Towards the end of last year, our PwC global Islamic finance team wrote “Growing gains: managing Islamic banking risks*” which is available free from our website by following the link. This document was based on our team’s experience of working with Islamic banks in Malaysia, the Gulf Cooperation Council countries and the UK.
After our document had been published, I came across “Islamic Banks and Financial Stability: An Empirical Analysis” which is a working paper available on the International Monetary Fund (IMF) website. Two things struck me about the working paper.
(1) Discursive coverage of risks
The working paper spends a few pages explaining the risks (based upon a literature search) that Islamic banks face. This approach is of course different from my colleague’s approach of writing from actual experience. However, there is a significant overlap in the risk discussion in the two documents.
(2) Statistical risk analysis
The core of the working paper is to use a dataset of conventional and Islamic banks’ financial data, and to compute Z-scores as a measure of the stability of the banks concerned. The working paper sets out the methodology in detail, both explaining the data set and the way in which the Z-score is calculated. Very briefly, the higher the Z-score, the less likely the bank concerned is to become insolvent.
Statistical conclusions
The authors of the IMF paper divide their universe of banks into four categories, depending on whether the bank is Islamic or conventional, and whether it is large (total assets over US $1 billion) or small (total assets below that figure).
After excluding outliers (being in this case the 1st percentile and the 99th percentile) the authors conclude from the computed Z-scores that:
- large conventional banks are more stable than large Islamic banks
- small Islamic banks are more stable than small conventional banks.
The authors then carry out regression analysis using a number of actual and dummy variables and conclude that the regression analysis supports the above initial conclusions.
How reliable is the above statistical analysis?
I asked my colleague Cameron Evans from PwC Thailand, who like me contributed to the PwC risk paper, to look at the IMF working paper for me. He pointed out that the sample sizes for the four categories discussed above vary significantly. For example, there are 3,248 observations of conventional banks and only 520 observations of Islamic banks. Including or excluding the outliers also has a material impact.
Accordingly, Cameron did not consider that too much emphasis should be put on the risk differences between Islamic and conventional banks, based on this present data set. Of course additional data might allow more robust conclusions to be drawn.
What is more striking is the difference in risk characteristics between small and large Islamic banks, with the large Islamic banks having a significantly lower Z-score and therefore being more risky. This indicates that as an Islamic bank grows, it needs to give increasing attention to risk, and in particular to the key risks discussed in the PwC paper.
Mohammed Amin
Recent Comments