DOC

In banking research, the determinants of net interest margins are

By Jane Bell,2014-04-16 21:53
20 views 0
In banking research, the determinants of net interest margins are

    DETERMINANTS OF BANK NET INTEREST MARGINS OF

    SOUTHEAST ASIA

    by

    Jude S. Doliente

    College of Business Administration

    University of the Philippines Diliman

Mailing Address:

    Room 201-B

    College of Business Administration

    University of the Philippines Diliman

    1011 Diliman, Quezon City

    Philippines

Telephone No.:

    (632) 928-45-71

    Fax No.:

    (632) 920-79-90

    E-mail Address:

    jude.doliente@up.edu.ph

Acknowledgments:

    This paper was written while the author was an exchange student at the University of Limoges (France) through the funding of Duo-France Program. The author would like to thank Carlos Bautista, Ma. Socorro Gochoco-Bautista, Celine Crouzille, Pedro de Ocampo, Jr., Erlinda Echanis, Laetitia Lepetit, Philippe Rous, and Helen Valderrama. The usual disclaimer applies. An earlier version of this paper has already been submitted to an ISSI journal.

    DETERMINANTS OF BANK NET INTEREST MARGINS OF

    SOUTHEAST ASIA

    Abstract

    We investigate the determinants of net interest margins of banks in four Southeast Asian countries. We use the dealer model (Ho and Saunders, 1981) and run a two-step regression. Results of the first regression indicate that the net interest margins in the region are partially explained by bank-specific factors namely operating expenses, capital, loan quality, collateral and liquid assets. Second step regression results show that the region’s net interest margins are largely explained by the non-competitive

    structure of the region’s banking system and manifest some sensitivity to changes in

    short-term interest rates. Finally, we find evidence that bank spreads declined after 1997 thus reflecting the profit squeeze experienced by the region’s banks due to extensive loan

    defaults in the aftermath of the Asian currency and banking crises.

JEL Classification: G21

    Keywords: Banks, Interest Margins, Spreads, Southeast Asia

Section I. Introduction

    In banking research, the determinants of net interest margins (bank spreads) are empirically well explored. Results strongly suggest that net interest margin determinants vary across countries and among regions of the world. For instance, studies on banking systems of developed countries show that net interest margins have significant positive relationships with a banks level of capital, loan loss provisions, implicit interest payments, and interest rate volatility (Ho and Saunders, 1981; Saunders and Schumacher, 2000). These results are considered benchmarks because banks in developed countries operate in mature financial systems. On the other hand, a study of Latin American bank spreads rarely confirmed and even contradicted some of the benchmark results (Brock and Suarez, 2000). For example, loan losses and bank capital were shown to have significant negative relationships with bank spreads in some Latin American countries. These anomalous findings were partly explained by distortions caused by inadequate regulatory systems that allow weak banks to continue operating, unreliable financial reporting practices that result in misstated bank capital, and extensive government guarantees that encourage excessive risk taking among banks.

    In Southeast Asia, little is known about the determinants of its banks’ net interest

    margins. Since banks are the major source of financing in this region, the level of net interest margins is an important policy variable for it indicates how efficiently banks perform their intermediary roles of collecting savings and allocating funds. Curiously, and although in varying degrees, the banking industries of Southeast Asia exhibit similarities in market openness, regulatory stance, extent of government intervention,

     1

    1 In fact, some of these lending practices and the influence of macroeconomic policy.

    shared commonalities became vulnerabilities that caused a systemic crisis in 1997 which decidedly undermined the solvency of this region’s banking systems. An empirical investigation that considers some of these commonalities may demonstrate their unique effects on Southeast Asian bank net interest margins.

    The objective of this paper is to investigate the determinants of the region’s net interest margins while taking into consideration bank-specific factors, namely: collateral, capital, liquid assets, operating expenses and loan quality. Four Southeast Asian countries - Indonesia, Malaysia, Philippines and Thailand - constitute the sample of this paper. There is the impression that while the financial system of the region is indeed continually evolving, weaknesses still persist in the areas of financial reporting, regulation and government interference (Gochoco-Bautista, 1999; Kane, 2000). Thus, as between those for developed and Latin American countries, one would expect findings for Southeast Asia to be closer to Latin America especially that both regions have similar problems that plague their financial systems.

     The rest of the paper is organized as follows: Section II reviews the literature on net interest margins. Section III describes the variables, data and the empirical specification. Section IV presents and interprets the estimation results. Finally, section V concludes.

Section II. Review of Related Literature

    The analysis of net interest margins is an attempt to measure the cost of financial intermediation; that is, the difference between the gross cost paid by a borrower to a bank

     1 For an extensive discussion of these similarities see Chou (1999), Gochoco-Bautista (1999), Kawai and Takayasu (1999), and Oh (1999).

     2

    and the net return received by a depositor (Brock and Suarez, 2000). Generally, high interest margins are taken to be unfavorable because they lead to disintermediation. Low deposit rates represent unattractive returns for maintaining deposit accounts hence discouraging savings. High loan rates, on the other hand, make the cost of funds increasingly prohibitive to potential users thereby inhibiting investment activity. Nevertheless, while high net interest margins have usually been associated with inefficiency, they may also contribute in the strengthening of a country’s banking system

    (Saunders and Schumacher, 2000). This happens when profits earned from high spreads are being channeled by banks to their capital bases. For example, high spreads and healthy capital ratios were both observed among Colombian banks (Barajas, Steiner, and Salazar, 1999).

    On the other hand, very low spreads cannot always be taken positively especially in liberalized but inadequately regulated environments where certain mechanisms ensuring the closure of or intervention in poorly capitalized or unstable banks are absent. If weak banks are allowed to continue operating, there is the likelihood that they will adopt the strategy of offering lower loan rates to gain additional market share or to grow out of their troubles. This was presumed in some Latin American countries in the period after financial liberalization reforms were instituted in the region over the last decade (Brock and Suarez, 2000).

    There are at least two modeling frameworks for net interest margins. The first framework, which is considered here, is the Ho and Saunders (1981) dealer model. This model has been extended and modified by McShane and Sharpe (1985), Allen (1988) and Angbazo (1997). It has also been applied in different settings by Ho and Saunders (1981), Saunders and Schumacher (2000), Brock and Suarez (2000) and Drakos (2003).

     3

    The alternative model is the firm theoretic approach developed by Klein (1971) and Monti (1972). This model views the banking firm in a static setting where demands and supplies of deposits and loans simultaneously clear both markets. Following the same line of research, this framework was further explored by Zarruk (1989) and Wong (1997). Finally, there is the specification and estimation of Barajas, Steiner, and Salazar (1999) that can also be categorized under the firm theoretic approach.

    Under the dealer model, the net interest margin (s) is given by the following

    equation:

    12s;RQ (1) 2

    The first term, , is the ratio of the intercept (α) and the slope (β) of the

    symmetric deposit and loan arrival functions of the bank and is a measure of the bank’s

    2risk-neutral spread. Another interpretation of the ratio is that it provides some measure of the effect of market structure (i.e. monopoly rent) in the determination of the spread because if a bank faces relatively inelastic demand and supply functions, as expressed by

    a high ratio, then it may be able to exercise monopoly power.

     The second term is a first order risk-adjustment term and depends on three factors: (1) the bank management’s coefficient of risk aversion (R), (2) size of bank transactions

    2(Q), and (3) the instantaneous variance of the interest rate on deposits and loans (σ).

    This implies that, all things equal, the greater the degree of risk aversion, the larger the size of transactions and the greater the variance of interest rates, the larger bank margins are.

     2 For the derivation of the model, see Ho and Saunders (1981).

     4

    Applying the dealer model however entails the recognition of the effects of certain market and institutional imperfections that distort the observed net interest margin. Unfortunately, these factors cannot be directly incorporated in the dealer model. Thus, to control for the effects of these factors, a two-step regression is performed (Ho and Saunders, 1981; Saunders and Schumacher, 2000; and Brock and Suarez 2000). In the first regression, the hypothesis is that the observed net interest margin will comprise of the “pure spread” that is constant across banks plus the effects of certain market and

    institutional imperfections. Varied sets of institutional and market imperfections were

    3 In the process, the estimated considered in existing empirical applications of the model.

    intercept in the first regression is the “pure spread and is treated as the dependent

    variable in the second regression for the estimation of the effects of market structure ()

    2and interest rate volatility (σ). The effects of risk aversion (R) and size of bank

    transactions (Q) are not considered since they are not seen to change as fast as interest rates (Ho and Saunders, 1981).

Section III. Method

Empirical Specification

    We propose to analyze the determinants of Southeast Asian net interest margins following a two-step regression. The first step controls for the effects of collateral, operating expenses, decline in loan quality, capital, and liquid assets on the observed net interest margin to determine the “pure spread.” The second step is to analyze the effects

     3 Ho and Saunders (1981) considered implicit interest payments, loan loss, and bank capital. Saunders and Schumacher (2000) considered implicit interest payments, bank capital, and opportunity cost of reserves. Brock and Suarez (2000) considered operating expense, loan loss, bank capital, and liquid assets. Finally, Drakos (2003) considered liquidity, loan loss provisions, leverage, and interest rate risk.

     5

    of market structure, interest rate volatility, and possible country and time effects on Southeast Asian banks’ “pure spread.”

    Following Brock and Suarez (2000) and using panel data from 1994-2001 for each country, we perform the first step pooled regression as:

    , (2) NIM;D;X;uityyjitjitityj

    where:

     is the published net interest margin of bank i at year t; NIMit

     is the intercept of the regression and also the estimate of the “pure spread” component

    of the NIM for all i at year 1994;

     is a dummy variable that captures yearly time effects for the years 1995-2001. Dy

    Consequently, the annual estimates of the pure spread for the years 1995-2001 are estimated as ; ;y

     is a vector of control variables (collateral, operating expenses, loan quality and Xjit

    capital and liquid assets) for each bank i at year t; and

     is the residual. ui

    This first regression yields 32 estimates (4 countries and 8 years) of the “pure spread” which are used in the second step regression.

    As in Saunders and Schumacher (2000), the second step regresses the “pure

    spread” against interest rate standard deviation (a measure of volatility) and country dummies (D, D and D), with Malaysia as the base country, to account for differences TPI

    in market structure of the four countries. In addition, a crisis dummy () is included Dz

    where from 1998 onwards the dummy equals 1. Hence, the second regression is of the form:

     6

    2, (3) (;(D;(D;(D;(D;(;utzTpIm012345m

    where:

     is a time series of the pure spreads of country m; tm

     is the intercept and it reflects the effect of the market structure on the pure spread of (0

    Southeast Asia whereby a significantly high value indicates a noncompetitive market structure;

    2 is the annual variance of the short term interest rate of country m; and m

     is the residual. u

    The Variables

    Similar to the existing empirical applications of the model, this study tries to account for the effects of market or institutional imperfections to isolate the behavior of the “pure spread. In the first regression we assume that the imperfections manifest themselves through the movements of 5 variables, namely: (1) collateral, (2) operating expenses, (3) loan quality, (4) capital requirements and (5) liquidity.

     A cursory review of balance sheets of Southeast Asian banks shows that banks report real and other properties owned or acquired (ROPOA) as a major line item. ROPOA arises when banks foreclose existing mortgages (collateral) that secure nonperforming loans. The significant presence of ROPOA can be explained by the region’s credit culture which can be typified as collateral-based. This suggests that banks

    are willing to lend as long as the loan is sufficiently covered by collateral, usually in the form of real estate (Dickinson and Mullineux, 2002). However, in the long-run, despite providing full collateral cover, when a client fails to service the debt, the bank is forced to foreclose the existing mortgage thereby causing ROPOA to rise.

     7

     The impact of collateral is relevant in the determination of the spread although it is uncertain whether this variable has a decreasing or increasing effect. To be sure, when banks foreclose the collateral mortgage and thus hold ROPOA, they forego opportunity costs and incur maintenance and documentation costs, real estate taxes, and all other forms of carrying costs. As such, there is a possibility that banks are recouping these costs through the high spread that they demand.

    On the other hand, providing sufficient collateral may also serve to reduce a bank’s lending rate, thus the interest spread. This result is not totally surprising since collateral tends to reduce the loss exposure of the bank in case of loan defaults. In a lending environment where there is some difficulty in challenging the financial projections and creditworthiness of a borrower (Kane, 2000), sufficient collateral is crucial in mitigating the risk of loss. Moreover, it can also be argued that banks actually recover foregone carrying and opportunity costs in holding ROPOA when the assets are disposed of in the future. It is not uncommon for banks to set a floor price for the disposal of ROPOA. This floor price usually assures the recovery of carrying costs and other opportunity costs plus profit. The ratio of non-earning assets to earning assets proxies for the effect of collateral.

    The decline in loan quality is the second factor considered. The possible impact of this variable in interest margins can be seen in two ways. First, this would reflect the extent to which banks increase operating expenses as they intensify monitoring of loans and incur additional expenses for working out or selling these loans. Second, the impact may also reflect the additional risk premium charged by banks for the financial costs of forgone interest revenue (Barajas, Steiner, and Salazar, 1999). This variable is

     8

Report this document

For any questions or suggestions please email
cust-service@docsford.com