Interest rate margins: a decomposition of dynamic oligopolistic conduct and market fundamentals

Abstract

We propose a model in which the evolution of interest rate margin (markup) in banking is the outcome of two major components: (i) dynamic oligopolistic conduct and (ii) dynamics of market fundamentals. The model is specified such that oligopolistic dynamics are separated from the dynamics of fundamentals. Consistent with the theory, we employ the error-correction model which generates results indicating that margins are significantly different from the traditional measure once fundamentals are filtered out.

Acknowledgements

We thank Aharon Meir Center for Banking at Bar-Ilan University for financial support. We are grateful to Pedro Pita Barros, Michael Beenstock, Ben Eden, David Genesove, Margaret Slade, Yossi Spiegel, and seminar participants at the Bank of Israel, The Hebrew University, the 27th E.A.R.I.E conference in Lausanne. The participants in the 2005 Berlin conference on Bank Relationship, Credit Extension and the Macro-Economy. We also thank the participants in the Second Tel-Aviv Workshop on Industrial Organization and Anti-Trust, especially Sofronis Clerides, Chaim Fershtman, David Gilo, Ariel Pakes and Bob Willig raised very important and constructive issues.

Notes

¹ In the sequel we define fundamentals as well as oligopolistic dynamics more precisely.

² Domowitz (1993) p. 215, articulates this phenomenon as follows: ‘Attributing observed departures of price from marginal cost to generic descriptions of imperfect competition or market power can be seriously misleading’. Borenstein et al. (1999) show, in the context of electricity markets, that if opportunity cost exceeds that of production cost, positive markup by itself is not a proof of market power abuse.

³ Lukacs (2000) decomposes the time varying behaviour of price-cost margins in the UK manufacturing sector into inter and intra industry components and finds that most of the variance over time is attributed to size rather than industry effects. Ryan (2000) studies the effects of imperfect competition (price-cost margins) on cyclicality of total factor productivity growth rates.

⁴ As Klette and Griliches (1996) argue, ‘one problem with the empirical research … is that the estimated markup is critically dependent on the assumption of constant returns to scale. If the firms in fact are operating with short run decreasing returns to scale, their markup estimates might be an artifact due to the erroneous assumption of constant returns’ (p. 3). Berg and Kim (1994) show that the measurement of firm's technology and efficiency is crucially dependent on the (non) treatment of conduct.

⁵ In a recent analysis of the relationship between market power and bank risk, Covitz and Heitfield (1999) find that the seriousness and degree of a bank's moral hazard determines the direction of the price-cost margins.

⁶ Humphrey and Pulley (1997), Berger and Mester (2001) and Valverde et al. (2001) use terms like ‘external business environment’ or ‘economic (business) conditions’ to represent externally initiated adjustments, which we term fundamentals.

⁷ Chirinko and Fazzari (1994) use the Lerner index and Ghosal (2000) employes concentration ratios both of which may or may not accurately depict market power.

⁸ We should emphasize that Chevalier et al. (2000) are able to test their proposed model by nesting it along with cyclical firm-conduct (tacit collusion) models a la Rotemberg and Saloner (1986) and Rotemberg and Woodford (1992) and those a la Bils (1989) which draw on demand elasticities.

⁹ It is believed that changes in banks decision rules are not continuous but rather discrete. Humphrey and Pulley (1997) for instance, average their data for individual banks over three successive four-year intervals believing ‘it unrealistic to assume that profit maximizing behaviour is manifested annually’.

¹⁰ Krugman (1991) and Svensson (1991) use this technique to model exchange rate dynamics.

¹¹ In Chevalier et al. (2000), supermarkets decisions regarding the type of goods to advertise (and commit to a low price) depends on the states of demand and consumers reservation price in each of the states.

¹² See Geroski (1992) for application in which prices respond to cost and demand shocks.

¹³ See Svensson (1991) for the application of Brownian motion in the context of exchange rate and interest rate variability, and Dixit (1989), in the context of firms’ entry and exist decisions under uncertainty. Dixit (1989) also discusses alternative pricing functions such as a mean reverting process.

¹⁴ See Appendix A for derivation.

¹⁵ The fundamentals consist of variables which are nonfirm specific. Therefore, these variables do not display information on interfirm rivalry.

¹⁶ An increase in e represents a devaluation of the local currency vis a vis the US dollar.

¹⁷ Risk averse customers reduce the demand for unindexed credit in a response to increasing inflation, however, the substitution effect would dictate an increase in demand since credit is cheaper.

¹⁸ The law of motion, described later in the article, is such that maximizing contemporaneous profits is compatible with the maximization of the infinite sum of the discounted present and future profits. Thus, no time inconsistency exist in that respect.

¹⁹ We specify deposits as an input in the usual manner.

²⁰ Note that α = 1/(1 − rr) with rr being the reserve requirement.

²¹ To arrive at the markup, marginal operating costs have to be deducted from s.

²² Like Hall (1988) we make no parametric assumption about the cost function, and use the wage rate (which accounts for over 75% of costs) as a summary statistic for marginal cost. Bils (1987) uses marginal wage cost as a proxy for marginal cost in his study of cyclical behaviour of marginal cost and price.

²³ Note that we allow for nonlinearities between the fundamentals and the process F_t via the second term in (32).

²⁴ Note that in the calculation of δ and σ, according to Equation 27, we appropriately substitute f_t with the time path F_t .

²⁵ The data on the output gap are deviations of the potential GDP from the actual GDP. The former was constructed using an estimated GDP production function for the Israeli economy.

²⁶ Computed data on the expected inflation in Israel are available from the capital market, where nonlinked government bonds and inflation-linked government bonds with the same maturities are traded simultaneously.