A STUDY OF LOAN COST BEHAVIOR EN CONSUMER “NANCE COMPANIES Thank Fm the Degree of Ph. D. MECHIGAN STATE UNWERSITY Eugene Edward Comiskey 3965 LIBRARY Michigan Stats Unchrsity guess Il/I'li/W/‘I‘IT/li/ilf/Ifi/llflli/W/IWWWW/ll ' 3 1293 10605 4129 This is to certify that the thesis entitled A Study of Loan Cost Behavior in Consumer Finance Companies presented by Eugene Edward Comiskey has been accepted towards fulfillment of the requirements for Ph.D. degree in Businegs Administration W can/flame Major Erhfessor Date ____J-191.1$.t_5_._l.9.65A 0-169 fox. ABSTRACT A STUDY OF LOAN COST BEHAVIOR IN CONSUMER FINANCE COMPANIES by Eugene E. Comiskey The principal objective of this dissertation is to contribute to our understanding of the influence of (1) average loan size_and (2) scale of operation on cost behavior in consumer finance companies. Current a literature reveals disparate views concerning the likely impact of loan size and scale of operation upon unit costs. The basic methodology employed in the study is multiple regression and correlation analysis of cost and other operating data from approxi- mately four hundred branches of a major consumer finance company. In order to isolate the individual influences of average loan size and scale of operation upon unit costs, a cost behavior hypothesis has been constructed and tested. The hypothesis was formulated on the basis of a thorough literature review and numerous discussions with.members of the consumer finance industry. Care was exercised in constructing the hypothesis to insure that the cost determinants included were relatively independent of each other. The basic behavioral hypothesis is presented below in regression formulation: * I=A+BX1+CX2+DXB+EX4+FX5+GX6 Eugene E. Comiskey Estimated direct branch operating cost per account P H‘* II I Regression constant 8-6 8 Net regression coefficients X1 = Average loan size = Average number of accounts outstanding = Delinquency measure Factor prices index 3 Loan mix = New application acceptance rate ck“ uf‘ c?‘\J* zd‘ II The behavioral hypothesis has been tested with four cross sections made up of a total of three hundred and eleven mature branch offices. The basic predictive or explanatory power of the hypothesis is reflected in the coefficient of determination (R2) fer each of the cross sections. These coefficients revealed that the hypothesis is effective in explaining approximately two-thirds of the variance in observed unit cost (Y) for three of the four cross sections. The hypothesis proved somewhat less effective with the fourth cross section. In addition to developing the cost behavior estimates, the relation of the findings to prior studies and their possible managerial and regu- latory applications and implications have been considered. The managerial applications discussed include: (1) Cost control of the decentralized operating structure of consumer finance companies. (2) The decision tool of incremental or marginal analysis. (3) Intra-firm resource allocation. Eugene E. Comiskey In addition to the managerial applications, the results of the study are also related to the following regulatory features: (1) Rate structure graduation. (2) Rate structure levels. (3) Restrictive licensing of branches. During the course of the study a number of recommendations for further research have been made. The present study has considered only the issue of "scale economies" in consumer finance branches. A fruitful area for further research would be a consideration of the somewhat broader concept of‘figm scale economies. That is, to what extent are their econ- omies or efficiencies inherent in a‘figm with a large number of branches? Further research along the lines of the present study would be useful in providing corroboration which could enhance the more general applic- ability of the study results. Additional efforts designed to further the predictive or explanatory power of the cost behavior hypothesis developed in this study would also be valuable. These efforts might provide additional or substitute explanatory variables. For example, it is likely that part of the unexplained variance in unit cost (Y) reflects the short-run influence of personnel additions. The addition of a "dummy" variable (Xi = 0 when there have been no personnel addi- tions during the year and Xi = 1 when personnel have been added) to accommodate this condition may improve the overall explanatory power of the original cost behavior hypothesis. A STUDY OF LOAN COST BEHAVIOR IN CONSUMER FINANCE COMPANIES By Eugene Edward Comiskey A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF.PHILOSOPHY Department of Accounting and Financial Administration 1965 —-+---_.. -_ J I ‘ gram ' ACKNOWLEDGMENTS The author wishes to express his appreciation to his doctoral committee: Dr. Charles Lawrence, chairman, and Dr. Floyd W. Windal and Dr. Richard F. Gonzalez, committee members. Each has made a valuable personal contribution to the successful completion of this dissertation. Thanks are also extended to Dr. James Don Edwards, Chairman of the Department of Accounting and Financial Administration, for marshalling the academic, financial, and personal support necessary to sustain and make fruitful the author's educational experience at Michigan State. Gratitude is also expressed to Dr. Robert ML Johnson for first introducing the author to the research possibilities in the field of consumer credit and for providing substantial counsel during the course of this study. Special-thanks are due the company which provided the data for the empirical content of the study; it would have been impossible without their full and complete cooperation. Finally, though words are far from adequate, I express my appre- ciation to my wife, Marlene, and my children, Beth, Paul, Patrick, and Mary, for the many sacrifices which they have made so that I might complete my studies. Also, I express my sincere appreciation for the encouragement and support provided by my parents. 11 TABLE OF CONTENTS ACKNOMIEmmNTS . C O l O I O O O O C O O 0 LIST OF TABLES e e o e e e e e e e e e e 0 LIST OF FIGURES AND EXHIBITS . . . . . . . Chapter I INTRODUCTIO N C I O C O O O O O I 0 Introduction to the Problem. . . . . Statement of the Dissertation Problem Significance of the Study. . Hypothesis . . . . . . . . . . . Methodology. . . . . . . . . . . Plan of the Dissertation . . . . Limitations of the Study . . . . II THE CONTEXT OF THE STUDY . . . . . Introduction . . . . . . . . . Reason for Small Loan Laws . . Economic Rates of Charge . . . Importance of the Industry . . . Inter-Institutional Cost Comparison. Sources of Funds . . . . . . . . . . Earnings Performance . . . . . . III LITERATURE REVIEW. . . . . . . . . Introduction . . . . . . Cost Accounting Studies. . . . . Other Studies. . . . . . . . . . Joel Dean. . . . . . . . . . . . Miller Upton . . . . . . . . . . LoringFa-rwalleeeeeeeee IV EMPIRICAL RESULTS. . . . . . . . . The Data and the Methodology . . General Empirical Results. . . . The Size-Cost Relationship . . The Number of Accounts-Unit Cost Relationship. 811mm. 0 O O O O C C O O I I O I O C O C O 0 iii I O O O . O I C O Q C O O I Page ii 123 Chapter V ANALYSIS OF THE EMPIRICAL RESULTS. . . . . Relation to Other Studies. . . . . . . . Implications and Applications. . . . . . Managerial Implications and Applications Regulatory Implications and Applications VI SUMMARY AND CONCLUSION . . . . . . . . . . Em ImRAPHY O O O O I O I C O C O C I I O O D O 0 APPENDIX C O I I I O C C C U O O O D O O C C D I 0 iv Page 124 12h 136 136 1% 153 160 165 Table 1.1 1.2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 209 2.10 2.11 3.1 3.2 4.1 #.2 “'03 LIST OF TABLES Discounted-CashpFIOW'Rates of Return on Household Investment in an Automatic Washer and a Dryer. . . . . Bank Rates on Short-Term Business Loans . . . . . . . . . . Estimated Required Rates on Loans of Different Sizes. . . . Uses of Funds by Consumer Finance Companies, End of 1959. . Distribution of the Number of Loans Extended by Purpose, 1963 O C O O O O O O O O I O O O O O O O O O O O O O 0 Percentage Distribution of the Mbnthly Income of Borrowers from Consumer Finance Companies, Fbr 1963. . . . . . . Total Consumer Instalment Credit Outstanding, 19b9, 1959, and 1964 O O O O O O O C O O C O O O O O O O O O O O 0 Personal Instalment Cash Loans Outstanding, 1949, 1959, and 1964 O O O C O O O O O I O O O O O O O O C O O O 0 Components of Gross Finance Charges on Consumer Credit, byTypeOfLender,l959eeeeeeeeoeeeeeee Average Size of Contracts Acquired and Outstanding, 1959. . Percentage Distribution of Liabilities, Capital and Surplus for Finance Companies Reporting to NCFA Questionnaires Earnings as a Per Cent of Assets Used and Useful. . . . . . Net Earnings of Michigan Consumer Finance Companies, 1960-1963eeeeoeeoeeeeeeeeeeeeeee Loan Profitability by Size. . . . . . . . . . . . . . . . . Loan Profitability by'Maturity. . . . . . . . . . . . . . . Statistical Summary of General Empirical Findings . . . . . Increase in Unit Cost (I) For One Dollar Increase in AverageLoanSize(xl)eeeeeeeeeeeeeeee Statistical Results of the Measurement of the Effect of Average loan Size upon the Cost of Risk Assumption . . Page 37 39 #0 42 43 49 52 55 73 73 102 106 112 lllf‘llll‘l’ll (It; Table 5.1 5.2 6.1 Page Classification of Loans Made by Security. . . . . . . . . . 127 Decrease in Charge-Off Rate per $100 Increase in Average 10am Size. 0 O O O O l C C O O O D O O C O O O O I O C 129 Increase in Unit Cost (I) For $100 Increase in Average Loan Size(xl)eeeeeoeeeeeeeeeeeeleeee Decrease in Charge-Off Rate Per $100 Increase in Average Ipan Size 0 I 0 O O O O C O O I l O O U I O O O C O O O 157 Figure 1.1 1.2 1.3 1.4 1.5 3.1 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 LIST OF FIGURES AND EXHIBIT Percentage Relationship Between Loan Size Loan Costs. 0 O O O O C O O O I O O O O 0 Size and Rate of Charge Relationship Two Break Even Point Case . . . . . . . . Scale Economies Case. . . . . . . . . . . No Scale Economies Case . . . . . . . . . Number of Accounts--Unit Cost Relationship Loan Loan Loan Loan Loan Loan Loan Loan JoelmanStUdyeeeeeeeeeee S and Size and Average Unit Cost Relationship cmssseCtionAeeeeeeeeee Size and Average Unit Cost Relationship crOSSSGCtionBeeeeeeeeee Size and Average Unit Cost Relationship CNSSSBCtionCeeeeeeeeee Size and Average Unit Cost Relationship crOSSSGCtiOnDeeeeeeeeee Size and Charge-off Rate Relationship crOSSSGCtiOnAeeeeeeeeee Size and Charge-off Rate Relationship CI‘OSSSeCtiOnBeeeeeeeeee Size and Charge-off Rate Relationship cmssseCtiOnCoeeeeeeeee Size and Charge-off Rate'Relationship crOSSSeCtionDeeeeeeeeee Number of Accounts and Average Unit Cost Relationship crOSSSeCtiOnAeeeeeeeeeeeeeeee Number‘of Accounts and Average Unit Cost Relationship CrOSSSOCtionBeeeeeeeeeeeeeeee vii Page 11 12 19 2O 21 79 107 108 109 110 113 114 115 116 119 120 Figure Page 4.11 Number of Accounts and Average Unit Cost Relationship Cross seetion C O O O O O O O O C C C O C O O O O O O 121 4.12 Number of Accounts and Average Unit Cost Relationship Cross Section D . . . . . . . . . . . . . . . . . . . 122 Exhibits Page 2.1 Composite Statement of Income and EXpense . . . . . . . . . 34 3.1 Determination of Acquisition and Servicing Costs. . . . . . 79 viii CHAPTER I INTRODUCTION ‘Introduction to the problem In a free enterprise economy resources are allocated among alter- native employments through the operation of the price system. Prices are determined through the interplay of supply and demand. Our economy is basically free enterprise and prices are established largely on the basis of demand-supply interplay. Some prices are, however, established in whole or in part by governmental fiat. This study is concerned with one such price--the price of personal instalment cash loans from con- sumer finance companies. Qppgsition to interest takigg Since the earliest times the price for a loan of money has been subject to a variety of restrictions and prohibitions. Some of the earliest recorded opposition took the form of prohibitions against the charging of £52 price 33 all for a loan of money: If you lend money to one of your neighbors among my people, you shall not act like an extortioner toward him by demanding interest from him. Exodus 22:24 Father Thomas F. Divine, in his excellent study on interest, notes that the early opposition to interest-taking was generally not based up- on an analysis of interest itself. Rather, it was founded upon one's obligation to be charitable and upon the particular circumstances and exigencies of the times: To lend gratuitously was looked upon as an obligation of fraternal charity, a duty of mercy and generosity which the rich owed to the poor. . . . . .interest was viewed only as a problem of social justice. The problem of commuta- tive justice, i.e., of equivalence of value in exchange of present for future goods, remains quite untouched. An additional argument against allowing a price to be charged for a loan was based upon the concept that money is barren. The "barreness thesis" was supported by a formidable scholar in the person of Aristotle. Eugen V. Bohm-Bawert, in his classic work Capital and Interest, summa- 2 rized the "barreness thesis" as expounded by Aristotle: What this positively amounts to may be summed up thus: money is by nature incapable of bearing fruit; the lender's gain therefore cannot come from the peculiar power of the money; it can only come from a defrauding of the borrower. . . Interest is therefore a gain got by abuse and injustice. In addition to the nature of the above arguments, it is important to consider the manner in which the lending environment was typically characterized. The following are two frequently mentioned character- istics:3 1Thomas F. Divine, Interest--An Historical and Analytical Study in Economics and Modern Ethics Milwaukee: The Marquette University Press, 1959 , PP- 9-10. 2Eugen V. Bohm-Bawert, ital and Interest, Translated with a preface and analysis by William Smart {London and New Ybrk: Macmillan and 00., 1890), p. 17. 3Divine, op.cit., p. 65. (1) Isolated exchange between individuals, and (2) Lending for consumption purposes to a needy borrower. In such a context it was argued that the opportunity for exploitation of the borrower was so great that prohibitions against charging any price on a loan of money were required. Mitigation of opposition Time, more searching analysis, and the force of economic circumstance mitigated opposition to charging a price for a loan of money. With the emergence of an impersonal competitive capital market, the argument based upon the exploitation of the borrower in isolated exchange lost some of its force. Also, the prominence of loans fer commercial purposes weak- ened the arguments that were promised with the "barreness thesis." Calvin made the point in the fellowing words:u In the face of the new conditions and circumstances of economic life, wherein both lender and borrower gain by a loan at interest‘lzt a pricé7, it is not a violation. . .to charge for the use of money. Since the borrower of funds for commercial ventures could be expected to employ (or attempt to employ) them productively, it is clearly not cor- rect to call money "barren.” Even in the case of a loan for consumption purposes, the barreness thesis is seen to be unsound. The borrower is able, with the loan, to accelerate his consumption. Since by seeking a loan the borrower exhibits a preference for present as opposed to future consumption, it can be said that he reaps a gain or derives util- ity from the present consumption which is facilitated by the loan. There» “Divine, op,cit., p. 87. fore, even the consumption loan is "productive" in the sense of utility creation. A very pertinent study, recently reported upon in The Journal of Finance, developed estimates of the rates of return on consumer dur- ables.5 Personal instalment cash loans often facilitate, either directly or indirectly, the purchase of consumer durables. The rates of return calculated in the study were developed with the "discounted cash flow" technique. In this procedure, the rate of return (r) represents the interest factor which, when used as a discounting rate, will equate the outlay required to acquire the asset with the net cash benefits which it generates. The estimated rates of return on an automatic washer and dryer ranged up to 27.1 per cent depending upon estimates of useful life, rate of usage, and other factors. A sample of the results of this study is presented in Table 1.1. On the basis of results such as those presented in Table 1.1, the authors of the study concluded that:6 The high rates of return on physical investment also mean that a rational basis exists for con- sumer credit. Individuals can increase their real income by borrowing, just as businesses can increase their profits through leverage. As long as the debt can be easily handled, the additional real income produced by borrowing to buy the fixed asset‘ISr durable goqd7 should make its purchase easier. SJ. V. Poapst and W} R. Waters, "Individual Investment: Canadian Experience," The Journal of Finance, XVIII (December, 1963), pp. 647- 665, and J. V. Poapst and W; R. waters, "Rates of Return on Consumer Durables," XIX (December, 1964), pp. 673-677. 6J. V. Poapst and.kh Rm haters, "Individual Investment: Canadian Experience," p. 649. TABLE 1.1 DISCOUNTED—CASH-FLOW RATES OF RETURN ON HOUSEHOLD INVESTMENT IN AN AUTOMATIC WASHER AND A DRYER (Per Cent) Useful Life Loads Per week 6 8 10 6 4.9 14.0 22.5 8 7.7 17.0 25.4 10 10.4 19.2 27.1 Source: J. F. Poapst and W} R. waters, "Rates of Return on Con- sumer Durables," The Journal of Finance, December, 1964, p. 674. In the case of borrowing to acquire assets for which an explicit rate of return could be developed, based upon differential cash flows, income would be increased if the rate of return exceeded the cost of borrowing. If we define the rate of return on the asset or durable as (r) and the cost of the borrowing as (i), then income is increased where (r) >.(i). Borrowing becomes an economically rational choice which is consistent with consumer maximization theory. Where the application of an explicit rate of return analysis built on differential cash flows is not feasible, a decision as to the economic soundness of borrowing to facilitate an expenditure will depend upon the relationship between the borrower's rate of time preference and the cost of borrowing. Unlike a rate of return built upon differential cash flows, the rate of time preference does not lend itself to explicit numerical formulation. Nevertheless, if we allow (t) to represent some subjective . A w-T i measure of the consumer's rate of time preference, then borrowing be- comes a maximizing decision (economically rational) and the borrower's real income is increased when (t) :>(i), where (i) represents the cost of borrowing. In the light of the above, it is not surprising that for all practical purposes opposition to charging a price for a loan of money has disappeared. Indeed, since the very earliest times money‘hgg been loaned and a price charged for its use, in spite of the opposition cited on the foregoing pages. The view remains, however, that while it is just that a price be charged for a loan of money, the allowable price ought nonetheless be limited. Price should still be limited The fundamental position that the allowable price on a loan should be restricted is reflected in the general interest rate limitations in force in almost all states. These general interest rate limitations are usually referred to as usury laws. Their contemporary justification is based mainly upon the imperfect character of the market fer certain types of loans.7 In a recent study entitledwgmerican's Small Loan Problem, W. H. Simpson stated that "in certain spheres within the economic struc- ture competition works so imperfectly that it cannot be depended upon 7For a discussion of imperfect competition in the market fer con- sumer credit, see T. O. Yntema, "The Market for Consumer Credit: A Case in 'Imperfect Competition,'" Annals of the American Academ of Political and Social Science, March, 193 , pp. 79- 5 and C. W. Phelps, "Monopolistic and Imperfect Competition in Consumer Loans," The Journal of Marketi , April , 1944. \ ME .al k d lite Ho 1 1% to regulate prices in the interest of consumers or of society in general.” Simpson goes on to characterize the market for small loans as being an example of imperfect competition and a market in which "public regulation must be resorted to to prevent the extortion of the public by unwarranted 9 rates.” Unlike Simpson, in that he does not stand in the advocates role, Upton has offered the following rationale for price restrictions on loans:10 From an economic standpoint the justification of usury laws is based upon the presumption that competition is ineffective in establishing a fair price for credit. Therefore it is presumably neces- sary for the state's police power to be envoked to protect its economically weaker members. . . The uninformed or necessitous borrower, in other words, must be protected from unscrupulous lenders who would take advantage of their preferential bar- gaining position. . . These views stand in striking contrast to those of which reject 11 laws which restrict or fix the price of money: There is no reason for a law which fixes the rate of interest. The rate ought to be, as is the price of all commodities, fixed by the agreement between the two contracting parties and by the relation of supply and demand. 8William Hays Simpson, America's Small Loan Problem - With Special Reference to the South (published by the author in cooperation with the University of South Carolina, 1963), p. 113. 9Ibid., p. 115. 10Miller Upton, "The Economics of Fair Charges For Consumer Loans," Missouri Law Review, June, 1951, p. 275. 11Turgot, Memoirs sur les Prets d'Ar ent, 536. (This section trans- lated by and quoted from F. W. Ryan, Usurz and Usury Laws (New York: Houghton Mifflin Co., 1924), p. 50. The general interest rate limitations have not proved to be a prob- lem in larger business loans since the market(competitively established) rates have typically held below the general interest rate maximums, which range from 6 to 12 per cent per annum. Below is a schedule of sample bank interest rates on short-term business loans, in the range of from $1000 to $200,000 for the period 1956 to 1964, which illustrates this point. TABLE 1.2 BANK RATES ON SHORT-TERM BUSINESS LOANS (Per cent per annum) Size of loan Area (thousands of dollars) and All H period loans 1-10 10-100 100-200 200 and over Year: 19 large cities: 1956 4.2 5.2 4.8 4.4 4.0 1957 4.6 5.5 5.1 4.8 4.5 1958 4.3 5.5 5.0 4.6 4.1 1959 5.0 5.8 5.5 5.2 4.9 1960 5.2 6.0 5.7 5.4 5.0 1961 5.0 5.9 5.5 5.2 4.8 1962 5.0 5.9 5.5 5.2 4.8 1963 5.0 599 5.5 5.2 4.8 196“ 5.0 5.9 5.6 5.3 4.8 Source: Federal Reserve Bulletin, February, 1965, p. 285. Character of price limitations In the area of smaller personal instalment cash loans to individuals, the general interest rate maximums have presented a problem. Experience has shown that the rates of from 6 to 12 per cent per annum are inadequate to attract capital into the personal loan field. Consequently, it has been necessary to provide an exemption from the general interest rate maximum in order to attract legitimate capital into the industry. Fbr any particular state the exemption has typically taken the form of a schedule of rates of charge which are greater than the general interest rate limitation. That the small loan industry has not been granted a complete exemp- tion from interest rate limitations reveals the persistence of the view that the borrower needs the protection of the state in order to prevent his exploitation. The traditional view has been that protection, in the form of price restrictions, is necessary in order to prevent moral usury. Ryan, in his excellent work on Usggy and Usugx Laws, has described moral 12 usury as follows: Mbral Usugy The taking advantage of the ignorance or necessitous condition of the needy borrower so as to get him into a hard bargain and exact from him unduly high charges. Cost based limitations Arguments for price restrictions designed to prevent ”unduly high charges" imply that the lender recoups something more than his costs plus some reasonable profit. It follows, then, that a rational approach to establishing the schedule of maximum charges would be to determine the costs incurred by the lender in providing loans. The rates would be designed to allow the lender to recover his costs plus a reasonable profit. 12F. W. Ryan, Usugy and Usury Laws (New York: Houghton Mifflin Co., 1924), p. xii. Efforts to apply the above concept prove difficult for two impor- tant reasons: (1) While one can readily speak of the costs of‘g loan from a conceptual viewpoint, there are significant practical obstacles to the development of such cost information. (2) When loan cost infer- mation has been developed it has generally revealed that the rates of charge required to cover the costs on smaller loans have been extremely high, by any traditional standards. Early in the development of small loan legislation Dr. M. R. Neifeld, noted consumer credit scholar, observed that the foregoing point prevented a strictly cost based ap- proach to establishing small loan rate schedules:13 It has been shown again and again that strictly scientific allocation of costs to make each loan stand on its feet is impossible and that some form of compromise must be included in whatever rate is determined upon from the point of view of the public policy of the state. Graduated versus siggle-rate schedule In the initial draft of the uniform Small Loan Law the recommended rate was a flat 3% per cent per month on the outstanding balance of loans of $300 and less. The Sixth Draft of the Uniform Small Loan Law broke from the flat rate tradition and recommended a two step rate which was graduated on the basis of loan size. The basic schedule recommended was 3% per cent per month on that part of any loan balance not exceeding $100 14 and 2% per cent per month on that part exceeding $100. One of the 13M. R. Neifeld, Ihe Personal Finance Business (New York: Harper and Brothers, 1933), p. 299. 14 Sixth Draft of the Uniform Small Loan Law, January 1, 1935, note 14. Jm --“~ {xi-WK h ll important reasons given for the introduction of graduation in the rate sched- 1 ule was based upon the cost differences among loans of different sizes: . . .the interest burden in dollars upon borrowers of large sums is high. While it is not considered socially desirable that very small loans should bear their full share of operating costs, it has seemed proper to provide a less inequitable dis- tribution of these costs than was possible under a flat maximum rate. The essence of the logic underlying the substitution of a graduated for a flat rate schedule is that lending costs are non-proportional with respect to loan size. That is, the dollar costs of a given loan do not increase proportionately with increases in loan size. Therefore, given the same maturity, the cost of a $1000 loan would be less percentagewise than the cost of a $100 loan. This view of the behavior of lending costs with respect to loan size is portrayed graphically in Figure 1.1 below. Figure 1.1 PERCENTAGE RELATIONSHIP BETWEEN LOAN SIZE AND LENDING COSTS Lending Costs As A Per Cent 0f Loan Size Loan Size *~'*“-’H_— 12 The slope of this line is determined by the behavior of lending costs with changes in loan size. In the present context, lending gggtg is used in a broad sense and would include all costs of the lending operation including the cost of money and of risk. In order to avoid discriminating against larger loans it is there- fore necessary that the rate structure be graduated downward with loan size as illustrated in Figure 1.2 below. Figure 1.2 LOAN SIZE AND RATE OF CHARGE RELATIONSHIP Rate Of Charge Loan Size If the degree of graduation in the above schedule were based wholly upon the behavior of lending costs with respect to loan size, it would be determined by the loan cost behavior schedule in Figure 1.1. While almost all states have established small loan rate structures which are graduated on the basis of loan size, there is almost a complete absence of actual studies of how lending costs vary with loan size in the small loan industry. ms" _ 13 Statement of the dissertation problem The preceding section reviewed the development of regulation in the area of loan pricing. It concluded by observing that small loan rate structures are graduated on the basis of loan size and that the essential logic underlying such graduation is the non-proportionality of costs with respect to loan size. While the concept of non-proportion- ality of costs with respect to loan size is widely accepted, there have been almost no studies which attempt to fbrmulate an explicit quantita- tive measurement of the size-cost relationship. This study shall attempt such a measurement. The problem The basic dissertation problem can be stated as follows: To deter— mine and to measure the behavioral relationship between loan size and lending costs in the Consumer Finance industry. For the purposes of this study, lending costs will be broken down into three categories: (1) Qperatipg costs-~a11 costs incurred in carrying on the small loan business exclusive of the costs related to risk assumption and forbearance. (2) Risk assumption costs--the rate or dollar amount of net charge offs. (3) Money costs--a proxy measure of the firm's average cost of capital. The principal analytical and empirical problems will center around the effort to determine the effect of loan size upon the operating and risk assumption cost elements. On p_ppippi grounds it is clear that the money cost associated with loans of varying sizes will be proportional to their size. Hence, it can be seen that the nonpproportionality of l4 lending costs with loan size springs from the behavioral relationship between loan size and the operating and risk assumption cost elements. Significance of the study Absence of comparable studies In their study for the National Bureau of Economic Research, Personal Finance Companies and Their Credit Practices, Ralph Young and Associates observed that information on cost behavior of loans by size and maturity was surprisingly not available: In view of the controversy [basing the level and graduation in the rate structure on the costs of servicing loans of differing sizes and maturitiep7 it is surprising that there is no adequate information available on the actual costs of making loans of various sizes and duration. The few cost accounting studies that have been made are not sufficiently de- tailed fer the purpose (emphasis added). The writer's research indicates that the observation that "there is no adequate information available on the actual costs of making loans of various sizes and duration" is still valid. Unlike the author's of the above statement, the present writer is not surprised that such informa- tion has been unavailable. The common or joint character of most of the principal cost inputs in a small loan operation make the development of information on ”the actual costs of‘making loans of various sizes and duration" extremely difficult. Managerial and regplatopy value From a purely academic viewpoint, it could be said that the present l6Ralph Young and Associates, Personal Einance Companies and Their Credit Practices (New York: National Bureau of Economic Research, 1940), w p. e 15 study has significance since it involves the application of analytical tools to resolve a problem which has no apparent p prio_1_'1 solution. In addition to academic merit, however, the study should be of practical importance. The results of the study should be applicable in the general areas of (1) internal policy formulation and (2) public policy formation and regulation of the industry. An understanding of a manufacturing or retailing firm's cost be- havior is essential to the formulation of rational (wealth maximizing) managerial policy. This is no less true for consumer finance companies. The fact that they do not manufacture or retail a physical product, in the conventional sense, does not reduce the need for an understanding of the firm's cost behavior. In a sense the Consumer Finance industry does deal in a product; it acquires money on a wholesale basis and then re- tails it in packages of differing size and maturity. A knowledge of the effect of the size of the package on its cost is just as important for the small loan as for any other manufacturing or retailing activity. The fact that the price at which money can be retailed by consumer finance companies is regulated might cause one to conclude that a study of cost behavior is therefore of little value for consumer finance com- panies. Dr. M. R. Neifeld, noted authority on consumer finance companies, anticipated this criticism in one of his works which included a basic treatment of cost analysis in lending activities:17 Offhand, there would appear to be small possi. bility of benefit in studying costs of individual 17M. R. Neifeld, Neifeld's Manual on Consumer Credit (Easton Pen- nyslvania: Mack Publishing Company, 19615, p. 544. 16 loan transactions because interest and finance charges are set by law. However, while the price of credit cannot be altered by the grantor, cost analysis can direct efforts to the more truly profitable segments. Some particular pipp or type of loan may be more truly profitable than others (emphasis added). The prospective institution of price (rate) discrimination by con- sumer finance companies also highlights the significance of this study of cost behavior. In recent years spokesman for the industry have indi- cated that a near term development in the competitive behavior of con- sumer finance companies would be an attempt to tailor rates in accordance with differences in the costs of servicing different borrowers. Principal attention has been directed at rate discrimination based upon the lender's assessment of the difference in‘pipk associated with loans to different borrowers. However, it is increasingly clear that an additional basis for discrimination will be the behavioral relationship between lending costs and loan size. This is becoming more important with the increasing maximum.loan sizes which have been authorized for the industry in a num- ber of states. In regard to the above issue of price discrimination by loan size, a recent study reported upon in The Journal of Finance is particularly illuminating.18 This study was restricted to commercial bank lending and it employed statistical procedures to measure the marginal cost of lending and the marginal cost of risk‘pyllppp‘pipg. The study revealed that "differences in the marginal rate of interest charged by commercial 18George J. Benston, "Commercial Bank Price Discrimination Against Small Loans," The Journal of'Finance, XIX (December, 1964), pp. 631-643. _ u' gal—i ~11 1? banks on large and small loans can be explained largely by differences in the marginal cost of lending and in the marginal cost of risk."19 This study suggests that commercial bank rate discrimination by loan ‘pipg is based upon a knowledge of the cost differences involved. If consumer finance companies are to emulate commercial banks and adopt the competitive practice of price discrimination, it is apparent that they will need to improve their understanding of the behavioral rela- tionship between both lending and risk costs 22g loan size. More adequate information on loan cost behavior is also of impor- tance in the area of public regulation of the Consumer Finance industry.20 One of the principal ingredients in the current mix of public regulation of the industry is the rate structure. Its level and graduation should reflect, at least in large part, the underlying costs of servicing loans of different sizes. As explained in the preceding section, rate struc- tures‘ppg presently graduated on the basis of loan size. Such graduation is presumably based upon some concept of the size-cost relationship. However, the acknowledged absence of information on this particular relationship makes the degree of rate structure graduation difficult to evaluate. To defend the prevailing rate structure of a particular state because the industry is able to earn a satisfactory return has little appeal. At the very least, it ignores the issue of inter-borrower equity. 19Ibid., p. 642. 20This statement is not to be taken as an endorsement of the basic concept of public regulation of the Consumer Finance industry. If the industry is to be regulated and the rate structure is to be part of that regulation, then it should be as carefully designed as possible. 18 A departure from an equitable degree of rate structure graduation is suggested by statements which assert that in some states a loan may become too large to be profitable. The comprehensive National Consumer Finance Association study of the Consumer Finance industry contained such a statement:21 Moreover, under some state laws, the rate per- mitted on larger loans is too low for such larger loans to be profitable. The above statement suggests the possibility of two break even points in terms of loan size. Such a situation would require a particular com- bination of (1) rate structure graduation and (2) cost behavior with re- spect to loan size. This hypothetical situation can be portrayed graphi- cally in Figure 1.3. Scale economies dispute All of the foregoing considerations combine to establish the impor- tance and potential fruitfulness of the proposed inquiry into loan cost behavior in consumer finance companies. In addition to shedding some light on the cost-size relationship, the present study, as will be re- vealed in the section on methodology, will necessarily deal with the effect of scale of operations on loan costs. This relationship has frequently been used to justify a restrictive licensing practice in the Consumer Finance industry. It has been argued that economies of scale inherent in the operation of a small loan operation call for restrictions in the number of units in a particular community. This is necessary to ZlThe Consumer Finance Indust , a monograph prepared for the cam. mission on Money and Credit by the National Consumer Finance Associa- tion, (Englewood Cliffs: N. J., 1962), pp. 58-59. «ream-h ' - ,..._,« V ‘ .. .. q , - ‘.’ 4 . I I .’ I . . . A 19 Figure 1.3 TWO BREAK EVEN POINT CASE ,-$ ._ Revenue Cost Loss * Profit : I RevenueN I l Costs 8 a +Ib(5) I I l Loss : I t ' I : l I : I a O O ' ' I I I I I I : Per-loan costs : I I I I O | Loan Size ! q, \i/ lst Break Even 2nd Break Even a = Loan costs unaffected by loan size b Coefficient of change in costs given change in loan size S Loan size ** The linear cost relation is only assumed for purposes of illustration and is not suggested as being indicative of actual cost behavior with respect to loan size. 20 insure sufficient business so that each can achieve the most efficient or optimal scale of operation. Standing in opposition to the position that the industry is subject to economies of scale is the view that it is one of relatively constant unit costs (the absence of economies of scale): . . .the average size of the loan office ‘Ppp is only one—fifth to one-third the optimum size of office which would operate with minimum unit cost. There is, however, little saving that comes Con from having larger offices because the in- ._- dustry.[small loap7 is basically one of more or less constant costs. Graphically, these apparently opposing views would appear as follows: Figure 1.4 $ SCALE ECONOMIES Unit Cost Long Run Scale 22 To 00 Yntema, OE:Cito, P. 84. 23Carl A. Dauten, The Consumer Finance Industpy in a Qypppic Econ- ppy, Consumer Credit Monograph Nb. 2 St. Louis: American Investment Company of Illinois, 1959), p. 73. 21 Figure 1.5 $ NO SCALE ECONOMIES Unit Cost Long Run Scale This study should make possible some assessment of the opposing views of cost behavior with respect to scale of operation in the in- dustry. Hypothesis The fermulation of a testable hypothesis concerning the various determinants of lending costs in the industry is essential if the in- fluence of one particular determinant is to be isolated and.quantified. The formulation of such a hypothesis requires an intensive study of the character of the particular business being studied. A review of the nature of the consumer finance industry and its operating character- istics has been carried out by the writer. It forms the basis for the development of the cost behavior hypothesis needed to implement the measurement of the "size-cost" relationship. 22 The basic approach to developing a testable cost behavior hypothesis is to determine those factors (variables) which would appear to explain cost differences among different operating units. A review of the avail- able literature, discussions with members of the industry, and an examina- tion of selected annual reports of industry members revealed a somewhat mixed view regarding the effect of loan size upon operatipg and‘pipk assumption costs. The operating cost category covers cost incurred mainly in carrying on activities such as investigation, receiving and recording of payments, collection, and various other activities. The views un- covered by the writer include: (1) Loan size does not affect the investi- gative, bookkeeping, and collection activities and hence does not affect operating cost levels per se. (2) Loan size increases somewhat the fore- going activities and hence operating costs. (3) Loan size decreases operating costs somewhat by reducing the investigative and collection activities. A sample of these views from the available literature is provided below: Effect Of Loan Size On eratin Costs No effect As the accompanying tabulation shows, the average loan has increased 76 per cent over the past five years. Such loans yield a better net return be- cause they can be serviced at the same unit cost as smaller loans, produce a larger gross income per account, and generally have a lower delinquency and loss rate.2 ' 24 Thrift Investment Corporation, Annual Repprt 1264, p. 5. 'I" t.‘ 23 Some increase It costs about pp much to investigate a credit applicant if he wishes to borrow $50 as if he wants to borrow $1000. The expenses of trying to collect past-due accounts are not affected much by the size of the account. The costs of investigation, bookkeeping and collection are almost ppp same on a loan of $100 as one of $500.26 Some decrease Many costs tend not to vary with the size of loans; consequently, costs as a per- centage of loan value decrease with loan size. . . In addition, the absolute size of the costs of investigating credit worthi- ness may be greater £25 small borrowers (less for larger borrowers). . . Since the principal objective of the study is to measure the size- cost relationship, loan size must necessarily be included in the hypothesis. The import of the above is to suggest that the relationship between a unit cost measure, such as average cost per account, and loan size may be mod- erately positive, negative, or equal to zero. Factor price differences As with most empirical cost studies, one must determine whether or not differences in factor prices (personnel, rent, etc.) influence the 25Carl A. Deuten, Financi The American Consumer, Consumer Credit Monograph No. 1 (St. Louis: American Investment Company of Illinois, 1956): P0 3’40 26Family Finance Corporation, Annual Repprt 1264, p. 9. 27L. Laudadio, "Size of Bank, Size of Borrower, and the Rate of Interest," The Journal of Finance, xVII (March, 1963), p. 20. This particular statement is made with reference to bank lending. However, it could just as easily have been made with respect to lending by con- sumer finance companies. 24 level of costs among the offices included in the study. Consequently, a measure of differences in factor price levels must be incorporated into the cost behavior hypothesis. Delinquency The control of delinquency entails considerable effort in small loan operations. To the extent that the level of delinquency varies among dif- ferent branches, there will be either more or less effort expended in delinquency control and collection activities. Added effort entails addi- tional costs. Thus, some measure or index of delinquency level must also be included in the cost behavior hypothesis. Loan mix Borrowers are commonly classified by consumer finance companies according to whether they are (l) a new borrower, (2) a former borrower, or (3) a present borrower who is being extended additional money. It is clear that ". . .it cannot be assumed that the cost of investigation is as great for present and former borrowers as it is for new borrowers; and also other items of cost, particularly risk, may be expected to vary among old, present and new customers."28 Members of the industry have stated to the writer that "two to three times the effort is entailed in making a loan to a new versus a present borrower." Therefore, a measure of "loan mix" should be an important index to include in the cost be- havior hypothesis. An operating unit doing a greater than average 28Ralph Young and Associates, Personal Finance Companies and Their Credit Practices (New York: National Bureau of Economic Research, 1940), p. 132. 25 percentage of its lending to new borrowers should, other things equal, have higher unit costs. Application acceptance rate The percentage of loan applications from new borrowers which a particular office accepts is a function of a number of factors. For ex- ample, on theoretical grounds it can be said that the level of the rate structure in a particular state will influence the percentage of new applications accepted. That is, if in state A the general level of the rate structure is higher than in state B, then one would expect that applications would be accepted in State A which would be rejected in State B.29 Within a particular state the rate of acceptance of applications will be strongly influenced by the particular location of an office. An office located in a high middle income suburb will generally have a higher acceptance rate than an office located in a lower income section of a city. In addition, the overall acceptance rate will be influenced by the‘pppppg of new applications. Loan applicants who are referred to a consumer finance office by retailers generally are granted loans much less fre- quently than applicants from other sources. Thus, an office with a high portion of retail referrals among its loan applicants will, other things equal, show a lower rate of application acceptance. 29An effbrt to provide empirical documentation on refutation of this view is presently underway by Maurice Goudzwaard, a Michigan State doctoral candidate. 26 In terms of developing the cost behavior hypothesis, it appears that variations in the rate of acceptance of new applications may have an influence upon the level of unit cost among consumer finance branches. Hence, a measure of the rate of acceptance of new applications becomes part of the cost behavior hypothesis. Number of accounts On page 20, it was observed that there is some difference of opinion regarding the influence upon "long-run" unit costs of differences in the number of accounts handled by an office. The question resolved itself into two basic positions: (1) Scale economies are slight or nonexistent-- a constant unit cost position. (2) Scale economies are strong-- a declin- ing long run unit cost position. An early study by Joel Dean, which will be reviewed in some detail in chapter three, concluded that there were scale economies which derived basically from indivisibilities in the human factor input. However, Dean's study included in the loan office sample offices which had been in operation for one year and longer. It is clear that by today's standards one year is not sufficient time for an office to "mature" and Dean's study, which revealed scale economies, cannot be relied upon since part of the declining unit costs must simply be attri- buted to a fuller utilization of the short-run capacity. In view of this controversy, it is clear that the‘pppppp.p£ accounts should be included in the cost behavior hypothesis. The hypothesis Based upon the above discussion, the explanatory cost behavior hypothe- sis becomes the following: v v; - -r-'I-_ "‘1 27 Unit Cost = F(Average Loan Size, Average Number of Accounts, Loan Mix, Application Acceptance Rate, Factor Price Level, and Delinquency) This writer's views on the probable direction of the relationship between unit costs and the various explanatory variables are outlined below: Positive Average loan size Loan mix Factor price level Delinquency Negative Average number of accounts Application acceptance rate Methodology The basic methodological procedure employed in this study flows from the statement of the dissertation problem and the dissertation cost be- havior hypothesis. The basic objective of measuring the influence of loan size upon operating costs will be accomplished with the above stated hypothesis and the statistical technique of multiple regression and cor- relation analysis. The data for the study are derived from the cost and other operating records of a major consumer finance company for the year 1964. All mature offices of the company located in feur states are included in the study to make up fbur different cross sections. Upon investigating the character of the available cost data, it was discovered that certain costs were allocated to the branch offices on the basis of a combination of (1) nmmber and (2) dollar amount of accounts outstanding. Therefore, in order to avoid possible distorting effects due " I ' ‘ J II I A ,It.m;; ‘r‘fl, 28 to these accounting allocations, adjustments were made to exclude such cost elements. The basic methodology and the character of the dependent and inde- pendent variables will be further discussed in chapter four. In statisti- cal formulation, the basic operating pppp hypothesis can be represented by the following regression equation: * Y = A + BXl + CX2 + qu + EX“ + FX5 + 0X6 * Y Estimated direct branch operating costs per account A Regression constant B—G = Net regression coefficients X1 = Average loan size outstanding X2 = Average number of accounts outstanding X3 = Delinquency measure X“ = Factor price measure X5 = Loan mix X6 = Application acceptance rate An estimate of the influence of loan size upon risk assumption cost, as measured by the rate of loans charged off, will be developed from a simple regression of average loan size upon the charge off rate. * Z 8 H + 1X1 III Z = Estimated charge off rate H = Regression constant Xi = Average loan size outstanding = Regression coefficient--average loan size 29 Plan of the Dissertation This chapter has included an introduction to the dissertation prob- lem, a statement of the dissertation problem, its significance, the hypothesis and a brief statement of the methodology. Chapter two will present a overview of the Consumer Finance industry which represents the context of the study. This chapter will include brief discussions of the reasons for small loan laws, the importance of the industry, an inter- institutional cost comparison, the industry's asset composition, fund sources and earnings performance. Chapter three will present a critical review of the more important related studies. This will provide an op— portunity to explain further the choice of methodology and to point up the contributions which have been made in the past and how the present study is designed to further and to build upon prior studies. Chapter four will present the results of the empirical cost study. This chapter will include additional identification of the independent and dependent variables employed in the cost hypotheses. Chapter five will draw upon the empirical results of chapter four and consider their relation to previous studies plus their managerial and/or regulatory implications. Chapter six presents a summary and conclusion. Limitations of the Stugy It must be emphasized at the outset that the character of the data employed in this study places certain limits upon the applicability of the findings. As noted, the data source is one large consumer finance company. The data source is therefore not a random sample and cannot be viewed as strictly representative of the industry in general, except to 30 the extent that the branches of company under study may be typical of the industry as a whole. However, the company's size necessarily sets it apart from many industry members. In addition its policies and operating procedures, which influence cost behavior, may be somewhat different from those of the rest of the industry. In the writer's opinion, however, variations in policy and/or Operating procedures be- tween the company under study and other industry members is unlikely to be significant. In the foregoing regard it is particularly pertinent to quote a statement by Loring Farwell which prefaced his own study of cost behavior in the Industry. It stresses the basic similarity in the operating characteristics of small loan units:30 Essentially the same things are done in the same way with the same types of personnel and equipment in all offices. There are differences in detail. . . But, by and large, students of the field indicate that a descri tion of any one small loan [consumer financ§$. office will serve well to describe any other. In view of the above, the writer feels that it will be necessary to make some guarded generalizations from the results of the present pilot study. 30Loring Chapman Farwell, "Cost and Output Relationships of Con- sumer Instalment Credit Agencies: Based on Reports of Indiana Licensees," Northwestern University unpublished dissertation, 1955, p. 5. CHAPTER II THE CONTEXT OF THE STUDY Introduction In almost all states the lending of limited amounts of money at effective annual rates of charge in excess of state usury limitations is authorized by special small loan laws.1 These laws are at once permissive and regulatory in nature. They offer legal immunity from the operation of the state usury laws and at the same time impose a variety of restrictions upon the licensing and operation of small loan (consumer finance) companies. In most states the licensing requirements and restrictions upon mode of operation follow the general outline of the Uniform Small Loan Law. The first draft of this model law was developed in 1916 by the Department of Remedial Loans of the Russell Sage Foundation. The major licensing provisions require that the applicant meet a minimum capital requirement, be of good character and competent, and that the license, if granted, promote the "convenience and advantage" of the community in which the business of the applicant is to be conducted. The restrictions upon operations include limitations on rates of charge, loan size and maturity, advertising content, security which can be accepted, and many 1Arkansas and the District of Columbia are the exceptions. 31 32 others. There are also various record keeping and reporting requirements. Reason fer Small Loan Laws State small loan laws, patterned after the Uniform Small Loan Law, were developed in response to the need to make possible the legitimate lending of smaller sums at rates of charge sufficient to allow profitable operation. In the absence of special provisions to allow charging rates in excess of state legal maximums, the demand for small loans (originally defined as amounts of $300 or less) was served by extra-legal operators who became known as l22£.§22£5§° Rates charged by these lenders ranged well beyond those necessary to provide a reasonable return. In addition to the extreme rates charged by the loan sharks, part of which repre- sented a risk premium for operating an illegal business, their mode of operation often included various ferms of threat, subterfuge, and intimi- dation. Economic Rates of Charge Nature An "economic" rate of charge on a loan of money is one which is sufficient to allow the lender to recover the costs entailed by the lending activity. "Costs" in this sense would be the "economic" as opposed to the "accounting" concept of costs and would include a return to the equity capital consistent with the risk inherent in the business. 2 Roger S. Barrett, Com ilation of Consumer Finance Laws (Washington: National Consumer Finance Association, 19525. 33 Contrasted with interest It should be noted that the above reference is to economic "rates of charge" and not to economic rates of "interest.” In general, the term ”interest” covers the payment made to the lender for forbearance. Theoretically, the rate of interest on a riskless loan would be equal to the rate of return which the lender‘gggld have earned in a comparable riskless endeavor. Where the loan is not riskless, the appropriate rate of interest would equal the rate of return which the lender could achieve in an investment of comparable risk. An economic rate of charge has been defined above as one which is sufficient to allow the lender to recover the costs entailed by the lending activity. The cost of forbearance represents only one of the cost elements which the rate of charge must be sufficient to cover. The rate of charge must also be sufficient to cover the costs associated with the investigative, collection, and other servicing activities. In addition the charge must be sufficient to cover the risk and losses entailed in a lending operation. Therefore, it is inappropriate to compare the rate of charge on a small loan with some historical concept of a £323 rate of interest. As Professor Robert'w. Johnson has observed: On business loans it is entirely acceptable to call the finance charge 'interest' because the dominant component is the charge for forbearance. Hewever, the further one departs from a large, single-payment, riskless loan, the further the charge made on the loan departs from what is properly called interest. Because the major component of a finance charge is for service 3Robert W; Johnson, Methods of Stati Consumer Finance Char es (New York: Columbia University Graduate School of Businesss, p. IE. 3n and risk, it is more properly viewed as a service charge. A rough approximation of the cost of forbearance would be the amount of interest paid on borrowed funds. It will represent an understatement to the extent that some portion of the funds loaned are derived from equity sources. No explicit cost for such funds is included in conventional accounting presentations, and of course no explicit cost exists. They are, however, not costless from an economic viewpoint. Bearing in mind this limitation on the amount of interest paid as a measure of the cost of forbearance, the above point concerning the predominance of "non-forbear- ance" costs in the personal lending business is illustrated in the followa ing composite statement of income and expenses. Exhibit 2.1 COMPOSITE STATEMENT OF INCOME AND EXPENSES Calendar Year 1963 Personal Property Broker Business Pertaining to Loans of Less Than $5,000 Income Percentage to Average month-end loans of less Amount than $5,000 Charges collected on PPB loans of less than $5,000 $60,690,9u7.u9 11.05 Earned precomputed charges on PPB loans of less than $5,000 60,021,91#.08 10.92 Other interest and return on invest- ment 115,704.#2 .02 Insurance commissions 1,165,181.16 .21 Collections on personal property broker loans of less than $5,000 which were previously charged off 683,676.70 .12 Income from credit life insurance #2h,266.l9 .08 Other income 1 l 88 68 21 Total income $12h,239,078.73 22.31 35 Exhibit 2.1 (con't) Expenses * Percentage to Average month-end loans of less Amount than $5,000 Advertising $3,838,512.27 .70 Auditing u65,767.00 .08 Bad debts charged off on PPB loans of less than $5,000 1,612,459.59 .29 Provisions for bad debts reserves on PPB loans of less than $5,000 12,203,030.20 2.22 Depreciation 1,153,805.81 .21 Insurance and fidelity bonds 427,456.81 .08 Legal fees and disbursements 692,471.33 .13 Postage and express 1,216,566.60 .22 Printing, stationery and supplies 1,186,727.73 .22 Rent, lights, heat and janitor 5,024,253.20 .91 Salaries of office managers and employees 25,750,369.51 4.69 Salaries of owners, partners and executives 1,247,430.10 .23 Administrative bookkeeping and collection expense 3,020,926.26 .55 Taxes, other than income taxes 2,266,879.69 .41 License fees 311,468.33 .06 Telephone and telegraph 4,082,922.52 .74 Travel 1,163,578.70 .21 Home office expense allocation to California operations 7,348,555.88 1.34 Other expenses 5,255,852.22 .22 Total expenses before interest and income taxes $76,969,033.75 14.01 Interest paid 22,528,113.16 4.10 Provisions for income taxes 12,128,804.22 2.21 Total expenses $111 .18 20. 2 Net profit--loans of less than $5,000 $12,563,127.58 2.29 Net profit--loans of $5,000 or more 2 8 2 0 .8 .52 Net profits--combined totals $l5,435,l79.45 2.81 Source: Annual Report u n erations of licensed Finance Comp nies for the Ye r 1 , State of Califbrnia, Division of Corporations, p. 10. 36 The relative importance of "non-forbearance" costs is evidenced by the fact that they total 16.22 per cent of average loans outstanding (14.01 + 2.21) whereas interest paid (the forbearance cost proxy) amounts to only 4.10 per cent. If the net profit percentage is used as an ap— proximation of the cost of equity funds, the forbearance cost would amount to 6.91 per cent (4.10 + 2.81) versus 16.22 per cent for non-forbearance costs. Traditional maximum rates inadgguate "Restrictions imposed upon maximum charges for the use of money per- mitted lending in large amounts, but in practice prohibited lending in small amounts. Because of the high proportionate expense and the risks involved in loans of small sums, they were not commercially profitable at the conventional maximums."u The basis for the foregoing view is reflected in Table 2.1 which presents estimates of the monthly and annual rates of charge on unpaid balances which would be necessary to cover all costs--which includes a profit allowance. These rates were developed by a major consumer finance company from its cost data. This schedule of rates is presented only for the purpose of illustrating the view that conventional limits on finance charges were inadequate to allow the con- duct of the small loan business on a profitable basis. The development of these rates required assumptions to be made concerning the proper allocation of costs. Alternative allocative schemes would have produced a somewhat different schedule of required rates of charge. Consequently, “F. B. Hubachek, Annotations on Small Loan Laws (New York: Russell Sage Fbundation, 1938), p. 1. Because most of the loan costs are held to be "perbloan" as opposed to "per-dollar," loan costs will represent a higher proportion of small as opposed to large loans. NJ! 37 TABLE 2.1 ESTIMATED REQUIRED RATES ON LOANS OF DIFFERENT SIZES Required Rate Required Rate Loan Size per month per annum $ 25 7.3% 87.6% 50 4.3 51.6 100 2.8 33.6 200 2.2 26.4 300 1.8 21.6 400 1.7 20.4 500 1.6 19.2 Source: A major consumer finance company. these rates are not to be interpreted as representative. The required rates in the above table are those whichnwgglg‘bg.gggg§- ‘gggy to allow each loan to be profitable in its own right. They are not illustrative of the‘ggtggl rates charged on the various loan sizes. The per annum equivalents of the above monthly rates range from a low of 19 to a high of 88 per cent per annum. These rates are well above the con- ventional interest rate maximums which have traditionally ranged from 6 to 12 per cent per annum.5 It should be recognized that the rate structures contained in state small loan laws have seldom been designed to make the smaller loan sizes profitable in their own right. Rate structures were established in part with the view that part of the profits on larger loans would be used to 5It should be noted that the yield on a loan made under the limita- tions of between 6-12 per cent could approach 12 to 24 per cent if made on an add-on or discount basis and repaid in instalments. 38 6 subsidize the losses on smaller loans. This view is highlighted below: If licensed lenders are to perform the func- tions which the statute contemplates, they should be willing to meet the requirements of the borrower who is in need of a $25 or $50 loan. To fix a rate which would make very small loans profitable would not be feasible because such a rate would necessarily be pro- hibitive. The Department believes that the licensed lenders must recognize that the State will not continue indefinitely to permit them to earn excessive profits on larger loans unless they utilize a portion of such income to sup- port a reasonable volume of very small loans. . . The statement that lenders must apply excess profits from large loans to cover losses on smaller loans means that borrowers of larger amounts must be discriminated against. They must pay "rates higher than is justi- fied by their pro-rate share of the costs. In other words, price discrimi- 7 nation must not only be accepted but encouraged." Hopefully, the present study will make it possible to judge the magnitude of the price discrimi- nation entailed by the regulatory philosophy expressed above. Impgrtance of the Industgy Industgy defined In most of the literature companies which operate principally under State small loan laws are referred to as "consumer finance" companies. Consumer finance companies are defined by the Federal Reserve Board as- 6Second 3 ecial Re rt of Sn erintendent of Banks Relative to " Licensed Lenders (New York: State Banking Department, 1&1), pp. 19-20. 7Miller Upton, "The Economics of Fair Charges For Consumer Loans," Missouri Law Review (June, 1951), p. 287. 39 ". . .engaged principally in.making personal loans under State small loan laws." The industry which is made up of these companies is termed the Consumer Finance industry. Compgsition of business The Consumer Finance industry specializes in.making personal instal- ment cash loans under state small loan or comparable laws. In addition to direct personal cash loans, consumer finance companies also carry some automobile and other consumer goods paper. Table 2.2 below presents a breakdown of the uses of funds by nine major consumer finance companies for 1959. This asset composition reveals that consumer finance companies TABLE 2.2 USES OF FUNDS BY CONSUMER FINANCE COMPANIES, END OF 1959 (Per cent) Ra e of Ratios Mean Item Distribution Maximum Minimum Earning assets, net 87.? 4.6 82.0 Consumer credit 86.5 94.4 80.0 Automobile paper 1.8 15.7 0 Other consumer goods paper 6.8 21.5 0 Personal loans 77.9 94.3 53.7 Other 1.2 5.8 0 Cash and bank balances 9.0 14.6 2.7 Other assets 3.3 4.5 2.2 Total 100.0 -- -- Source: Paul F. Smith, ns or Credit Costs 1 u - (Princeton, New Jersey: Princeton University Press, 1924), p. 7. K . on l—I_ '- do have some diversification in their lending activities. In particular, note that one company had only 53.7 per cent of its net earning assets in personal loans. Types of expenditures facilitated The distribution of loans made by consumer finance companies accord- ing to the purpose which the borrower gives for the loan is presented in Table 2.3. TABLE 2.3 DISTRIBUTION OF THE NUMBER OF LOANS EXTENDED BY PURPOSE (1963) Purpose Per Cent To consolidate existing bills 48 Travel, vacation, education expense 8 Automobile purchase or repairs 11 Home furnishing and appliances 3 Household repairs 3 All other purposes ‘_gz Total 100 Source: Estimated by the National Consumer Finance Association from companies which operated 4,370 offices. Reported in S. Lees Booth, Finance Facts Yearbook 1265 (washing- ton: NCFA, 1965). p. 51. Industpz impprtance--gualitative It has been observed that the Consumer Finance industry developed in response to a need for a fairly compensated and adequately supervised source of small personal instalment cash loans. It is difficult to assess the importance of this industry to our society and our economy. From a social viewpoint, it is clear that state provisions for legally sanctioned ‘ - k 1 41 sources of small loans have gone a long way toward the elimination of the abuses characteristic of the loan shark era.8 In addition, there are the benefits which accrue to those who are able to secure funds for emergencies and numerous other worthwhile purposes. Moreover, where (l) the funds are used to acquire some durable good whose rate of return is greater than the cost of the money or (2) where the loan facilitates an expenditure whose rate of time preference is greater than the cost of the funds, borrowing proves a positive benefit which allows the borrower to (a) maximize his wealth or (b) maximize his subjective utility. Income groups served Table 2.4 presents a distribution of loans extended by consumer . finance companies on the basis of the monthly income of the borrowers. It is important to note that 66.4 per cent of the loans were made to bor- rowers whose monthly incomes were in excess of $400. This indicates that the services of the industry are not confined to low income groups. General economic effects There is considerable agreement concerning the influence of the cone sumer financing industry (all financial institutions which extend consumer credit) on our economic growth and development. The central position is that the availability of instalment financing has facilitated consumer expenditures and stimulated our economic growth. Also, the expansion in the market for consumer durable goods facilitated by instalment credit 8Harold Johnson, "Nebraska Hes No Loan Shark Problem Today." J,A.A. Burnquist, "A Regulatory Small Loan Law Solves Loan Shark Problem." B.J. Lenihan, "Progress In Consumer Credit In Kentucky." Law and Con- tem ra Problems, Vol. VIII, No. 1 (Winter, 1941). 42 TABLE 2.4 PERCENTAGE DISTRIBUTION OF THE MONTHLY INCOME OF BORROWERS FROM CONSUMER FINANCE COMPANIES, FOR.1963 Monthly Income Per Cent of of Borrowers Loans Extended $0.00-100.00 0.3 100.01-200.00 3.0 200.01—300.00 10.4 300.01-400.00 19.9 400.01-500.00 25.0 500.01-750.00 30.1 750.01-l,000.00 9.1 1,000.01 and over '_222 Total 100.00 Source: National Consumer Finance Association survey data for comp panics operating over 1,700 offices in 1963. Reported in S. Lees Booth, fiinance Facts Yearbook 1265, p. 52. has made possible the attainment of economies of scale which have re- sulted in lower costs to the producers and lower prices to ultimate con- sumers e Industry’importance-equantitative From.a quantitative viewpoint the importance of the Consumer Finance industry is indicated by its share of outstanding consumer instalment credit. Table 2.5 shows the percentages of consumer instalment credit held by the various financial institutions. As the Table 2.5 reveals, commercial banks are the largest single holder of consumer instalment credit at the end of 1964. Credit unions have shown significant growth in their share of outstanding credit over 43 TABLE 2.5 TOTAL CONSUMER INSTALMENT CREDIT Dec. 31 Dec. 31 Dec. 31 1949 1959 1964 Outstandings (in millions) $11,590 $39,245 $59,397 Percentages Held By: Commercial Banks 38.3% 38.8% 40.3% Sales Finance Companies 25.4 26.3 24.9 Consumer'Finance Companies 9.3 8.5 8.5 Credit Unions 3.8 8.4 10.9 Other Financial Institutions 3.1 3.5 2.9 Retailers 20,1 14.5 12.5 Total 100.0% 100. 0% 100. 0%, Source: Federal Reserve Board. the period 1949 to 1964. Consumer finance companies held 8.5 per cent of outstanding consumer instalment credit which places them in fifth place among the 6 institutions covered. This fact alone might cause one to conclude that consumer finance companies are of little importance among consumer instalment credit grantors. However, since the principal activity of the industry is concentrated in the extension of personal loans, a better measure of the industry's overall importance is revealed by its share of this market. Table 2.6 presents a breakdown on the holdings of personal instalment cash loans. In this presentation the consumer finance subsidiaries of sales finance companies are combined with other consumer finance companies to yield a closer representation of the holdings of the Consumer Finance industry. Table 2.6 shows that in the relevant market-~personal instalment cash loans-~the Consumer Finance industry is the principal credit grantor. TABLE 2.6 PERSONAL CASH.LOANS Dec. 31 Dec. 31 Dec. 31 1949 1959 1964 Outstandings (in minions) $2,431 $9, 386 $16,071 Percentages Held By: Consumer Finance Companies 37.5% 30.4% 27.6% Sales Finance Companies .8 10.1 12.6 "Consumer Finance Industry" 43.3% 40.5% 40.2% Commercial Banks 37.6 34.1 33.5 Credit Unions 11.9 17.6 20.2 Other Financial Institutions 2,2 2,8 6.1 Total 100.0% 100.0% 100.0% Source: EFoderal Reserve Board. This market share has been fairly stable over the period 1949 to 1964. Credit unions have shown a marked increase in their share of outstanding personal cash loans. The long run competitive position of the small loan industry will be conditioned in part by the cost of its product. What is the general relationship between the level of product cost in the Consumer Finance industry and other financial institutions? How can the cost differences among the various financial institutions be accounted for? An effort to shed some light on these questions will form the content of the following section on ”Inter-Institutional Cost Comparisons.” Inter-Institutional Cost gpmparison To effect a comparison of costs among financial institutions, it is necessary to use some common index or measure of cost. A possible measure u — ' Cl 45 of relative institutional costs is the gross finance charges per one hundred dollars of average outstanding loan balance. This type of mea- sure was employed in a recently published National Bureau study conducted by Paul F. Smith.9 A summary comparison of the cost of consumer credit at four finan- cial institutions based on Smith's study is presented in Table 2.7. Reasons for cost differences Table 2.7 reveals that the cost of consumer credit, as measured by the gross finance charges per $100 of outstanding credit, was highest at consumer finance companies ($24.04) and lowest at the Federal credit unions ($9.13). What accounts for this considerable range in cost? A number of important factors or conditions contribute to the variation in operatipg e enses, which were four times higher at consumer finance companies than at Federal credit unions and more than three times higher at consumer finance companies than at commercial banks. These factors include differences in:10 (l) The method of acquiring business, whether directly from.the public or indirectly through dealers. (2) The character of the risks assumed. (3) The average size of contract. (4) The type of credit. (5) Institutional differences. 9Paul F. Smith, Consumer gradit stts 1242-52 (Princeton, New Jersey: Princeton University Press, 19 . 10 Paul F. Smith, op.cit., pp. 79-85. TABLE 2.7 COMPONENTS OF GROSS FINANCE CHARGES ON CONSUMER CREDIT, BY TYPE OF LENDER, 1959 (dollars per $100 of average outstanding credit) Stockholder-Owned Institutions Nine All Consumer Ten Sales Nine Federal Finance Finance Commercial Credit Item Companies Companies Banks Unions Gross finance charges 24.04 16.59 10.04 9.13 Dealer's share of gross finance charges .17 2.95 .62 0 Lender's gross revenue 23.87 13.64 9.42 9.13 Operating expenses 14.25 7.74 4.17 3.30 Salaries 6.45 3.47 2.33 1.77 Occupancy costs 1.09 .43 .23 .06 Advertising .89 .31 .34 .07 Provision for losses 1.98 1.46 .28 n.a. Actual losses (1.70) (1.11) (.15) .38 Other 3.84 2.07 .99 1.02 Nonoperating expenses 9.62 5.90 5.25 5.83 Cost of nonequity funds 3.97 4.02 1.50 .12 Income taxes 2.73 1.07 1.33 0 Cost of equity funds (lender's profit or net income) 2.92 .81 2.42 5. 71 Dividends 2.31 .48 1.49 4.00 Retained .61 .33 .93 .98 Services to owners -- -- -- .73 Source: Paul F. Smith, Consumer redit Costs 1 4 - (Princeton, New Jersey: Princeton University Press, 19 ), p. 78. 47 Method of agguiripg business Generally, sales finance companies acquire their paper from dealers (indirect paper) while consumer finance companies deal directly with their customers (direct paper). As Table 2.7 reveals, approximately 18 per cent of the gross finance charges of the sales finance companies represents the dealer's share. This "dealer's share" is typically termed the "dealer reserve" in the trade. The dealer's reserve has been described as follows:11 To compensate a dealer for originating the business, to reimburse him for extra clerical costs, and to build up a fund against which losses may be charged, a reserve fund‘Zdealer's reserv§7 is set up out of the finance charge. Consumer finance companies acquire their paper principally on a direct basis from their customers (borrowers). This difference in the method of acquiring business is reflected in part in the greater relative advertising expenditure per $100 of credit made by consumer finance com- panies. Consumer finance companies spent almost three times as much per $100 of credit outstanding as do either sales finance companies or come mercial banks. In addition, acquiring business direct from the customers requires that consumer finance companies locate in such a manner so as to be convenient to their customers. This may well entail more outlets relative to an institution (such as Sales finance companies) where the principal method of acquiring business is indirect. This factor alone would tend to cause costs to be higher at consumer finance versus sales finance companies. llInstalment Credit Committee, The American Bankers Association, Instalment Credit Fundamentals (New York: American Bankers Assoc., 196“) 9 Pa 29. Character of risk assumed In addition to differences in the method of acquiring business, the character of the risk assumed by the various financial institutions also exerts an influence on their relative costs. Table 2.? reveals that the rate of net charge-offs (charge-offs net of recoveries) ranged from a high of $1.70 dollars per $100.00 of average outstanding credit at con- sumer finance companies to a low of $0.15 at commercial banks. This represents objective evidence that consumer finance companies have their business concentrated in a more risky borrower group. This does not mean that there is no duplication among the financial institutions in the type of borrower served. Consumer finance companies do make some loans which would be just as readily made by other financial institutions. In considering the effect of greater risk assumption by consumer finance companies on their relative costs, it should be recognized that the difference in the net charge-off rate between consumer finance com- panies and other financial institutions does not present a complete picture of the added costs due to accepting more risky business. Some of the added costs are lodged in salary expense due to the fact that servicing a more risky borrower group entails additional investigative and collection effort. Additional effort in these areas is reflected mainly in personnel costs. As Smith noted:12 If all the costs associated with variations in risk could be isolated, risks would undoubtedly play a substantial part in explaining differences in operating costs among lenders. lemith, op.cit., pp. 81-82. ‘Qifferences in loan size Another factor which contributes to the explanation of the differ- ences in operating costs among lenders is the difference in the average size loan dealt in by the different lenders. The differences in the average size contract acquired and outstanding for the institutions and companies included in Paul Smith's study are presented in Table 2.8 TABLE 2.8 AVERAGE SIZE OF CONTRACTS ACQUIRED AND OUTSTANDING (1959) Average Size of Contrggt Type of Institution Acquired Outstanding Nine consumer finance companies $ 436 $346 Ten sales finance companies 896 700 Nine commercial banks 1,031 723 All federal credit.unions 593 553 Source: Paul F. Smith, Consumer Credit Costs 1 4 - (Princeton, New Jersey: Princeton University Press, 19 ), p. 83. Consumer finance companies have the lowest average size contracts acquired and outstanding. This fact contributes to their position as the highest cost consumer financing agency. This derives from the fact that costs are not proportional to size. As the loan size increases, for any financial institution, the total costs as a per cent of the loan balance decline. Therefore, other things equal, the institutions with the smaller loan size will have higher costs (using the measure of dollars 50 per $100 of average outstandings) than institutions that deal in larger consumer credit contracts. Table 2.8 is based upon 1959 data. In the intervening time period the average size contract acquired and outstanding have increased con- siderably for consumer finance companies. This is apparent from a review of company annual reports as well as the annual reports on consumer finance companies compiled and published by most states. A study by the National Consumer'Finance Association indicates that for 1963 the average contract acquired by consumer finance companies was about $524 and the average out- standing about $447.13 These figures reflect the influence of the limita- tions on maximum loan size and maturity of the various state small loan and related laws. Since the above figures reflect the averages of com. panies which operate in different states with different limitations on loan size and.maturity, they do not reveal the actual averages of con- tracts acquired and outstanding by state. In a state where the limita- tion on maximum loan size is below the average for the nation as a whole, one will find the loan averages below the above cited figures. Where the maximum loan size is greater than the national average, the loan averages will also tend to be above the national average reflected by the above figures. Ohio, with a $2,000 maximum, showed an average loan made of' $649 and outstanding of $563 against the comparable averages of $524 and $447 shown above from the NCFA study}!4 Illinois, with an $800 maximwm, showed averages of $467 and $391 for 1963.15 13s. Lees Booth, op,cit., p. 55. 14 . Re rt and Roster of Small Loan Licensees - State of Ohio (Columbus: Division of Securities, 1964), p. 5. l 51 Anal sis of Re rts (Illinois: Department of’Financial Institutions, Division of Consumer Finance), p. 5. Types of credit handled The types of credit handled by an institution will affect its cost levels. For example, it is usually held that certain types of consumer financing arrangements are more costly than others. Therefore, part of the variation in costs among the financial institutions may be explained in terms of the different types of consumer financing arrangements which are emphasized. Institutional factors Finally, institutional differences exert an influence on relative costs. In the case of credit unions, it is common for them to occupy quarters provided by the sponsoring organization at little or no cost. Also, some of the clerical and other administrative activity may be done on a voluntary basis by members. These factors favorably affect the credit union's relative operating cost position. The effect of the maximum loan size limitations on costs at consumer finance companies (an institutional factor) has already been discussed above. In addition, various added reporting and record keeping requirements imposed by state small loan laws are an institutional factor contributing to cost differ- ences among the different credit granting agencies. Nonoperatipg e§penses In the area of’nonoperating expenses, there is considerably less variation among the four institutions (see Table 2.7). "Differences in costsllsonoperating7 among the four types of institutions reflect pri- marily different sources of funds and, in the case of federal credit unions, exemption frmm income taxes.” 16Smith, op,cit., p. 85. 52 Sources of Funds Consumer finance companies employ a variety of fund sources. The distribution of liabilities, capital, and surplus in Table 2.9 is based upon annual surveys conducted by the National Consumer Finance Associa- tion. TABLE 2.9 PERCENTAGE DISTRIBUTION OF LIABILITIES, CAPITAL AND SURPLUS FOR FINANCE COMPANIES REPORTING TO NCFA QUESTIONNAIRES Balance.Sheet.Item 1960 1963 Notes payable to banks 18.6 16.5 Commercial paper and other short-term debt 5.3 7.9 Deposit liabilities and thrift certificates (short-term) 5.5 2.3 Long-term debt (excluding subordinated debentures) 35.2 36.7 Subordinated debentures and notes 8.7 9.7 Other liabilities 4.6 4.6 Capital and surplus 22.1 22.4 Source: NCFA surveys reported in S. Lees Booth, Finance Facts Yearbook 1265 (Washington: NCFA, 1965), p. 0. Among consumer finance companies there has been a shift in the struc- ture of fund sources, most of which occurred during the 1950's. The de- gree of change has not been uniform over all company sizes. However, the general character of the change in capital structure has been in the direction of greater reliance upon debt funds and among debt funds the emphasis has 17 shifted to long-term debt and away from short-term bank sources.. For l7For comprehensive treatment of consumer finance fund sources see John M. Chapman and.Frederick W; Jones, Finance Companies: How and Where The Obtain Their Funds (New York: Columbia University, Grad. School of Business, 1959) and James R. Longstreet, "The Capital Structure of Con- sumer Finance Companies" (unpublished Ph.D. dissertation, Dept. of Eco- nomics, Northwestern University, 1956). 53 the nine major consumer finance companies included in Smith's study, non-equity sources of funds increased from 67 to 75 per cent over the 18 period 1949-59. Short term sources declined in importance from 44 1 per cent in 1949 to 33 per cent in 1959. 9 Among medium and smaller size companies the shift in fund sources over this same period was generally from equity to a greater reliance upon debt. Within the group of debt instruments the shift has been to "subordinated deben- tures." In an article on subordinated debentures, Professor Johnson 20 observes the following regarding their character and use: The key feature that distinguishes subordinated debentures from the usual debenture is the su- bordination agreement. Under this provision the debentures are usually subordinated in the event of liquidation or reorganization to any existing and future debt that is defined in the indenture as senior debt. hhth respect to senior creditors, subordinated debentures become in effect part of the equity base. On the prospects for continued employment of subordinated deben- tures, the NCFA study on the Consumer Finance Industry predicted that the widespread use of subordinated debentures among medium and smaller size companies would continue because "small and mediumesize companies need this type of financing as a source of long-term debt and to im- 21 prove their borrowing position for senior financing." lamth, OE‘Cj-te, p. 1300 19;pid., p. 128. 2oRobert‘W. Johnson, "Subordinated Debentures: Debt That Serves as Enuity," The Jopppg; of Eipgnpg, X, No. 1 (March, 1955), PP- 235. 21The Consumer Finance Industpy (Englewood Cliffs: Prentice-Hall, 1962), p. 53. 54 In general, in terms of basic capital structure developments, it can be expected that the increased employment of debt will continue for the following reasons:22 (1) The stability and earnings records of the consumer finance industry has justified the use of more debt in relation to equity as a sound financial practice. (2) The increase in debt improves the earning power of these companies. (3) Competitive pressures in the market for consumer loans exert a strong force for reduction in rates and increases in ser- vices provided with loans. Such compe- titive pressures can only be met by more efficient operations or by increasing the ratio of borrowed funds to net worth. Each of the above points include some reference to the earnings of the industry, its stability, growth, or maintenance. The following sec- tion will review briefly the industry's earnings performance and in par- ticular the role of financing strategy in industry earnings performance. garnipgs Performance Ihe zardstick An effort to assess the earnings performance requires the selection of some measure of return. Commonly used bases are earnings as a per cent of stockholders' equity and earnings as a per cent of total assets. In the consumer finance industry a traditional earnings measure has been earnings as a per cent of assets used and useful. Generally, the earnings measure employed is earnings after taxes but before taxes. This is, at zglpig., pp. 52-53. 55 least in part, a regulatory by-product. In determining the adequacy of earnings, it is held that earnings after taxes but before interest is the appropriate measure since "interest is not a cost of doing business but rather is a bybproduct of the capital structures of the various 23 finance companies." Earnipgs performance The study of the Consumer Finance industry, prepared by the National Consumer Finance Association, revealed that "Net earnings of the consumer finance industry as a percent of assets used and useful has declined steadily almost without exception in states for which data are available."24 Table 2.10 presents a summary of the NCFA study results on earnings per- formance by states (high, median and low). TABLE 2.10 EARNINGS AS A PER CENT OF ASSETS USED AND USEFUL Year Median High Low 1950 7.30 10.07 4.60 1958 5.51 6.76 3.19 Source: The Consumer Finance Epgustpy (Englewood Cliffs: Prentice- Hall, 1 2 , p. 53. 23Thomas G. Gies, C. V. Fricke and Martha Seger, Consumer Finance Comppnies In Michigan (Ann Arbor: Bureau of Business Research, Uni- versity of Michigan, 1961), p. 79. 2“The Consumer Finance Industry, op,git., p. 78. 56 A study by the Stanford Research Institute revealed a similar earn- ings trend. In addition, the Stanford study indicated that the average earnings performance of the consumer finance industry was exceeded by twenty-nine of thirtyhthree industries included in the study.25 Infor- mation on the earnings of Michigan consumer finance companies reveals the same downward trend in earnings throughout the Fifties. In addition, the persistence of the earnings decline into the Sixties is highlighted. TABLE 2.11 NET EARNINGS OF MICHIGAN CONSUMER FINANCE COMPANIES, 1960 - 1963 (After taxes but before interest) As Per Cent of As Per Cent of Year Average Loans Outstanding Average Invested Assets 1960 6.74 6.31 1961 6.49 6.12 1962 6.13 5.77 1963 5.65 5.36 Source: Michigan State Banking Department, Abstract of Repgrts of Regplatopy Loan Licensees. The declining rates of earnings on assets reflect a combination of the effect of (1) increasing costs and (2) the graduated rate structures in effect in almost all states. All other things equal, the increases in consumer finance company costs would tend to reduce earnings as a 25 Com site Earni 5 Performance of The Ma or Consumer Finance Compgnies, (Menlo Park, California: Stanford Research Institute, 1959). ' . w 57 per cent of assets. In addition, the increases in average loans made and outstanding, given the downward graduation in small loan rate struc- tures, reduces the average rate of charge collected on outstanding loans. Given the above conditions, the industry finds itself squeezed between ”a decreasing rate of charge collected and increasing costs. . ."26 The impact of declining earnings on returns to stockholders' equity has been mitigated through the increased employment of borrowed funds-- financial leverage. The National Bureau of Economic Research study by Paul F. Smith showed a decline in net profits as a per cent of equity funds (for the 9 major consumer finance companies included in his study) from 14.5 per cent in 1949 to 12.1 per cent in 1959.27 This decline in return to equity is clearly less pronounced than the general rates of decline in earnings on either average invested assets or average loans outstanding. The role of financial leverage The effectiveness of financial leverage in.maintaining returns to equityholders will depend upon the relationship between the rate of earn- ings on assets and the cost of borrowed funds. It is a fundamental finan- cial management principle that the return earned on total assets must . exceed the cost of borrowed funds if financial leverage is to enhance the return earned on the owner's equity. The effectiveness of financial leverage in consumer finance companies will depend upon the behavior of 26The Qpnsumer Egnance Industgy, p. 73. 2 7Smith, op,cit., p. 20. (1) the cost of borrowed funds as well as upon the behavior of (2) other lending costs and (3) the average rates of charge earned on loans. The behavior of all of the above elements suggests that the maintenance of " returns will ‘become increasingly difficult in the future. Costs cone tinue to increase; average loan sizes increase; and with such increase the average rate of charge collected continues to decline, reflecting the downward graduation in rate structures. In addition, the cost of borrowed funds shows indications of being on the increase.28 "The average rate for directly placed finance company paper with maturities of 3 to 6 months increased from 3.75 Per cent to 4.25 between last fall and this spring. Bank rates on loans to finance companies have also risen during this period."29 All of these factors point up problems which the industry must meet to maintain its earnings performance. In addition to the foregoing, the increased competition in the personal loan market from commercial banks and credit unions will present a challenge. 28As has been noted earlier, the higher average loan size does have a cost reducing aspect since costs do not increase proportionately with loan size. This mitigates somewhat the effect of the reduction in the average rates of charge collected due to the interaction between higher average loan sizes and graduated rate structures. 29Federal Reserve Bulletin, June, 1965, p. 798. 5! CHAPTER III LITERATURE REVIEW Introduction Any scholarly effort must necessarily entail a careful review of the pertinent literature. The literature pertinent to the present study is found in a wide range of sources: finance and accounting journals, economics journals, company studies, corporate annual reports, trade association monographs, banking journals, and unpublished doctoral dis- sertations. The relevant literature can be divided into two general categories: (1) Discussions and illustrations of loan cost accounting technique and methodology. (2) More general statistical and nonpstatistical works on industry and firm cost behavior. This literature review will be divided on the basis of the above classification. The objective will be to highlight the work that has already been done in the area of loan cost behavior and to emphasize further the particular contribution which the present study is designed to make. Many more works have been reviewed than will be covered here. Some useful works are not reviewed where in the writer's judgment the same essential points have been made in other works which are reviewed. It should be emphasized that, particularly in the case of'more lengthy studies, the review is confined primarily to the material which is directly applicable to the author's study. Therefore, a judgment as 59 60 to the total contribution of the study should not be drawn on the basis of the review included in this chapter. Cost Accounting Studiesl Cost accounting and financial institutions The traditional concern of cost accounting has been the development of product cost information for purposes of asset valuation-income deter- mination, pricing, and cost control. A review of contemporary cost accounting textbooks reveals that emphasis is almost exclusively on the cost accounting problems of manufacturing enterprises. Cost accounting problems of financial institutions (such as consumer finance companies) are seldom covered. This might suggest that cost accounting methodology is not applicable to financial institutions. However, it is the writer's opinion that the application of cost accounting in financial institutions has been slow because financial institutions have not been compelled to lRepresentative cost accounting and other closely related studies: Gilbert W. Urban, ”Time and Dollar Costs in Consumer Financing," National Association of Cost Accountants Bulletin (NAA), April, 1955, pp. 10 3- . Haro d . Randall, "The Basic Cost Factors in Instalment Lending," Time Sales Financing, May, l9h9, pp. 7-8, and 20-21. fllyging The Cnst Factors of Instalment Lending (New York: Instalment Credit Committee, The American Bankers Association, 19h6). G. Schuyler Blue, "Determining Bank Instalment Credit Costs," Banki , September, 1951, pp. 36-37 and 106. Develoning Cost Infbrmation on Instalment Credit Qnerations (New York: Instalment Credit Committee, The American Bankers Association,” 196“). J. T. Arenberg, "Consumer Credit Net Yields." An address pre- sented at The Instalment Lending Conference of the Illinois Bankers Association, 196“, 16 pages. Know Your rni s on Instalment Loans (New York: American Bankers Association, l930§. Harold E. Randall, ”Cost Accounting fer Financial Service is Just as Necessary as It is in Manufacturing," The Journal of Accountanc , July, 1949, pp. 73-7h. W2 L. Brown, "Instalment Loan Cost Analysis Includes variety of Items," The Industrial Banker, April, l9fi5, pp. 5—7. 61 develop cost accounting systems or procedures fer asset valuationpincome determination purposes. In a practical sense, a manufacturer is com- pelled to develop a cost accounting or cost determination system in order to cost inventory and to determine income. The same is not gen- erally true of financial institutions. In addition to not being compelled because of the asset valuationp income determination requirement to develop cost information, financial institutions have also faced an additional problem which has retarded their application of cost accounting methodology. With respect to the unit of product, financial institutions have a preponderance of common or joint costs. Direct unit costs are negligible. This is in contrast to a manufacturing enterprise where direct materials and direct labor commonly form an important segment of total unit cost. This is an imp portant problem for the financial institutions and one which has cons tributed to the lack of emphasis on cost accounting. Character of studies reviewed The cost accounting studies reviewed here are in the conventional cost accounting tradition in that they are directed to the development of unit cost information. The basic objective which they share is that of developing estimates of the costs of loans of different classes, sizes, and maturities. In most cases ”cost" includes not only the costs incurred within the confines of the instalment lending activity but also various elements of indirect costs--indirect with respect to the instalment lending activity. 62 Income determination per so was not the reason for the cost account- ing studies reviewed here. The objective of the studies was the develop— ment of unit cost information on loans of different classes, sizes, and maturities. The ngnnnn for developing such cost information appears to have varied depending upon the time and the particular type of financial institution. The early development of cost information on instalment loans by banks seems to have been designed as a form of banker-education program to curb the practice of rate cutting. A l9h6 publication of the American Bankers Association, nnnlyning The Cost Factors 0f Instalment Lending, stressed the need to know loan costs in order to prevent unwise rate cuts:2 Cost factors are too often overlooked or under- estimated and may result in unstudied lowering of instalment loan rates. Competitive factors may influence lower rates but good management would not consent to a rate reduction unless all the pertinent cost elements were considered. In addition to warning against rate cuts which would result in fail- ure to recover costs, it was also stated that "Experience shows very little, if any, justification for cutting rates where volume remains within figures found to exist in all but a few larger banks."3 This statement suggests either a relatively inelastic demand schedule for instalment loans or perhaps counter price cutting by competing banks and other financial institutions. 2Anal 1 The Cost ctors of Instalment Lendi (New York: In— stalment Credit Committee, American Bankers Association, l9h6), p. 3. 3nngu p.5. 63 In the case of consumer finance companies, the development of unit cost information appears to have been more of a "special study” interest; unit cost data were developed primarily for rate negotiation purposes. There has been very little effort to develop unit cost information on either a continuing or periodic basis for managerial purposes. In this regard, the writer recently made the following completely obvious remark to an executive of a major consmmer finance company: "Unit cost data on instalment loans does not exist naturally. The executives response was that: "They don't exist unnaturally either.” The import of this remark will become more apparent as this review of cost accounting studies progresses. Functional cost classification The basic approach to developing unit cost information on instalment loans, regardless of whether they are made by a bank, consumer finance company, sales finance company, or credit union, is similar in most major details among the studies reviewed. The costs of the instalment lending activity are first related to a functional base and then to the unit of product--the loan. The basic functional activities are divided among: (1) Acquisition (2) Servicing (3) Risk assumption (4) Forbearance This functional classification is not completely homogeneous. Func- tions 1 and 2 involve mainly personnel activity. Conceptually, 3 (risk assumption) entails personnel effort in addition to the inert act of 6h assuming the risk of loss entailed in the lending activity. This point was made in chapter two when it was observed that net charge offs (real- ized risk assumption) were an incomplete measure of the costs which could reasonably be ascribed to the risk assumption function. Part of the acquisition function is made up of the investigative effort which is undertaken to hold the amount of risk assumed to tolerable levels. The servicing activity includes delinquency control which also is related to controlling realized risk assumption. However, the difficulty of isolating the costs which are conceptually related to risk assumption has resulted in the use of net charge-offs as a measure of the cost of the risk assumption function. Likewise, forbearance is not a function which entails personnel effort. Its cost is also only approximated. In a theoretical sense the cost of forbearance is equivalent to the opportunity cost entailed by lending the money versus some alternative employment of the funds in a venture involving comparable risk. The practical concession to the theoretically correct has generally been to employ the cost of borrowed funds as the measure of the cost of the forbearance function. Unit cost develonnent The foregoing functional cost classification is relatively easy to conceptualize: it is much more difficult to implement with the cost and operating data of'an operating unit. One of the chief problems derives from the joint or common character of most costs incurred in the instal- ment lending activity. The development of unit costs necessarily re- quires a considerable amount of estimation and allocation. Some studies 65 which develop loan unit costs employ fairly detailed time study tech- niques as the basis for developing (l) the functional cost breakdown and (2) assigning the functional costs to the unit of output--the loan. Other studies minimize time study effort and.make many allocations and assignments of cost on a "best-guess" basis. The following general procedure, which can entail a minimum of time study, has been recom- mended to develop unit costs for instalment loans.“ Step One: Ascertain the amount of each expense of the branch which relates to the acquisition of loans and the amount which relates to the servicing of loans. Step Two: Compute the acquisition cost per loan by dividing total acquisition expenses by the number of loans made during the most recent fiscal year. Compute the servicing cost per payment by dividing total maintenance expenses by total payments processed during the most recent fiscal year. Step Three: Compute those costs which vary with the amount of each loan; namely, a loss experience factor and a 'cost of’funds employed' factor. Step Four: Compute the cost of a loan by adding the ac- quisition, servicing, loss and cost of money factors. Steps one and two would require that the branch expenses, both direct and allocated, be divided on the basis of their relationship to the ac- quisition and servicing activities--broadly defined. Exhibit 3.1 presents such a division of costs between acquisition and servicing. “The procedure outlined is adapted from an approach outlined for developing unit costs for instalment lending departments of commercial banks by J. T. Arenberg of Arthur Andersen & Company and presented at The Instalment Lending Conferences of the Illinois, Indiana, and Iowa Bankers Associations. The address was titled "Consumer Credit Net Yields." EXHIBIT 3.1 DETERMINATION OF ACQUISITION AND SERVICING COSTS (Year Ended December 31, l9__) Amount Relati To Total Acquisition Servicing Direct Branch Expenses: Salaries $56,000 $30,000 $26,000 Depreciation and maintenance 12,000 6,500 5,500 Advertising 4,500 u,500 ‘ Telephone and postage 4,000 300 3,700 Stationery and supplies 3,500 500 3,000 Payroll taxes 1,100 600 500 Legal and audit 1,000 — 1,000 Insurance 200 #00 500 Totals $83,000 $42,800 $40,200 Indirect (allocated) Branch Expenses: 1 000 16 00 1h 00 Totals 114 000 m m Acquisition cost per loan based on 5000 loans made in 19__ 11.90 Servicing cost per loan payment based on 60,000 payments processed during 19__ - .91- Note: The above illustration is adapted from J. T. Arenberg, "Consumer Credit Not Yields," an address presented at the Instalment Lend- ing Conferences of the Illinois, Indiana, and Iowa Bankers As- sociations. The dollar amounts of the various individual and the total expenses arelnnn presented as being representative of actual costs for a consumer finance operation of the above scale. 67 The division of cost elements between the acquisition and servicing functions could be based upon very laborous time studies or the division could be accomplished with a minimum of time study and a maximum of sub- jective, but informed, estimates. The additional effort invested in the time studies should be related to the resulting increase in accuracy and the value of such added accuracy. In addition, in assessing the amount of refinement which should be incorporated into the cost determination, it should be recognized that the resulting unit cost data represent only a tool ”which serves primarily as a guide in establishing loan policies and practices. . . The costs produced by the procedures outlined, then, must be viewed in the light of what they are intended to portray; namely, as reasonably realistic indications of cost."5 Step three calls for a determination of a loss experience factor and a cost of funds employed factor. These cost factors will represent the cost of th°.£$§§ assumntion and forbearance functions respectively. Some average loss experience rate can be employed as the measure of the cost of risk assumption. The shortcomings of such a measure to reflect the real costs of risk assumption have already been noted. The factor for the cost of funds employed will be an estimate of the cost of bor- rowed funds. The shortcomings of this measure have also been discussed previously. The following cost of funds and loss factors will be used for illustrative purposes. Like the foregoing cost data, they are used only for purposes of illustration and are not presented as representative: 5Arenberg, on,cit., p. 3. 68 Loss experience factor 1.5 per cent of average outstanding balance. Cost of funds factor 5.0 per cent of average outstanding balance. hhth.the data in Exhibit 3.1 and the above assumed cost factors, it is possible to develop estimates of unit costs of loans of various sizes and maturities. In symbolic notation, the total unit cost of a loan can be represented as follows: Ci = 2(A1 +'31 + Fa *'31) C1 Total cost of loan i A1 Acquisition cost of loan i S Servicing cost of loan 1 Fi Forbearance cost of loan i R1 Risk assumption cost of loan 1 Based on the assumed cost data, the acquisition cost of loan (1) (or any other loan) is $11.90. The servicing cost is the number of paya ment periods during which loan (1) is to be outstanding times the monthly servicing cost of $0.91. The forbearance and risk assumption cost fac- tors must be applied to the average balance outstanding on loan (i) to yield the forbearance and risk assumption cost elements. An approxima- tion.of the average amount outstanding on an instalment loan is given by: X = L N + 1 i i 2N Average balance outstanding on loan (1) Initial amount of loan (1) Zara?“ Number of periods to payout for loan (i) 69 On the basis of the above, the cost of a $1,000 loan with a 24 months maturity would be determined as follows: C = 11.90 + .9l(2h) + (.015 + .05) (1000 2“ + 1) i 2 x 25 Ci = $62.60 The unit costs of loans of varying sizes and maturities would be developed by substituting the appropriate size and maturity variable in the above formula. It should be noted that in the above formulation, which is typical of the various studies reviewed by the writer, acquisition and servicing costs are implicitly assumed to be unaffected by loan size. In addition, the use of the same cost factor for risk regardless of loan size implies that risk is proportional to size. That is, bad debt experience is the same for loans of $100 as for loans of $1,000. The available evidence on bad debt experience by loan size is inconclusive and somewhat dated.6 Bad debt experience by loan size may well vary among different lenders because of differences in lending policy. Further discussion of this point is deferred to chapter four. It should be realized, however, that to the extent that the acquisition, servicing, and risk assumption func- tions are affected by loan size the foregoing methodology yields some- ,what questionable results. If the effect of size is slight, the fore- going procedure may give sufficiently reliable results. 6See W. D. Robbins, nsumer Instalment loans (Columbus, Ohio: Bureau of Business Research, OSU, 1955), p. 103. David Durand, Risk Elements in Consumer Instalment Financi (New York: National Bureau of Economic Research, 19 l , pp. 9-50. 7O {ggeakeven.loan size annlication In most of the studies reviewed, the step taken after the cost infor- mation is developed for each of the functions (acquisition, servicing, risk assumption, and forbearance) is to develop breakeven loan sizes. As with any breakeven construct, breakeven loan size occurs when costs and revenues are equated. Given a particular rate or rates of finance charge, the breakeven loans for different maturities could be developed through successive approximations. Where a single add-on or discount rate prevails, the breakeven loan size can be determined with a break- even formula. The following breakeven loan size is based upon the estimates of acquisition, servicing, forbearance, and risk assumption costs presented on the preceding page. In addition, an add-on finance rate of 8 per cent per annum is assumed with a 24 month maturity. The breakeven amount (A) is that amount of cash disbursed by the lender to the borrower. It is not the face of the note (A'). Formula (1) is used to determine the breakeven loan (A). Formula (2) yields the face of the note (A').7 (1) A = 2m éCa + nCs) n - C + Cr n + 1 (2) A' = A (1+in/m) A = Breakeven loan - cash invested by the borrower A' Face amount of breakeven loan T,— 7For a further development-of these formulas, see M. R. Neifeld, Neifeld's Manual on Consumer Credit (Easton, Pennsylvania: Mack Pub- lishing Company, 1961), Chapter 23. Ca = Acquisition cost Cs = Servicing cost per month or per payment Cf Forbearance or money cost Cr = Risk assumption cost i = Annual rate of finance charge n = Number of payments in the contract m = Number of payment periods in a year 24 (11.90 + 24 x .91) 2 (.08 x 24) - .015 + .05 (24 + 1) E A' = 24 O E Gross revenue = A' - A 3 $424.07 - $365.58 = $58.49 The breakeven loan is defined as that loan for which costs and reve- nues are just equated. A check on the breakeven loan determined above is provided below: Check: Gro ss Revenue $58 . 49 Total Costs: Acquisition $11.90 Servicing 21.84 .Ri 8k 5 o 72 **Forbearance 12,03 §§.42 Profit - o - *Risk assumption cost is determined by: Cr x A n +’1 x 2 2n **Forbearance cost is determined by: Cf x A n + 1 x 2 2n 72 Based upon the above cost formulation, loan profitability is a function of size and.maturity. Large loans and loans with longer maturi- ties are more profitable than smaller loans and loans with short maturi- ties. This is illustrated in Tables 3.1 and 3.2. The foregoing illustration of the cost accounting procedure used in developing loan unit costs is typical of the studies reviewed. Some are based upon detailed time studies, however, the outline of the metho- dology remains the same. Costs initially recorded on an‘nnjggn basis are reclassified according to their relationship to the different func- tional bases. The functional costs are then assigned to the unit of product (loan) as illustrated in the preceding pages. A question which can be asked at this point is what are the uses and limitations of cost information such as the foregoing? Uses and limitations The unit cost information developed by the above procedures must be interpreted and used with care. It should be recognized, for example, that the unit cost figures represent the ultimate in the application of "full absorption" cost accounting. ‘All costs have been assigned to the product. This procedure may yield a good estimate of the.§2§él cost of a loan, however, it creates certain problems in employing the data as a basis for decision making. An example of the above difficulties is provided when one observes that Table 3.1 indicates that loans of less than $500 with 12 month maturities are unprofitable. Should management act on this information by establishing a poliey that no loans of'less than $500 in the 12 month 73 TABLE 3.1 LOAN PROFITABILITY BY SIZE 12 Monthfnoan Item $100 $500 $1000 $2000 Gross revenue 8.00 40.00 80.00 160.00 Costs: Acquisition 11.90 11.90 11.90 11.90 Servicing 10.92 10.92 10.92 10.92 Risk .81 4.05 8.10 16.20 Forbearance 2.2; l}.55 2 .10 .20 Total costs 26,54 40,42 .02 25.22 Profit (loss) (18.34) (0.42) 21.98 66.78 Note: Computed with cost estimates developed on prior pages. Calcula- tions employ a 8 per cent add on finance rate. TABLE 3.2 LOAN PROFITABILITY BY MATURITY $5OOLgan Item 12 mos 24 mos 36 mos Gross revenue 40.00 80.00 120.00 Costs: Acquisition 11.90 11.90 11.90 Servicing 10.92 21.84 32.76 Risk 4.05 3.91 3.85 Forbearance 15,55 15,02 12,85 Total costs 40.42 50,62 61.56 Profit (loss) (0.42) 29.33 58.64 Note: Computed with cost estimates developed on prior pages. Calcula- tions employ an 8 per cent add-on finance rate. 74 maturity class be accepted? The answer may well be no for two important reasons: (1) On a short-run basis the added costs of accepting a 12 month loan of $500 are much less than the calculated average unit cost which has been developed using a form of the "full absorption" costing technique. Therefore, on a marginal cost-marginal revenue basis, making the loan is probably consistent with profit maximization tenets--marginal cost is less than marginal revenue. (2) The breakeven and other loan profitability formulations reviewed contain an implicit assumption that the loan under consideration will be the totality of the relationship of the borrower and the lender. Evidence on the operations of consumer finance companies shows that a large percentage (varying up to 70 per cent) of loans represent the extension of additional money to present borrowers. On the additional extension, the acquisition function will entail a negli- gible amount of effort. In a sense, the cost of the initial acquisition effort has been extended over to the additional financing. A basic tenet of proper income determination is that costs are matched with the revenues which they generate.8 The initial acquisition cost contributes to the generation of revenues beyond those of the initial loan when refinancing with the extension of additional money is involved. Since a majority of the loans extended involve subsequent refinancing, it is clearly inappropriate to treat the initial acquisition cost as an expense applicable in whole to the revenue of the initial extension. 8 For an authoritative discussion of this point see chapter 5 of hh A. Paton and A. C. Littleton, Introduction to Co rate Accounti Standards (Ann Arbor: American Accounting Association, 1940). 75 As Paton and Littleton observed:9 Any type of cost may be 'deferred' if it originates in a justifiable expenditure and represents a factor from which a future benefit or contribution can reasonably be anticipated. The above points highlight the fact that care must be exercised in acting upon unit cost developed with the techniques described on the foregoing pages. In addition, it suggests that a basic reformulation of the unit cost calculation is required in view of the high proportion of refinancing characteristic of consumer finance operations. The basic modification required would be to develop account mortality data which could be used as a basis for distributing the initial acquisition cost over the expected average number of loan contracts.10 The effect of this adjustment will be to reduce the loan unit cost and to lower the breakeven loan size as otherwise calculated. With respect to the issue of the assessment of acquisition costs, it is interesting to observe the affinity between the proper treatment from both the theoretical accounting and practical decisionpmaking viewpoints. Relation to the nresent stugy The review of the essentials of the cost accounting studies reveals that an implicit assumption is made concerning the effect of loan size on.lending costs--exclusive of forbearance and risk assumption costs. 91bid., p. 65. 10For a related application of this concept, see Allan Richard Drebin, Savings Accounts and Commercial Bank Earnings (Ann Arbor: Bureau of Business Research, Graduate School of Business, University of Michi- gan, 1963): P0 36. 76 It is assumed that loan size does‘nnn affect lending costs. The present study will allow an assessment of this assumption. Unlike a cost ac- counting approach, the method used.here to measure the size-cost rela- tion will entail no allocative techniques. The cost-size relation will be determined through the application of statistical procedures. QEher Stngnes In addition to the cost accounting studies, several other more gen. eral industry studies have been.made. The three reviewed here were all doctoral dissertations. 11 Joel Dean In his 1936 doctoral dissertation on the behavior of average and marginal cost, Joel Dean included a study of the long-period average and marginal cost curves of consumer finance offices. The data for the study were derived from the cost and other operating reports of 150 branch offices of a consumer finance chain. New offices (those operating for fewer than 12 months) and offices which had grown rapidly during the year under study were eliminated "on the ground that their cost figures were seriously distorted by short-run adjustments."12 This adjustment was viewed necessary in order to make the annual cost data reflect mainly long-period cost behavior. One of the principal arguments supporting the position that the annual cost data would reflect long-period cost behavior rested on the L llJoel P. Dean, «1 Statistical. Eboamination of the Behavior of Aver- age and Marginal Cost," University of Chicago unpublished dissertation, 1935- 12mm. , p. 363. 77 l flexibility of cost inputs with changes in plant scale: 3 One reason for believing that this finance com- pany exemplifies long run cost behavior is that the fixity of output exceeds the period of fixity of the cost components. Loans, which represent the output of this corporation, are made for a period of 20 months. Hence this output is flexr ible (downward) only over a rather long period of time. . . In contrast to the sluggish adjustments in the number of open accounts, most items of expense appear to be adjusted fairly quickly. As a consequence, the scale of plant, except rented space, can be accommodated relatively easily to the remarkably stable rate of opera- tion. Given the flexibility of cost inputs and the fact that new offices and offices which had undergone rapid growth were excluded from the study, it was reasoned that the cost schedules developed from the remaining 143 mature branches would represent long-period cost scheduleszli+ By regarding each of these branch offices as a snapshot of a typical establishment as it increases its scale of plant, curves of long- period marginal and average cost may be drawn up from simultaneous observation of offices representing a range of plant scale. . . Included in the costs studied were only those which could be clas- sified as direct branch expenses. Costs included were rent, salaries, local taxes, depreciation, etc. Home office supervision, federal taxes, interest and advertising were excluded on the grounds that their particu- lar level at a given branch office was the result of somewhat arbitrary accounting allocations. 15 131b1d., p. 362—363. 14 Ibid., p. 361. 15mm. , p. 365. 78 Methodology Dean employed graphic multiple correlation analysis in developing his average and marginal cost curves. The dependent variable (Y) in this analysis is the average cost per open account. Marginal cost is the increase in cost associated with an increase of one account. Five independent variables were used in the multiple correlation analysis: X2 Size of branch office, expressed as the total number of open accounts X Delinquency. represented by the ratio of delinquent accounts to total open accounts X Size of city in which branch is located X New business, represented by the number of loans to 5 new plus former customers, expressed as a percentage of total open accounts X3 Renewals, represented by the number of renewed loans made during the year, reduced to a percentage of total open accounts The primary measurement which is desired from the above is the rela- tionship between the dependent variable and the measure of office scale-- XE total number of open accounts. The delinquency measure is introduced because it is believed that more delinquency in a branch, other things equal, would be reflected in higher costs due to the control effort re- quired. The size of city (Kn) is introduced as a proxy for a factor prices variable. The measure of new business (X5) is introduced into the analysis "to allow for the short-run departures from.squilibrium of the mature offices. . ."16 An index of renewals (XE) is introduced in the.belief that a high per cent of renewals in an office would entail 161mm , p. 363. 79 added personnel effort and be reflected in higher costs. The use of this variable may well duplicate the delinquency measure (X3) since the level of renewals and the level of delinquency are no doubt highly cor- related 0 Findings The principal result of Dean's analysis was to reveal the classical downward sloping long-period average cost curve. The general shape of this curve is presented in Figure 3.1 below: Figure 3.1 NUMBER OF ACCOUNTS-UNIT COST RELATIONSHIP Unit Cost 1000 1500 2000 2500 3000 Number of Accounts Source: Based upon Dean's results, p. 367. 80 The usual interpretation of a declining long-period (run) average cost curve is that it implies the existence of economies of scale. The forces causing the LAC‘ZIong run average cost curv§7 curve to decrease as the firm increases its output and scale of plant are called 'economies of scale.‘ In particular, the economies are said to derive from increased possibili- ties for division and specialization of labor as the scale of plant is increased. Also, the possibility of employing advanced technological developments and/or larger machines is usually cited.18 Dean's own interpretation of his results is somewhat at variance from that which would be suggested by the above statements. In addressing himself to the reason for the downward slope in the average cost curve, Dean notes that the "causes for net internal economies of large scale production, such as appear to be present here, are little discussed by economists. . ."19 He goes on to suggest that factor divisibilit , rather than the above cited factors, provide the explanation for the observed scale economies:20 A common, and practically important, expla- nation is in terms of limited divisibility of the factors of production. This precludes the possibility of attaining the same advantageous proportioning of the factors at a small scale as is achieved at a large scale. In this firm, for example, the indivisibility of a human being may be an important factor. . . 17Richard H. Leftwich, e Price S stem a Resource Allocation (New York: Holt, Rinehart and Winston, 1961), p. 156. 181b1g., p. 156. 19Dean, on,cit., p. 368. 20Ibid., p. 368. 81 In addition to the downward sloping average unit cost schedule, the study revealed the expected relationships between average unit cost and the other independent variables. A high level of delinquency in a branch (X3) was associated with higher costs. The same was true for the level of renewal loans in a branch (X6). Both of these variables reflect delinquency level in a branch. Delinquency control requires added effort which should be reflected in turn in higher costs, other things equal. The relationship between size of city (Kn) and unit cost was also positive. This reflects the fact that "Branch offices operated in large cities have higher average costs than those in small cities. . . because of higher rentals and salary rates.21 The relation between unit cost and the measure of new business (X5) is negative: the greater the proportion of new loans in the business, the lower the average unit costs. Dean provides no satisfactory explanation for this relationship. Moreover, the direction of the relationship (negative) is contrary to the relationship which this writer would expect (see page 78). Relation to the nresent stung The present study includes some of the same relationships which Dean measured in his 1936 dissertation. Therefore, it will be of value to compare the behavioral relationship between average unit cost and the scale of office (as measured by the munber of accounts) revealed by Dean's study with that of the present writer's study. The writer's study should provide a better test of this particular relationship since the definition of a mature office is much more realistic (one year in 21am. , p. 369. 82 Dean's study versus four years in the present study). It could be argued that much of the initial decline in Dean's average cost schedule is simply the result of including branches in the study which are still in the proc- ess of adjusting their output to the initial factor input. Also, the present study will provide a check on the negative relation between unit cost and new business (X5) revealed in Dean's study. 22 Miller Upton Miller Upton's dissertation was entitled "The Importance of Direct Costs in the Granting of Consumer Instalment Credit." As the title imp plies, the area investigated extended beyond personal instalment cash loans to include retail instalment credit. The objectives of Upton's study are reflected best in the following two statements: (1) What is sought is an understanding of the relative importance of direct and indirect COStS .23 (2) What is desired in the present study. . . is a knowledge of the relative importance of the various expense items and their adaptability to changes in business volume.2 Upton derived the data for his study from the published reports of the state of Illinois on its licensees. These reports present only aggre- gative data; no data on individual operating units is presented. The time period covered is 1930 to 1946. While it is difficult to determine just 22Miller Upton, "The Importance of Direct Costs in the Granting of Consumer Instalment Credit," Northwestern university unpublished dis- sertation, June, 1948. 2311213., p. 1. zuij-do’ P. 270 83 how the results of the study might be affected, it is worth noting that the period 1930 to 1946 included our greatest depression and our biggest war. Since the times were not typical, it might well affect amt pro- jections that one might make based upon Upton's particular data. Methodology Unlike the preceding study by Dean and the following study by Farwell, Upton did not rely primarily upon statistical methodology in his investigation. His approach included (1) a careful study of the operating characteristics of instalment lending agencies, and (2) tabu- lar and graphical presentation of various expense items and expense groupings and their behavior with changes in business volume. Findings Early in his study, and after a very thorough exposition on the character of the lending activity, Upton concluded that: "All things considered it would seem that the direct cost items are minor from a value standpoint."25 Direct costs in this particular context refers to those costs which can be related directly. and without allocation, to the unit of output--to the loan or instalment sale contract. Upton's concept of direct costs includes those costs which are incurred because a loan is made. This view is reflected in the following remarks which include an itemization of the likelyngingnn.ng§§§:26 There are very few direct costs involved in the granting of an.individual loan. Most unit costs 25_me., p. 56. 26 . . Ibido, p. 217-218. are necessarily determined and allocated on a functional basis rather than a per loan basis. The few expense items that may be looked upon as direct from.the standpoint of the unit of loan granted are supplies, postage and express, recording and acknowledgement fees, some travel expenses, and some legal fees and disbursements. The above conclusion provides the answer to Upton's first dissertation objective of determining the relative importance of direct costs. The second objective involved a determination of the relative importance of various expense items and their adaptability to changes in business vol- ume. The finding in this regard was that salaries are the principal operating expense followed by bad debts, advertising, and rent in the order given. In addition, it was determined that over the seventeen- year period studied there were only minor changes in the importance of the major operating expense items when expressed as a percentage of average loan balance or in terms of a unit of loan granted.27 Upton also concluded that the principal factor inputs (labor and.money) are quite adaptable to changes in business volume. combining the foregoing observations with the position that (1) a unit of adequate efficiency may be attained relatively easily because of the limited application of mass production techniques, and (2) that the basic productive components--labor and capital--are highly mobil, Upton concludes that "the industry would seem to possess the attributes of one of relatively constant unitcosts."28 27113191.. p. 218. game. , p. 219. 85 The position on constant unit costs is applied to an excellent analysis of the "convenience and advantage" licensing requirement. Among others, the convenience and advantage licensing provision is sup- ported by a hypothesis concerning unit cost behavior in consumer finance branches. Basically. the restriction on the number of licensees in a particular community is justified on the basis that the existence of too many units would prevent the units from attaining efficient size. Efficiency is measured in terms of unit cost levels. Upton takes the position that scale economies are slight and that the cost behavior justification for convenience and advantage licensing is unsound.29 Instead of the business possessing decreasing cost tendencies, its cost elements and behavior are much.more in keeping with constant cost conditions. Therefore, limitation of the number of firms does not result in.lower unit cost. . . Relation to the nresent stugy Upton's study is one of only a very few which attempt to deal with the cost behavior of consumer finance companies. Upton's study is based primarily upon aggregative reports of licensees in one state. One of Upton's important conclusions concerns the behavior of average unit cost as the scale of operation is expanded. His position is that the character of the lending function does not lend itself to significant opportunities for economies from the specialization of labor. In addition, most of the major cost inputs are shown to be very adjustable to changes in volume. Hence, a conclusion of relatively constant unit costs is reached. The present study should provide additional evidence on this important issue. 29Ibid. , p. 233. 86 30 Loring Earwell Farwell's dissertation, completed in 1955, parallels in certain respects and extends in others, the study by Miller Upton. Like Upton, Farwell draws upon the aggregative industry reports of the state regu- latory bodies. Farwell uses the data for Indiana whereas Upton used Illinois reports. The Farwell study was based upon the aggregative reports of the Indiana Department of Financial Institutions for the years 1933, and 1935 to 1950 inclusive. Farwell outlined the basic objective of his study as follows:31 The objective of this study is measurement of association between the costs of doing business in the field of consumer instalment credit and the output of credit service. In somewhat different words, the purpose is estimation of the average long-run unit costs of small loan and time sale credit. 'Methodology The basic methodology employed in the study is statistical--multiple regression and correlation analysis. Farwell developed a cost behavior hypothesis on the basis of a review of prior cost accounting studies. The cost studied are confined primarily to operating costs which include all costs with the exception of interest, bad debts and federal income taxes. .It was observed that:32 _ 30Loring Chapman Farwell, "Cost and Output Relationships of Con- sumer Instalment Credit Agencies: Based on Reports of Indiana Licensees," Northwestern University unpublished dissertation, 1955. 31Ibid., p. 1 and p. 53 respectively. 32Ibid., p. 158. 87 An hypothesis, already accepted for cost accounting purposes. . . shows direct association of this cost.[3perating coq§7 with number, size and duration of loans, i.e., the per contract‘zchuisition7, per dollar.[;isk and forbearancé7, and per payment‘zgervicing7 functions. This is not a completely accurate interpretation of the cost account- ing studies. Farwell's definition of onerating122§§ excludes money costs and the bad debt cost as well as income taxes. The costs related to loan size in the cost accounting studies typically include only money (for- bearance) and bad debt (risk assumption) costs. The cost accounting hypothesis is later translated into an algebraic formulation:33 (1) C = R(a + bT + cS') C Total operating cost R Number of loans granted within a year T Average duration 3' Average outstanding balance per loan a Per contract costs b Per payment cost c For dollar costs Since operating costs (C) excludes money costs and bad debts, (cS') :nust represent the operating costs incurred on account of differences in Iloan.size. On this point, however, Farwell notes that in the small loan :industry "it is believed that no more effort or material is expended on Ilarge than on small loans within the range of size characteristic of the 33Ib1d., p. 158. *“I_ v. 7" E I 88 "34 small loan industry. As a consequence, the cost equation is reduced by the elimination of (cS') to: (2) C = R(a + bT) The cost and other data are also adjusted to accommodate the effect of changing price levels over the period studied. The operating cost series was adjusted by dividing through with salary cost ratios. The index of duration (T) was also modified to represent the ”average loan 35 '0 service per loan granted. . . T' = Average number of loans outstanding guring the ynar Number of loans granted during the year With the above adjustment, the average operating cost per loan granted, in constant dollars, becomes: (3) 0/3 =0" =a+bT' This formulation is further refined to accommodate the short-run influence of deviations of the number and average size of’loans granted from their averages for the period studied.36 (4) 0" = a +‘bT' + cR* +-dS"I R"I Deviation of number of loans granted from the average S“I Deviation in.average loan size from the average “313,, p. 166. 35Ibid., p. 168—169. Ibid.. p. 187. 89 Findings Employing the foregoing variables (T', R*, and 8*) in a multiple regression and correlation routine, yielded the following regression equation:37 (5) c" = 2.11 + 12.531" .. 0.005523: + 0.04543. The interpretation given the elements in the above equation are the following: The regression constant, $2.11, represents the operating cost per loan granted. It is equivalent to acquisition cost per loan. $12.53 is twelve times the perupayment cost. That is, if T' were equal to one (the average loan duration being one year), then the average servicing cost per loan would be $12.53. The coefficient of R*, -0.00552, is the adjustment in unit cost attributable to each loan granted above or below the average per licensee for the period studied. The coefficient of S‘, +0.0454, is the adjustment in operating cost per unit for each dollar loaned above or below the average for the period. The multiple correlation coefficient achieved with the above formulation is R = .993. Some statistical and internretative difficulties While it cannot be denied that Farwell's statistical analysis yields some impressive multiple correlation coefficients, the interpretation provided seems somewhat debatable. During the course of his analysis, Farwell notes that loan size can be eliminated.from the analysis on two separate counts: (1) Loan size is associated with the number of'loans 37 . Ibig., p. 187. granted and, therefore, on statistical grounds, either variable could be used to represent move- ments in both.38 (2) Loan size is not an indepengent variable in de- termining onerating costs. In an effort to analyze Farwell's results, this writer developed an additional regression employing Farwell's basic series of data on number of loans granted, average loan size, average duration, and average operating cost per account. The particular multiple regression and correlation program employed yielded the simple correlation coefficients on all comp binations of the above variables. The association between the number of loans granted and average loan size was reflected in a correlation co- efficient of +.75. Given that such a high correlation exists among two independent variables being proposed for use in a multiple regressionp correlation formulation, one must be sensitive of the interpretative problem which the condition.of'multicollinearity creates. Multicol- 1inearity is the condition of high correlation among independent vari- ables. It presents serious problems in determining the separate influ- ence of the intercorrelated independent variables upon the dependent variable. Johnston has explained the concept as follows?"0 This‘lfiulticollinearity7 is the name given to the general problem which arises when some or all of the explanatory variables in a relation are so highly correlated one with another that 3823.59... p. 166. 39Ibid., p. 166-167. In reality, the substitution would be appro- priate only if number and size were perfectly correlated. The degree of association between loan size and number of loans granted is reflected in a correlation of +.75, calculated by the writer. “OJ. Johnston, Econometric Methods (New York: McGrawaHill, 1963), p. 2010 91 it becomes very difficult, if not impossible, to disentangle their separate influences and obtain a reasonably precise estimate of their relative effects. Brennan further notes that:Ul Though the regression equation remains valid for prediction of y from both x and z together, the effect of a change in x (or z) upon y when z (or x) is conceptually held constant cannot be determined. In the regression formulation at the top of page 89, Farwell in- cludes R* and 8‘. These are respectively the deviations of loans granted and average loan size from the average for the period studied. Since number of loans granted and average loan size are highly intercorrelated, a series based upon deviations from such an average would likewise be highly intercorrelated. As noted, the multiple correlation coefficient yielded with equation (4) is .993. This shows an extremely high correlation between the ob- served values for cost per account (C") and the predictor variables (T', R*, and 8*). Based on the foregoing R, R? (the coefficient of determination) would be .986. The usual interpretation of this statistic is that it represents the percentage of the variance in the dependent variable which is explained by_the independent variable_(s).l+l2 In view of the high intercorrelation between number of loans granted (R) and average loan size (S), and hence among (R*) and (8*), it appears that Farwell errs in attempting to ascribe some individual effect on the dependent variable (0"), to (R*) and (8*). The statement that "in the alMichael J. Brennan, Prefa e o conometrics (Cincinnati: South- Western Publishing Co., 1960 , p. 3 1. “ever a discussion of R? see Brennan,,22;£$£or Po 333' 92 short run the cost per loan is increased (in constant dollars) $0.00614 for each loan below an average granted. . .and $0.044O for each dollar loaned above the constant dollar average. . ." is not justified statisti- cally.43 In addition to the interpretative problem which centers around the intercorrelation of the independent variables, Farwell's interpretation of the value of the regression constant in equation (5) is debatable. In translating the cost accounting studies into an algebraic cost be- havior, Farwell presented equation (1) in which (a) was representative of ”per-contract" costs. (1) C = R(a + bT + cS') Equation (1) is a perfectly legitimate algebraic formulation of the cost behavior hypothesis presented in the cost accounting type studies. The questionable interpretation concerns the issue of whether or not the constant term in Farwell's multiple regression equation is equivalent to (a) in equation (1) above or (a) in.aquation (4). An illustration of difficulty in this regard is found when the constant term is developed while using a slightly different regression formulation. Employing Farwell's basic detailed data (i.e., the basic data and not the deviations from.the averages) on number of loans granted, average loan size, average duration, and average operating cost per loan (C", R, S', and T') in a multiple regression-correlation routine yielded the following regression equation, i.e. , (6): 43 Farwell, on,cit., p. 196. 93 (6) cM -- 6.0525 + 9.3128?! - .00001379R + .01217985' 3 - .96 (5) cH - 2.11 +12.53T' - ,oo5szne + .045“. R = .99 The predictive value of (6) is quite similar to that of (5). A consistent interpretation of equation (6) would be that the per contract cost would be $6.05. This is contrasted to a per contract cost of $2.11 determined in equation (5). This example seems to illustrate the hazard inherent in the interpretation of the regression constant term as being equivalent to per-contract costs. Relation to the nresent stugy Farwell's study used a time series approach based upon aggregative industry reports for a single state. The writer's study uses a cross sectional approach based upon the branch operating data of a major cone sumer finance company. The two studies are similar in some respects. Both use a cost per account as the basic unit cost measure. In addition, the studies each employ a measure of number and size of accounts in their cost behavior hypotheses. It will, therefore, be of significant corrobo- rative value to compare the behavioral relations determined in the present Study with these common predictive variables. In addition to the corraborative value of Farwell's study, it also provides considerable insight into the general problem of developing statistical measures of cost behavior. CHAPTER IV EMPIRICAL RESULTS The Qnta and the Methodology The data The basic empirical content of this study is drawn from the cost and other operating reports of almost four hundred branch offices of a major consumer finance company. These branch offices were distri- buted among four different states. Only relatively mature branches were included in the study. The basic cutoff point used as an approximate measure of maturity was four years. All branches included in the study have been in operation for four or more years. The reader should contrast the maturity measure employed in this study with that used in Joel Dean's study of cost behavior in consumer finance branches. Dean excluded only branches which had been in opera- tion for less than one year. This writer's research shows that offices which have been in operation for only one or two years have uniformly high average unit costs. This reflects the fact that it takes some time to generate a volume of lending activity which is reasonably con- sistent with the initial factor input. Including immature branches in a study'of cost behavior will intro- duce a bias which would produce results suggestive of substantial econo- mies of scale. The average unit cost schedule developed from data which 94 95 includes immature branches will show a greater initial slope than the same schedule developed from data which does exclude immature branches. In addition to branches in operation for less than four years, certain other branches were excluded because they were used for person. nel training activities. All other branches located in the four states were included in the study. Since the cost and other operating data are drawn from a single company, they have a uniformity and homogenity which would be difficult, if not impossible, to attain if data from a number of different comp panies were used. Methodology The basic methodology employed in this part of the study was out- lined briefly in chapter one. It is restated and expanded somewhat in the next several pages. In addition, the manner in which measures of the various independent variables were developed is outlined. The main analytical device used in the empirical portion of this study is multiple regression and correlation analysis. The various statistical calculations were performed with the assistance of the Michi- gan State University Control Data 3600 computer. The multiple regression and correlation program used (AES Four Core) was developed by D. F. Kiel and W. L. Ruble of the Michigan State University Agricultural Experiment Station. The multiple regression and correlation routine is used to test the cost behavior hypothesis presented in chapter one. This particular hypo- thesis was developed on the basis of a thorough search of the available 96 literature and discussions with members of the Consumer Finance industry. From a methodological standpoint, the development and testing of the cost behavior hypothesis is crucial to the basic dissertation objective of providing a measure of the effect of loan size upon lending costs. An alternative approach would be to develop a simple regression of loan size upon lending costs. Such an approach is clearly unsatisfactory since it ignores the influence and impact of other important variables upon the relationship measured. The basic cost behavior hypothesis is restated below, followed by an.outline of’the manner in.which.measures of the various independent variables was achieved. It should be noted that the following specifi- cation of the cost behavior hypothesis differs from that presented in chapter one; I has been substituted for I. u, the disturbance term, has also been added. These items are also explained following the specification of the cost behavior hypothesis in regression equation form: (1)Y=A+EC1+CXZ+DX3+EXu+FX5+GX6+u Direct branch operating cost per account Regression constant = Average loan size Average number of accounts outstanding Delinquency'measure Factor prices index excref‘r?‘ p *4 II “N I Loan.mix X6 = New application acceptance rate B-G 8 Net regression coefficients u 8 Disturbance term The development of the measures on the above variables is outlined below: Y This is the dependent variable which it is hypothesized can E to X6. Y represents a measure of average unit cost. It is developed by dividing the average number of open accounts during the year into direct branch operating costs. Direct branch operating cost includes basically all locally incurred and paid expense items. It excludes certain allocations of home office cost elements. In addition, it does not include interest costs, federal and state-local taxes, or bad debt charge-offs. be predicted or enplained with the independent variables X: I hath the above adjustments, it is believed that most arbitrary influences upon the behavioral cost relationship which is to be measured have been eliminated. This is the measure of the average size loan outstanding. It is developed by dividing the average number of open accounts into the average dollar amount of loans outstanding. This variable is a monthly average of the number of open accounts or loans outstanding. This variable is a measure of the level of delinquency in each of the individual branches. It is employed in two dife ferent forms in the statistical analysis. The measure which produces the best "fit" is adopted. (1) The first delinquency measure is the charge-off rate for each of the individual branches for the year under study. The measure is developed by dividing the dollar amount of loans charged off by the average amount of loans outstanding during the year. (2) The alternative delinquency measure is the dollar amount of doubt- ful accounts outstanding expressed as a percentage of the average amount of loans outstanding during the year. It is possible that differences in factor price levels within a state will account for some of the differences in unit cost levels among branches located in a given state. This variable, 98 X , represents a factor price index designed to accom- modate this possibility. The price index developed is based upon differences in wage levels authorized by the company studied for branches at different locations within the state(s). Because of the importance of perb sonnel costs among branch operating costs, and in view of the likely close relationship between the level of labor and most other factor prices in a particular area, the differences in authorized wage levels were used as the basis for a general factor price index. In developing the factor price index the lowest level of salary scale authorized was assigned the value of 100. Higher salary levels were expressed as a percentage of the base value to form the index. Level one 100 Level two 106 Level three 113 Level four 121 This is the measure of loan.mix. It reflects the importance of loans to new'borrowers among the total number of loans made during the year. It was developed by dividing the number of loans made during the year to new borrowers by the total number of loans made. This is an index of the rate of acceptance of n23 loan applications. It was developed by dividing the number of new loan applications for the year which were granted by the total number of new loan applications received. This element in the foregoing regression formulation is generally referred to as the disturbance term.1 The dis- turbance term may take on either positive or negative values. It reflects the fact that the behavioral hypo- thesis cannot form a complete specification of all vari- ables which influence or which determine the value of I. In the present case it is apparent that differences in managerial skill at the branch level will influence cost levels. However, no feasible way of developing a useful measure of managerial skill is available and hence this variable is not included. Its impact will, however, be reflected in the disturbance term (u), as will other factors which affect costs but which are not included in the behavioral hypothesis. lFbr further discussion of the disturbance term see J. Johnston, Econometric Methcgs (New York: McGraw-Hill, 1963), pp. 5—7 and pp. 107- 99 The above regression formulation is designed primarily to facilitate the measurement of the effect of loan size and scale of operation upon the level of direct branch operating costs per account. An estimate of the influence of loan size upon the cost of risk assumption, as measured by the'nnpg of loans charged-off, is also desired. Such a measurement will make possible a more general assessment of the effect of loan size upon lending costs. The cost accounting studies reviewed in chapter three were based upon a functional cost classification which the writer symbolized as follows: C1 = F(Ai + 31 + F, + R1) The cost behavior hypothesis represented by equation (1) is designed to isolate primarily the effect of loan size upon acquisition and related servicing activities-.. A, + 8,. Forbearance cost (F1) is unaffected by loan size. That is, it will be proportional to loan size. The following regression will provide an estimate of the effect of loan size upon the cost of risk assumption (R1), as reflected by the charge-off rate: (2) Z = H + 1x, +lu 8 The charge-off rate 8 Regression constant HEN 8 Regression coefficient Average loan size x1 The regressionsrepresented by equations one and two were developed for four different cross sections. The cross sections were made up of 100 the branches in each of the four different states. This division of branches into four cross sections on the basis of state lines is desir- able for the following reasons: (1) Developing four cross sections and handling the statistical calculations separately make it possible to check for consistency in the statistical parameters among the four cross sections.2 (2) Achieving meaningful results could not be assured by merging cost and other operating data of branches located in different states. Regulatory provisions vary from state to state and manifest themselves in operating results. Merging the cost and other operating data from the four states would introduce additional variation which would likely distort rather than reveal important relationships. In the absence of transformations of the independent variables, the multiple regressionpcorrelation routine develops least squares‘lingnn regression relationships. To test the appropriateness of a linearity assumption, scatter graphs of the relationship of the various independent variables to the dependent variable were prepared. Only in the case of the relationship between I and X2 (average accounts outstanding) did a nonlinear assumption appear to be called for. In order to accommodate this nonlinear relationship, the regressions were developed for all four cross sections with X: in three different forms: (1) simple arithmetic form, (2) asea logarithmic transformation, and (3) in a second degree 2The data employed in the empirical portion of this study are popu- lation values. It is conventional to refer to population characteristics as parameters and to sample characteristics as statistics. 101 exponential form.3 In all cases the nonlinear fits produced significant improvements in the predictive power of the regression, as reflected by the multiple correlation coefficients of the four cross sections. The logarithmic and exponential transformations modify the specification of the cost behavior hypothesis, in its regression equation form, from that reflected in equation one: Logarithmic (3)Y=A+BX1+ClogX +Dx 2 3+EX1++FX 5 + GX6 +11 gnpnnential (4)Y=A+Bx1+cx2+nx:+Ex3+qu+Gx5+1n6+u In equation three, Cth2 is substituted for CX2. In equation four, 0X2 + UK: is substituted for C12. General Empirical Results The basic results achieved with the cost behavior hypothesis are pre- sented in Table 4.1. The regression coefficients are followed by their standard errors in.parentheses. Both R (the multiple correlation coefe ficient) and R2 (the coefficiert of determination) are included and have their usual interpretation. R is the coefficient of correlation between .the observed values of.Y'and the values predicted by the.multiple regression 3For a discussion of the mathematical specification of different functional relationships, such as are represented in the logarithmic . and exponential transformations above, see Mordecai Ezekiel and Karl A. Fox, Methods of Correlation nd Re ression Anal sis (New York: John Wiley & Sons, 1959), pp. 70-74. Amaoaum.nueuaeumv muoodOfiumooo nonmmoamom sopH Aboao.ov Ambmo.ov Aomma.ov Asmmo.ov m memo.o mmm~.o onmm.o mamm.o was seen x Amoo.ov Asmo.ov Aamo.ov Ammo.ov a made made some need noose coca serene x nons.ov Arms.ov “Hmoa.ov Aamm.ov . m Hoo.H HmN.H wron.o mam.a oneneee.aososesaaon x $893 Nomm.mu measoooe Ho Acne—«E omeage Ho m3 NM . a. Amsoooooo.ov Aesoooooo.ov Aaroooooo.ov l. m“ maoooooo.o momooooo.o oamooooo.o senses. nauseous no genes: oweaoa< WM Admoo.ov AoNoo.ov Ammoo.ov wsfiusepmaso oooo.o- Hoao.o- memo.o- nessooos co senses oneness Na Annoo.ov hamootov Aemootov Amoco.ov $00.0 H.840 {No.0 38.0 mean 83..” amenable. HM a o m 4 . mgHQZHh gHmHE g8 ho Hgm SHamHafium H14 as A '- 3 or R m2 assesses no noose 3.N mete. NNJ 3% 3.2593 Me .393 232% .. mm . mo . mm. .3. Aeofivafieflopoo . m. no neonates m . l 1 m mm. mm« mm. mm. Aesofioamooo sown-Backhoe oamggv m S.ma mafia madm Hmém essence scammouwom 3398 330.8 “28.8 Ammaoé QNHH.ou oumoqou mmnH.OI moma.ou oven w acquiesce soapeofiaamd. 302 M sofi “unease H.838 npsouaceooo senescence Afnoov H...“ an. 104 equation. R2 measures the percentage of variance in Y (the dependent variable) which is explained by all of the independent variables in the regression equation. Predictive value of the_nypothesis The general effectiveness of the cost behavior hypothesis is reflected best by the coefficients of determination (R2) for the several cross sections. These coefficients range from .53 for state D to .69 for state C. Therefore, the cost behavior hypothesis can be said to explain from 53 per cent to 69 per cent of the variance in cost per account (Y) among the four cross sections. For the cross sections A, B, and C, an average of 66 per cent of the variance in cost per account (I) is accounted for by the cost be- havior hypothesis. An average of 34 per cent is, therefore, left un- explained. Behavior of hyppthesis elements In chapter one the expected direction of the relationship between unit costs and the explanatory variables was presented. These antici- pated relations did develop in the regression equations for each of the four cross sections. Positive Average loan size Delinquency'measure Factor prices index Loan.mix index \3‘4rldTrj‘ Negative Xi Average number of accounts outstanding X6 New application acceptance rate The above relationships between unit costs (Y) and the independent variables (Xi to X6) are reflected in Table 4.1 by the sign, positive or negative, which attaches to the individual net regression coeffi- cients. It is significant to observe that the direction of relation- ship between the dependent variable (Y) and the independent variables (Xi to X6) is wholly consistent among the four cross sections. The Size-Cost Relationship In chapter one the major objective of this study was established as that of "determining and.measuring the behavioral relationship be- tween loan size and lending cost in the consumer finance industry.” The results presented in Table 4.1 provide both a determination and measurement of this relationship with respect to direct branch operating costs. The sign of the net regression coefficient fer average loan size (Xi) is positive for each of the cross sections. The behavioral relationship between loan size and direct branch operating costs per account is, therefore, determined to be positive. The measurement of the size-cost relationship is provided by the value of net regression coefficient of average loan size (Xi) for each of the cross sections. The net regression coefficient of average loan size (Xi) provides an estimate of the increase in unit cost (I) which accompanies each dollar of increase in average loan size in a branch. This measure of the size-cost relationship is summarized in Table 4.2. The slope of the size-cost relationships presented in Figures 4—1 to 4-4 reflect a measure of the marginal direct lending costs (MDLC). These particular two dimensional graphic relationships are developed 106 TABLE #.2 INCREASE IN UNIT COST (Y) PER ONE DOLLAR INCREASE IN AVERAGE LOAN SIZE (X1) State Unit Cost Increase A $0.02u6 B 0.0244 C 0.0151 D 0.0058 Source: Table 4.1. by holding variables (X2) to (X6) at their mean values while allowing average loan size (X1) to vary. Fbr each cross section the relationship is extended'ggly'over the range of average loan size which is present in the underlying observations.“ The slope of the relationships reflect the increase in direct branch operating costs (as they have been des- cribed in this study) due to the increase in average loan size. Marginal cost terminology is generally applied to cost increases associated with increases in the output of a fimm. It should be noted that in the present case loan size is treated as one of the measures of the output of a con- sumer finance branch. It must be emphasized that the cost-size relationships presented in Table 4.2 and graphed fer each of the cross sections in Figures 4—1 to 4-4 are based upon inter-branch comparisons.e That is, they.reflect the uFor the mean values of all variables included in the study, see Appendix A. Average Unit Cost 50 4O 30 20 10 109 Figure 4-3 LOAN SIZE AND AVERAGE UNIT COST RELATIONSHIP CROSS SECTION C 400 450 500 Average Loan Size Average Unit Cost 110 Figure 4-4 50 30 20 LOAN SIZE AND AVERAGE UNIT COST RELATIONSHIP CROSS SECTION D 10 I 7’, 550 600 650 Average Loan Size 700 750 800 ll} MDLC which is incurred, other things equal, when the average loan size in a branch is increased. It is not a‘gigggp measure of the intra- branch MDLC associated with‘lgggg of differing sizes. It is contrasted in this regard with the cost accounting studies which were designed to develop intra-branch estimates of the costs of loans of differing sizes. The MDLC represents only a partial measure of the size-cost rela- tionship. The influence of loan size upon the cost of risk assumption, as reflected by the charge-off rate, has yet to be considered. A simple regression of loan size upon charge-off rate was developed in an effort to estimate the above relationship. The results achieved with this regression are summarized in Table 4.3. Only one independent variable, average loan size (Xi), is employed in this regression. R is therefore equivalent to the simple correlation between the charge-off rate and average loan size. The indicated relationship between average loan size (X1) and the cost of risk assumption is uniformly negative fer the four cross sec- tions. The strength of this inverse relationship varies considerably between cross sections A and C (-.12 and -.l9) and cross sections B and D (-.54 and -.57). The size-risk assumption cost relationships are presented graphically in.Figures 4-5 to 4—8. The slopes of the size-risk assumption cost relationships in Figures 4—5 to 4-8 reflect an estimate of the marginal risk assumption cost (MRAC). They represent the change in the cost of risk assumption, as measured by the charge-off rate, when the average loan size is increased. As with the measure of the MDLC, it must be emphasized that the present measures are based upon inter-branch comparisons. That is, they reflect the MRAC ... me me an ace 35:93 No nonasz an. _ we. we. ea. ma. eyesaeee no eeeee eeeeeepm 9H6 0.1m “mi $.N 9:338 scammonmom 2 x “A an. . 9n - an. . NH. .. eeeeeaee8 533280 a Amaoo.ov Aowootov Audaciov Aoaoo.ov H Haoo.o- emootou mnoo.o- maoo.o- eeae need emeeesa x a o m a Amaoanm 69383.3 35.3.3908 scammoumom 50H §H§mm< Ma ho amoo may 20 mNHm 240A $0553 ho E E ho gain: may ho magma gHamHaSam m...“ Ea 113 F i 9 ur e 4 - 5 Per Cent 1; LOAN SIZE AND CHARGE-OFF RATE RELATIONSHIP CROSS SECTION A 3 2 1 400 450 500 550 600 Average Loan Size 650 - «Wm lIulllll-Illll 116 Figure 4-8 Per Cent 4 LOAN SIZE AND CHARGE-OFF RATE RELATIONSHIP CROSS SECTION D 3 2 l _/l I 550 600 650 700 750 800 Average Loan Size ; ". T-":'."_"‘5 "q _ 93v . 9.- 117 which is incurred when the average loan size in a branch is varied. The Number of Accounts-Unit gpst Relationship In chapter one it was observed that there is some controversy regard- ing the behavior of unit costs when.the scale of operation, as measured by the number of accounts, is expanded. The regression results presented in Table 4.1 allow a determination and.measurement of the number of accounts-unit cost relationship. For each cross section the relationship between I’Cunit cost) and X2 (number of accounts in its various forms, i.e., arithmetic, logarithmic, and exponential) is determined to be nega- tive. The measurement of the behavioral relationship between unit cost (I) and the average number of accounts (X2) is visualized best when presented in a two dimensional graphic formulation. As with the loan size-cost relationship, this presentation is achieved by holding the independent variables other than.Xé at their mean values while allowing Xé to vary. In developing the regressions upon which the following number of accounts-unit cost relationships are based, the average number of accounts (X2) was employed in three alternative forms: simple arithmetic, logarith— mic, and second degree exponential. The results presented in Table 4.1 represent the regressions and the particular specification of (Xé) which resulted in the best "fits," as measured by the multiple correlation coefficients (R). The best "fits” for cross sections A, B, and D were achieved with the number of accounts in second degree exponential form. Figures 4-9, 4—10, and 4-12 are therefore based upon the following equation: 118 ._ 2 .. _. .. ._ I = A + Bxl + cx2 + nx2 + Ex3 + Ex“ 6X5 + HX6 In the above formulation,'i.indicates that the particular variable is held at its mean value in developing the two dimensional graphic presentation. X2 and x: arezallowed to vary. For each cross section the degree to which X2 and X2 are allowed to vary is determined by the range in number of accounts in the underlying observations. That is, the relationship is not extrapolated beyond the relevant range of ob- servations on number of accounts (X?) for each cross section. The best "fit" for cross section C was achieved with the number of accounts (X2) in logarithic form. Figure 4-11 is therefore based upon the following equation: * _ - — — — Y a A + 3x1 + Clog;2 4- BIB + Ex4 + EXS + ex; Figures 4-9 to 4-12 show that considerable reductions in unit cost are associated with expansions in the average number of accounts in a particular branch. For cross sections A and B, it appears that some of the branches have expanded beyond the optimal (lowest cost) size of operation and have incurred certain diseconomies of scale which manifest themselves in increasing unit costs. For cross sections C and D there is no evidence of expansion.beyond efficient size. 119 Figure 4-9 Average Unit Cost NUMBER OF ACCOUNTS AND AVERAGE UNIT COST RELATIONSHIP CROSS SECTION A 45 35 25 15 500 1500 2500 3500 4500 5500 Average Number of Accounts Average Unit Cost 55 Figure 4-IO 45 35 25 NUMBER OF ACCOUNTS AND AVERAGE UNIT COST RELATIONSHIP CROSS SECTION B 500 1500 2500 3500 Average Number of Accounts 4500 5500 Average Unit Cost 55 121 Figure 4-II 45 35 25 15 NUMBER OF ACCOUNTS AND AVERAGE UNIT COST RELATIONSHIP CROSS SECTION C 500 1500 2500 3500 4500 5500 Average NUmber of Accounts 122 Figure 4-I2 Average Unit Cost 55 NUMBER OF ACCOUNTS AND AVERAGE UNIT COST RELATIONSHIP CROSS SECTION D 45 35 25 1., 500 1500 2500 3500 4500 5500 Average Number of Accounts liil‘l III. I‘" II t y I i... ‘1. J m A brief sketch of the basic empirical results of this study has been presented on the preceding pages. The basic cost behavior hypothe- sis was shown to be effective in accOunting for an average of two-thirds of the variance in the dependent variable (Y, cost per account) for cross sections A, B, and C. The hypothesis was less effective with cross section D and accounted for only 53 per cent of the variance in Y among the branches making up the cross section. The direction of the relationship between I and the independent variables (Xi to Kg) was consistent with that predicted in chapter one. The relationship between loan size and unit cost was determined to be positive. The relationship between the number of accounts and unit cost proved to be negative for‘gll cross sections up to approximately 2500 accounts. Beyond 2500 accounts cross sections A and B revealed a posi- tive (increasing cost) relationship; for C and D the relationship re- mained negative. The regression of average loan size upon the charge-off rate yielded somewhat mixed results. The regression coefficients for the four cross sections all indicated an inverse relationship between average loan size and the risk assumption cost proxy (the charge-off rate); however, there was considerable variation in the indicated strength of the association between loan size and charge-off rate. The following chapter will provide a further analysis of the empirical findings outlined in this chapter. In particular, it will consider (1) their relationship to the findings and views expressed in prior studies, and (2) some of their possible managerial and regulatory applications. A a mafia r.-. c m . iwififidufiflffi$fi mm CHAPTERV ANALYSIS. OF THE EMPIRICAL RESULTS In chapter four the basic empirical findings were outlined. In the Present chapter these findings are analyzed and evaluated in terms of their relation to the findings of other studies, their managerial, and their regulatory implications and applications. Relation to Other Studies The basic empirical results presented in chapter four involved three important relationships: (1) the relationship between loan size and unit cost, (2) the relationship between loan size and the cost of I‘fiask assumption, and (3) the relationship between the number of accounts and unit costs. The findings concerning these relationships are dis- cussed below in terms of their agreement or lack thereof with views or findings presented in other studies. Loan size and unit cost In both chapter one and chapter three various views concerning the likely relationship between loan size'and lending costs were presented. In this particular context, M m excludes costs associated with risk assumptionand forbearance. Discussion centered on direct branch Operating costs. The views expressed regarding the influence of loan size upon g; (as contrasted with cost per dollar) operating costs cOVBred all possibilities: loan size increases unit operating costs; 124 125 loan size decreases unit operating costs; and loan size has no effect upo 1': unit operating costs. In his study which included a consideration of cost behavior in co nsumer finance branches, Joel Dean did not include loan size as an 1‘ 1daependent determinant of unit costs. Farwell observed the following 1 0° hcerning the likely position of loan size in a cost behavior hypothesis: . . .there is doubt whether operating cost actually is associated directly with size in this business. In the industry, it is believed that no more effort or material is expended on large than on small loans within the range of : sizecharacteristic of the small loan industry. . . .for the long-run it appears preferable to accept the concept that loan size is not an in- dependent factor in the determination of operat- ing cost. All of the cost accounting studies reviewed by the writer contain the implicit assumption that loan size is not an independent determinant or operating costs--i.e., of the acquisition or servicing elements in the basicgcost equation: Ci 8 A1 +'Si + R1 + F1. The findings outlined in the preceding chapter support the position tllat loan size is; an independent factor in determining unit operating costs. For cross sections A to D, the marginal direct lending costs (MDLC) per $100 increase in average loan size ranged from $0. 58 to $2.46. That is, the increase in average unit cost (Y) associated with an increase of $100 in average loan size (XI) ranged from a low of $0.58 f0? cross section D to a high .of $2.46 for cross section A. ‘ d1 g _1Farwell, ep,gt., pp. 166.167. It should be noted that Farwell 1 d incorporate a measure of the deviation of average loan size from a °n8~period average as a short-run determinant of cost behavior. 126 It must be emphasized again at this point that this investigation is a .0539. or pilgp study. As with any case or pilot study, great care mass-t be exercised in generalizing upon the results. The findings of thi s study do support the position that loan size is an independent co 3“: determinant. This in no sense mp that other views concerning the effect of loan size upon lending costs are wrong. It may be that th 6 positive size-cost relation of this particular study is attributable to operating policies or procedures unique to the compamr under study. Q 80, it may be that regulatory features of the states studied exert an influence. mile the statistical procedure reveals the positive size-cost Ji‘efl.ationship, it does not explain why it exists. In the writer's opinion, the positive relationship is explained, at least in part, by the additional inVestigative effort devoted to larger loan sizes plus the predominance of security entailed by the larger loan sizes. That is, it seems clear that the total investigative effort will be greater on a loan of $1000 Versus a loan of $200. In addition, it is probably significant that mo st larger loans entail the taking of some form of security. Verifica- tion of the existence of the security and the recording of the chattel involve additional effort which reflects in turn in unit cost levels. The following schedule illustrates the point that larger loans generally inmlve the taking of some form of security. The character of this. par— ticular schedule is representative of a much larger number which were ll}! 127 mm: 5.1 CLASSIFICATION OF LOANS MADE BY SECURITY ,1 i Loans made during the year secured by Average loan Mattel mortgages on both household goods and motor vehicles $1,191.69 mattel mortgages on household goods 773.01 mattel mortgages on motor vehicles 660.49 Oklattel mortgages on other chattels 510.42 other security 801.30 Comaker, endorsed, or guaranteed 372.014 Unsecured (signature loans) 279.88 ‘ Source: Stateof Ohio, 5% Remrt and Roster of Small Loan Licensees, 1963, p. . M1161: studies are needed to corroborate the present findings 93; to I‘a.:'l.se doubts about their general applicability. .L-Oan size and risk assumption cost In chapter three it was observed that the unpirical evidence on the ef-‘f‘ect of loan size upon the cost of risk assumption, as measured by the charse-off rate, was somewhat mixed. Most of the evidence is somewhat dated. ‘ The works reviewed show all combinations of possible relationships betWGen loan size and risk assumption cost, i.e., positive, negative, and no I‘eilation between loan size and charge-off experience. 128 This study revealed an inverse relationship, of varying strength, between average loan size and the cost of risk assumption, as measured by the charge-off rate (see Figures ll—S to ll-8 of the preceding chapter for a graphical presentation). The reduction in charge-off rate per $LOO increase in average loan size is presented in Table 5.2. The inverse relation between average loan size and the charge-off rate can be restated in marginal analysis terminology: decreasing mar- 81ml risk assumption cost (MRAC). Over the relevant range of average loam size, increases in average loan size have resulted in reductions in the cost of risk assumption. It should be emphasized that this does not mean that the absolute dollar amount of risk assumption cost is reduced. The dollar'amount of loans charged off may increase or de- c:E‘ease as the total amount of loans outstanding and their average balance are increased. As with the size-cost results, a warning with respect to the general applicability of the size-risk assumption cost results must be expressed. These results are likewise the product of the £13; study. They may be conditioned by the policies and procedures of the comparw under study and the regulatory provisions of the particular states studied. Further s“indies are needed to corroborate the present findings with respect to the size-risk assumption cost relation 2; to raise doubts as to their “re general applicability. In the light of the results presented in Table 5.2, it is interesting to review a statement concerning the influence of loan size upon loss ex- perience :2 \ 2Thrift Investment Corporation, ma; Report 1&3, p. 5. ‘I .llll‘l TABLE 5.2 REDUCTION IN CHARGE-OFF RATE PER $100 INCREASE IN AVERAGE LOAN SIZE State Charge-off Rate Decrease A .12 B .55 c . 36 D .71 Source: Table #.3 As the accomparwing tabulation shows, the average loan has increased 76 per cent over the past five years. Such loans. . .gen— erally have a lower delimuency and loss rate. The above statement is consistent with the results achieved with the company under study. A somewhat related study in the field of com- InGarcial banking yielded results which are also consistent with those Presented in Table 5.2. The results of this particular study showed that the marginal cost of risk (the change in the dollar amount of charge-offs per $100 change in average loan size) ”declines as the aVGrage size of loans increases.”3 As with the size-cost relationship, the statistical results do not pmvide an answer as to why the observed relationship-decreasing mar- gin‘ll cost of risk assumption with respect to average loan size--exists. 36eorge J. Benston, ”Comercial Bank Price Discrimination Against 96 Loans: An mpirical Study,” The Journal of Finance, XIX (December 130 Che. rge—off experience is somewhat better on larger versus smaller loans. On an ex-post basis, this means that borrowers of larger mo mats are held to a higher standard of creditworthiness than would be necessary to make their loss experience equal to that experienced on smaller loan sizes. Whether this apparent policy, be it explicit ° 1‘ implicit, is justified from a managerial viewpoint would depend upon the additional revenues which are forgone by this higher standard which is imposed in making larger size loans. These forgone revenues would he ed to be compared with the costs avoided to judge the managerial SOlmndness of the policy. In addition to the above argument, it may be that charge-off experi- el'1<.=e on larger loans reflects greater persistence in collection efforts on larger loans. If this be the case, it would account not only for the loan size/charge—off relationship, but it would also provide a par- tial explanation for the positive loan size-direct cost per unit rela- tionship of Figures 14-1 to #4. The reduction in charge-offs on larger loans may be offset by additional direct branch lending costs which are 1I'lcurred in additional investigative and/ or collection efforts on behalf of larger loan sizes. .Number of accounts and unit cost A third important relationship developed in this study is that be- ‘ tween the average number of accounts in a branch and its unit cost. Prior “The reference to better charge-off experience on larger loans is in Part an inference drawn from the observed relationships between the u“Digs-off experience and average loan gig. In addition, it is based an examination of a schedule of cumulative charge-off experience :ydloan size prepared by the company under study. The schedule showed eGilded improvement in charge-off experience on larger loan sizes. Till Ill lll‘lll {1' All! I 131 studies have presented somewhat disparate views concerning the likely cha racter of this relationship. The present study reveals decreasing unit costs for all four cross se ctions up to approximately 2000 accounts. Beyond this point, cross Se c‘tions A and B are characterized by increasing unit costs; unit costs £0 1‘ cross sections C and D continue to decline. These findings are surprisingly similar to those revealed by Joel Dean almost thirty years ago. The unit cost schedule developed by Dean d3 clined up to between 2000 and 2500 accounts and then flattened out cc) hsiderably. This unit cost behavior approximates very closely that I“4“3‘realed by cross sections C and D. It is unlike cross sections A and B in that no segment of increasing unit cost is revealed. The findings of both Dean and the present writer appear to be at Odds with views expressed in other studies. Miller Upton wrote that:5 Instead of the business possessing decreasing cost tendencies, its cost elements are much . more in keeping with constant cost conditions. On this same general topic, Carl Dauten wrote that:6 There is, however, little saving that comes from having larger offices because the in- dustry is basically one of more or less con- stant.costs. 5Miller Upton, ”The Importance of Direct Costs in the Granting "of consumer Instalment Credit, ”, Northwestern University unpublished dis- sertation, June 1948, p. 233. 6 Carl A. Deuten, financing The American Qnsumer, Consumer Credit for!) aph No. 1 (St. Louis: American Investment Comparw of Illinois, 956 , p. 73, 1.1-i. 1 II" 132 The above views are contrasted with an early statement by T.O. Yntema Wkli ch implies a contrary position:7 . . .the average size of the loan office is only one-fifth to one-third the optimum size of office which would operate with minimum unit cost. This disparity between findings is not easily reconciled. The ob- fie Q‘tive fact is that an inverse relationship exists between unit costs and the average number of accounts outstanding in a branch. This has b3 en demonstrated. The particular point of disagreement appears to Center around the question of whether this cost behavior is indicative of "scale economies,” in the conventional economic sense. Two frequently cited sources of "scale economies” have been:8 (1) Increasing possibilities for labor division and specialization. (2) Increasing possibilities of using advanced technological developments. These factors are said to manifest themselves in a downward sloping long-run average cost curve for the firm. The long-run average cost curve "Shows the least possible cost per unit of producing various outputs when 9 the firm has time to build any desired scale of plant." A downward S3-ijing long-run average cost curve is taken as evidence of economies of scale. —.— In attempting to reconcile the differing views on the issue 7 T. O. Yntema, "The Market for Consumer. Credit: A .CaSe in 'Imq perfect Competition,” Annals of The American Academ of Political and and Social Science, March 1953, p. 51:. aLeftwich, op,cit., p. 156. 9Ibid’. , p. 154. 133 of ”scale economies,” the character of the cost schedule relating unit costs to the average number of accounts must be considered. In particular, is this schedule equivalent to a long-run average cost schedule? It is clear that Dean so regarded the relationship. This is re- flected in the following remarks:lo By regarding each of these branches as a snapshot of a typical establishment as it increases its scale of plant, curves of long-period marginal and average cost may be drawn up from simultaneous observation of offices representing a range of plant scale. Miller Upton criticized Dean's procedure fer developing what Dean advanced as a long-run average cost schedule. In particular, he objected to the use of an‘gutput’measure as a proxy'fOr‘ggalg. Upton emphasized that the use of an output measure as a proxy for scale would be justi- fiable only ”when the major productive components are perfectly divisible so that there is perfect correlation between output and scale."11 This restriction is unduly severe and it has not been fellowed by researchers who have investigated the topic of ”scale economiesU in the related field of commercial banking. Recent studies of scale economies in banking have employed various ferms of output measures as scale proxy measures. Benston's recent work has employed the number of accounts and the number of loans.12 Gramley's 10Dean, M" p. 361. nUpton, 02.01te, p0 1380 12George J. Benston, "Economies of’Scale and Marginal Costs in Banking Operations,” The National Banking Review, June 1965, pp. 507- 549. 131+ study on scale economies in banking employed total assets (essentially an output measure) as a scale proxy.13 These procedures represent a reasonable approach to the problem of measuring long-run average cost schedules. They reflect the fact, unacknowledged by Upton, that em. pirical studies of scale economies cannot provide‘digggt,measures of firm scale. In place of direct measurement, firm scale is measured inferentially. It is implicitly assumed that a larger level of output is achieved with an increase in the basic factor inputs or in the scale of the firm. It is generally true that a larger level of output, as measured by the number of accounts, is achieved by increasing the basic factor in- puts which determine the "scale" of the consumer finance branch. These added factor inputs generally take the form of additional personnel. The human-factor input is characterized by considerable indivisibility. While it is true that personnel can sometimes be added on a less than full time basis, they are not divisible to the extent that the personnel input can be added in increments necessary to handle a single account (or in incre- ments sufficient to handle as few as 25 to 50 accounts for that matter). Due to the acknowledged indivisibility of the human input, it is likely that some of the measured decreases in unit cost simply reflect the fuller utilization of an.initially underemployed factor. This par- ticular type of short-run adjustment is reflected in the cost schedule developed between unit costs and the average number of accounts. If these short-run adjustments to factor inputs could be isolated and 13 Lyle E. Gramley, A Stu of Scale Economies In Banki , Kansas City: Federal Reserve Bank of Kansas City, 1532. 135 removed from the cost schedule, the effect would be to reduce its slope. It seems highly unlikely, however, that the removal of such short-run conditions would be sufficient to produce the constant cost condition (a horizontal cost schedule) advanced by other writers. The import of the above analysis is to support the position that the operation of larger branches (using the average number of accounts as a scale proxy) produces certain economies which manifest themselves in a decreasing unit cost condition, over some particular range. It is uncertain just how these economies are attained. They may result in part from the traditional sources of scale economies discussed earlier. FUrther research is needed to isolate their precise origin. It is important to emphasize that the discussion of "scale economies” has been confined strictly to consumer finance branches. The broader and equally engaging issue oflgizm scale economies has not been touched. That is, to what extent are there economies associated with the operation of a larger firm, i.e., of a‘figm with a large number of branches. There is certain evidence to suggest that a larger firm achieves certain econp omies in the acquisition of funds. Its size makes it a better financial risk and hence makes it possible to acquire funds at a lower cost. It seems likely that there would be certain economies in the area of ad- vertising effort. A firm with a large number of branches in the area covered by a particular media has a distinct advantage over the firm with a small number of units in the same area. The firm with the larger num- ber of branches achieves a certain form of "scale economy” by being able to spread the cost of reaching the area over a greater number of branches. 136 The above ideas on firm scale economies are simply suggestive. It seems to be clear that some very fruitful research could be undertaken in this particular area. Implications and Applications The preceding discussion has attempted to relate the basic empirical findings of this study to the views and results presented in.prior works. This section will entail additional analysis of the empirical results in an effort to assess their possible implications and applications. A useful division is on the basis of likely managerial and regulatory impli- cations and applications. The following treatment uses this breakdown. Managerial implications and Applications The empirical findings of this study, which were outlined in chapter four and discussed further in the above section, have at least three possible managerial applications: (1) Cost control of the decentralized operating structure of consumer finance companies. (2) Application of the decision tool of incre- mental or marginal analysis. (3) Intra-firm resource allocation with respect to decisions concerning optimal sized opera- ting units. Control applications The empirical portion of this study entailed the construction and testing of a cost behavior hypothesis. * + Y-A Bll+cx2+nx3+Ex4+Fx+Gx 5 6 --:~=-.~1 we...” . '- 137 Fbr three of the four cross sections, the above‘ppggl accounted for or explained an average of two-thirds of the variance in unit costs (Y).14 The operations of consumer finance companies involve considerable decentralization. This creates in turn substantial control problems. Any device which can facilitate the evaluation of the operating per- formance of the branches would be a significant contribution to the management function in consumer finance companies. hath further refinement the above cost behavior model could play an important role in evaluating cost levels in the company's branches. Such refinement would be directed at improving the predictive or explana- tory power of the basic model. A basic approach to controlling the decentralized operations characteristic of consumer finance companies is through application of the principle of ”management by exception." The cost behavior model could contribute to the application of the management by exception principle. The net regression coefficients (B to G) and the regression constant (A) make up the basic elements of the predictive model. These coefficients represent a form of average relationship between the individual inde- pendent variables and the measure of unit cost. Incorporating the indi- VidDAI values of a particular branch on variables Xi to X6 in the regres- sion equation (the predictive model) generates an estimated value for * unit cost (I). The application of the “management by exception” principle luSince the basic cost behavior hypothesis has been tested and found to have significant predictive or explanatory power, it is appropriate to refer to it as a model in the present discussion. 138 a enters at this point. The estimated or average value for (Y) is com- pared with the actual value of unit costs fer the branch (Y). Where significant differences are found between (Y) and (Y), additional in- vestigation would be called for. It should be emphasized that the application of the model produces average unit cost estimates. They should not be interpreted as a measure of good or acceptable performance; they are not the equivalent of care- fully constructed standard costs. This fact should not, however detract in any way from the general usefulness of the predictive model and the estimated unit costs. A comparison of the estimated with the observed value of unit cost for a particular branch will still highlight cost levels which are out of line from average relationships and which warrant additional investigation. The following will illustrate the application of the control tech- nique described above. The regression equation (predictive model) developed for cross section C will be used:15 i = 29.42 + °°°15(X1) — 9.88 Log(Xé) + 1.23(x§) + 0.198(Xu) + 0.2820(5) - 0.0870(6) The predictive model will be applied to five different branch offices. The values of each of the independent variables for the five branches are listed below: m 35}. 52. 3‘2. 2 32 i6. 1 491 3012 1.52 106 12 41 2 505 2365 1.52 106 16 35 3 385 5253 1.35 121 18 38 4 412 2726 4.91 121 19 35 5 481 511 .46 121 20 41 ‘1 5This regression equation is drawn from Table #.l. 139 Incorporating the values on the independent variables X1 to X6 in the above regression equation (predictive model) yields the estimated * * * values of unit cost (I). Y, Y and Y'- Y are presented below for the five branches: I * Branch 3 g Y - Y 1 25.18 25.94 0.76 2 27.80 27.78 -o.02 3 25.92 25.85 -0.07 4 31.55 37.06 5.50 5 36.57 47.39 10.82 The above results show that branches four and five have actual unit costs which are well out of line from.average relationships. Further ins vestigation would be called for to discover the reason(s) for this dis- parity. If warranted, corrective action could then be taken. Incremental analypis A second possible application of the study findings is in the general area of incremental or marginal analysis. The particular application could entail an analysis of the likely impact of an increase in average loan size upon branch profitability. Or, the effect of an expansion in the average number of accounts could be considered. The behavioral relationships developed in this study have a direct application to the issue of determining the effect which increases in average loan size are likely to have upon branch profitability. The impact of’any change upon profitability can be determined if the changes in costs and revenues can be predicted. The results of the study provide a basis for making estimates of the change in costs; other information could be drawn upon to estimate the likely change in revenues. 140 The impact upon profitability of an increase in average loan size is determined by first estimating the expected increase in branch costs. That is, what will be the marginal or incremental cost of an increase in average loan size (MOS)? MCS will be made up of three elements: MDLC Marginal direct lending costs MCRA Marginal cost of risk assumption MCF Marginal cost of ferbearance (money) NOS 8 MDLC + MCRA + MCF Estimates of the elements in the above equation are developed in the following manner: <1) mm = axle) Xm 8 Change in average loan size B 8 Net regression coefficient of Y1 (2) m - 5:3 (29.7 - [£3 (16117 = Original average loan size New average loan size J‘o pf-J‘o n ' Charge-off rate at I; O I; 3 Charge-off rate at Xi Note: The first element in the above represents the average dollar amount of loan charge-off given the new average loan size. The second element represents the average dollar amount of’loan charge-off at the original loan size. The difference between the first and second ele- ment is the increase (decrease) in the cost of risk assumption per account due to the change in.average loan size per account. 141 (3) MCF = Xm (i) dXi = Change in average loan size i = Assumed cost of funds To illustrate the application of the above formulation, the regres- sion elements of cross section C will be used (these are found in Tables 4.1 and h.3). The estimates will be made for an assumed $100 increase in average loan size (Xm). In order to develop the illustration fully, an assumed cost of funds of 6 per cent per annum is used. .Qgpg: x; a $400 x; = 1.72 per cent * xi = $500 x; - 1.36 per cent * dxl = $100 i = 6.00 per cent B -- .015 . Computed from Table 4.3. M3 MCS = Xm (B) + 1?; (19.7 - [X3 (x117 +dxl (1) M08 = 100(.015) + £0136(5oo)_7-50172(4oo)_7 +1oo(.06) MC 8 1.50 - .08 + 6.00 3 M05 = $7.42 t“ lé’fia r5641: 1‘ ls" l) 1 ' ~ -— . a q ’ 1.15 - r" -' - $2.; , 142 The MCS of $7.42 represents the added costs per account which result from a $100 increase in average balance per account. It is not the in- crease in unit cost (Y) which was used in earlier analysis. Y included only direct lending costs and excluded the costs of risk assumption and forbearance (money). MCS includes all three of these elements. It must be emphasized that the foregoing estimate of'marginal cost contains an important implicit assumption. It is assumed that the average relationships reflected by the regression coefficients will be pertinent to the projected increase in average loan size. These average historical relationships could be altered if some unusual effort or activity were employed in attempting to expand the average loan size. For example, assume that an.g££ppp to expand the average loan size in a branch entailed a reduction in the normal standards of creditworthiness. This would dis- tort the normal relationship between average loan size and the cost of risk assumption, as measured by the charge-off rate. Applying the average historical relationship between average loan size and the charge-off rate to estimate the cost change would, therefore, be inappropriate. On the two preceding pages an estimate of the marginal cost associ- ated with an increase in average loan size was developed. To assess the likely impact upon branch profitability, marginal revenue would also need to be estimated. A possible method of providing an estimate of marginal revenue would be to approximate the decrease in the average rate of charge collected. The "average rate of charges earned or collected" is developed by dividing income for the year from finance charges by the average total loan balance outstanding during the year. This particular statistic is 143 provided in almost all reports by states on the operations of consumer finance licensees. It is generally divided by 12 in order to express the average rate of charges earned or collected on a monthly basis. Since consumer finance companies make loans under rate structures which are graduated downward, on the basis of loan size, an increase in the average loan size will produce a reduction in the average monthly rate of charges earned or collected. An estimate of this reduction is necessary in order to estimate marginal revenue (MRS). Strictly for the purpose of developing this illustration fully, it is assumed that the interaction between the graduation in the rate structure and the increase in average loan size reduces the monthly rate of income to 1.75 from 2.00 per cent. The estimate of MRs is developed below: MR = [k' (xi) - k° (1:3)] x 12 S = Initial average loan size ($400) New average loan size ($500) We .3“. die ll 8 Monthly rate of income at average loan size of $400. (2.00 per cent) 3' ll Monthly rate of income at average loan size of $500. (1.75 Per cent) Results: MRS = .[-0.0175(500) - 0.0200(400)';7 x 12 MRS = $9.00 144 The estimate of marginal revenue can be combined with the estimate of marginal cost to provide an approximation of the impact of the change in average loan size upon branch profitability: ms($9.oo) .. MCS($7.42) = $1.58 It must be emphasized that the above results are not to be inter- preted as typical or representative. While the behavioral relations for estimating the MDCL and the MCRA were based upon the results of cross section C, the figures for the cost of funds and for the decline in the monthly rate of income are simply assumed figures and are not based upon any empirical data. Given more factually based estimates of the cost of funds and the prospective reduction in the monthly rate of income, the foregoing analysis could provide reasonably reliable estimates of MCS and MRS. This could then be combined to estimate the impact on branch profitability of a change in average loan size. Intra-finm resource allocation A third.managerial implication of the study results concerns the intra-firm.allocation of resources. The initial establishment and subse- quent expansion of a consumer finance branch requires the commitment of scarce firm resources. The study results suggest that in some cases a misallocation of firm resources may have occurred. The misallocation takes the form of an excessive commitment of resources to particular branches--i.e., over expansion. The basis for the above view is found in the relationship between unit cost and the average number of accounts (the scale proxy) for 145 cross sections A and B. The cost schedules developed for cross sections A and B (Figures 4—9 and 4—10) suggest that some of the branches may have been expanded to greater than optimal size. Optimum size in this particular context refers to the average number of accounts at which unit cost is minimized. Optimum size for cross section A and B branches is achieved in the range of approximately 2000 to 2500 accounts. Be- yond this range, the empirical evidence suggests that unit costs tend to increase. In economic terminology, diseconomies of scale set in when branch size is increased beyond 2500 accounts. If certain realities can be temporarily ignored, the above results would suggest that management consider the establishment of additional operating units when a branch (in cross sections A and B) expands to beyond 2500 accounts. The emphasis is upon consider in the preceding sentence and it is meant to highlight the fact that 2500 accounts should not be viewed as some magical cutoff point for further expansion of any particular branch. Twentybfive hundred accounts would simply represent a bench mark beyond which the commitment of additional resources, to facilitate expansion, would be carefully evaluated. One of the realities which would influence decisions concerning optimal size units is the restrictive licensing features of’many state small loan laws. Even if it were determined that further expansion would result in.aubstantial diseconomies of scale, restrictive licensing practices might prevent the establishment of additional operating units. In addition to restrictive licensing, prospective volume considera- tions would temper the application of any scale cutoff criterion. Even though a branch may have expanded beyond an optimal size, prospective 146 volume may be insufficient to justify an additional operating unit. If an additional branch were established, it would likely operate at such a small scale that it would have relatively high unit costs. In addition, if some of the new branch's volume were drawn from business handled by the original branch, it is possible that the original branch might be moved back up the average cost schedule and therefore incur higher unit costs. The above discussion emphasizes that a scale cutoff criterion should be employed only as a general guide to optimal branch size. Nevertheless, it does appear that it could serve a useful function in the intra-firm allocation of scarce resources. With the completion of the discussion of some of the managerial implications and applications of the study results, some of the possible regulatory implications and applications may now be considered. Regulatopy Implications and Applications The operations of consumer finance companies are subject to consider- able government regulation. This regulation generally includes limita- tions upon rates of charge, loan size, loan maturity, security which can be accepted, and restrictions upon the establishment of additional operat- ing units. Many of these features are built upon cost behavior concepts. In the following pages some of these regulatory features will be considered in the light of the study results. These will include: (1) Rate structure graduation. (2) Rate structure levels. (3) Restrictive licensing. 147 Rate structure graduation Early in the study it was observed that the introduction of graduated rate structures was designed to provide a more equitable distribution of the underlying costs of providing loans of different sizes. The meaning of "equity" in this particular context is that individual loans bear finance charges which are reasonably consistent with their costs. While the graduated rate structures do achieve a‘pppg equitable relationship between cost and charge, there is clearly no pretense that present rate structures have eliminated discrimination against larger loans. The results of this study have not provided direct estimates of the costs of loans of varying size. Rather, the effect of differences in average loan size upon the level of direct lending costs per unit has been estimated. These estimates of the effect of differences in average loan size in a branch are not unrelated to the problem of determining the effect of different size loans upon costs. A higher average loan size in a branch implies a generally higher structure of individual loan sizes. Therefore, the influence of increases in average loan size can provide an indirect measure of the effect of larger individual loans upon lending costs. For illustrative purposes the graduation represented in the following rate structure will be considered: 3-2-15 ($300-$600) under this rate structure 3 per cent per month may be charged on the outstanding balance below $300, 2 per cent on the balance between $300 and $600, and l per cent on the balance over $600. The dollar charges allowed on 12 month loans of $100, $300, $500, and $1000 are listed below: Loan Size 5100 $200 $500 $1000 Maximum charges 20.48 61.56 95.96 154.16 The net regression coefficients developed in this study for average loan size show a positive relationship between direct branch costs per unit (Y) and average loan size (Xi). The largest increase in unit cost per $100 increase in average loan size was approximately $2.50. If this information is used as a rough measure of the difference in direct branch costs attributable to differences in individual loan sizes, a substantial degree of discrimination against larger loans is implied. For example, the charge allowed on a $300 loan is approximately three times that allowed on a loan of $100. The difference in the charge allowed on the $100 versus the $400 loan is $41.08. Assume that a rea- sonable estimate of the added cost of risk and forbearance (money) is 10 per cent of the $100 increase in average outstanding balance of the $300 over the $100 loan. Deducting this added cost of risk and for- bearance from the gross difference in allowable charge leaves $31.08 ($41.08 - $10.00 8 $31.08). For no discrimination to exist against the $300 loan, this difference of $31.08 would need to represent a reasonable estimate of the additional costs (other than risk and for- bearance) attributable to the loan of $300 versus the loan of $100. The size-cost relationships developed in this study (see Figures 4—1 to 4-4) imply additional costs of no such magnitude. cats unlf 108.] con. ace cos add fun dis cle M loa out for ind! con con: cha: ifw 149 It should be emphasized that the above analysis does.pgp imply discrimination against the larger loan in an absolute sense. The indi- cated discrimination is only relative to the smaller loan. It is rather unlikely that absolute discrimination is entailed in the case of a $300 loan under the foregoing rate structure. A review of recent reports by consumer finance regulatory bodies reveals an average‘pppgl cost per account (a unit cost figure not restricted to direct branch lending costs) which ranges from approximately $65.00 to $75.00 per annum. In addition, this unit cost figure does not include the cost of borrowed funds (the cost of forbearance proxy). The maximum charge allowed on a $300 loan under the above schedule is $61.56. Therefore, absolute discrimination is clearly not indicated. Absolute discrimination must clearly be confined to loans which generally range beyond $300 in size. Relative discrimination appears to exist across the entire range of loan sizes. It is important to note that the inequitable distribution of costs outlined above is imppsed upon the industry by the'lpggl and economic force of the authorized rate structures. From a legal standpoint, the industry is prohibited from charging rates on smaller loans which are consistent with any reasonable estimate of their costs. The economic consequence of the foregoing is that the excess of costs over authorized charges on smaller loans must be covered by the charges on larger loans if the industry is to avoid operating at a loss. l L . \ It i} - 3. ,I‘ ‘1 n. l.- r . £1; _ 7 l > . 11 igiha‘” H .5 ' , - |_ ‘ _. 7 1,1. , _. ‘._, l i ‘4 , _- ' n "l ‘ ’~ I ' K ‘_ . . ' Ia -' _-. _ a... :- 150 Rate structure levelg There is an appreciable amount of variation in the general level of rate structures among the various states. The results of this limited study suggest that such variation is not surprising in view of cost dif- ferences among the states. Over the four states covered in this study, there was a difference of 32 per cent in average direct branch costs per account between the highest and lowest cost state. Higher lending costs are reflected, in general, in higher rate structure levels. For the highest cost state included in the study, the finance charge allowed on twelve month loans of $300, $500, and $800 were respectively 21, 19, and 20 per cent higher than the charges authorized for the lowest cost state. Based upon the study results, these inter-state differences in cost levels appear to be explained, in general, by the same variables which account for differences in inter-branch cost levels. While there were some exceptions, the two higher cost states (cross sections A and B) had generally'higher average values on the variables which were positively related to unit costs (average loan size, delinquency, factor prices, and loan.mix) and lower average values on the variables which were nega- tively related to unit cost levels (average number of accounts and new applications acceptance rate). Restrictive licensing A final regulatory feature to which the study results can be applied is the practice of restrictive licensing of consumer finance branches.l6 16For a comprehensive analysis of the pros and cons of restrictive licensing see: Miller Upton. "An Economic Appraisal of Convenience and Advantage Licensing by Small-Loan Statutes.” The Journal of Business, June 1952, pp. 249-263. 151 The existence of scale economies has been offered as partial justifica- tion for restrictive licensing. It has been argued that the number of branches in a community must be limited py ppg §_t_g_t_g if business volume is to be sufficient to allow existing units to achieve an efficient size. This study has revealed scale economies at the branch level. It might be concluded, then, that the results tend to support one of the arguments, if not the basic concept, of restrictive licensing. However, it should be recalled that the study revealed a range of branch scale characterized by diseconomies as well as a range characterized by econ- omies (see Figures 4—9 and 4—10). Therefore, it can be seen that restrictive licensing has the potential for impairing, as opposed to improving, resource allocation in the Consumer Finance industry. An impairment of resource allocation would result in cases where existing branches are expanded beyond optimal scale because restrictive licensing prevents the establishment of additional operating units. Barring a completely omniscient regulatory body, this latter result would seem almost as likely as an outcome which would serve to improve the alloca- tion of resources within the industry. * I The import of the above is to indicate that the practice of restric- tive licensing has the potential for impairment as well as possible im. provement of resource allocation in the industry. In assessing the merit of this regulatory feature, it should be observed that the benefits from its implementation are quite uncertain. The‘ppppp associated with the implementation of restrictive licensing are much more certain. Restric- tive licensing provisions require that consumer finance companies support applications for new licenses by demonstrating that an additional branch 152 is justified in a particular community. The development of such evidence entails a certain cost to the consumer finance company. The regulatory body also incurrs certain costs in reviewing the additional details which must support applications in "restrictive licensing" states. In evalua- ting, in part, the overall worth of this particular regulation, these certain costs should be related to the rather uncertain benefits. CHAPTER VI SUMMARY AND CONCLUSION Summary The principal objective of this study has been to contribute to the understanding of cost behavior in consumer finance companies. In particular, a determination and measurement of the influence of (l) aver- age loan size and (2) scale of operation upon unit costs was sought. A careful review of the pertinent literature revealed disparate views concerning the likely impact of average loan size and firm scale upon unit costs. The initial approach to providing estimates of the influence of average loan size and branch scale upon unit costs entailed the appli- cation of’more traditional cost accounting or cost determination methodology. Hewever, this approach was abandoned when the predominance of joint or comon costs in the operation of consumer finance branches was revealed. It was determined that the application of traditional cost determination procedures would entail numerous allocations of common or joint costs which would inject a somewhat arbitrary element into the study results. In addition, the breadth of the study would be severely restricted since traditional procedures would require considerable onpsite investigation. This would limit the number of branches which could be included in the study. In addition, the practical problem of attempting to avoid interfering with daybto-day branch operations 153 154 would be substantial. In view of these and other considerations, it was decided that an alternative methodology was needed. After a review of some of the earlier works on cost behavior in the industry and related works in the field of commercial banking, it was decided that the statistical methodology of multiple regression and correlation analysis offered a useful alternative to traditional cost determination procedure. The application of the statistical methodology would (1) entail no arbitrary allocations of joint or com- mon costs; (2) minimize on—site investigation and interference with daybto-day'branch operations; and (3) make it feasible to include a fairly large number of branches, located in different states, in the study. In order to employ the multiple regression and correlation.method- ology it was necessary to formulate a hypothesis concerning the deter- minants of branch unit costs. In particular, it was necessary to develop a predictive hypothesis which would explain unit cost variance among different branches. This would in turn.make possible an assess- ment of the individual influence of (1) average loan size and (2) scale of operation upon branch unit costs. The formulation of the behavioral hypothesis was accomplished by a thorough investigation of the pertinent literature. The pertinent litera- ture included studies in the related field of commercial banking. In addition to literature review, discussions with numerous members of the consumer finance industry proved most helpful. Caution had to be exer- cised in developing the hypothesis in order to insure that the cost 155 determinants would be independent of each other. The basic behavioral hypothesis and its elements are presented below: a . - Y = A + Bil + CX2 + DX3 + EX“ + FXS + GX6 Estimated direct branch operating cost per account > N* I! 8 Regression constant g! C) II Net regression coefficients I Average loan size Average number of accounts outstanding Delinquency'measure = Factor prices index = Loan mix 8 New application acceptance rate ' ll The rationale underlying the use of each of the variables listed above and their development was outlined in chapters one and four. It is too lengthy to summarize again at this point. The above hypothesis was tested with data for four cross sections made up of over three hundred consumer finance branches of a major con- sumer finance company. Each cross section was made up of branches from a single state only (cross section A includes only branches from state A, cross section B includes branches from state B, etc.) The basic predictive or explanatory power of the hypothesis was reflected in the coefficient of determination (R?) for each of the four cross sections. These coefficients revealed that the hypothesis was effective in explaining 156 approximately two-thirds of the variance in unit costs (I) for three of the four cross sections. It was somewhat less effective with the feurth cross section. The relationship determined between average loan size and unit costs (Y) was positive for all cross sections. The strength of this associa- tion is revealed in Table 6.1 which presents the increase in unit cost estimated for a one hundred dollar increase in average loan size (X1). TABLE 6.1 INCREASE IN UNIT COST (Y) PER $100 INCREASE IN AVERAGE LOAN SIZE (x1) Cross Section Unit Cost Increase A $2.h6 B 2.4# C 1.51 D 0.58 Source: Computed from Table h.l The relationship between average loan size and the cost of risk assumption, as measured by the charge-off rate, was also estimated. The method employed was a simple regression of charge-off on the average loan size. The relationship between average loan size and the charge-off rate was negative fer all cross sections. The results are summarized in Table 6.2. The change in charge-off rate per $100 change in average loan size is shown. ,. . - . .._—‘.i‘. .‘ '- - i ,' WIFE} i". ‘_ I‘ "i.“ f". 1'” T *H fa“: 157 TABLE 6.2 DECREASE IN CHARGE-OFF RATE PER $100 INCREASE IN AVERAGE LOAN SIZE - Cross Section Charge-off Rate Decrease A 0.12 B 0.55 c 0.36 D 0.71 Source: Computed from Table 4.3 The relationship between unit cost and the number of accounts (the scale proxy) was only'partly'consistent among the four cross sections. For cross sections A and B scale economies, as measured by the decline in the average unit cost schedule, were indicated up to approximately 2000 to 2500 accounts (see Figures h—9 and h—lO). Beyond this range scale diseconomies were indicated. For cross sections C and D average unit cost declined over the entire range of scale represented in the underlying observations. However, as with cross sections A and B, substantially all of the economies of scale were achieved when branch size was expanded to approximately 2500 accounts. In addition to developing the cost behavior estimates summarized above, the relation of the findings to prior studies and their possible managerial and regulatory applications were considered. Three managerial applications of the study results were discussed: Illlrlll'v‘l'l: l‘i‘sl 158 (1) Cost control of the decentralized operating structure characteristic of consumer finance companies. (2) Application of the decision tool of incre- mental or marginal analysis. (3) Intra-firm.resource allocation with respect to decisions concerning optimal sized operat- ing units. In addition to managerial applications, the results of the study were related to the following regulatory features: (1) Rate structure graduation. (2) Rate structure level. (3) Restrictive licensing provisions. During the course of the study a nwmber of recommendations for further research have been made. The present study has considered only the issue of ”scale economies" in consumer finance branghe . A fruitful area for further research would be a consideration of the somewhat broader concept of‘figg scale economies. That is, to what extent are their economies or efficiencies inherent in a I}!!! with a large number of branches? Further research along the lines of the present study would be useful in providing corroboration which could enhance the more general applicability of the study results. Additional efforts designed to further the predictive or explanatory power of the cost behavior I hypothesis developed in this study would also be valuable. These efforts might provide additional or substitute explanatory variables. For example, it is likely that part of the unexplained variance in unit cost (I) reflects the short-run influence of personnel additions. The addition ,__._. v 159 of a "dummy" variable (Xi = 0 when there have been no personnel additions during the year and Xi = 1 when personnel have been added) to accommodate this condition may improve the overall explanatory power of the original cost behavior hypothesis. ggnclusion This study has provided some empirical determinations and measure- ments of cost behavior relationships in consumer finance companies. Hopefully, this will contribute to filling the existing void surrounding our understanding of cost behavior in consumer finance companies. If it does, the writer's efforts will have been amply rewarded. BIBLIOGRAPHY Books an no ra hs American Bankers Association. Analzgipg The Cost Factors of Instalment Lendipg. Prepared by the Instalment Credit Committee. New York: Amarican Bankers Association, 1946. . fivelopipg Cost Information on Instalment Credit Operations. Prepared by the Instalment Credit Committee. New York: American Bankers Association, 1964. . Instalment Credit Fundamentals. Prepared by the Instalment Credit Committee. New York: American Bankers Association, 1964. ____fi . Know Your Earnings on Instalment Loans. Prepared by the Instalment Credit Committee. New York: American Bankers Association, 1960. Roger S. Barrett. Compilation of Consumer Finance Laws. Washington D. 0.: National Consumer Finance Association, 1952. Nathan A. Batchker. Manual on Instalment grey. Cog utations for Yields, Char es Rebates and Earni 5. Boston: Bankers Publishing Co., 19%. S. Lees Booth. finance Facts Year Book 1265. Washington D. 0.: National Consumer Finance Association, 1 5. Michael J. Brennan. Preface Io Epmmetrics. Cincinnati: Southwestern Publishing Comparw, 1 0. John M. Chapman and Frederick W. Jones. flmnpe Coppanies; How and Where flex Obtain Their Ms. New York: Columbia University, Graduate School of msiness, 1959. Comsite Earnipgs Performance of the Major Qnsumep finance Companies. Menlo Park, California: Stanford Research Institute, 1959. Marshall C. Corns. Practical gpst Apcountipg Er Banks. Boston: Bankers Publishing Comparv, 1959. Ernst A. Dauer. (Emparative Qperati_pg Merience of Consumer Instalment Financipg Agencies and Comercial Banks, 1929-01. New York: National Bureau of Economic Research, 1 . 160 161 Carl A. Danten. The Consumer Finance Industpy in a Qypppic Economy, Consumer Credit Monograph NO. 2. St. Louis: American Investment Company of Illinois, 1959. Allan Richard Drebin. Savipgs Accounts and Commercial Bank Earnipgs. Ann Arbor: Bureau of Business Research, School of Business Admins istration, University of Michigan, 1963. David Durand. Risk Elements in Consumer Instalment Financipg. New York: National Bureau of Economic Research, 1941. Mordecia Ezekiel and Karl A. Fox. Methods of Correlation and Regression Analysis. New YOrk: John Wiley and Sons, 1959. Thomas G. Gies, et a1. Consumer Finance Companies in Michigan. Ann Arbor: Bureau of Business Research, School of Business Administration, University of Michigan, 1961. Lyle E. Gramley. A study of Scale Eponomies in Bankipg. Kansas City: The Federal Reserve Bank of Kansas City, 19 2. F. B. Hubachek. Annotations on Small Loan Laws. New Ybrk: Russell Sage Foundation, 1938 . Robert WL Johnson. Methods of Statipg Consumer Finance Charges. New York: Columbia University Graduate School of BusineSs. J. Johnston. Econometric Methods. New York: McGrawaHill, 1963. . Statistical Cost Analysis. New YOrk: McGrawbHill, 1960. Richard H. Leftwich. The Price S stem and Resource Allocation. New York: Holt, Rinehart and Winston, 1961. M. R. Neifeld. Neifeld's Manual on Consumer Credip. Easton, Penn.: Mack Publishing Company, 19 l. . Trends in Consumer Finance. Easton, Penn.: Mack Publishing Company. 19 - William Hayes Simpson. America's Small Loan Problem. Published by the author in cooperation with the University of South Carolina, 1963. Paul F. Smith. Cost of Providipg Consumer Credit 1202 to 1222. New York: National Bureau of Economic Research, 19 . Tynan Smith and Robert Johnson. ”Operating Characteristics of Consumer Credit Institutions." Consumer Instalment Qpedit part 1, volume 1, Growth and Impprt. ‘Washington D. 0.: Federal Reserve System, 1957. 162 Andrew C. Stedry. Budget Control and Cost Behavior. Englewood Cliffs, No 1].: Prentice-11311, Ines, l O. The Consumer Finance Industpy. A monograph prepared by The National Consumer Finance Association for The Commission on Money and Credit. Englewood Cliffs, N. J.: Prentice-Hall, 1962. Thomas H. Williams and Charles H. Griffin. The Mathematical Dimension of Accountan y. Cincinnati: South-Western Publishing Company, 1964. Lowell C. Yoder. The Consumer Finance Industry in Florida. Gainesville: University of Florida Press, 1957. Ralph A. Young and Associates. Personal Finance Companies and Their Credit Practices. New York: National Bureau of Economic Research, 19. Articles and Periodicals "Accounting for a Finance Company." The Journal of Accountancy, March 1932, PP- 233-234- Charles E. Allen. "Introduction to Model Building on Account Data." N,A.C.A. Bulletin, June 1955, pp. 1340-1343. George J. Benston. "Commercial Bank Price Discrimination Against Snall Loans: An Empirical Study." The Journal of finance, December 1964, pp. 631-643. . "Economies of Scale and Marginal Costs In Banking Operations." The National Bankipg Review, June 1965, pp. 507- 549. G. Schuyler Blue. "Determining Bank Instalment Credit Costs." Bankipg, September 1951, pp. 36-37 and 106. W. L. Brown. "Instalment Loan Cost Analysis Includes Variety of Items." The Industrial Banker, April 1945, pp. 5-7. Sydney Cottle. "Earnings Performance of the Consumer Finance Industry. " 7 The Journal of Finance, September 1960, pp. 387-407. ”Higher Loan Ceilings for Consumer Finance Companies." Personal Finance Law Quarterl Re rt, Winter 1962, p. 6. Robert W. Johnson. "Subordinated Debentures: Debt That Serves as Equity." The Journal of Finance, March 1955. {g “was. 163 Harold Johnson. "Nebraska Has no Loan Shark Problem Today." Law and Contemporary Problems, Winter 1941. L. Laudadio. "Size of Bank, Size of Borrower, and the Rate of Interest." The Journal of Finance, March 1963, pp. 20-28. A. M. Levenson. "Interest Rates and Cost Differentials in Bank Lending to Small and Large Business." Review of Economics and Statistics, May'l962. O. C. Lorenz. "Consumer Credit Operating Costs." Bankin , May 1940, pp. 34-35. Richard Mattessich. "Mathematical Models in Business Accounting." The Accounting Review, July 1958, pp. 472-481. Clyde William Phelps. "Consumer Finance Company Charges II." The Journal of Marketi , July 1952, pp. 22-36. J. V. Poapst and W} R. Waters. "Rates of Return on Consumer Durables," The Journal of'Finance, December 1964, pp. 673-677. Harold E. Randall. "Cost Accounting for Financial Service is Just as Necessary as It is in Manufacturing." The Journal of Accountancy, July 19499 PP o 73.740 ______t . "Cost Accounting in Commercial Banks." N,A,C.A. Bulletin, March 1941, pp. 764-765. . "The Basic Cost Factors in Instalment Lending." Time Sales Financipg, May 1949, pp. 7-8 and 20-21. Miller Upton. "The Economics of Fair Charges For Consumer Loans.” Missouri Law Review, June 1951, pp. 274-291. . "An Economic Appraisal of Convenience and Advantage Licensing by Small-Loan Statutes.” The Journal of Business, June Gilbert W; Urban. "Analyzing and Controlling Consumer Loan Costs." Burro hs Cleari House, January 1956. . "Time and Dollar COSts in Consumer Financing." N.A.C.A. Bulletin, April 1955, pp. 1063-1068. Theodore O. Yntema. ”The Market for Consumer Credit: A Case in Im- perfect competition." Annals of the American Acadgpy of Political and Social Science, March 193 . R. Youngdahl. "The Structure of Interest Rates on Business Loans at Member Banks." The Federal Reserve Bulletin, July 1947, pp. 803-819. 164 Doctoral Dissertations Joel P. Dean. "A Statistical Examination of the Behavior of Average and Marginal Cost." Unpublished Ph.D. dissertation, University of Chicago, 1936. Loring Chapman Farwell. "Cost and Output Relationships of Consumer In- stalment Credit Agencies: Based on Reports of Indiana Licensees." Unpublished Ph.D. dissertation, Northwestern University, 1955. Ernest I. Hanson. "An Analysis of the Revenue Recognition Problem of Consumer Finance Companies Using Precomputation." Unpublished D.B.A. dissertation, Michigan State University, 1963. Miller Upton. "The Importance of Direct Costs in the Granting of Con- sumer Instalment Credit." Unpublished Ph.D. dissertation, Nerth- western University, 1948. Other Sources J. T. Arenberg. "Consumer Credit Net Yields." An address presented at the Instalment Lending Conference of the Illinois Bankers Associa- tion, 1964. Reproduced and made available by Arthur Andersen & Company, Chicago , Illinois . The Uniform Small Loan Law. Various drafts. Summary State Reports on the Operations of Consumer Finance Licensees. _Various States. Thrift Investment Corporation. Annual Report 1264. Family Finance Corporation. Annual Report 1264. Nu.mmu dw.mma ma.mmw mm.mmw szooom non pmoe ommao>< H mm m: mm mm open commodooom wommmmwdflmmw.3oz mu 9 S 2 mm 3:8 so is 53 mm a: was was an 5.3 82a Sousa ex em.H em.H om.H Ho.m ease non mmouomumnu 0H.N mo.m mo.m mm.m pace hem masseuse Homeoson «moaammoa.hocoowmfiaon mm amm- ommm weaa Heed wodooovmooo composes Ho moons: omsaop< wx who» Nmea Hmma nae» mafioamumnoo oudm doom oweao>< fix Gropmvm U ovmfim m 0....de < 09.3% mZOHmmmmumm AdZHm 2H QmGDAUZH mmudeMds ho mmbqgs 24H: <_Nanmmm< "WETTNNTNETES