Determinants of Operational Efficiency and Total Factor Productivity Change of Major Cambodian Financial Institutions: A Data Envelopment Analysis During 2006–13

ABSTRACT

We examine the determinants of efficiency and the total factor productivity (TFP) change of major financial institutions in Cambodia using data envelopment analysis (DEA). We apply two-stage bootstrapping estimation method to a panel data of twenty-seven financial institutions in the period of 2006–13. The empirical results reveal that the size and ownership structure of financial institutions are significantly correlated with the efficiency and TFP growth of banks. The efficiency of domestic institutions is found to be better than that of their foreign counterpart, and there is no significant difference in TFP growth between domestic and foreign institutions.

KEY WORDS:

• bootstrapping analysis
• Cambodian financial institutions
• DEA
• operational efficiency
• TFP change

Funding

This study was financially supported by a Grant-in-Aid for Scientific Research from the Ministry of Education and Science (Scientific Research C, No. 25380283).

Notes

1. Throughout this article, the term “financial institutions” is used to refer to three types of intermediators: commercial banks, specialized banks, and microfinance institutions.

2. For example, most foreign-owned banks are small, and there is an apparent correlation between bank size and bank ownership.

3. The banking sectors in these Asian countries are less developed as well, although mostly different in many aspects, such as number of banks, banking law, and the roles of public banks and foreign-owned banks. In contrast to these countries, the banking sector in Cambodia has less regulation.

4. For a decomposition of the Malmquist productivity index, see Lovell (1993).

5. The variables used in the DEA model vary slightly across recent studies, even though these studies use the same concepts to define the input and output variables in bank activity. This is mainly a result of the limitations in available data for each country.

6. The variables used in the DEA model vary slightly across recent studies, even though these studies use the same concepts to define the input and output variables in bank activity. This is mainly a result of the limitations in available data for each country.

7. As we will show in Table 7, the diversification ratio in Cambodia is relatively high on average.

8. Although the intermediation approach was used in previous studies, value-added and operating approaches are better in Cambodia because some Cambodian banks, especially those that have newly entered the sector, tend to rely on equity to expand their loan portfolios. Thus it is likely that the efficiency scores would be overestimated if we were to adopt the intermediation approach.

9. This proxy is a combination of indicators of profitability, leverage, and return volatility. If this variable increases, it indicates a decrease in overall risk exposure and an increase in the stability of financial institutions (Ariss 2010).

10. This indicator has been widely used in many recent studies (Ariss 2010; Boyd and De Nicole 2005).

11. These studies use the ratio of equity to total assets to investigate the relationship between resilience and efficiency. However, it is vague as to what the ratio represents as an increase in the ratio of equity assets also implies an increase in payments to shareholders. Sometimes the payment to shareholders as a percentage of total equity becomes higher than payments to borrowing as a percentage of total borrowing. Thus, an increase in the ratio of equity to total assets does not always mean an increase in the resilience of bank operation. Furthermore, Cambodian banks generally have a high capital adequacy ratio due to the dollarized operational environment, and they already have a sufficiently high capital adequacy ratio.

12. Normally, ordinary least squares (OLS), a censored model, and a truncated model are mainly used in the literature on the two-stage DEA approach.

13. In our sample, the foreign shares of paid-up capital fluctuated and the foreign-owned dummies changed in several banks during the period in question. Thus the effect of the foreign-owned dummy was estimated, even in the fixed-effect model.

14. The further explanation of estimating the standard errors using the bootstrapping method is comprehensively described in Efron and Tibshirani (1993).

Additional information

Funding

This study was financially supported by a Grant-in-Aid for Scientific Research from the Ministry of Education and Science (Scientific Research C, No. 25380283).