Bank interest margins in OECD countries

Abstract

This paper extends the literature on bank interest margins by providing empirical evidence using panel data covering the banking sector of fourteen OECD countries. Each country's banking sector is treated as a single representative firm viewed as a national risk-averse dealer setting loan and deposit rates to balance the random arrivals of loan requests and deposit supplies. We find that national industry margins are influenced by market power, operational cost, risk aversion, interest rate volatility, credit risk, volume of loans, implicit interest payments and quality of management.

Introduction

Through the process of taking deposits and making loans, banks help the economy allocate funds from savers to borrowers in a more efficient manner (Brock & Suarez, 2000). The reward earned by the banking sector during this process depends largely on the average interest margin, which is commonly defined as interest revenue minus interest expense, per dollar of assets. This paper seeks to explain the factors that determine interest margins. Regularly updating our knowledge of interest margin determinants is valuable for a number of reasons, including monitoring changing trends in bank efficiency through time, and evaluating whether bank margins are providing effective price signals to market players. In addition, margin behaviour potentially has significant policy implications: for instance, Claeys and Vander Vennet (2004) argue that high interest margins may reflect an inadequate regulatory banking environment.

Bank interest margins can be analysed on four levels. At the most ‘micro’ level, margins are determined at the highly disaggregated level of business division within each bank. Next, the ‘meta’ approach uses the banking conglomerate as the unit of analysis. Third is the ‘macro’ approach which employs the combined industry data in a given country. And fourth there is the ‘continental’ approach where banking sectors in a given economic region of the globe are considered in aggregate. This paper uses the third approach. We model interest margins at the level of each country's banking sector, treating the banking sector in terms of a single representative agent, and are interested in the determination of industry margins on a national basis. Our industry-wide approach complements another recent study of bank interest margins conducted at the individual bank level which stressed the differences between institutions (Maudos & Guevara, 2004). Our paper builds on that work by emphasising differences between systems, not institutions. By conducting tests at the nationally aggregated level, our work makes the contribution of providing a vital second opinion to existing studies conducted at the micro level.

Our view is that the differences between countries’ banking sectors are as pertinent as differences between individual banks within that country. A particular benefit of our approach is that it takes seriously the view of the productive nature of the banking firm (Lerner, 1981), emphasizing cross-border differences in the production costs associated with the process of intermediation. While production costs may be fairly similar for banks within a given country (common wages rates, property values etc) they can be expected to show wider variability between countries. Moreover, from an econometric perspective, it also allows the role of fixed effects across panel data to be tested. Since financial deregulation in the late 1970s, banks within each of the countries studied here have seen significant rationalisation of their domestic banking industries via the market for corporate control. The effect of this more competitive environment has been that players within each market largely mimic one another and often act ‘as one’, and therefore the approach of treating each national banking sector as a single agent is both reasonable and interesting. Econometrically, we effectively consider each industry is like a national ‘team’, competing globally but with its own cost structure and regulatory framework.

The paper is based on the standard well-specified general model of bank interest margins. Based on the dynamic dealership model developed by Ho and Saunders (1981) and its extensions, the representative bank is viewed as a dynamic risk-averse dealer aiming to maximize expected utility, setting loan and deposit rates to balance the random arrivals of loan requests and deposit supplies.¹ This study employs single stage estimation to explain the pure margin as well as the unit-specific components in the regression. The empirical estimation covers banks of fourteen OECD countries over the period 1987 to 2001, the most recent cross-country data available. We employ a data set that is reasonably consistent across countries in regard to its institutional definitions and accounting classifications. Market power, operational cost, risk aversion, volatility of the interest rate, credit risk, economies of scale, implicit interest payments, quality of management and opportunity cost are the factors tested for determining banks’ interest margins.

The rest of the paper is organized as follows: Section 2 reviews the relevant literature on determinants of bank interest margins. Section 3 outlines the methodology and discusses the data. Section 4 develops the empirical results, and Section 5 concludes the study and discusses the implications.