The Determinants of Sukuk Market Development

ABSTRACT

The objective of this article is to empirically investigate the structural, financial, developmental, institutional, and macroeconomic determinants of Sukuk market development for a sample of 13 countries over the period 2001–2013. We employ the Generalized Method of Moments (GMM) procedure to tackle the problems of endogeneity of lagged dependent variable, heteroscedasticity, and serial correlation in the residuals. Our results suggest that a combination of structural, financial, and institutional factors seem to exert a significant effect on Sukuk markets. Indeed, larger economic size, higher proportion of Muslims in the population, better investment profile (IP), and lower corruption are associated with larger Sukuk markets, while higher interest rate spread is negatively related to Sukuk market development.

KEY WORDS:

• dynamic panel
• GMM estimation
• Islamic Finance
• Sukuk

Notes

1. Islamic Financial Services Board Stability Report (Islamic Financial Services Board 2015).

2. AAOIFI is a nonprofit organization that prepares auditing, governance, and Sharia standards for companies interested in pursuing Shariah compliance.

3. AAOIFI Shariah Standards for Financial Institutions (2008).

4. As a result, Bangladesh, Bermuda, Cayman Islands, and Luxembourg have been eliminated from our sample.

5. We are thankful to an anonymous referee for suggesting this link between bond markets and Sukuk market development.

6. To see this, simply lag Equation (2) by one period.

7. Furthermore, even standard panel data estimators with fixed or random effects are not appropriate for estimating models like Equation (2). For example, fixed-effects panel data regressions yield biased estimates since correlation between the transformed lagged dependent variable and the transformed error term is still there (Baltagi 2001).

8. However, Arrelano and Bond (1991) show that when the error term, ε_i,t, is not serially correlated, then the differenced error term, $\mathrm{\Delta }{\epsilon }_{i,t}$, should display first-order autocorrelation but no second-order autocorrelation.

9. Using Monte-Carlo simulations, Windmeijer (2005) confirms that the corrected variance closely approximates the finite sample variance of the two-step System GMM estimator.