 n=23 (L3H (030) 44 3.42 0.58 0.066 0.79 n=23 (Len (013 45 5.97 0.36 0.312 0.46 n=23 (15m (026) Regressions computed using ordinary least squares (OLS) unless DW and F-ratios are not shown, when first-order autoregression (ARI) techniques were used. SE - standard error of the regression, R2 is adjusted except in the ARI regressions, F-rat.- F- statistic of the regression, DW - Durbin-Watson statistic. The sample institutions may be separated into four groups according to the results of t-tests performed on the slope coefficient (Table 15). For a bank to behave as a pure monopoly, a1 should be equal to 0. Pure competition implies a1 = 1. The third group consists of “imperfectly competitive” institutions, in which neither of these hypotheses can be rejected. The fourth (and smallest) category is composed of institutions in which both Hypothesis 1 and 2 are rejected; these results are unexplainable under this framework. "Imperfectly Competitive" Banks 94 Table 15. Institutions Classified by Market Structure AUT - so + 01(AUTX) Hypothesis 1: a1 = 0 FAILURE TO REJECT implies monopolistic behavior Hypothesis 2: a1 - l FAILURE TO REJECT implies competitive behavior FAIL TO REJECT H1: 01 - o REJECT Al: 01 ¢ 0 REJECT H2: 01 ¢ 1 FAIL TO REJECT H2: 01 a 1 "Monopolistic" Banks "Competitive" Banks 1 Anchor Savings Bank 3 Chase Manhattan 6 First American Bank 5 Citibank 8 Marine Midland Bank 9 Ben Franklin Savings 6 Loan 11 Exchange National Bank 12 First Chicago 18 First Interstate Bank 14 LaSalle National Bank 25 Bank of New England 15 Northern Trust Bank 38 Standard Federal Bank 17 Bank of America 40 Bank One 20 Great Western Bank 45 Society Bank 21 Home Federal Savings 6 Loan 22 Sears Savings Bank 23 Wells Fargo Bank 27 First Mutual of Boston 29 State Street Bank and Trust 31 Comerica 33 First of America 35 Great Lakes Savings Bank 36 Manufacturers National Bank 37 National Bank of Detroit 39 AmeriTrust 42 Fifth Third Bank 43 First National Bank REJECT H1: 011 ¢ 0 FAIL TO REJECT H1: 0.1 “ 0 REJECT H2: (11 ¢ 1 FAIL TO REJECT H2: 01 a 1 bub "Unclassified“ Banks 2 Bowery Savings Bank 10 Continental Illinois 4 Chemical Bank 13 Harris Trust 8 Savings Bank 7 Manufacturers Hanover 16 Talman Home Savings 6 Loan 1 Central Trust 19 4 Provident Bank First Nat'l Bank Monterey 24 Bank of Boston 26 Boston Five Cents Bank 30 Union Warren Savings Bank 32 Empire of America 34 First Federal of Michigan a 95% significance interval. Decisions to accept or reject the stated hypotheses based on a two—tailed test at. 95 Automobile loan rates at seven of the nine banks classified as “MOnopolistic” appear to follow the movements of the prime rate and suggest a simple form of markup pricing. The remaining two banks, First Interstate and Bank of New England, have near-insignificant coefficients on any of the explanatory variables. Twenty-one of the 44 banks in the sample have automobile loan rates which behave in accordance with a competitive market. These banks are fairly evenly dispersed across geographic regions and among size classes, demonstrating that competition is more widespread than suggested by earlier authors. A cautionary note: this is a small sample (44 out an estimated 14,000 banks in the United States) examined over a rather short period (25 months). Of the five banks in the third group, only Banks 2, 41 and 44 have coefficients on the competitive variable between 0 and I and can properly be labelled “Imperfectly Competitive”. Chemical Bank and Manufacturers Hanover do not fit this classification, as their coefficients on the competitive variable are significantly greater than 1. The final group of banks designated “Unclassified," are a mystery. A third are from Chicago, and a third are from Boston, but no other distinguishing characteristics are apparent. In summary, the empirical evidence shows that the automobile loan market is better characterized by a competitive model with respect to automobile loan rate behavior. 96 Attempts to investigate the influence of institutional location and size were thwarted by the heterogeneity of behaviors within a given classification. As indicated by the high F-ratios in Tables 16 and 17, intraregional and intra- size class variation was too great to allow continued examination of interregional or inter-size class differences. Table 16. Intraregional variation of Auto Loan Rates Unrestricted: AUT - £01,338k + 81(1'80) + 82(BUS)+ (33(AUTX) 4» (4i Restricted: AUT - 010 + 81mm) + [32(805)+ 83(AUTX) + 81 Hypothesis: All banks within a given region behave in sufficiently the same manner as to allow "pooling" of the data Number of . . Region Institutions F-ratio Dec131on New York City 8 69.13 REJECT Chicago 8 61.83 REJECT California 7 26.18 REJECT Boston 6 75.90 REJECT Detroit 8 27.74 REJECT Cincinnati 7 13.14 REJECT 97 Table 17. Intra-Sise Class variation of Auto Loan Rates Unrestricted: AUT - Edknsk + 81(1'80) + 82(BUS)+ B3(AUTX) + 81 Restricted: AUT - (10 + 81mm) + 82(BUS)+ 83(AUTX) + 81 Hypothesis: .All banks within a given size class behave in sufficiently the same manner as to allow "pooling" of the data Size Number of Class Assets ($) Institutions F-ratio Decision 5 $ 45 to 100 billion 4 20.77 REJECT 4 25 to 50 billion 5 60.86 REJECT 3 10 to 25 billion 7 29.56 REJECT 2 5 to 10 billion 11 70.46 REJECT I l to 5 billion 13 26.55 REJECT DEPOSIT RATES The response of an institution’s deposit rate to changes in the security rate, business rate and competitors’ rates is shown in Table 18. Banks 18 and 19 did not offer money market accounts during the sample period and were excluded from the study. 598 Table 18. Deposit Rate Regressions MMA - do + d1(rso) + 02(803) + a3(AUTX) + 81 Bank do “1 012 013 SE R2 F-rat . ow 1 3.01 -0.33 0.53 0.07 0.041 0.56 0.74 n=23 (1.44) (0.17) (0.24) (0.13) 2 2.81 -0.71 1.38 -0.27 0.040 0.57 24.85 1.17 n=22 (0.82) (0.14) (0.28) (0.14) 3 -0.99 -0.49 .95 0.20 0.047 0.75 5.19 n-24 (2.03) (0.19) (0.31) (0.21) 4 0.97 -0.38 .86 0.03 0.030 0.80 0.08 n=24 (1.27) (0.14) (0.21) (0.13) 5 1.13 -0.27 0.81 -0.02 0.040 0.64 4.06 1.13 n=24 (1 14) (0.13) (0.19) (0.11) 6 1.43 -0.59 1.04 0.00 0.058 0.48 13.81 1.04 n=24 (1.19) (0.16) (0.23) (0.12) 7 -0.82 -0.39 0.93 0.15 0.074 0.56 4.77 1.02 n=24 (1.48) (0.18) (0.26) (0.16) 8 1.24 -0.60 1.16 -0.07 0.040 0.64 20.30 1.32 n=24 (0.98) (0.13) (0 19) (0.11) 9 3.38 -0.21 0.35 0.08 0.012 0.72 5.55 n=25 (0.64) (0.09) (0.15) (0.05) 10 -0.14 0.23 -0.31 0.62 0.031 0.45 2.65 1.14 n=25 (1.23) (0.14) (0.21) (0.14) 11 -0.33 0.10 0.05 0.47 0.012 0.75 1.31 1.35 n=25 (0.71) (0.09) (0.12) (0.07) 12 -1.26 0.26 -0.22 0.64 0.010 0.82 9.92 1.29 n=25 (0.66) (0.08) (0.11) (0.07) 13 3.64 0.20 -0.34 0.33 0.015 0.37 4.49 1.32 n=25 (0.89) (0.09) (0.13) (0.09) 14 -1.07 0.04 0.01 0.60 0.010 0.80 0.22 1.07 n=25 (0.71) (0.09) (0.12) (0.08) 15 0.46 0.10 -0.01 0.43 0.014 0.61 1.02 1.39 n=25 (0.85) (0.10) (0.14) (0.09) 16 1.76 0.04 -0 09 0.39 0.005 0.80 2.18 n=25 (0.75) (0.06) (0.09) (0.07) 17 1.25 -0.25 0.41 0.21 0.012 0.58 9.67 1.95 n=25 (0.81) (0.08) (0.11) (0.07) 20 1.42 -0.03 0.00 0.38 0.011 0.65 0.20 1.25 n=25 (0 74) (0.08) (0.10) (0.07) 21 5.99 -0.10 0.00 0.02 0.023 0.56 1.05 n=25 (0.92) (0.12) (o 17) (0.03) (Tamils) 11!. (contd. ) . MMA - a0 + 01(r80) + 02(805) + d3(AUTX) + 81 Bank (to 011 (12 d3 58: 9.2 F-rat . DW 22 -0.75 0.00 0.08 0.50 0.007 0.85 0.00 1.16 n-21 (0.61) (0.07) (0.11) (0.06) 23 '2.03 -0.16 0.50 0.36 0.027 0.62 1.65 2.20 n=25 (1.24) (0.13) (0.17) (0.11) 24 -1.72 -0.06 0.86 0.13 0.053 0.79 0.13 1.27 n-25 (1.78) (0 17) (0.23) (0.15) 25 0.12 -0.13 0.41 0.30 0.023 0.73 1.30 n=25 (1.49) (0.12) (0.17) (0.12) 26 -0.31 -0.40 0.69 0.31 0.031 0.54 11.03 1.48 n=25 (1.51) (0.12) (0.17) (0.13) 27 -1.86 -0.33 0.85 0.28 0.042 0.67 5.48 1.58 n=25 (1.57) (0.14) (0.19) (0.13) 28 -1.09 -0.06 0.44 0.35 0.025 0.71 0.34 2.09 n=25 (1.05) (0.11) (0.15) (0.09) 29 -3.39 -0.14 0.58 0.48 0.035 0.71 1.10 1.33 n=24 (1.45) (0.13) (0.18) (0.13) 30 -4.00 -0.49 1.32 0.26 0.112 0.64 4.0 1.59 n=25 (2.60) (0.25) (0.35) (0.25) 31 '3.35 -0.15 0.34 0.59 0.015 0.77 1.73 1.53 n=22 (1.26) (0.11) (0.15) (0.11) 32 -0.31 0.16 -0.09 0.46 0.028 0.40 1.22 1.54 n=23 (1.46) (0.15) (0.20) (0.13) 33 -4.19 0.00 0.04 0.77 0.026 0.74 0.00 1.99 n=23 (1.41) (0.14) (0.19) (0.13) 34 -1.54 -0.02 0.03 0.57 0.015 0.73 0.03 1.31 n=23 (1.08) (0.11) (0.15) (0.10) 35 -2.20 -0.08 0.00 0.68 0.016 0.81 0.46 1.38 n=23 (1.11) (0.11) (0.15) (0.10) 36 -1.84 -0.19 0.44 0.42 0.009 0.85 5.34 1.81 n=23 (0.80) (0.08) (0.11) (0.07) 37 -4.28 —0.56 1.22 0.34 0.031 0.83 12.98 2.22 n=23 (1.54) (0.16) (0.21) (0.14) 38 -1.83 0.05 0.01 0.57 0.009 0.81 0.33 1.29 n=23 (0.82) (0.08) (0.11) (0.07) 39 2.06 0.10 0.08 0.16 0.009 0.77 1.70 0.97 n=17 (0.55) (0.08) (0.12) (0.06) 40 4 90 «0.32 0,53 -o.11 0.186 0.15 0.48 0.75 n-13 (2.15) (0.46) (0.54) (0.20) 100 Table 18. (contd.). MMA - do + a1(TBD) + a2(BUS) + d3(AUTX) + ti Bank do (:1 EL (13 SE R2 F-rat. DW 41 4.09 -0.10 -0.03 0.22 0.076 0.16 0.23 .66 n=23 (1.27) (0.20) (0.30) (0.10) 42 4.55 -0.45 0.86 -0.20 0.153 0.03 2.12 47 n-23 (1.89) (0.31) (0.48) (0.19) 43 -1.74 -0.13 0.36 0.40 0.035 0.74 0.77 19 n=23 (0.90) (0.15) (0.23) (0.09) 44 0.84 0.26 -0.11 0.29 0.023 0.71 4.50 .37 n=23 (0.74) (0.12) (0.19) (0.07) 45 2.56 0.01 -0.70 0.62 0.107 0.57 0.00 .56 n=22 (1.59) (0.27) (0.41) (0.16) Regressions computed using ordinary least squares (OLS) unless DW is not shown, when first-order autoregression (ARI) techniques were used. SE - standard error of the regression, R2 is adjusted except in the ARI regressions, F-rat.= F-statistic for the hypothesis Ho: 52 - 82 - 0, DW - Durbin-Watson statistic. Rypothesis 5: Deposit Rates React to Innovations in Security Rates, but NOT Business Loan Rates and Competitors’ Rates According to the theoretical model, deposit rates should respond to Changes in open market security rates and be unresponsive to changes in the prime rate or the rates charged by competitors on consumer loans. This assertion is tested with an F-statistic: if it is possible to accept the joint hypothesis [32 = Ba a 0 then further investigation of the the security rate coefficient is warranted. The results shown in Table 18 are mixed; of the 43 financial institutions examined, the null hypothesis is rejected approximately one- third of the time. The model’s predictions are realized in the remaining two-thirds of the sample. There seems to be no pattern describing the group of rejecting banks. 101 Bypothesis 6: Deposit Rates Increase with Increases in Security Rates As discussed earlier, tests of individual coefficients are conducted using t-statistics. Table 19 summarizes the signs of the significant coefficients from the deposit rate regressions. The coefficient on the security rate proxy is positive in 22 of 43 regression equations. Eypothesis 7: Deposit Rates are Unaffected by Changes in Business Loan Rates It is not possible to draw meaningful conclusions regarding Hypothesis 7, as the t-statistics indicate the null hypothesis is rejected in 24 out of the 43 regression equations and accepted in the remainder. Bypothesis 8: Deposit Rates are unaffected by Changes in Consumer Loan Rates Hypothesis 8 is rejected for 28 of 43 banks, and unable to be rejected by 14 of 43 banks,suggesting that deposit rates are correlated with consumer loan rates at the majority of banks in the sample. 102 Table 19. Siged Coefficients of mosit Rate Refiessions MMA - do + comm) + 0.2 (BUS) + (13(AUTX) + 81 £1 (12 a3 a0 BANK NAME + 0 + 0 44 Provident Bank + - + + 13 Harris Trust and Savings Bank + — + - 12 First Chicago 0 + + 0 23 Wells Fargo Bank 0 + + 0 28 New World Bank for Savings 0 + + - 29 State Street Bank 0 + + - 31 Comerica 0 + 0 + 42 Fifth Third bank 0 + 0 0 24 Bank of Boston 0 + 0 0 25 Bank of New England 0 0 + + 16 Talman Home Savings and Loan 0 0 + + 20 Great Western Savings 0 0 + + 39 Ameritrust 0 0 + + 41 Central Trust 0 0 + 0 10 Continental Illinois 0 0 + 0 11 Exchange National Bank 0 0 + 0 14 La Salle National Bank 0 0 + 0 15 Northern Trust Bank 0 0 + 0 22 Sears Savings Bank 0 0 + 0 32 Empire of America 0 0 + 0 34 First Federal Savings Bank 0 O + 0 45 Society Bank 0 0 + - 33 First of America 0 0 + - 35 Great Lakes Savings Bank 0 0 + - 38 Standard Federal 0 0 + - 43 First National Bank 0 0 0 + 21 Home Federal Savings and Loan 0 0 0 + 40 Bank One - + + 0 17 Bank of America - + + 0 26 Boston Five Cents Savings Bank - + + 0 27 .First Mutual of Boston - 1 + - 36 Manufacturers National Bank - 1 + - 37 National Bank of Detroit - + o + 1 Anchor Savings - + 0 + 9 Ben Franklin Savings and Loan - + 0 0 3 Chase Manhattan - + 0 0 4 Chemical Bank - + 0 0 5 Citibank - + 0 0 6 First American Bank - + 0 0 7 Manufacturers Hanover - + 0 0 8 Marine Midland Bank - + 0 0 30 Union Warren Savings Bank - + - + 2 Bowery Savings Entries represent coefficients significantly different from zero at a 95% confidence level. ‘+’ indicates the coefficient is significantly greater than zero, \_I indicates the coefficient is significantly less than zero, and ‘0’ indicates the coefficient does not differ significantly from zero. 95‘ confidence interval for banks in italics. Hypothesis 1 is rejected at the 103 DISCUSSION This section discusses the results presented in the previous chapter and attempts to resolve inconsistencies between the predictions of the theoretical model and observed behaviors. Before beginning a discussion of these tOpics, however, it is important to note that one of this dissertation’s biggest strengths is also one of its biggest drawbacks. This statement refers to the use of microeconomic time series data for the empirical testing. The use of individual, rather than aggregate, bank data is justified by both theoretical and practical considerations, however, it may lead one to “ ... miss the forest for the trees.” Theory suggests individual firms seek to maximize profits; it is unlikely that aggregates pursue such enlightened self interest. Further, bank managements are concerned primarily with decision-making at their own institutions, and much less with the results of aggregate behaviors. However justified, the decision to examine the behavior of individual financial institutions makes interpretation of the results rather difficult, for in every test, some banks' results agree with the theoretical predictions and others do not. Because no clear-cut answer to the working hypotheses generated in Chapter V was found, the remainder of this discussion will focus on reasons why the results might NOT support the theoretical model developed in Chapter III. 104 Three possible explanations will be discussed: 1) Is the model wrong? 2) Is the data suspicious? 3) Is the world different than theory? IS TB! NDDEL‘WRONG? If the theoretical model developed in Chapter III has been incorrectly specified or if its solution contains mathematical errors, then it would not be surprising for the empirical results to differ from the predicted theoretical responses. Operating under the belief that the model has been correctly manipulated and solved leads one to question the validity of its underlying assumptions. A crucial assumption is that consumer loan markets are imperfectly competitive, i.e. banks have some price-setting ability. Recall that a controversy exists amongst academics and practitioners concerning bank market structures. The theoretical model assumes a bank’s consumer loan volume results from the interplay of endogenous pricing decisions and competitors' offerings, although it does not specify the exact nature of this interaction. The strong positive effects of competitors’ rates on a many banks’ automobile loan rates suggests this market more closely approximates the pure competition model and might be better described with another approach, more similar to the portfolio theory models reviewed earlier. The results of regressions run against competitors’ rates alone are indicate that the assumption of imperfect competition is untrue for most banks in the sample. 105 The failure of the theoretical model to explain the behavior of deposit rates was not wholly unexpected. A bank offers many types of deposit accounts, each with different characteristics, and it is naive to presume that a single account could capture the behavior of the overall aggregate. In defense of these non-results, it should be noted that the accurate description of the liability side of a bank’s balance sheet was NOT a primary consideration in the formulation of the model. Other researchers have investigated questions of deposit-rate determination in much greater depth, usually by including simplifying assumptions on the asset side (Weber, 1966; Klein and Murphy, 1971; Sealey, 1980; Flannery, 1982). IS Inn DATA.SUSPICIOUS? The strengths and weaknesses of the data used to test the empirical model have already been discussed. Non- sampling errors due to inaccurate collection or transcription of the individual banks’ data and the difficulties inherent in the choice of a single proxy for an institution’s security portfolio were cited. Money market account rates are surely an oversimplified proxy for aggregate bank deposit rates. A few additional points might also be raised. First, the sample period is relatively short, with only twenty-five observations spanning a period only slightly longer than two years. Also, this was not a highly volatile period, and the variation in loan rates was quite small. The automobile loan 106 rate of one bank did not change at all during the sample period! The small sample size and low variability of the data lead to decreased confidence in the estimation of the “true” results. The composition of the available data set is also an important issue. The sample institutions came from six metropolitan areas and were primarily composed of large banks. Indeed, seven of the ten largest banks in the country were included in the sample. Only a few large savings and loans or savings banks were included; credit unions were notable by their absence. It is quite possible different results would be obtained if the theoretical model were tested with data from smaller institutions in more rural markets. IS THE WORLD DIFFERENT FROM THEORX? The British philosopher Alfred North Whitehead is reputed to have remarked, “Seek simplicity m and distrust it.” Hopefully, the theoretical model captured the essential characteristics of a bank’s decision—making process while ignoring extraneous details. However, it is very likely that certain important information was excluded in the interest of maintaining the model's tractability. For instance, the model assumes bank managements are risk-neutral and attempt to maximize end-of-period profits. If these assumptions are not met in practice, the empirical tests are flawed. One result of the empirical tests has been to identify consumer 107 loan markets as being much farther along the continuum from monopoly to competitive markets. Other possible ways in which the real world might differ from its theoretical construct include institutional constraints, such as time lags in the adjustment of rates. Another possible complication is associated with the variable used to proxy the consumer loan rate. Perhaps the interest rates charged on new automobile loans is not a good proxy for the consumer loan rate? The theoretical model is mum on the subject of non-interest rate compensation. Of the three possible explanations for the divergence of the empirical results from the theoretical predictions, this last explanation, “the world is different from theory,” is the most appealing. First, the theoretical model is a simplification of the real world. Financial institutions make many types of loans in addition to the two specifically examined in the model. Mortgage lending is an important outlet for many banks’ funds. Second, the markets for which we have data are much more competitive then previously thought. A theory-of-the—firm model of the type used in this dissertation is perhaps not an appropriate description of bank behavior in these markets. Finally, the maximization of short-term profits may not be the relevant objective function. Other considerations, e.g. regulation, strategic policy and managerial satisfaction may be important. ‘VI. .SUHUIAJUY AJflD CXNNCHHUSIKHES This study reviewed and examined the theoretical and practical issues surrounding the determination of consumer loan rates at commercial banks. It developed a microeconomic model of the banking firm emphasizing the consumer loan pricing decision and tested this model with time series data from individual United States banks. This chapter summarizes the major conclusions and implications of this research. Chapter I conceptualized and defined the central research issue of this dissertation: Although lending to consumers is a rapidly expanding area of bank activities, very little theoretical or empirical work has been conducted on the pricing of consumer loans. The relevant consumer credit literature was reviewed in Chapter II and key issues were identified. It was argued that consumers are both knowledgeable and responsive regarding the interest rate charged on consumer loans. A discussion of existing models of the banking firm and empirical observations of bank loan markets led to the choice of a monopoly, or ‘theory-of-the- firm' model for modelling consumer loan rates. The closing paragraphs of the literature review summarized the results of the few empirical investigations of bank loan rate behavior and briefly reviewed the relevant deposit rate literature. 108 109 THE THEORETICAL MODEL The model developed in Chapter III extended earlier work by allowing a bank to hold more than one type of loan portfolio. Previous studies assumed banks used the monies collected from deposits to make “loans”. Cost functions for administering the various loan and deposit portfolios were included as draws against bank revenue, although the requirement that the second-order conditions be met obviated their usefulness. The bank’s choice variables were the consumer loan rate and the rate paid on bank deposits, while security rates, the business loan rate, and the rate on consumer loans offered at competing institutions were taken to be exogenous and beyond the bank’s immediate control. The model’s solution was stated only in general term, although testable hypotheses of the predicted signs of its coefficients were generated. Consumer loan rates were predicted to respond positively to positive innovations in security rates and competitors' rates and negatively to increases in the business loan rate. The rate paid on deposits was dichotomized from lending rate decisions and predicted to increase as security rates increased. Changes in the business loan rate or the rate offered on competitive consumer loans were predicted to have no effect on deposit rates. 110 EMPIRICAL RESULTS The hypotheses generated by the theoretical model were tested with classical linear regression techniques and probit analysis, although the latter technique did not yield meaningful results. The results of the hypotheses stated in Chapter IV are summarized in Table 20. lll mcofimwn omflum> c9 sauna ocm Hamfin anon mum mxcmn mmumu .nuouflummfioo no mmmcansn >n.vouowmwm .0 oumnnmn.m 00 HO .muOufiquE nmumu mmmcmno 2Q noun“ mmOCHmsn CM mcauomflmu amCOflnsHocoo waamuwcmo 902 mum mwumu nanommo omuommwm 902 mum mmumu uflmoamo .m sewn on pasoflwwfin ancowusufiuncfi we me am an Qm9mMOU< o w. 9 "om DUOHHOOCH ma mmumu 2MB coauQESmmm Osauumm a pa: . . omEoo an mommeuoc. c . IOUAHQ unwocwsawcfi mxcmn we 90 Na um macmCOMumamu oc .mumwwwoca mmumu cmoa “wasmcoo .v nouauwQEoo vacuum amxcmn «e no mm um Qflcmcoaumaou m>H9Hm0m m . o N we “ox Q2 Q Q a moamacman nmumu nflnmcuo s Hows usufim ma um m>fluano a»: . . . Ga a3 ecu Umuomnmu omam mxcmn .nxcmn we mo ma um Qflcmcoflumamu oc cmoa mmocflmsn an mwmmmuwmfismwmo .m mcwuomflmu m was we cm>om anxcmn we no a um chncoaumawu m>9940m2 unmouomp means a H no mOfiaowuuoa cmoa Huesmcoo o w d : Hanan w>mn on Use» memn nauam ma um m>flunmmc maucmoflwflcoflm mcauomflmu m .mumumuwucH .nxcmn ve we mm on chncoaumamu oc moumu huflusowm c9 nonmmuocw cues . umuwh MO COMHQOOXO nuw3 umxcmfl vv MO m um anmCOMUMHOH m>H9HmOm OmmeOCfl mmumu CMOH HOESMCOU N o a mu n Nd u H6 "0: mxcmn Hmcowmmu summed was we can hogan mmumu .muoufluwQEoo 0cm noun“ ecu cw nxcmn unwaamsn moanmawm> omens up omuowmmm mmwcamsn .mmumu muflusomn c9 03» can an owummoom 9aamumcwm was nmumu :moa unanncoo mcowum>occfl an omuomdefl 902 we nfimmnuoahc Hasz ancoflusuwuncfl av no mm 9n om90momm mum nmumu cmoa Hoesmcoo .H mxmazmm messmmm mHmmzeoa»= chunon. nanonuomaw mo mafia—mom .oa 09.399 112 cued umfisncoo xcnnncso nus: pennaouuoo one nouns nanommp mamas: .coaumcnamxm 02 nouns cmoa unocfinsn Eouu powanuocoap one noun» nanoaoo nouns ucsooom assume usage :9 advance mcwcamamxo um naadnu9 can» umuumn adu:o«an nocon-a unocsocsu coo assay am we om um o>auqnoa haucaowuficoMn «nxcmn mm mo m um anncoauoaou 02 wanna mu no as an o>quqnoc aauanonuscosn «oxcmn an no ma us mwcncoaueaou 02 mama“ am no v no o>auuuoc aducoowmwcman «nxcmn an «0 mm um aqcncoaunaou 02 o I mu ”0: nouns tououwuoaaoo cw oco«uu>occ« ou poundoucs one nwunu nanoaoo O I N“ no: nounu ceoa onocdnsn cw acoauo>occa on nauseous: one nouou uwnofioo o w an no: nousu auausooo ca nonnouoca baa: . onoouocq nounu nanoaon .m mxmumm ufimommp mcflmaamsm mo cofiuocsm umoo n onox 09a0muuoa cmoH Hussmcoo man mcaoa>uom can mcaumcamwuo mo coauocsw umoo u onbx OAHOHDuOQ cued mmmcflmsn ecu mcfiofl>uon can mafiumcamauo mo Goduocsu unoo u Amcex mucmewuasvmu >ufloasvaa new o>ummmu How can: mnemommp mo ommucuouom u 0 upon mwwuausomm magnumxume n ma mnemommp seen we mEsHo> u m munu uamommp u on manoH mmmcfimsn mo oE=Ho> n m mums awed mmocfimsn u an mecca noesmcoo we oEsao> n 0 upon cnoH qusmcoo u on "mums: m :6 82x I 32x I 8.0x I can I Hm I o I Ed I :7 + m3 + can u x "ceauocsm m>auumflno on» Hamuom .Axu .mu .muvm u «on can .xu .au tun.“ n «on macauocsm COADSHOm can as mucmaowmumoo on» down mamsosmanencs on use sedan adequapcoo usuaoupcoomm on» an pmfiaqefi macauofiuummu mc9 .0“ can on embannwum> geamfiouo ozu on» you mca>90m >HmsomcmuHsEam new ones on Hands mCOAuLocoo ampuOIumwau on» no maneucououmap Haven 0:» mcauuom an oocfimuno mg m umumnno c9 oonHo>mo Hopes ceaumuasaxme pamoum noduomloco ecu on ceausHOm oc9 XMHh GZHKZ‘Q :8 b0 Hun—0: AaHagfllag ‘ .‘ ”HG—aflmmd 117 '.J. . ... ..I .wnrmhlquvIIIJI‘II-II.NI., I 85 am cum a cum m I IquIW Am»: OIIQMIQIXMIQIIWMHINMIAUIHVHIHE I ohm mm ohm om ohm cum oumm ohm I oum ANANV DiagllnvlmlulxlawlmlmmHIIWIQImHllmlm.H+IHIVMUH+UIIm—..m. “ewe emec9 .mc0aufiocoe wepuOIumuHm 03» 0» emau e>am meaneaue> ceamaeep osu ec9 .cofiusHOm m.aeooe ecu on mcfloeea mmeue ecu Hfieueo memem mcazoaaow mac9 .meEsao> meauausoem one nanomeo on me reevaoco emecu some soaaom mceoa mmecamsc one neesmcoe mo mae>em acfiuaeflerIuauoue ec9 .eueu nanomep ecu use eueu ceoH measecoo ecu ewe meaneaue> coinaueo ntxcen ec9 .o A 322x o A 82.x o A £38 .63 .I. a o A 322x o A 2.2.x o v 0.63m o A 33mm innate I m o A 8.0.2“ o A 8.0.x o A 398 o v #38 ixuato I o "mauaefiaaxe oeueum ewe uofi>ecen xcen unoce macauQEsnme acazoaaow ecu oce 118 Em 8QO as am y mum mum AS: ovllIIIIIIIHIIoII+IIIIauI8IZmuIIIII 8 Town as an ...m OH Oummm Ohm mm m m wum oumoml IUIIIMML. Um m U “mum I “mum Gflovum+ mRNm mm+mmequm~m+ oxwm +omequ0~mIR~m .EsmeeE ammoua e eusmce ou ones cecu mmea en Hess A940 use and. escaumvcoo HepuOIocouem ec9 Om cum GHQ 35 ouaIIIouIAUICmuvllll as am 5.. mm 0|..Hm+ om UHm OH“ :1: ou0+IIIImuImva|m I IoullIlll .58 Am om .5 eosooum Ou pemceuueeu ec ceo ceacz 119 came. Am9uflope eue ecofiuocou ueoo ecu “H 120 on o I 323 + 523 + mush? I SIM + 8:: QHmN Om on GHQ SuI am :IIonmIouImfisI HWI; Ima +£23Ific N xum oum oum oum H x I I I I as: o SAW. + SnI mm + 36 mm om + £23 + cum mm wum om Uri 3m U I I I I m I I I m I I no 0mm + 8m a A. + as a Iii; emu .mcamceuueeu one oueu Henge maemuuem mnouo ecu mcfiesmmm 121 um 85 cm mlulm I eue 3 on ma wumm wIII I IlaulmuAUI: Id n O A IMN I g I mH I QH I IAIUUIm I MN I UH Una amN + 00m INIIomI cum ua mew a Co 0 o .gICImM. IqummIn-WRWISIanIMIum IRIInquuvaMI o “I—uwma So Sm ”a Sm Wm om Sm am a. 3:. I C Imwm I «up .M. an? 0 40.. N + a I I new .xauuee uceaouwmeoe mag ecu no emue>ca ecu 9c mcfihamauHSEeum a ._ ta XHU O 0 AU I HVQIMI 0H0 QHIIMN +AI I OH I IHAU I Hv Mum I O WIIHG nub nu u 0 oh I H uuv : cum ma mum I“? a pa nm a IJNIIISIS IIQIS an H mm elm. 2.. o 8 as mm al.,..- .cofiueuoc xauuee cm EeueHm ceauesge mucu ochmeudxm a . a I 395 o 0 .MI mup cu ou mu u ~ xu cu 0 cum xHU IUImWfi MMNI. .oINIWIoHImHAUIvaIM-IMI + N on an a cum c 3:: muoT an; Immm + .mMmmfl I an I mucv HMI' Im.0I + A .I. want 2 a 2 ... am as am am Sm IILIm. Im.:Im.~.fl-: -..Ie-ial..m meespoum mEueu mmouue @causcuuummo one ocuuauzlem o o HAUIHUWMHquINmIquuIn—ou MINIMI AIInquu ”MI c nu a u a u u a 4| .. ....-Z..- MI. .....i ...... If: ...... i... ...... i T... ...-i.-. ...“... me>flm cmsoucu mcmaamfiuasz 123 Amaav Amficv chad. $22 I mH I " QHU xuU mmm mHU cum + 0 ohm m . XHm mHm IHW " UH 58 mulm + 96 fl .8 3m a .Euom Heuecem ca .9940 one Amacc mcfimmeuaxm GHQ GHQ 0m 0H I mH A0 I. HUNT IIIN + mwm I amml I A 0cm H 9.6 om Mum UI.H@N I ImmIImHImHVIPI UIIImHIUHvUImI OImN mxm m~m xm N u & euecs H may“? I :aImm a I We x UH UHm UHm a" ...... I. II: .2... ...; -_I muaseeu Hecam ecu ee>wm mEueu unannoum one mafiaumHQEmm 124 .mmumu .mnoufiummeoo Ucm mumu cmoH mmmcmfisn may ca mmwcmno ou m>fimcommmucz can mwumu >uauaowm Ho :ofiuucsw m>auwmoa n ma mama uflmommn was .mumu cmoH mmwcamsn mnu mo ceauocsm w>«ummm: m can mwumu .muouwummeoo new mumu aufiunomm man no cofiuocsw m>wufimoa n mg on Umuommxw mum mmumu cmoH “wasmcoo o x U U MSSB .O A ..I..Hl®. Dam O A ..Imlm: ~O V [MW ..O V .IHIW ~m30aufl§fimmfl msdflflOE m2...» EOHM .0>fiuflm0nm am om mm om mum A new & mmaHQEa Essfixme uauoum m uom macauavcoo umUuOIncoomm may no coHuommmaumm xH MH mH is Guam 7% 77:-QIMTfiM xu xu u m u o m 8?: WM m xuwmm Tan «Mm AM|%+ ohm x «mm umnu ucmummmm ma ufi APPENDIX B IF..- 125 Inna... 1 d .8320 38 5:8; 3&3 25.3.25 58.9513 52.2.5.3 H20 8 8820 033 «H d .8820 3: 835 32... 53.52 25.25; 08.81~ H5 8 and 153.52 8:205 S d .8320 3.2 58.3.. 33: 2525.2 8185.2 $5.35.: 50 8 30:33 2355280 3 AH .8820 «3235:: EU gm 53 a 8:33 52:3.“ :8 a 22 .oaouusm 3:... 583.2... .8: «2.35.0 20.68.02 53.2.5.8 82 8 2:8 0:33: 2:3: o a: .250 fie» .52 .8: 25.25... :5: 25.313 25.2.8.3 2563.8. 02 8 3.652 £33082...» 2. 22 .38 fie» so: :58 25.2: :53 2.053 «8.35.2 25.2.”; 02 8 25m 503.5 03:. w .5 £36 58» 392 :5: 85.25.: .32 25.2.5.8 25.25.32 25.53.52 02 8 353.30 a a. .230 58» so: .2: 25.255 :53 2553.2 25.2%.? 2548.3. 02 B .33 ”82.20 c E .230 58» :oz :8 25.33” .0. 2:8 58.33: » 25.2.5.8 0853.8 02 no 500925. 320 m E .208 “now .52 2553.5 02 mam 25m 853m bosom a E .flomfiuoz 2.5.2.5... m 0.2 mum 25m 853m 3:22 2 223 523200 mz000 x00» :02 I o»z .0uoH 6:0 moca>0m I am .xcmn mch>0m #0202: I mm: .xcon H0fi00weeoo I no .muoaaoo «0 000000000 0« «musoqu 004 .wmazumzuo nmuooaoca mamaca 0000 .H0 uwnewooo we 00 000 0002000 .xoonucuz ucwxcum n.a-0zuz beam "Cannon 128 me .0308 .205 000.02 :0: 03.3 30.2... 02.30 28 8 0000 000030 0.. no #003308 33. £0.80 .83 20.02.. 30.3.0.0 30.02.; 28 B 0.000 033.600 3 no 4003038 :00. 02.000 300. 03.2... 02.2.8.0 00068.0 28 8 300 0030002 0030 3 no 4003808 :5 2.0.2.0 32.0. 03.00.... 80.2.0.0 30.0.3.0 28 8 0000035. 003.0 «0 00 4003850 200. 80.3.. 300. 02.80 000.080 08.3.0.0 28 8 0030. 00330 8 so .0308 3:; 000.20 :62 02.82. 80.2.0.0 80.0310 28 6 08 0:00 3 2o 6008.58 .80. 2.3.08.0 :03 80.2.0.0 80.806 0.3.08.0 28 00 0030.322 00 Hz .88. 02.0.3.0 .80 000 300 00030 0030000 00 Hz .0338 .8: 000.02. :00. 02.2.3... 08.80.00 02.80.: 0.00 00 0338 no 300 003302 2.0 mz