Wednesday, December 08, 2010

Do Sukuk Attract Riskier Borrowers?

After reading the paper "Are Islamic Investment Certificates Special? Evidence on Post-Announcement Performance of Sukuk Issues", I was left with significant questions about what information the study provided.  Given the lack of data on sukuk and the limited universe of sukuk issues, it was unlikely to reach a significant conclusion that would illuminate the costs and benefits for sukuk issuers, but the report is a good data point to have, even if I have questions for me about the statistical tests used in the study.

The paper, written by Christophe Godlewski and Laurent Wiell of the EM Strasbourg Business School at the University of Strasbourg and Rima Turk-Aris of Lebanese American University, attempts to determine whether sukuk are beneficial for issuers (and their shareholders).  They use a test to determine whether there is a significant impact on the share price of companies (compared to the market) when they announce sukuk versus conventional bonds.  The test focuses upon the period plus and minus two days from the issuance.  They limit the analysis to Malaysian corporate sukuk because the data (and secondary market activity) in sukuk in other regions is lacking.

The general finding is that conventional bond issuance has no statistically significant impact on a company's share price, while sukuk issuance has a negative impact on a company's share price.  The significance is relatively weak (10% significance level), but that is to be expected for the relatively limited data set they are working with.

The authors' conclusion, despite their earlier point that most sukuk are structured to be nearly identical to conventional bonds in economic outcome, is that the profit-sharing structure creates adverse selection problems (companies with worse prospects are more likely to issue a profit-sharing debt instrument than companies with better prospects).  They did find that the companies issuing sukuk were generally worse performing than conventional issuers, had higher debt levels and they conclude that the adverse selection problem is the main reason for their results.

With due credit for creating an interesting analysis based on limited data, there are many questions that arise from their analysis.  First is why they did not try to account for the performance metrics of the company (based on its underlying fundamentals) to reduce the likelihood that their result of significant adverse market reaction to sukuk issuance was due to the companies fundamentals rather than just the issuance of a sukuk.  The specification of their models is not explicitly provided, which makes it difficult to know what control variables they included in the model.  For example, they could have used total debt to assets as a way to control for the riskiness of the company (or another metric).

A more fundamental problem is that their conclusion does not fit the actual makeup of the sukuk market.  The authors' explanation of the preferability of sukuk over conventional bonds for companies with poor business prospects (or even just more risky prospects) would be justified if sukuk were actually structured as profit-sharing investments (a quasi-debt security).  However, this is not how sukuk are structured.

Sukuk are structured (whether they are ijara-based or pre-AAOIFI mudaraba/musharaka or another structure) to replicate the economic outcome of a conventional bond.  That means that they are structured so that the company issuing the sukuk is essentially signing up for a periodic payment over several years and then a balloon payment at maturity to redeem the sukuk.  The periodic payment is most commonly set based on the prevailing interest rates.  From an economic perspective (the perspective of the company's shareholders), the company does not create any different situation whether they use a sukuk or conventional bond.

The interesting test of this hypothesis would be to take the same or a similar data set and re-run the analysis and see whether the outcome is identical with additional control variables included.  Specifically, does the finding that companies that issue sukuk have worse stock market performance (and conventional bond issuers have no significant change in performance) remain robust if the specification includes firm-specific control variables to pick up the poorer performance of sukuk issuers? Unless the sukuk involved are structured to actually expose the investors to greater risk than a conventional bond, the reported results will come into question.

The adverse selection problem that is possible with Islamic finance is a serious problem and has led to the industry adopting products that closely replicate conventional debt.  However, with the product mix being as it is, it is difficult to argue that there remains  an adverse selection problem for Islamic finance and this paper essentially tries to do just that.  If the data set allows it, one way to test for the effect would be to try to find whether the effect depends on whether a sukuk is a mudaraba or musharaka versus an ijara (or other debt-based sukuk).  The data limitations for sukuk remain difficult within the context of statistical study, but it still leads me to have questions about whether the methodology used in this report is adequate for providing a result that is meaningful for the industry.


Bernardo Vizcaino CAIA said...

Thanks Blake for this detailed commentary.

The study could have been more specific, for instance:

- It could have taken into account the different types of sukuk (there are 2 or 3 different types that dominate the MY market).
- It should have also taken into account that in Malaysia there has been a clear shift in the types of sukuk being issued over the past 3-4 years.
- Consider individual sukuk issuance outside of Malaysia even if they were few and far in-between, or other markets that are as liquid (Pakistan? UAE?)
- Look deeper at reasons (other than adverse selection) - perhaps consider higher transaction costs (structiring fees), greater difficulty in raising cash (placement),

Nevertheless, I appreciate the question being raised (i.e. "Do Sukuk Attract Riskier Borrowers?") since I know of a few instances where sukuk has attracted some overtly risky underlying assets, some of which would have not raised a penny in the conventional space anway. Also, it is good to see the use of non-parametric tests although more could have been done with the data as suggested in Blake's commentary.


Blake Goud said...


Thanks for your comments. I think you raise some valid points. Now that you mention it, there could be an analysis of other markets because what they tested was not based on secondary market liquidity in sukuk. They looked at the movement of the equity price (the reaction to the sukuk vs. conventional bond issuance) of the companies issuing sukuk and many of the GCC issuers are liquid at least in terms of the market for their stock.

Also, I agree that the structure could make a difference and the shift in preferred structure could have an impact. However, when push comes to shove, almost all sukuk are equivalent (or intended to be so) to conventional bonds in the creditors' rights and the structure of issuance, periodic payments and redemption amount. It makes little difference whether the sukuk is an ijara or a (pre-AAOIFI ruling) mudaraba or musharaka.

Even if the conclusion of adverse selection is based on the Malaysian market alone, the natural next step would be to compare the issuer attributes of sukuk issuers versus conventional issuers in terms of debt-to-assets and other metrics that are intended to measure the company's riskiness. It would be unfortunate if Islamic finance and sukuk became a last resort for high yield issuers or companies with less favorable future prospects. That more than any other challenge in sukuk markets could marginalize sukuk as an asset class.


Bernardo Vizcaino CAIA said...

In the back of my mind I can think of a few sukuk out there (mainly in western countries) which fit into this notion of being too risky (i.e. not viable as conventional paper) and I just wish the paper had addressed those as well. In a sense I feel they reflect this 'riskiness' far more than what you have in the Malaysian market.

Ironically it is difficult to argue this since these periphery sukuk are one-off products and/or isolated incidents... making it impossible to perform any relevant empirical tests on them.

However, as you rightly pointed out there is much more analysis that can be done and hopefully this paper is just the first of many. This can help yield some suggestions/answers and eventually encourage the development of a wider family of sukuk products.