# The Impact of Regulatory Measures on Commercial Bank Interest Rates: A Micro Analysis of the Barbados Case.

WENDELL MCCLEAN [*]

This study estimates the impact on commercial banks' interest-rate behavior of the more pervasive regulatory measures adopted by the Central Bank of Barbados. The results indicate that the cash ratio, the stipulated government securities ratio, and the savings deposit rate floor significantly impacted the loan rate for every bank. Generally, the deposit rate for any given bank has been responsive to fewer policy variables than the loan rate. The loan rates, though generally responsive to all policy variables other than the bank rate, have exhibited very low elasticities. The results indicated that the ceiling on the average lending rate, when it existed, depressed loan rates by less than 1 percent on average. This is largely attributable to the Central Bank's policy of adjusting the ceiling in line with market trends. (JEL E40)

Introduction

Since the establishment of the Central Bank of Barbados in 1972, commercial banks have been subjected to a wide variety of regulatory controls. [1] From August 1978 to the present, the Central Bank has prescribed the minimum interest rate payable by commercial banks on savings deposits. This interest rate floor has been applied to every class of deposit from March 15, 1994. Ceilings were imposed on deposits, in accordance with maturity and size of deposit, from October 1973 to October 1982. A ceiling was also imposed on the average loan rate from May 1976 to August 1991. There was a floor on the prime lending rate from May 1976 to June 1984.

Several restrictions have also been imposed on the composition of the assets portfolio of commercial banks. In addition to a minimum cash reserve ratio, commercial banks are required to hold a stipulated minimum amount of government securities, expressed as a proportion of deposits. Sectoral restrictions and other specific restrictions targeting the personal and distributive sectors have frequently been imposed. Commercial banks have also been subjected to foreign exchange restrictions, including Central Bank directives regarding the holding of foreign assets.

The impact and efficacy of these regulatory measures have been investigated in Williams [1996] and Worrell [1997]. The former focused on bank performance, as measured by profitability and market share, vis-a-vis other financial institutions. The latter addressed the impact of regulation on commercial bank interest rate and portfolio decisions. However, these studies were based entirely on hypothetico-deductive reasoning rather than econometric analysis. This study estimates the impact on commercial banks' interest-rate behavior of the more pervasive regulatory measures adopted by the Central Bank of Barbados.

This paper is organized as follows. The second section outlines various regulatory measures used since the inception of the Central Bank. The third section will discuss the theoretical perspective that has informed the study and presents a model of the banking firm. The fourth section describes the data set and presents the results of unit root tests and pair-wise Granger causality analysis. The fifth section reports the results obtained from a system of reduced-form equations that accord with our theoretical model. The study concludes in the sixth section with some evaluative comments on the efficacy of Central Bank regulation of commercial banking activity in Barbados.

Regulation of the Banking System

The complexities that risk and uncertainty impose on commercial bank assets and liabilities management have been compounded by Central Bank regulation. Since its inception in 1972, the Central Bank has imposed various restrictions on commercial banks as part of the government's stabilization program. These regulatory measures include cash reserves, security requirements, credit controls, and interest rate ceilings and floors.

A legal cash reserve requirement for commercial banks was introduced by the Central Bank in 1973 at 2 percent of the commercial bank deposit liabilities. This had to be met in the form of a noninterest-bearing cash deposit with the Central Bank. In September 1974, the ratio was increased to 4 percent and banks were allowed to include till cash as an eligible reserve asset. The ratio was increased to 6 percent in 1974 and to 8 percent in 1977, where it remained unchanged until May 1993 when it was reduced to 6 percent.

The Central Bank has also stipulated that commercial banks maintain a percentage of their total deposit liabilities in government securities (treasury bills and debentures). This ratio was introduced at 1 percent in 1973 and was increased on 13 occasions until it stood at 25 percent in October 1991. Thereafter, in September 1992, it was reduced to 23 percent and has remained at this level.

Credit controls have been used extensively by the Central Bank since its inception. In 1974, foreign borrowing by commercial banks to finance domestic credit operations was restricted. All such borrowing had to be referred to the Central Bank. This restriction remains to date. From time to time, the Central Bank has placed ceilings on the distributive and personal sectors. These ceilings were often expressed as a percentage of some previous outstanding level of credit to the particular sector. The Hire-Purchase, Credit Sales, and Hire-Control Act came into force in 1975. This gave the Central Bank power, under ministerial supervision, to control credit by issuing statutory orders. Such statutory orders issued by the Central Bank included stipulations on the minimum percentage down payment and maximum repayment period for items sold under hire-purchase contracts.

Rediscounting commercial bank paper with the Central Bank was introduced in 1973. Over the years, the Central Bank has varied its rates for discounts and advances in pursuit of its monetary objectives.

Prior to 1978, the government of Barbados, under the provisions of the Interest Rate Act 1970-47, was responsible for fixing the maximum rates to be paid on deposits with commercial banks. In 1978, the Central Bank attained this responsibility. The Central Bank placed floors on the savings deposit rate and ceilings on the term deposit rate and the weighted average loan rate. The ceiling on the term deposit rate was removed in 1982, as was the floor on the prime lending rate in 1986. The ceiling on the average lending rate was removed in 1991. The floor on the savings deposit rate remained in effect until March 1995 when it was replaced by a floor on all interest-bearing deposits. By June 1984 only three interest rate restrictions were in place and by August 1991, only two were in place. Only one restriction remains to date.

[II.sub.i] = [r.sub.1i][L.sub.i] + [r.sub.b][B.sub.i]--[r.sub.2i][D.sub.i],

Theoretical Perspective

There are two approaches that have been used extensively in literature to model commercial bank behavior. The earlier and still more widely used approach is rooted in the Markowitz-Tobin portfolio theory. This approach has been adopted by such authors as Hyman [1972], Parkin [1970], Pyle [1971], Hart and Jaffee [1974], and Kane and Malkiel [1965]. Proponents of the portfolio-theoretic approach generally assume that assets and deposits markets are perfectly competitive. This assumption, in turn, implies that the relevant behavioral mode in both of these markets is quantity setting. Hart and Jaffee [1974] have acknowledged that rate-setting behavior cannot be adequately treated within a portfolio model.

However, it has been argued that the quantity-setting assumption is not generally applicable to deposit markets. Sealey [1980] has contended that deposit markets are virtually always highly concentrated and that under such conditions, it is normal for commercial banks to set interest rates and face random deposit levels. Klein [1970] and Klein [1971] questioned the portfolio approach on the grounds that financial intermediaries operated in imperfectly competitive markets. They showed that some basic theorems of portfolio theory are not applicable to imperfect market structures. Sealey and Lindley [1977] have asserted that "the inadequacy of this approach stems from the total omission of production and cost constraints in determining the equilibrium output mix and scale size of the financial firm." We, therefore, submit that the portfolio approach is especially inappropriate in the context of Barbados, with its underdeveloped financial markets and oligopolistic commercial banking system.

The alternative approach adopted in this study focuses on commercial banks as regulated financial services firms. This approach has been exemplified by studies such as Pesek [1970], Klein [1970], Benston [1973], Sealey and Lindley [1977] and King [1986]. In the context of the Barbados regulatory environment, a basic representation of the profit function of a banking firm is shown as:

[[pi].sub.i] = [r.sub.1i][L.sub.i] + [r.sub.b][B.sub.i] - [r.sub.2i][D.sub.i], (1) where [[pi].sub.i] is profit at bank i; [D.sub.i] are deposits with bank i; [L.sub.i] are loans with bank i; [r.sub.1i] is the average rate of interest on loans at bank i; [r.sub.2i] is the average rate of interest on deposits at bank i; [r.sub.b] is the rate of interest on government securities; and [B.sub.i] is the nominal value of government securities held by commercial bank i.

In a regulated environment characterized by a ceiling on the loan rate of interest and a floor on the deposit rate of interest, there are four possible scenarios:

1) Case 1 has a binding ceiling and a nonbinding floor;

2) Case 2 has a nonbinding ceiling and a binding floor;

3) Case 3 has a binding ceiling and a binding floor; and

4) Case 4 has a nonbinding ceiling and a nonbinding floor.

Case 4 is tantamount to the absence of interest rate regulation. In Case 1, the loan rate of interest would predominantly be determined by regulatory measures and other exogenous variables. In such circumstances, commercial banks would determine the interest rate on loans prior to the interest rate on deposits. The banks would also choose the level of loans that is consistent with constrained profit maximization. Prior to March 1994, Cases 2 and 3 could not have been generally applicable in Barbados because of the absence of a floor on the interest rate on time deposits. Hence, to the extent that interest rate regulation has had an effect on commercial banking behavior in Barbados, Case 1 would have characterized the Barbadian experience. However, in this regard, please note the official view of the content of interest rate regulation. Worrell [1997] states:

"The Central Bank of Barbados' direct interventions with respect to interest rates reflected the underlying market conditions: The interest rate rose when credit conditions tightened and fell when credit eased....Spreads between deposit and loan rates were adjusted in line with the banks' requirements. The loan rate was controlled in name only; exemptions were pervasive and the actual was frequently above the 'controlled' maximum."

[r.sub.1i] = f([phi],[r.sub.1-i]); [f.sub.[r.sub.1-i]] [greater than] 0 , (2)

Examining a related issue, Williams [1996] concluded:

"The hypothesis of cost-axiomatic pricing is supported by the results for the banking system in Barbados. Results suggest accommodation by regulators in their approach to the setting of interest rate ceilings and floors and contradicts the conventional literature which posits that interest rate regulation impacts adversely on bank profitability."

The foregoing observations suggest that the Central Bank interest rate policy has been implemented in a manner that has afforded commercial banks with a high degree of discretion in regard to fixing the loan rate of interest. Hence, it is possible that Case 4 closely approximates the Barbadian experience.

Given Case 1, as qualified by Williams [1996] and Worrell [1997], the interest rate on a commercial bank's loans may be represented in functional terms as:

where [phi] is a vector of policy and other exogenous variables, including foreign interest rates, and -i are banks other than bank i.

In Barbados, commercial banks normally fix the interest rate on deposits and accept all deposits. For an individual bank, the supply function for deposits may be represented as:

[D.sub.i] = D([r.sub.2i], [r.sub.2Ai], TD); [D.sub.[r.sub.2]], [D.sub.TD] [greater than] 0, [D.sub.[r.sub.2A]] [less than] 0 , (3)

where [r.sub.2i] is the average rate of interest on deposits at bank i; [r.sub.2Ai] is a vector of interest rates on alternative financial assets that may be held by the public, including deposits at other banks; TD are total deposits within the banking system; and (--) indicates exogenity.

Assuming that commercial banks are desirous of holding no excess reserves, [2] and abstracting from nondeposit sources of funding, then profit maximization may be represented as subject to the constraint:

[L.sub.i] = (1 - [[rho].sub.c] - [[rho].sub.s]) [D.sub.i] , (4)

where [[rho].sub.c] is the required cash reserve ratio, and [[rho].sub.s] is the stipulated government security ratio.

From the foregoing, abstracting pro tempore from (2) obtains the Lagrangean Function: [3]

and

[L in script form] = [r.sub.1i] [L.sub.i] + [[r.sub.b][[rho].sub.s] - [r.sub.2i]]D([r.sub.2i], [r.sub.2Ai], TD) + [lambda] [[L.sub.i] - D([r.sub.2i], [r.sub.2Ai], TD) (1 -[[rho].sub.c] - [[rho].sub.s])] . (5)

Equation (5) implies the following first order conditions for profit maximization:

[delta][L in script form]/[delta][L.sub.i] = [r.sub.1i] + [lambda] = 0 , (6)

[delta][L in script form]/[delta][r.sub.2i] = -D([r.sub.2i], [r.sub.2Ai], TD) + [[r.sub.b][[rho].sub.s] - [r.sub.2i] - [lambda](1 - [[rho].sub.c] - [[rho].sub.s])][D.sub.[r.sub.2i]] = 0 , (7)

and [delta][L in script form]/[delta][lambda] = [L.sub.i] - D([r.sub.2i], [r.sub.2Ai], TD)(1 - [[rho].sub.c] - [[rho].sub.s]) = 0. (8)

By differentiating (6), (7), and (8) totally, we obtain:

d[r.sub.1i] + d[lambda] = 0, (9)

[D.sub.[r.sub.2i]] [[[rho].sub.s] [dr.sub.b] + [r.sub.b]d[[rho].sub.s] + [lambda](d[[rho].sub.c] + d[[rho].sub.s]) - 2[dr.sub.2i] - [dr.sub.2Ai]] - [D.sub.TD]dTD - (1 - [[rho].sub.c] - [[rho].sub.s]) [D.sub.[r.sub.2i]] d[lambda] = 0, (10)

and

d[L.sub.i] - (1 - [[rho].sub.c] - [[rho].sub.s])([D.sub.[r.sub.2i]] [dr.sub.2i] + [D.sub.[r.sub.2Ai]] [dr.sub.2Ai] + [D.sub.TD]dTD) + [D.sub.i](d[[rho].sub.c] + d[[rho].sub.s]) = 0. (11)

Equations (9) and (10) imply that:

[dr.sub.2i] = 1/2 [[[rho].sub.s][dr.sub.b] + (1 - [[rho].sub.c] - [[rho].sub.s])[dr.sub.1i] + [lambda]d[[rho].sub.c] + ([lambda] + [r.sub.b])d[[rho].sub.s] - [dr.sub.2Ai] - 1/[D.sub.[r.sub.2i]]([D.sub.TD]dTD)]. (12)

Equations (11) and (12) may be expressed in general functional form as:

[r.sub.2i] = r([r.sub.b], [r.sub.1i], [[rho].sub.c], [[rho].sub.s], [r.sub.2Ai], TD); [r.sub.b], [r.sub.[r.sub.1i]], [r.sub.[[rho].sub.c]], [r.sub.[[rho].sub.s]], [r.sub.[r.sub.A2i]] [greater than] 0, [r.sub.TD] [less than] 0, (13)

and

[L.sub.i] = L([r.sub.2i], [r.sub.2Ai], TD, [[rho].sub.c], [[rho].sub.s]); [L.sub.[r.sub.2i]], [L.sub.TD] [greater than] 0, [r.sub.[r.sub.2Ai]], [L.sub.[[rho].sub.c]], [L.sub.[[rho].sub.s]] [less than] 0. (14)

Equations (2), (13), and (14) represent a recursive system of structural equations, with a causal ordering which indicates that the impact of regulatory measures on interest rates may be derived from the reduced-form variants of (2) and (13). However, in this regard, it should be noted that commercial banks satisfy their stipulated government securities ratio predominantly by holding treasury bills in the context where a high proportion of outstanding treasury bills are held by commercial banks. This situation and the Central Bank's tendency to allow market forces to determine the treasury bill rate (see Worrell [1997, pp. 31]) suggests that the treasury bill rate is endogenous, and the set of exogenous variables that enter a reduced-form equation for the treasury bill rate should essentially be the same as those included in the loan and deposit rates functions.

Data Definition and Properties

The data used in this study are monthly series from December 1976 to December 1996 and were obtained from the Central Bank of Barbados. The software package used in this study is Econometric Views 3.0.

Included in the data set are 13 endogenous variables: the treasury bill rate (TBR) and loan and deposit interest rates for six of the seven banks currently operating in Barbados. The banks are labeled B1 to B6. The loan rates are averages (ALR) and the deposit rates (RT) are modal values of three-month time deposits. A plot of both rates for each bank reveals similarity in the behavior of interest rates among the banks.

As indicated in our theoretical model, the set of exogenous variables comprises foreign interest rates, total deposits within the commercial banking system (AB DTO), and policy variables. The overnight London interbank offer rate (LIBOR ON) was chosen as representative of foreign rates. Five policy variables were included in our analysis: the bank rate (BR), the average lending rate ceiling (ALRC), the savings deposits rate floor (SDRF), the required cash reserve ratio (RHO), and the stipulated government securities ratio (SGSR). The ceiling on the average lending rate is the only policy measure that was not in force throughout the period of our study, 1986:01 to 1996:12. It was discontinued in June 1990. Hence, ALRC is a constructed variable comprising the interest rate ceiling for the period up to June 1990 and unit value ([e.sup.0]) thereafter. This enabled us to take logarithms of ALRC to obtain a modulated on-off variable. Preliminary analysis indicated that this variable outperformed an ordinary on-off dummy variable.

Order of Integration

We first investigated the order of integration of these variables using the augmented Dickey-Fuller and Phillips-Perron tests for nonstationarity. The results of both tests indicated that all variables, except RHO, were found to be integrated of order 1, I(1). Although the results indicate that the interest rates are I(1), for the estimation of the model, we entered them in their levels. The main reason for this stems from the discrete nature of the individual series. There are periods when the rates do not change, and when they do change, the movement is a discrete jump up or down. This has some implications for the above standard unit root tests. Over any period when the rate at an individual bank does not change, then [delta][X.sub.t] = [[epsilon].sub.t] = 0, and the second moment of [[epsilon].sub.t] does not exist--a violation of one of the assumptions of the standard tests.

Granger Causality

We next investigated Granger causality between pairs of interest rates. Granger causality is used here solely to ascertain whether loan rates can help forecast deposit rates and vice versa. Recall from our theoretical model that the loan rate of the same bank and the deposit rates of other banks are explanatory variables in the individual bank's deposit rate equation (see (12)). The reverse is not true of the loan rate equation (see (2)). Hence, the test of Granger causality is intended as a preliminary investigation of this hypothesis.

The results presented in Table 1 indicate that every interest rate variable is Granger-caused by at least one of the other variables. The general pattern of causation suggests a high degree of simultaneity in the relationships among the variables and, except in the case of the results for B6, is consonant with the structure of our theoretical model. For B1 to B5, there is unidirectional Granger causality between loan and deposit rates for the same bank, running from the loan rate to the deposit rate. In the case of B6, unidirectional causality also exists but in the opposite direction. However, any adaptation of the model to accommodate the causal patterns indicated in Table 1 would not alter the system of reduced-form equations implied by our theoretical model.

Another interesting insight offered from the pattern of Granger causality is that not all the banks operate in the same manner. B6 is a leader in the market. Its loan rate Granger-caused the deposit rate of all other banks, while its deposit rate is not Granger-caused by any other bank's loan rate. B3 is an unmitigated follower.

Econometric Results

Here, we report the results pertaining to a 13-equation system of reduced-form equations, comprising loan (Table 2) and deposit (Table 3) interest rate equations for the six commercial banks and an equation for the treasury bill rate (Table 4). Tables 2 and 3 represent equations, with diagnostic statistics also shown. The exogenous variables initially included in all 13 equations comprise the five policy variables, the foreign rate of interest, and total deposits within the commercial banking system. The system was estimated using an iterative seemingly unrelated regression procedure. This allows for the possibility of contemporaneous correlation between the disturbances in different interest rate equations. Each iteration of the seemingly unrelated regression reestimates the parameters of the model after transforming the equations to remove the correlation across the residuals. Iteration to convergence gives the maximum likelihood estimates.

The results indicate that the loan rate for all banks and the deposit rate for three of the six banks were repressed by the ceiling on the average interest rate on loans, albeit marginally so. The bank rate has had a significant effect on the loan rate of only one bank. This bank is the only bank whose deposit rate has not been significantly influenced by the bank rate. The cash ratio, the stipulated government securities ratio, and the savings deposit rate floor impacted significantly on the loan rate for every bank. However, these three measures had a less pervasive influence on the deposit rate. Generally, the deposit rate for any given bank has been responsive to fewer policy variables than the loan rate. The loan rates, though generally responsive to all policy variables other than the bank rate, have exhibited very low elasticities. This is evidently attributable to the constraining influence of the ceiling on the average lending rate. The loan and deposit rates have been most responsive to changes in the stipulated government securities ratio. The cash reserve ratio has had the second strongest effect.

All five policy variables have had a significant effect on the treasury bill rate. The average lending rate ceiling and the cash ratio have had effects on the treasury bill rate that are opposite to the respective effects on loan and deposit rates. All other policy variables have had a positive impact on the treasury bill and loan and deposit rates. All of the results are intuitively plausible.

Evaluation of Results

The ineffectiveness of the bank rate vis-a-vis commercial bank loan rates is one of the noteworthy findings of this study. The reason for this phenomenon is that commercial banks do not make substantial use of the Central Bank's rediscount window. Interbank borrowing is usually available to the extent necessary to enable any given bank to manage its liquidity requirements. Hence, the cost of Central Bank accommodation is largely inconsequential to the banks. In these circumstances, the Central Bank should pursue a policy in which the bank rate follows the market. Any attempt to use the bank rate to lead the market is likely to fail and could result in misalignment of the bank rate with other interest rates.

The ceiling on the average lending rate, when it existed, depressed the loan rate by less than 1 percent on average. This is largely attributable to the Central Bank's policy of adjusting the ceiling in line with market trends. The policy was discontinued as a concession to market liberalization forces. There is no evident reason for a return to the former policy.

The cash reserves and stipulated government security ratios have had a highly significant impact on interest rates and may be used effectively to target interest rates. However, these measures can also impact bank liquidity and the balance of payments. The savings deposit rate floor can be used to influence income distribution or the structure and level of interest rates. Hence, the appropriate use of these instruments should be determined case by case.

(*.) Central Bank of Barbados and University of the West Indies--West Indies.

Footnotes

(1.) The regulation of commercial bank interest rates predated the existence of the Central Bank.

(2.) This is a simplification whose implications should be investigated in a further study. Banks do carry significant and varying levels of excess reserves.

(3.) When we abstract from (2), [r.sub.li] is presented as a predetermined variable.

References

Benston, George J. "The Optimal Banking Structure: Theory and Evidence," Journal of Banking Research, 3, Winter 1973, Pp. 220-37.

Hart, Oliver D.; Jaffee, Dwight M. "On the Application of Portfolio Theory to Depository Financial Intermediaries," Review of Economic Studies, 41, January 1974, pp. 129-47.

Hyman, David N. "A Behavioral Model for Financial Intermediation," Economic and Business Bulletin, 24, 3, 1972, pp. 9-17.

Kane E. J.; Malkiel, B. G. "Bank Portfolio Allocation, Deposit Variability, and the Availability Doctrine," Quarterly Journal of Economics, 79, February 1965, pp. 113-34.

King, Stephen R. "Monetary Transmission: Through Bank Loans or Bank Liabilities?," Journal of Money, Credit, and Banking, 18, 3, 1986, pp. 290-303.

Klein, John J. "The State of Finance Field Methodology: A Discussion," Journal of Finance, 25, May 1970, pp. 322-5.

Klein, Michael A. "A Theory of the Banking Firm," Journal of Money, Credit, and Banking, 3, May 1971, pp. 205-18.

Parkin, Michael J. "Discount House Portfolio and Debt Selection," Review of Economics Studies, 37, October 1970, pp. 469-97.

Pesek, Boris P. "Bank's Supply Function and the Equilibrium Quantity of Money," Canadian Journal of Economics, 3, 3, 1970, pp. 357-85.

Pyle, David H. "On the Theory of Financial Intermediation, "Journal of Finance, 26, June 1971, pp. 737-47.

Sealey, C. W. Jr. "Deposit Rate Setting, Risk Aversion, and the Theory of Depository Financial Intermediaries," The Journal of Finance, 35, 5, 1980, pp. 1139-54.

Sealey, C. W. Jr.; Lindley, James T. "Inputs, Outputs, and a Theory of Production and Cost at Depository Financial Institutions," The Journal of Finance, 32, 4, 1977, pp. 1251-66.

Williams, Marion V. Liberalizing a Regulated Banking System: The Caribbean Case, Avebury, United Kingdom: Aldershot, 1996.

Worrell, DeLisle. "Role Models for Monetary Policy in the Caribbean: Comparing Caricom Central Banks," in Harold Codrington; Roland Craigwell; Cleviston Haynes, eds., Central Banking in Barbados: Reflections and Challenges, Barbados, West Indies: Central Bank of Barbados, 1997.

This study estimates the impact on commercial banks' interest-rate behavior of the more pervasive regulatory measures adopted by the Central Bank of Barbados. The results indicate that the cash ratio, the stipulated government securities ratio, and the savings deposit rate floor significantly impacted the loan rate for every bank. Generally, the deposit rate for any given bank has been responsive to fewer policy variables than the loan rate. The loan rates, though generally responsive to all policy variables other than the bank rate, have exhibited very low elasticities. The results indicated that the ceiling on the average lending rate, when it existed, depressed loan rates by less than 1 percent on average. This is largely attributable to the Central Bank's policy of adjusting the ceiling in line with market trends. (JEL E40)

Introduction

Since the establishment of the Central Bank of Barbados in 1972, commercial banks have been subjected to a wide variety of regulatory controls. [1] From August 1978 to the present, the Central Bank has prescribed the minimum interest rate payable by commercial banks on savings deposits. This interest rate floor has been applied to every class of deposit from March 15, 1994. Ceilings were imposed on deposits, in accordance with maturity and size of deposit, from October 1973 to October 1982. A ceiling was also imposed on the average loan rate from May 1976 to August 1991. There was a floor on the prime lending rate from May 1976 to June 1984.

Several restrictions have also been imposed on the composition of the assets portfolio of commercial banks. In addition to a minimum cash reserve ratio, commercial banks are required to hold a stipulated minimum amount of government securities, expressed as a proportion of deposits. Sectoral restrictions and other specific restrictions targeting the personal and distributive sectors have frequently been imposed. Commercial banks have also been subjected to foreign exchange restrictions, including Central Bank directives regarding the holding of foreign assets.

The impact and efficacy of these regulatory measures have been investigated in Williams [1996] and Worrell [1997]. The former focused on bank performance, as measured by profitability and market share, vis-a-vis other financial institutions. The latter addressed the impact of regulation on commercial bank interest rate and portfolio decisions. However, these studies were based entirely on hypothetico-deductive reasoning rather than econometric analysis. This study estimates the impact on commercial banks' interest-rate behavior of the more pervasive regulatory measures adopted by the Central Bank of Barbados.

This paper is organized as follows. The second section outlines various regulatory measures used since the inception of the Central Bank. The third section will discuss the theoretical perspective that has informed the study and presents a model of the banking firm. The fourth section describes the data set and presents the results of unit root tests and pair-wise Granger causality analysis. The fifth section reports the results obtained from a system of reduced-form equations that accord with our theoretical model. The study concludes in the sixth section with some evaluative comments on the efficacy of Central Bank regulation of commercial banking activity in Barbados.

Regulation of the Banking System

The complexities that risk and uncertainty impose on commercial bank assets and liabilities management have been compounded by Central Bank regulation. Since its inception in 1972, the Central Bank has imposed various restrictions on commercial banks as part of the government's stabilization program. These regulatory measures include cash reserves, security requirements, credit controls, and interest rate ceilings and floors.

A legal cash reserve requirement for commercial banks was introduced by the Central Bank in 1973 at 2 percent of the commercial bank deposit liabilities. This had to be met in the form of a noninterest-bearing cash deposit with the Central Bank. In September 1974, the ratio was increased to 4 percent and banks were allowed to include till cash as an eligible reserve asset. The ratio was increased to 6 percent in 1974 and to 8 percent in 1977, where it remained unchanged until May 1993 when it was reduced to 6 percent.

The Central Bank has also stipulated that commercial banks maintain a percentage of their total deposit liabilities in government securities (treasury bills and debentures). This ratio was introduced at 1 percent in 1973 and was increased on 13 occasions until it stood at 25 percent in October 1991. Thereafter, in September 1992, it was reduced to 23 percent and has remained at this level.

Credit controls have been used extensively by the Central Bank since its inception. In 1974, foreign borrowing by commercial banks to finance domestic credit operations was restricted. All such borrowing had to be referred to the Central Bank. This restriction remains to date. From time to time, the Central Bank has placed ceilings on the distributive and personal sectors. These ceilings were often expressed as a percentage of some previous outstanding level of credit to the particular sector. The Hire-Purchase, Credit Sales, and Hire-Control Act came into force in 1975. This gave the Central Bank power, under ministerial supervision, to control credit by issuing statutory orders. Such statutory orders issued by the Central Bank included stipulations on the minimum percentage down payment and maximum repayment period for items sold under hire-purchase contracts.

Rediscounting commercial bank paper with the Central Bank was introduced in 1973. Over the years, the Central Bank has varied its rates for discounts and advances in pursuit of its monetary objectives.

Prior to 1978, the government of Barbados, under the provisions of the Interest Rate Act 1970-47, was responsible for fixing the maximum rates to be paid on deposits with commercial banks. In 1978, the Central Bank attained this responsibility. The Central Bank placed floors on the savings deposit rate and ceilings on the term deposit rate and the weighted average loan rate. The ceiling on the term deposit rate was removed in 1982, as was the floor on the prime lending rate in 1986. The ceiling on the average lending rate was removed in 1991. The floor on the savings deposit rate remained in effect until March 1995 when it was replaced by a floor on all interest-bearing deposits. By June 1984 only three interest rate restrictions were in place and by August 1991, only two were in place. Only one restriction remains to date.

[II.sub.i] = [r.sub.1i][L.sub.i] + [r.sub.b][B.sub.i]--[r.sub.2i][D.sub.i],

Theoretical Perspective

There are two approaches that have been used extensively in literature to model commercial bank behavior. The earlier and still more widely used approach is rooted in the Markowitz-Tobin portfolio theory. This approach has been adopted by such authors as Hyman [1972], Parkin [1970], Pyle [1971], Hart and Jaffee [1974], and Kane and Malkiel [1965]. Proponents of the portfolio-theoretic approach generally assume that assets and deposits markets are perfectly competitive. This assumption, in turn, implies that the relevant behavioral mode in both of these markets is quantity setting. Hart and Jaffee [1974] have acknowledged that rate-setting behavior cannot be adequately treated within a portfolio model.

However, it has been argued that the quantity-setting assumption is not generally applicable to deposit markets. Sealey [1980] has contended that deposit markets are virtually always highly concentrated and that under such conditions, it is normal for commercial banks to set interest rates and face random deposit levels. Klein [1970] and Klein [1971] questioned the portfolio approach on the grounds that financial intermediaries operated in imperfectly competitive markets. They showed that some basic theorems of portfolio theory are not applicable to imperfect market structures. Sealey and Lindley [1977] have asserted that "the inadequacy of this approach stems from the total omission of production and cost constraints in determining the equilibrium output mix and scale size of the financial firm." We, therefore, submit that the portfolio approach is especially inappropriate in the context of Barbados, with its underdeveloped financial markets and oligopolistic commercial banking system.

The alternative approach adopted in this study focuses on commercial banks as regulated financial services firms. This approach has been exemplified by studies such as Pesek [1970], Klein [1970], Benston [1973], Sealey and Lindley [1977] and King [1986]. In the context of the Barbados regulatory environment, a basic representation of the profit function of a banking firm is shown as:

[[pi].sub.i] = [r.sub.1i][L.sub.i] + [r.sub.b][B.sub.i] - [r.sub.2i][D.sub.i], (1) where [[pi].sub.i] is profit at bank i; [D.sub.i] are deposits with bank i; [L.sub.i] are loans with bank i; [r.sub.1i] is the average rate of interest on loans at bank i; [r.sub.2i] is the average rate of interest on deposits at bank i; [r.sub.b] is the rate of interest on government securities; and [B.sub.i] is the nominal value of government securities held by commercial bank i.

In a regulated environment characterized by a ceiling on the loan rate of interest and a floor on the deposit rate of interest, there are four possible scenarios:

1) Case 1 has a binding ceiling and a nonbinding floor;

2) Case 2 has a nonbinding ceiling and a binding floor;

3) Case 3 has a binding ceiling and a binding floor; and

4) Case 4 has a nonbinding ceiling and a nonbinding floor.

Case 4 is tantamount to the absence of interest rate regulation. In Case 1, the loan rate of interest would predominantly be determined by regulatory measures and other exogenous variables. In such circumstances, commercial banks would determine the interest rate on loans prior to the interest rate on deposits. The banks would also choose the level of loans that is consistent with constrained profit maximization. Prior to March 1994, Cases 2 and 3 could not have been generally applicable in Barbados because of the absence of a floor on the interest rate on time deposits. Hence, to the extent that interest rate regulation has had an effect on commercial banking behavior in Barbados, Case 1 would have characterized the Barbadian experience. However, in this regard, please note the official view of the content of interest rate regulation. Worrell [1997] states:

"The Central Bank of Barbados' direct interventions with respect to interest rates reflected the underlying market conditions: The interest rate rose when credit conditions tightened and fell when credit eased....Spreads between deposit and loan rates were adjusted in line with the banks' requirements. The loan rate was controlled in name only; exemptions were pervasive and the actual was frequently above the 'controlled' maximum."

[r.sub.1i] = f([phi],[r.sub.1-i]); [f.sub.[r.sub.1-i]] [greater than] 0 , (2)

Examining a related issue, Williams [1996] concluded:

"The hypothesis of cost-axiomatic pricing is supported by the results for the banking system in Barbados. Results suggest accommodation by regulators in their approach to the setting of interest rate ceilings and floors and contradicts the conventional literature which posits that interest rate regulation impacts adversely on bank profitability."

The foregoing observations suggest that the Central Bank interest rate policy has been implemented in a manner that has afforded commercial banks with a high degree of discretion in regard to fixing the loan rate of interest. Hence, it is possible that Case 4 closely approximates the Barbadian experience.

Given Case 1, as qualified by Williams [1996] and Worrell [1997], the interest rate on a commercial bank's loans may be represented in functional terms as:

where [phi] is a vector of policy and other exogenous variables, including foreign interest rates, and -i are banks other than bank i.

In Barbados, commercial banks normally fix the interest rate on deposits and accept all deposits. For an individual bank, the supply function for deposits may be represented as:

[D.sub.i] = D([r.sub.2i], [r.sub.2Ai], TD); [D.sub.[r.sub.2]], [D.sub.TD] [greater than] 0, [D.sub.[r.sub.2A]] [less than] 0 , (3)

where [r.sub.2i] is the average rate of interest on deposits at bank i; [r.sub.2Ai] is a vector of interest rates on alternative financial assets that may be held by the public, including deposits at other banks; TD are total deposits within the banking system; and (--) indicates exogenity.

Assuming that commercial banks are desirous of holding no excess reserves, [2] and abstracting from nondeposit sources of funding, then profit maximization may be represented as subject to the constraint:

[L.sub.i] = (1 - [[rho].sub.c] - [[rho].sub.s]) [D.sub.i] , (4)

where [[rho].sub.c] is the required cash reserve ratio, and [[rho].sub.s] is the stipulated government security ratio.

From the foregoing, abstracting pro tempore from (2) obtains the Lagrangean Function: [3]

and

[L in script form] = [r.sub.1i] [L.sub.i] + [[r.sub.b][[rho].sub.s] - [r.sub.2i]]D([r.sub.2i], [r.sub.2Ai], TD) + [lambda] [[L.sub.i] - D([r.sub.2i], [r.sub.2Ai], TD) (1 -[[rho].sub.c] - [[rho].sub.s])] . (5)

Equation (5) implies the following first order conditions for profit maximization:

[delta][L in script form]/[delta][L.sub.i] = [r.sub.1i] + [lambda] = 0 , (6)

[delta][L in script form]/[delta][r.sub.2i] = -D([r.sub.2i], [r.sub.2Ai], TD) + [[r.sub.b][[rho].sub.s] - [r.sub.2i] - [lambda](1 - [[rho].sub.c] - [[rho].sub.s])][D.sub.[r.sub.2i]] = 0 , (7)

and [delta][L in script form]/[delta][lambda] = [L.sub.i] - D([r.sub.2i], [r.sub.2Ai], TD)(1 - [[rho].sub.c] - [[rho].sub.s]) = 0. (8)

By differentiating (6), (7), and (8) totally, we obtain:

d[r.sub.1i] + d[lambda] = 0, (9)

[D.sub.[r.sub.2i]] [[[rho].sub.s] [dr.sub.b] + [r.sub.b]d[[rho].sub.s] + [lambda](d[[rho].sub.c] + d[[rho].sub.s]) - 2[dr.sub.2i] - [dr.sub.2Ai]] - [D.sub.TD]dTD - (1 - [[rho].sub.c] - [[rho].sub.s]) [D.sub.[r.sub.2i]] d[lambda] = 0, (10)

and

d[L.sub.i] - (1 - [[rho].sub.c] - [[rho].sub.s])([D.sub.[r.sub.2i]] [dr.sub.2i] + [D.sub.[r.sub.2Ai]] [dr.sub.2Ai] + [D.sub.TD]dTD) + [D.sub.i](d[[rho].sub.c] + d[[rho].sub.s]) = 0. (11)

Equations (9) and (10) imply that:

[dr.sub.2i] = 1/2 [[[rho].sub.s][dr.sub.b] + (1 - [[rho].sub.c] - [[rho].sub.s])[dr.sub.1i] + [lambda]d[[rho].sub.c] + ([lambda] + [r.sub.b])d[[rho].sub.s] - [dr.sub.2Ai] - 1/[D.sub.[r.sub.2i]]([D.sub.TD]dTD)]. (12)

Equations (11) and (12) may be expressed in general functional form as:

[r.sub.2i] = r([r.sub.b], [r.sub.1i], [[rho].sub.c], [[rho].sub.s], [r.sub.2Ai], TD); [r.sub.b], [r.sub.[r.sub.1i]], [r.sub.[[rho].sub.c]], [r.sub.[[rho].sub.s]], [r.sub.[r.sub.A2i]] [greater than] 0, [r.sub.TD] [less than] 0, (13)

and

[L.sub.i] = L([r.sub.2i], [r.sub.2Ai], TD, [[rho].sub.c], [[rho].sub.s]); [L.sub.[r.sub.2i]], [L.sub.TD] [greater than] 0, [r.sub.[r.sub.2Ai]], [L.sub.[[rho].sub.c]], [L.sub.[[rho].sub.s]] [less than] 0. (14)

Equations (2), (13), and (14) represent a recursive system of structural equations, with a causal ordering which indicates that the impact of regulatory measures on interest rates may be derived from the reduced-form variants of (2) and (13). However, in this regard, it should be noted that commercial banks satisfy their stipulated government securities ratio predominantly by holding treasury bills in the context where a high proportion of outstanding treasury bills are held by commercial banks. This situation and the Central Bank's tendency to allow market forces to determine the treasury bill rate (see Worrell [1997, pp. 31]) suggests that the treasury bill rate is endogenous, and the set of exogenous variables that enter a reduced-form equation for the treasury bill rate should essentially be the same as those included in the loan and deposit rates functions.

Data Definition and Properties

The data used in this study are monthly series from December 1976 to December 1996 and were obtained from the Central Bank of Barbados. The software package used in this study is Econometric Views 3.0.

Included in the data set are 13 endogenous variables: the treasury bill rate (TBR) and loan and deposit interest rates for six of the seven banks currently operating in Barbados. The banks are labeled B1 to B6. The loan rates are averages (ALR) and the deposit rates (RT) are modal values of three-month time deposits. A plot of both rates for each bank reveals similarity in the behavior of interest rates among the banks.

As indicated in our theoretical model, the set of exogenous variables comprises foreign interest rates, total deposits within the commercial banking system (AB DTO), and policy variables. The overnight London interbank offer rate (LIBOR ON) was chosen as representative of foreign rates. Five policy variables were included in our analysis: the bank rate (BR), the average lending rate ceiling (ALRC), the savings deposits rate floor (SDRF), the required cash reserve ratio (RHO), and the stipulated government securities ratio (SGSR). The ceiling on the average lending rate is the only policy measure that was not in force throughout the period of our study, 1986:01 to 1996:12. It was discontinued in June 1990. Hence, ALRC is a constructed variable comprising the interest rate ceiling for the period up to June 1990 and unit value ([e.sup.0]) thereafter. This enabled us to take logarithms of ALRC to obtain a modulated on-off variable. Preliminary analysis indicated that this variable outperformed an ordinary on-off dummy variable.

Order of Integration

We first investigated the order of integration of these variables using the augmented Dickey-Fuller and Phillips-Perron tests for nonstationarity. The results of both tests indicated that all variables, except RHO, were found to be integrated of order 1, I(1). Although the results indicate that the interest rates are I(1), for the estimation of the model, we entered them in their levels. The main reason for this stems from the discrete nature of the individual series. There are periods when the rates do not change, and when they do change, the movement is a discrete jump up or down. This has some implications for the above standard unit root tests. Over any period when the rate at an individual bank does not change, then [delta][X.sub.t] = [[epsilon].sub.t] = 0, and the second moment of [[epsilon].sub.t] does not exist--a violation of one of the assumptions of the standard tests.

Granger Causality

We next investigated Granger causality between pairs of interest rates. Granger causality is used here solely to ascertain whether loan rates can help forecast deposit rates and vice versa. Recall from our theoretical model that the loan rate of the same bank and the deposit rates of other banks are explanatory variables in the individual bank's deposit rate equation (see (12)). The reverse is not true of the loan rate equation (see (2)). Hence, the test of Granger causality is intended as a preliminary investigation of this hypothesis.

The results presented in Table 1 indicate that every interest rate variable is Granger-caused by at least one of the other variables. The general pattern of causation suggests a high degree of simultaneity in the relationships among the variables and, except in the case of the results for B6, is consonant with the structure of our theoretical model. For B1 to B5, there is unidirectional Granger causality between loan and deposit rates for the same bank, running from the loan rate to the deposit rate. In the case of B6, unidirectional causality also exists but in the opposite direction. However, any adaptation of the model to accommodate the causal patterns indicated in Table 1 would not alter the system of reduced-form equations implied by our theoretical model.

Another interesting insight offered from the pattern of Granger causality is that not all the banks operate in the same manner. B6 is a leader in the market. Its loan rate Granger-caused the deposit rate of all other banks, while its deposit rate is not Granger-caused by any other bank's loan rate. B3 is an unmitigated follower.

Econometric Results

Here, we report the results pertaining to a 13-equation system of reduced-form equations, comprising loan (Table 2) and deposit (Table 3) interest rate equations for the six commercial banks and an equation for the treasury bill rate (Table 4). Tables 2 and 3 represent equations, with diagnostic statistics also shown. The exogenous variables initially included in all 13 equations comprise the five policy variables, the foreign rate of interest, and total deposits within the commercial banking system. The system was estimated using an iterative seemingly unrelated regression procedure. This allows for the possibility of contemporaneous correlation between the disturbances in different interest rate equations. Each iteration of the seemingly unrelated regression reestimates the parameters of the model after transforming the equations to remove the correlation across the residuals. Iteration to convergence gives the maximum likelihood estimates.

The results indicate that the loan rate for all banks and the deposit rate for three of the six banks were repressed by the ceiling on the average interest rate on loans, albeit marginally so. The bank rate has had a significant effect on the loan rate of only one bank. This bank is the only bank whose deposit rate has not been significantly influenced by the bank rate. The cash ratio, the stipulated government securities ratio, and the savings deposit rate floor impacted significantly on the loan rate for every bank. However, these three measures had a less pervasive influence on the deposit rate. Generally, the deposit rate for any given bank has been responsive to fewer policy variables than the loan rate. The loan rates, though generally responsive to all policy variables other than the bank rate, have exhibited very low elasticities. This is evidently attributable to the constraining influence of the ceiling on the average lending rate. The loan and deposit rates have been most responsive to changes in the stipulated government securities ratio. The cash reserve ratio has had the second strongest effect.

All five policy variables have had a significant effect on the treasury bill rate. The average lending rate ceiling and the cash ratio have had effects on the treasury bill rate that are opposite to the respective effects on loan and deposit rates. All other policy variables have had a positive impact on the treasury bill and loan and deposit rates. All of the results are intuitively plausible.

Evaluation of Results

The ineffectiveness of the bank rate vis-a-vis commercial bank loan rates is one of the noteworthy findings of this study. The reason for this phenomenon is that commercial banks do not make substantial use of the Central Bank's rediscount window. Interbank borrowing is usually available to the extent necessary to enable any given bank to manage its liquidity requirements. Hence, the cost of Central Bank accommodation is largely inconsequential to the banks. In these circumstances, the Central Bank should pursue a policy in which the bank rate follows the market. Any attempt to use the bank rate to lead the market is likely to fail and could result in misalignment of the bank rate with other interest rates.

The ceiling on the average lending rate, when it existed, depressed the loan rate by less than 1 percent on average. This is largely attributable to the Central Bank's policy of adjusting the ceiling in line with market trends. The policy was discontinued as a concession to market liberalization forces. There is no evident reason for a return to the former policy.

The cash reserves and stipulated government security ratios have had a highly significant impact on interest rates and may be used effectively to target interest rates. However, these measures can also impact bank liquidity and the balance of payments. The savings deposit rate floor can be used to influence income distribution or the structure and level of interest rates. Hence, the appropriate use of these instruments should be determined case by case.

(*.) Central Bank of Barbados and University of the West Indies--West Indies.

Footnotes

(1.) The regulation of commercial bank interest rates predated the existence of the Central Bank.

(2.) This is a simplification whose implications should be investigated in a further study. Banks do carry significant and varying levels of excess reserves.

(3.) When we abstract from (2), [r.sub.li] is presented as a predetermined variable.

References

Benston, George J. "The Optimal Banking Structure: Theory and Evidence," Journal of Banking Research, 3, Winter 1973, Pp. 220-37.

Hart, Oliver D.; Jaffee, Dwight M. "On the Application of Portfolio Theory to Depository Financial Intermediaries," Review of Economic Studies, 41, January 1974, pp. 129-47.

Hyman, David N. "A Behavioral Model for Financial Intermediation," Economic and Business Bulletin, 24, 3, 1972, pp. 9-17.

Kane E. J.; Malkiel, B. G. "Bank Portfolio Allocation, Deposit Variability, and the Availability Doctrine," Quarterly Journal of Economics, 79, February 1965, pp. 113-34.

King, Stephen R. "Monetary Transmission: Through Bank Loans or Bank Liabilities?," Journal of Money, Credit, and Banking, 18, 3, 1986, pp. 290-303.

Klein, John J. "The State of Finance Field Methodology: A Discussion," Journal of Finance, 25, May 1970, pp. 322-5.

Klein, Michael A. "A Theory of the Banking Firm," Journal of Money, Credit, and Banking, 3, May 1971, pp. 205-18.

Parkin, Michael J. "Discount House Portfolio and Debt Selection," Review of Economics Studies, 37, October 1970, pp. 469-97.

Pesek, Boris P. "Bank's Supply Function and the Equilibrium Quantity of Money," Canadian Journal of Economics, 3, 3, 1970, pp. 357-85.

Pyle, David H. "On the Theory of Financial Intermediation, "Journal of Finance, 26, June 1971, pp. 737-47.

Sealey, C. W. Jr. "Deposit Rate Setting, Risk Aversion, and the Theory of Depository Financial Intermediaries," The Journal of Finance, 35, 5, 1980, pp. 1139-54.

Sealey, C. W. Jr.; Lindley, James T. "Inputs, Outputs, and a Theory of Production and Cost at Depository Financial Institutions," The Journal of Finance, 32, 4, 1977, pp. 1251-66.

Williams, Marion V. Liberalizing a Regulated Banking System: The Caribbean Case, Avebury, United Kingdom: Aldershot, 1996.

Worrell, DeLisle. "Role Models for Monetary Policy in the Caribbean: Comparing Caricom Central Banks," in Harold Codrington; Roland Craigwell; Cleviston Haynes, eds., Central Banking in Barbados: Reflections and Challenges, Barbados, West Indies: Central Bank of Barbados, 1997.

Printer friendly Cite/link Email Feedback | |

Comment: | The Impact of Regulatory Measures on Commercial Bank Interest Rates: A Micro Analysis of the Barbados Case. |
---|---|

Author: | GREENIDGE, KEVIN; MCCLEAN, WENDELL |

Publication: | International Advances in Economic Research |

Geographic Code: | 5BARB |

Date: | Aug 1, 2000 |

Words: | 4413 |

Previous Article: | Malmquist Indices of Productivity Change in Portuguese Banking: The Deregulation Period. |

Next Article: | Regional Tests of Okun's Law. |

Topics: |