Market power versus efficiency under uncertainty: conventional versus Islamic banking in the GCC

Abstract

In this article we estimate the effect of concentration on intermediation margins in Gulf Cooperation Council's (GCC) Islamic and conventional banking under the assumption that margins are uncertain. The empirical model, which we formally derive from an expected utility maximization problem, allows us to test for risk aversion as well as competitive conduct in loan and the deposit markets. The model also yields an expression showing that the effect of concentration on margins is the sum of its respective effects on market power, marginal cost of intermediation and marginal cost of uncertainty. The expression allows us to test whether concentration is welfare enhancing, reducing or neutral. We find Islamic banks to be risk-averse and conventional banks to be risk-neutral. We also find that concentration is welfare-neutral in Islamic loans and deposits, welfare-enhancing in conventional loans and welfare-neutral in conventional deposits. We used Nonlinear Two-Stage Least Squares (N2SLS) and Nonlinear Three-Stage Least Squares (N3SLS) to check for robustness.

Keywords:

• Islamic banking
• conventional banking
• market power
• risk

JEL Classification:

• L13
• L11

Notes

¹Member countries of the GCC are Bahrain, Kuwait, Oman, Qatar, Saudi Arabia and United Arab Emirates (UAE).

²There are more than 300 Islamic banks in over 75 countries, including North America and Western Europe. However, most Islamic banks are located in the Middle East, particularly in the Gulf region, and co-exist with conventional banks except in Iran, Pakistan and Sudan, where all financial intermediation is performed by Islamic banks only. Islamic banks' assets grew at an annual rate of 15% over the past decade and were worth 300 billion US Dollars in 2008 (Chong and Liu, 2009).

³Concentration data compiled by the authors from BankScope.

⁴This section draws largely from Kettel (2010). For an in-depth look at the theory and practice of Islamic finance, see Iqbal and Mirakhor (2007). El-Gamal (2006) provides a brief and useful overview. Kuran (1995) provides an insightful essay on Islamic Economics in general.

⁵As for prior work on competition in banking in general, studies are too numerous to cite here. Suffice it to say that there are two main strands of literature. The first consists of earlier concentration-profitability studies (Gilbert (1984) surveys that literature) where a positive concentration-profit correlation is considered as evidence of market power. The second strand of literature, recently surveyed by Berger et al. (2004), uses structural models, otherwise known as New Empirical Industrial Organization (NEIO) models (Bresnahan, 1989), to examine competition.

⁶Under risk-neutrality, Equation 5 reduces to the standard profit equation for a conventional bank (Freixas and Rochet, 1997; Neven and Roller, 1999; Adam et al., 2002).

⁷The conventional bank's loan rate is included to account for the possibility that not all borrowers from Islamic banks are strict adherents to Sharia'a-compliant finance. If they were, then γ would be zero. Rather than imposing the zero restriction, we test it.

⁸Same as footnote 7. Replace ‘borrowers’ with ‘depositors’ and γ ^s with φ ^s .

⁹The definitions ‘interest income’ and ‘interest expense’ are used by BankScope for labelling purposes only. The labels do not imply that income received and expenses paid by Islamic banks are based on conventional interest.

¹⁰ p-values are denoted by three asterisks for statistical significance at the 1% level, by two asterisks at the 5% level, and 1 asterisk at the 10% level. Following convention, we use the 5% level as the cut-off point.

¹¹This may be due to a highly leveraged economy, at least in UAE, and dominance of foreign capital flowing chiefly to conventional banks.

¹²The sign of the market power effect is negative in the Islamic deposit market because the estimate of λ _D in that market is slightly larger than the lower theoretical limit of −1. As mentioned earlier, the confidence interval of λ _D also contains values equal to or larger than −1, implying a range of positive value for λ _D and, thus, a range of positive values for the market power effect in the Islamic deposit market also. We therefore conclude that increasing deposit concentration confers market power to Islamic banks.

¹³Because MG construct a yearly Lerner index and use it as a measure of market power, rather than estimating market power as a fixed parameter from the data, as we do, they conclude that market power increased in the loan market and decreased in the deposit market.