**Summary :**

The aim of my PhD thesis is: first, to check for differences in technology between the Islamic and the conventional banking systems in the Middle East & North Africa and South East Asia regions, and second, to investigate the impact of the banking productivity on economic growth in the MENA region.

**Research questions :**

- Are Islamic banks more cost efficient than their conventional counterparts?
- Are there differences in technology between Islamic and conventional banks?
- What is the impact of banking cost efficiency on economic growth in the MENA region? And does economic growth affect banking cost efficiency?

**Methodology :**

- For the research questions 1 and 2 : Stochastic Frontier Approach, and Stochastic Meta-frontier Approach for cost efficiency estimations (following Battese, 2002 and Bos & Schmiedel, 2007)
- For the research question 3 : Causality analysis using a two-step Generalized Methods of Moments (following Windmeijer, 2005)

**Conclusions :**

This Ph.D. thesis provided two novelties:

- A cross-country comparison between Islamic and conventional banking cost efficiency highlighting the quasi-inexistent technology gap between the two banking system
- The existence of a causality and reverse causality between banking productivity and economic growth in the MENA region

Three major remarks stem out of my results :

- Islamic banking will need more than just the replication and adaptation of conventional contracts for its development in order to achieve its main goal :
**promoting social welfare** **Efficient investments and regulation**in the Middle East & North Africa region should promote a higher banking productivity level and financial deepening

**Future research:**

Considering future research, I think it should be interesting to investigate the post 2008 financial crisis in emerging countries, and more precisely investigate to which extend Islamic banks have moderated the effect of the crisis in these countries.

I guess the main limitation in this case would be to gather relevant data in order to run a causality analysis. As some countries have a certain opacity around their data disclosure (at both the macro and micro levels).

I would cite for example the UAE which purposely was not publishing its macro data in the last period, fearing speculation consequences on the economy.

Another point of reference for future research is to assess the impact of Islamic banks' investments efficiency on the economy as a whole. This would be a very useful study if Islamic banks give access to their investment portfolio mix, which is not an easy task.

**Link to my PhD thesis**

## 6 comments:

Hi Walid,

My name is Viktorija Atanasovska, and currently I am a PhD student at Staffordshie University, UK.

Part of my thesis is devoted to efficiency estimation (I need efficiency as a variable in my latter model - Determinants of market structure).

At the beginning I was reading for SFA only, but recently I came across metafrontier and I read your thesis.

However, I am confused about the estimation procedure itself, i.e. whether metafrontier can be estimated in some the friendly user software :limdep, stata ... or it should be programmed in gauss/matlab?!

Please, I would be very grateful if you can help me and give me some comments.

Thank you very much.

Kind regards,

Viktorija

Hi Viktorija,

I personally used Matlab, the version I used contains a optimization tool GUI for linear and quadratic programming. To check your version of Matlab has the GUI type: "optimtool" and normally the GUI will be launched. All you need to do is to enter your matrices (of course a big deal of data preparation and reshaping is required).

MS Excel does LP very well, but is limited to 32 constraints (so if your model doesn't include more than that number, it might be easier to use MS Excel. There is another program GAMS (http://www.gams.com/) which is limited to 150 contrainst! and it's free to download (but I don't know much about it).

With regard to your model, what are your assumptions? as the Metafrontier cannot be a simple substitution to the SFA in itself, but could be used if you assume you have different frontiers which do not share the same techonology (which can be tested by the LR test).

I hope this helps.

Kind regards,

Walid Ghannouchi

Hi Walid,

Thank you very much for the quick response.

Unfortunately, I have never worked in GAUSS/MATLAB ... :(

I need the efficiency estimates for each bank for each year (1995-2009, so later I can use these estimates as independent variable in other model.

My country set comprises 15 transition economies, which are similar (ex-post communist countries), but I assume on different technological level (since some of them are already EU members, some of them not).

Please could you help me first by answering whether I can obtain by metafrontier efficiency for each bank for each year and whether there is some intuitive explanation for metafrontier and for its estimation (for dummies like my self).

Thank you very much. I hope you can help me and guide me in this process, if I decide to use this method.

Best wishes,

Viki

Hi Vik,

Here is my suggestion:

1/ obtain the single frontiers estimates for each country

2/ Run the Likelihood Ratio Test (LR Test) to check for technology gap

3/ If your LR test shows the existence of a technology gap, then you can start thinking of the Stochastic Metafrontier Approach as a methodology to estimate your cross country efficiency.

With regard to the optimization software you'll need to use: Matlab is not compulsory as I posted before, there is a number of software that could be a good alternative...

In my opinion, you shouldn't worry about knowing how to use the software or not, I personally never worked on Matlab before doing my PhD, so I just spent some time learning how to use it till I mastered it.

That's the essence of doing a PhD: you start something from scratch and work your way up till you gain knowledge and expertise in it.

Do not hesitate to contact me in any case. (you can invite me on Linkedin, I can communicate my email address there if you need further help)

All the best

Walid Ghannouchi

Hi walid,

My name is kanis fatama ferdushi and currently I am a phd student at Universiti sains malaysia.

my working field metafrontier analysis. befor i did metafrontier technical efficiency with SHAZAM.Actually i am interested about Mathlab.thats why i need your assistance.

with best wishes

kanis

Hi Kanis Fatama,

I don't think there will be any difference regarding the results. The optimization to run in order to obtain the metafrontier is quite straight forward and any maths package should provide the same results. In this context, Shazam and Matlab are just tools to run the optimization. I am not familiar personally with Shazam. Let me know if I can help you with Matlab.

Regards,

Walid

Post a Comment