The approach is scientifically sound. If you ask the same question, this model is very good to provide the answer. However, if i understand correctly, it does not differentiate situations when foreign banks enter the market because the local banking is obsolete, inefficient and in trouble. Are the foreign banks entries the cause, or the effect of local banking weakness? The model i know does not reveal that.http://www1.fee.uva.nl/fm/PAPeRS/Claessens/foreignentry_JBF.pdf

ISSUE: Claessen et al. model

The abstract of that paper tells us that "this paper the extent and effect of foreign presence in domestic banking markets." J. Banking & Finance, 25 (2001), 891. reading through the paper, it appear the second part is missing: "effect of foreign presence on domestic banking markets"---was not reported.

PERFORMANCE COMPARISON NOT EFFECT MEASUREMENT: The article talks about performance comparison on the basis of accounting ratio comparison. This is not the same as measuring the effect of foreign banks having on that of domestic banks. In order to capture the effect of foreign banks on domestic banks, two periods must be examine: (i) performance of domestic banks prior to foreign banks entry, and (ii) performance of domestic banks after the entry of foreign banks. The study appears to be large: 7,900 banks from 80 countries for a period of 7 years (1988 - 1995 "inclusive"). In these 7 years, the effect measurement could have been measured if, say, 1988 - 1995 (2 years---as an initial period where the performance of foreign banks are recorded) then year 1990 through 1990 as comparison of the effect of foreign banks on domestic banks. This is not done----there is a problem here because at 1998, there are already foreign banks present---a means to control for that part must be design.

EFFECT MEASUREMENT: This is not the same as performance comparison. This part is missing from the article. In order to measure the effect of foreign banks on the performance of domestic banks in the banking industry, a separate effect measurement (not performance comparison) must be done separately. As a possible test candidate, suppose we measure the differences in performance from year to year, this allows us at least try to use Poisson two-counts.

Z = (R1 - R2) / sqrt [(R1 / t1) + (R2 / t2)]

R1 = N1 / t1

R2 = N2 / t2

... where N1 and N2 equals to the observed counts at t1 and t2. One can define effect as +1 and non-effect = 0.

For comparison of the performance, we now have two groups of data: (i) domestic banks before and after entry of foreign banks; (ii) domestic and foreign banks performance after foreign bank entry.

I am thankful to you Dr. Paul for such an extensive analysis of Claessens et. al, According to your answer on effect measurement, you mentioned that two groups of study period, i) before foreign bank entry- Liberalization ii) After foreign bank entry- after Liberalization are necessary. But if you look at the Indian scenario, most of the foreign banks entered into India after 1991 i.e. the period of financial liberalization. but the fact is also that the first foreign bank has entered in India in 1853 i.e. Hong Kong and Shanghai Bank. India had 22 foreign banks present with 148 branches in 1990 and today there are 43 foreign banks present with 334 branches. 

So, In your view point, is it still worth to measure foreign banks effect on domestic performance by dividing the time period into two i.e. Pre and Post Liberalization or else. Is the poison two counts still the concrete way to measure effect of foreign banks entry on domestic banking system ?

I think you should consider entry of banks into India by differentiating strategies. i..e. Only Corporate Investment Banking , only High End client Treasury, M&A, Organic growth and others for instance. India would definitely need a custom analysis and time period studies may still provide unique data but it may be more worthwhile to choose comparisons of strategy performance and their consequent commitment or other macroeconomic variables you may be interested in.