Financial Crime World

Here is the article in markdown format:

Efficiency of Tunisian Banks: A Study Using Data Envelopment Analysis (DEA)

Background

The Tunisian banking sector comprises 30 banks, with 23 being onshore and 7 offshore. Among these, there are six publicly-owned conventional commercial banks and three specialized in Islamic banking.

Sample Selection

This study focuses on the three largest publicly-owned conventional banks and the seven largest private ones, which collectively hold over 85% of total bank branches, employ about 83% of total banking staff, account for an annual average of 87.3% of total banking assets, generate 89.2% of total gross banking income, provide 88.7% of total loans by banks, and collect 90.2% of total deposits in banks.

DEA Model

The study uses a three-input/four-output DEA model where:

  • Labor
  • Physical capital
  • Financial capital (deposits) are inputs to produce:
    • The bank’s portfolio
    • Loans
    • Interbank loans
    • Off-balance sheet commitments as outputs

Validation of DEA Model

The Efficiency Contribution Measure proposed by Pastor et al. (2002) is used to validate the selected DEA model, indicating that candidate variables are deemed relevant if more than 15% of bank-year observations have their technical efficiency changed by at least 10%.

Input and Output Variables

These are obtained from published balance sheets and income statements of each bank in annual reports released by the Tunisian Professional Association of Banks and Financial Institutions (TPABFI).

Descriptive Statistics

The study presents descriptive statistics for the input and output variables, highlighting the role played by Tunisian banks as intermediaries between savers and borrowers.

Determinants of Technical Efficiency

Technical efficiency estimates can be used to identify determinants in a censored Tobit model or to regress technical inefficiency directly against its determinants. However, it is essential to correctly specify the model to avoid biased parameter estimates and questionable inferences.