COMPETITION AND INTEREST RATE CEILINGS IN COMMERCIAL BANKING*

RICHARD STARTZ Regulations prohibiting the payment of explicit interest on demand deposits are gradually being eased. As banks switch from payment in the form of free services to explicit interest, both the level of money demand and the response of money demand to market interest rates will change. Banks are modeled here as being Chaniberlinian monopolistic conipe^tors. Equilibrium deposit interest rate relationships are found for markets botli with and without an effective interest rate ceiling and the behavior r of the two markets is compared. The elimination of deposit interest rate ceilings leads to increased money demand and an increased responsiveness of deposit rates to market interest rates. I. INTRODUCTION

What happens to the monetary system when regulations change to allow the payment of interest on deposits? Two answers are easy. Theclassic view is that money does not presently pay interest. If deposit interest is legalized,"lFe'3emancl tor money will increase. The "revisionist" view is that competitive pressures are strong, so that banks have effectively found ways to pay interest in spite ot regula^ tionsTExplicit payments will merely replace impli^iib payment^T^o there will be no significant economic impact. At least one of these easy answers is wrong. I argue in this paper that an intermediate position most accurately describes the behavior of the deposit market. A standard model of monopolistic competition is presented as a description of bank behavior. I argue first that such a model is needed to account for the observed behavior of banks. Second, I formally present a model of monopolistic competition and show how some existing empirical evidence can be used to bound its parameters. After this I use the model to analyze policy changes with regard to the payment of explicit interest. The central conclusion of this paper is that removal of the explicitJnterest_prohiHition willjorce the Banlung industry closer to~fully competitive behavio^^presuinably the desired resultl ihe monetary authority is interested in both the level of money demand and the marginal relation between money demand and
* The author would like to acknowledge helpful suggestions by Andrew Abel, Stanley Fischer, Mark Flannery, Paul Joskow, Aris Protopapadakis, Richard Schmalensee, members of the M.I.T. Sloan Workshop, members of the Wharton Finance Seminar, and two anonymous referees. Special thanks go to Steven Wiggins for many hours of fruitful discussion. Financial support from the Sloan Foundation and the Alcoa Foundation is gratefully acknowledged.
1983 by the President and Fellows of Harvard College. Published by John Wiley & Sons, Inc. The Quarterly Journal of Economics, May 1983 CCC 0033-5533/83/020255-11$02.10

QUARTERLY JOURNAL OF ECONOMICS

market interest rates. M{r,ro) is the public demand for money, where r is a typical short nominal interest rate and rr, is the nominal interest rate paid by banks on demand deposits. (Throughout the paper, extraneous arguments, such as income, are embedded in the constant term.) The partial derivatives of the money demand function are Mr < 0 and Mr I) > 0. Existing law fixes r^ at zero.^ Money demand is M(r,O), and dM/dr is just Mr. Suppose that thfe deposit rate ceiling is raised a small amount dr^. If it is assumed that banks continue to earn more from lending out deposits than the new ceiling allows depositors to be paid, perfect competition requires that banks pay the new, higher rate. Money demand will rise by Mmdro. The marginal relation between the level of money demand and the market rate will be unchanged (assuming that the partial derivatives are constant over the relevant range). Now suppose that the authorities either eliminate the deposit rate ceiling or raise it to the point where it no longer binds. Perfect competition requires zero profits, or, equivalently, that banks pay out all their investment earnings to depositors. Let b be the fraction of deposits that banks are able to invest. Abstracting from risk (and the multiplicity of market interest rates), one can see that the competitive deposit rate must be ro = dr. Consider the switch from a zero deposit rate ceiling to a competitively set rate. Money demand will increase by Mr^8r. The relation between money demand and the market rate will be less negative. The total derivative will increase to Mr dr (-) (+) The questions of interest to the monetary authority can be restated witb a more general formulation of the deposit rate equation. Suppose tbat we have A change in regulatory policy changes a and ^. The resulting change in the demand for money is AM = M,o(Aa + 5riAl3)).
1. The institution of NOW and "automatic transfer" accounts has increased the number of institutions that effectively have checking account authority and has raised the interest ceiling on some checkable accounts from zero to 5 percent. The 5 percent ceiling remains a binding constraint. Money market funds are very close substitutes for bank checking accounts and operate without limitation on interest rates.

COMPETITION AND INTEREST RATE CEILINGS

/257 _

The change in the total derivative of the money stock with respect to the market interest rate is dr I The monetary authority must be concerned with both the change in the level of money demand and the change in the marginal relation between money demand and the market interest rate. An increase in the level of money demand that is not accommodated will induce either a deflation or a drop in real aggregate demand. An increase in the money supply solely to accommodate a higher level of money demand may be misinterpreted as an inflationary excess growth if the impact of the regulatory change is not properly taken into account. The change in the marginal relation between money demand and the market interest rate means, in this case, that the LM curve will become steeper. The monetary authority will need smaller changes in money supply to produce a given change in market interest rates and real aggregate demand. The prohibition of explicit deposit interest is nothing more or less than a particular form of price controls, albeit one with macroeconomic consequences. The classical analysis of price controls fails to recognize the power of competition to enforce a market-determined solution. Price controls are evaded fully or in part as agents substitute quality, advertising, or other forms of nonprice competition in place of forbidden, open price competition. There is ample evidence that the total legal restriction on demand deposit interest has resulted in only a partially effective economic restriction. See, for example, Barro and Santomero [1972], Becker [1975], Keen [1979], Santomero [1979], and Startz [1979]. (Other empirical evidence on the extent of bank monopoly power can be found in Heggestad and Mingo [1976 and 1977].) Below, I model such "evasion" in the banking industry. (By substituting "price" for "interest rate" and "output" for "deposits," most of the analytic framework has been applied in other industries, most notably airlines. The model developed here draws heavily on models presented in Douglas and Miller [1974], Schmalensee [1976, 1977], Stigler [1968], and White [1972].) The principal objective of this paper is to use a microeconomic model of monopolistic competition to examine the effect on a and /3, and thfiiefore^nJheJgxeLgndjnterestsensitivity of money demanc of changesinregulations. most e^pe^jallv a tioiTbf the ban on explicit interest. The arguments proceed in the

258

QUARTERLY JOURNAL OF ECONOMICS

following fashion. I argue that tbe world presently operates witb partial evasion of deposit rate controls, and tbat tbe observed market behavior strongly suggests monopolistic competition as a model of bank deposit markets. I then set up a Cbamberlinian model in wbicb botb implicit and explicit interest may be paid. Solution of the model leads to one equation describing the deposit market under partial evasion of controls and a second equation for the case in wbicb controls are eliminated. Existing empirical evidence on the first equation is used to bound parameters in order to answer several policy questions, including a regime switch to a market-determined deposit rate.
II. MOTIVATING THE MODEL

Four "stylized facts" about banking, taken jointly, suggest a monopolistically competitive model of tbe bank deposit market and appear to rule out a purely competitive model. i. The imposition by law of an effective deposit rate ceiling creates an excess unit profit on deposits, and therefore an incentive for each bank to expand its deposit liabilities. ii. Banks are able to attract deposits tbrougb nonprice competition, tbat is, through the payment of "implicit" interest. iii. Banks do not compete away all potential excess profits. Charter requirements limit free entry. iv. As market interest rates rise and the profit margin on deposits increases, banks engage in more active nonprice competition. Migbt perfect competition serve as an adequate description of tbe bank deposit market? Perfect competition implies zero economic profit. (I abstract from risk so tbat economic and accounting profits are identical; see Klein [1971] for a bank model with risk.) Tbe legal prohibition of deposit interest would create profits, except tbat banks can compete away profits tbrougb nonprice competition. In tbe U. S. banking system, tbe cost of providing implicit interest bas been consistently below tbe revenue from investing deposits; therefore, tbe market must be only imperfectly competitive.^ Tbe assertion tbat tbe
2. Point iv rules out one final possibility favoring a competitive model. Suppose that only limited implicit payments are possible, though the market is perfectly competitive, for example, that the only avenue of nonprice competition is through the provision of flowered checks and consumers have satiated their desires along this dimension. We would see some implicit interest being paid, but less than a competitive amount. We observe that banks increase the level of implicit payments when market interest rates rise. If the banks can increase the implicit rate, they could have done so previously, proving that implicit payments had not reached a limiting point and that the market must be imperfectly competitive.

COMPETITION AND INTEREST RATE CEILINGS

259

implicit interest rate is below the competitive level, which I treat as a given, bas been shown econometrically in Startz [1979], and can also be seen in tbe implicit interest measures in Barro and Santomero [1972], Becker [1975], and Santomero [1979]. Tbese studies all tend to show tbat banks return to depositors between one tbird and two tbirds of tbe yield from investing deposit funds.
III. MONOPOLISTIC COMPETITION FOR DEMAND DEPOSITS

I develop bere a model of bank bebavior in tbe demand deposit market. Tbe model is Chamberlin's system of monopolistic competition [Cbamberlin, 1962] slightly extended to meet tbe present need. Tbe notion of modeling bank behavior by monopolistic competition appeared at least as early as 1938 [Chandler], and much of tbe work since then is summarized by Alhadeff [1967] in Cbamberlin's Festschrift. It is convenient to define several symbols: r rx Tm n D D' r'^,r'm tbe market interest rate explicit interest rate on demand deposits implicit interest rate on demand deposits number of banks total demand deposits deposits of bank i deposit rates offered by bank i.

Tbe demand faced by bank k, sbown in (1), is positively related to tbe rates it offers, positively related to tbe differences between its rates and those offered by competitors, and negatively related to tbe market interest rate. All tbe coefficients, including tbe intercept, are positive. Arguments tbat are not relevant to tbe problem, such as income, are subsumed in tbe intercept. I make tbe usual assumption of symmetry. All banks face identical demands (and will also bave identical costs):
(1) Z3^=^ n n -

n-

I i^k

The parameters a^ and a^ represent the ability of tbe feth bank to attract deposits from tbe entire market or increase its market share by offering higher deposit rates tban its competitors. Tbe parameters Ax, Am, and Ar represgnt the cbange in market demand for a change

260

'

QUARTERLY JOURNAL OF ECONOMICS

^ in tbe respective average interest rates. I make the usual assumption of gross substitution, tbat Ax> Ar. Summation of tbe n bank demand curves (1) yields the market demand curve: Tbe revenue and cost functions for tbe /eth bank are ^rD'' (revenue) r',D>'-\-C(r'^D'') (cost). Because it permits analytic simplification, I make tbe assumption tbat implicit interest is paid in strict proportion to tbe size of tbe demand deposit and tbe (untrue) assumption tbat the cost function is linear. Tbus, C(-) = criD''. Tbe profit of tbe /jth bank is (3) Tf'' = 8rD>' Services provided tbrougb nonprice competition are almost certainly less valuable to tbe consumer tban would be an equivalent payment in dollars;" therefore, c > 1.1 measure implicit interest in terms of its relation to consumer demand, specifically normalizing tbe measure so tbat A^ = A^, written hereafter simply as A. The notion that monetary payments are a stronger competitive tool tban is payment in the form of toasters and umbrellas, expressed above as c > 1, drives tbe results ofthe model. A rational consumer is more responsive to an increase in explicit interest tban to an equally costly increase in implicit payments, as long as tbe payment by provision of some auxiliary service is truly "implicit," in tbe sense that tbe consumer receives more of the auxiliary good than he would freely consume and tbat tbe auxiliary good cannot be resold. An increase of one dollar in explicit interest due to an increased deposit yields one times tbe marginal utility of income worth of utils. One dollar's worth of already overly abundant toaster yields less than tbe marginal utility of income. Only wben banks can make nonmonetary payments that exactly match each consumer's unconstrained consumption bundle are implicit and explicit interest economically equivalent. When tbe explicit interest ceiling is an economically meaningful prohibition, the increase in deposits corresponding to one dollar of explicit interest can be attracted only at the expense of more tban one dollar's worth of toaster.
3. Essentially, nonprice competition limits the consumer choice set. Keen [1979] estimated, for special checking accounts, that services costing a bank one dollar were worth 59 cents to the consumer. Note, however, that implicit payments are effectively tax-free to nonbusiness customers; see Kimball [1977]. ^

COMPETITION AND INTEREST RATE CEILINGS

261

Each hank attempts to maximize profits. This requires the hank to make some sort of assumption ahout the oligopolistic hehavior of its competitors. I shall adopt the Chamherlinian "large group" assumption. Each hank is small in the market of any given competitor; in the absence of overt or tacit collusion, hank k assumes that the other ri and r'^ are determined exogenously with respect to its own actions. There are two interesting policy regimes. In the first, hanks are restrained by an effective ceiling on explicit interest. Each bank chooses the rate of implicit interest that maximizes its profits. In the second regime, each hank may freely set its explicit interest rate. (The linearity of the model generates a knife-edge property such that a hank will never pay implicit and explicit interest simultaneously if it can choose without limitations.) The first-order conditions for these cases are given in (4) and (5), respectively:
(4)

= 0 = (br - r^ -

a. binding explicit interest ceiling

(5)

dir'' = 0 = {brdr

n no explicit interest ceiling.

By applying symmetry and solving (4) or (5) simultaneously with market demand (2), we find the expressions for the equilibrium implicit or explicit rate in (6) and (7), respectively:
(6)

Onin + A + A/n

cA - amn + A/n H . + A + A/n]

^r +

n + bA/n

\r

(7)

A-\- A/n

-I- A + A/n\ \ \r + baxn + bA/n a^n -i- A -t- A/n r.

Equations (6) and (7) are interior solutions. There are also side conditions that r^ and 7r be nonnegative."*
4. Notice that I take advantage of the assumed linearity of demand to produce linear functions in (6) and (7). Given the usual empirical observation that money demand is nonlinear, the analogous empirical response functions (6) and (7) will also be nonlinear in r and r^:.

/ 262 /

QUARTERLY JOURNAL OF ECONOMICS

Perfect competition is a special limiting case of (6) and (7). Tbe monopolistically competitive solution approaches pure competition^ as either n goes to infinity (unlimited entry) or as a^ and a^ go to infinity (competing banks are perfect substitutes). In eitber case tbe competitive limits are r^ = (dr - rx)lc and r^ = hr, respectively. For convenient reference, rewrite (6) and (7) as (6') and (7'): (60 (70 r,n.= Otm-'ymrx + ^mhr rx=ax + Mr.

In the competitive case, a^ and a^ go to zero and fim and ^x go to lie and 1, respectively. More generally we bave 1/c < 7 ^ < 1 and ^mh < 7m- A one-point increase in tbe explicit interest ceiling lowers T by less tban a point. A one-point decrease in the market rate also m decreases rm by less than one point. Equal increases in tbe market rate and tbe explicit ceiling rate generate a net drop in implicit interest. It is noteworthy that jS^ may be greater tban 1; tbat is, implicit interest may be more responsive to changes in tbe market rate tban perfectly competitive explicit interest would be. However, a little algebra shows tbat if tbe monopolistic competition is sufficiently competitive, in tbe well-defined sense that amn is sufficiently large, tben /3m will be less tban 1. Tbere is actually a fair amount of empirical evidence bearing on equation (6). The implicit rate is unobservable, but the cost of implicit interest crm can be measured, albeit with some difficulty. Tbe regressions reported in Startz [1979] and Figure 2 in Santomero [1979] both compare crm to br in circumstances in whicb tbe explicit rate was zero. Botb papers suggest estimates of c/3m on tbe order of one third to one half.
IV. T H E EFFECTS OF REGULATORY CHANGE

Tbe impact of a regulatory cbange is measured in tbis model by its effect on tbe position and slope of the money demand schedule. Three sorts of "deregulation" are of particular interest. The first change is an increase in tbe ceiling on explicit interest to a bigber, but still binding, rate. The gradual introduction of NOW and automatic
5. The limiting properties of the model depend on the assumption of constant returns to scale and on the particular form of the demand schedule. For a model of a similar formal nature see Schmalensee [1976]. The limit of a Chamberlinian model always has price go to average cost, but does not in general have price driven to marginal cost.

COMPETITION AND INTEREST RATE CEILINGS

263

transfer service accounts, and the dramatic expansion of these accounts following the Monetary Control Act, is precisely an increase in the ceiling rate, though limited to certain classes of transaction accounts. The second change is an increase in the number of competitors in the "demand deposit" market, as thrifts receive authority to offer transaction accounts. The third interesting change is the aholition of an effective ceiling rate. An increase in the explicit ceiling rate leaves the slope of the LM curve unaffected, but does increase the level of money demand. Implicit interest falls, but by less than the increase in explicit interest. Deposit demand increases hy A(l- 7m)Arj, which must be positive since 7m < 1^ Permitting thrift institutions to offer checkable accounts will, in most geographic areas, greatly increase the number of participants in the demand deposit market. Examination of (6) shows that the implicit interest rate increases monotonically with n, as long as the pure monopoly effect is small (i.e., a^n > A/n). Increasing the number of banks may either increase or decrease the responsiveness of r^ to r. /35 increases monotonically with n if CjS < 1 (and a^n > AIn) and decreases monotonically if the reverse is true. The empirical evidence quoted above bears directly on this question and indicates that increasing the number of banks will indeed lead to increased responsiveness of the implicit deposit rate to the market interest rate. The third reform, abolition of an effective interest rate ceiling is qualitatively different. Equation (6) no longer applies. Instead, deposit rates are controlled by equation (7). A freely floating rate will result in greater money demand than with a controlled explicit ceiling. By the argument above, any increase in r^ up to the highest binding level increases demand, and of course any further increases in r^ indicated by (7) further increase demand. Will money demand be more or less responsive to changes in market rates after the switch from implicit to explicit interest? Common sense would argue that it ought to be less responsive, since one would expect offsetting changes in the deposit rate to be larger when banks compete through price competition instead of nonprice competition. The difference in dD/dr is A8(l3x - ^m)- In general, the sign of i8x ~ ^m depends on all the parameters of the model. However, one suspects that explicit payment is at least as effective a competitive tool as is implicit interest: i.e., a^ ^ a^. Two sufficient conditions for l^x > fim are a^ = a^ or a^ > a^ and c^^ < 1- Since the latter condition is probably true, the LM curve is probably more steeply sloped following the removal of price controls.

264 \
<'.

QUARTERLY JOURNAL OF ECONOMICS

V. SUMMARY

If the market for demand deposits were inherently perfectly competitive, and if the formal regulation against the payment of interest could not be evaded through methods of nonprice competition, then the impact of abolishing the deposit interest prohibition would be easy to understand. Ceteris paribus, demand for money would increase, and the demand for money would become less (negatively) responsive to the market interest rate. Recognition of the practice of paying interest implicitly suggests that the effect of the change will be much less dramatic than a simple picture of pure competition might suggest. In this paper I have developed a simple model of Chamberlinian monopolistic competition. This model can account for the observed practices of nonprice competition and yields predictions for market behavior if explicit price competition were to be allowed. Specifically, the effect of increasing the number of banks competing in a market, the effect of raising, but not eliminating, the ceiling rate on explicit interest, and the effect of eliminating the ceiling rate on explicit interest, can all be predicted. Especially strong results can be obtained by using previous empirical estimates to put constraints on the possible range of various theoretical parameters. The combination of the theoretical model and the empirical observation of observable behavior allows prediction about policy changes prior to their institution. The discussion in this paper has focused on the two related questions of how various policies determine the level of the deposit interest rate and determine the relation between the deposit rate and market interest rates. By immediate extension, the effect of these policies on the level of deposit demand and on the relation between deposit demand and the market interest rate is also determined. -.-- With a binding ceiling on explicit interest, the implicit interest rate will be positive but below the competitive explicit rate. The implicit rate increases with the market interest rate, but by less than a competitive explicit rate would. An increase in the legal ceiling on explicit interest leads to a partially offsetting drop in the implicit rate, but the net result will be an increased total deposit rate. An increase in the number of banks competing in a market will increase both the implicit interest rate and its responsiveness to the market rate. "" Changing from our current system of no explicit interest to one in which the market determines the deposit interest rate will result in increased deposit demand and increased responsiveness of the

COMPETITION AND INTEREST RATE CEILINGS deposit rate to the market rate. The magnitude of these changes will be far less than that suggested by a simple change from a world of no interest to a world of payment determined by perfect competition.
UNIVERSITY OF PENNSYLVANIA

265

REFERENCES Alhadeff, David A., "Monopolistic Competition and Banking Markets," in Monopolistic Competition Theory: Studies in Impact, Robert E. Kuenne, ed. (New York: John Wiley, 1967). Barro, Robert J., and Anthony M. Santomero, "Household Money Holdings and the Demand Deposit Rate," Journal of Money, Credit and Banking, IV (May 1972), 397-413. Becker, William E., "Determinants ofthe United States Currency-Demand Deposit Ratio," Journal of Finance, XXX (March 1975), 57-74. Chamberlin, Edward H., The Theory of Monopolistic Competition (Cambridge, MA: Harvard University Press, 1962). Chandler, Lester V., "Monopolistic Elements in Commercial Banking," Journal of Political Economy, XLVI (Feb. 1938), 1-22. Douglas, G. W., and J. C. Miller, "Quality Competition, Industry Equilibrium, and Efficiency in the Price-Constrained Airline Market," American Economic Review, LXIV (Sept. 1974), 657-69. Heggestad, Arnold A., and John J. Mingo, "Prices, Nonprices and Concentration in Commercial Banking," Journal of Money, Credit and Banking, VIII (Feb. 1976), 107-17. , and , "The Competitive Condition of U. S. Banking Markets and the Impact ofStructuralReform,"Joumo(o/'Fmace, XXXII (June 1977), 649-62. Keen, Howard, "Household Demand Deposits: Do Nonpecuniary Payments Make a Difference," Ph.D. thesis, Bryn Mawr College, 1979. Kimball, Ralph, "The Effect of the Income Tax Structure on Explicit Service Charges on NOW Accounts," internal memorandum. Federal Reserve Bank of Boston, July 1977. Klein, Michael A., "A Theory of the Banking Firm," Journal of Money, Credit and Banking, III (May 1971), 205-18. Santomero, Anthony M., "The Role of Transaction Costs and Rates of Return on the Demand Deposit Decision," Journal of Monetary Economics, V (July 1979), 343-64. Schmalensee, Richard, "A Model of Promotional Competition in Oligopoly," Reuiew of Economic Studies, XLIII (Oct. 1976), 493-507. , "Comparative Static Properties of Regulated Airline Oligopolies," Bell Journal of Economics, VIII (Autumn 1977), 565-76. Startz, Richard, "Implicit Interest on Demand Deposits," Journal of Monetary Economics, V (Oct. 1979), 515-34. Stigler, George, "Price and Non-Price Competition," Journal of Political Economy, LXXVI (Jan. 1968), 149-54. White, Lawrence J., "Quality Variation When Prices are Regulated," Bell Journal of Economics, III (Autumn 1972), 425-36.