Hello Maan,

you should not determine using CRS or VRS in advance. As many DEA approaches do, you could run both models and compare the results.

e.g.: Article The Icelandic Economy: a victim of the financial crisis or s...

With an equal distribution of efficiency in CRS and VRS, the use of CRS is recommended. If the distribution is different you have to find out why, calculate MPSS and DRS/ IRS. Afterwords there has to be an explanation for the existence of DRS or IRS.

Best regards, Dominic

VRS is mostly recommended for commercial banks because if conditions are more close to CRS still you can capture in VRS.

Thank you, your help is really appreciated.

Dominic Loske

Salah U-Din

Hello,

The literature is offers mixed suggestions and findings on this and depends on the industrial organisation of the banking system you are examining (structure, condact etc) as well as on the sample itself.

There is a simple rule that I always suggest to my students on this. If the banks in your sample are quite homogeneous, all operate under similar conditions and if you think that it would be fair to compare all of them under the same e.g. loans to deposits ratio, then CRS is more appropriate. Now, if there are banks that may be new in the market and hence still struggling to grow, or if there are other banks that have branches even in the smaller villages of a country (perhaps a reason - e.g. publicly-owned), or have set up well-resourced business centers for special clients (large firms or the government), then VRS may be more appropriate (as it envelopes the frontier more closely). The reason in the latter case is because these extreme cases of banks, would be unfairly compared against the most productive scale size as it would be set out under CRS. So the simple question you need to ask if the assessment you are doing is a fair one.

A good review of the literature will provide you with the supporting evidence you require.

Thank you

I recommend VRS under BCC model for the efficiency of banks. You may decide on the bases of your choice of inputs and outputs.

You should run both BCC and CCR models initially and then examine the variability in your data. If variability is high go for BCC.

Ikram Ullah Khan Saeed Assani

Hello Sir, you mentioned that deciding the CCR or BCC model on the basis of the choice of inputs and outputs. My question is, how we can measure it? Thanks in advance.

Saeed Assani

I understand your point. Thank you so much.

Hi,

Q1. I have used the Python DEA package for analysis. I opted teh CRS and VRS with both input and output oriented method. But i could nt find which one is used (BCC or CCR)..how to find that when use pyDEA.?

Q2. I have got the DEA technical efficiency(TE)-CRS input oriented results and computed inefficiency (1- TE/TE) some of the score are going beyond 1 upto max of 15. In teh case while doing Tobit analysis by censoring dependent variable only between 0 and 1 is valid? as because we lose some of teh observation...

anyone can help with your thoughts....

Thanks,

RR