Profit Efficiency and Its Determinants: Evidence From Indian Commercial Banks

Abstract

This study examines profit efficiency and its determinants in Indian commercial banks during the post-reform period by using a stochastic frontier approach (SFA), specified by Battese and Coelli (1995). For the purpose of specifying the parameters in the SFA, translog functional form has been adopted. Assuming banking markets are not perfectly competitive, alternative profit function is estimated. Intermediation approach has been employed to define bank inputs and outputs. An unbalanced panel data of 103 commercial banks for the period 1996–2008 have been constructed for the empirical analysis. All the required data for the analysis were obtained from various publications of the Reserve Bank of India. The study found that profit efficiency of Indian commercial banks is increasing over the study period. However, on average, Indian banks could meet only three-fourths of their profit-generating potentialities relative to the best-practice bank, due to technical inefficiency, which is arising within the banks. Among the bank groups, it is revealed that the state-owned banks are relatively more efficient than their counterparts. The technical inefficiency effects model shows that bank-specific, market, and organizational characteristics play an important role in determining the profit efficiency of banks.

KEYWORDS:

• commercial banks
• profit efficiency
• determinants
• stochastic frontier approach
• India
• deregulation

Acknowledgments

The authors would like to thank the late Prof. K. Sham Bhat, Department of Economics, Pondicherry University, Pondicherry, India and Prof. B. Kamaiah, Department of Economics, University of Hyderabad, Hyderabad, India, for their valuable comments on the earlier version of this paper. The remaining errors are ours.

Notes

Source: Reserve Bank of India.

Notes: 1. All the monetary values are in real prices (1999–2000 = 100).

2. PBT = Profit Before Tax; LOANS = Loans and Advances; INVEST = Investments; NONINT = Noninterest Income; PL = Price of Labor; PPC = Price of Physical Capital; PPF = Price of Purchased Funds; CAP = Equity Capital; INDNPL = Industry Average NPL; AST = Assets; EQAST = Equity Capital to Asset Ratio; LOAST = Loans and Advances to Asset Ratio; LODEP = Loans and Advances to Deposits Ratio; NIIAST = Noninterest Income to Asset Ratio; LIQAST = Liquid Asset to Total Asset Ratio; DDTD = Demand Deposits to Total Deposits; RELNPL = Difference Between Bank NPL and Average Industry NPL; HHI = Herfindhal Herschman Index; STATE = Dummy Variable for State Ownership; FOREIGN = Dummy Variable for Foreign Ownership; GDP = Annual Growth Rate of Gross Domestic Product; and INF = Annual Rate of Inflation.

Notes: 1. SOB = State Owned Banks; DPB = Domestic Private Banks; and FB = Foreign Banks.

Values in parentheses show t-statistics.

3. ^a , ^b and ^c implies significant at 1, 5 and 10%, respectively.

4. PBT = Profit Before Tax; LOANS = Loans and Advances; INVEST = Investments; NONINT = Non-Interest Income; PL = Price of Labor; PPC = Price of Physical Capital; PPF = Price of Purchased Funds; CAP = Equity Capital; AST = Assets; EQAST = Equity Capital to Asset Ratio; LOAST = Loans and Advances to Asset Ratio; LODEP = Loans and Advances to Deposits Ratio; NIIAST = Non Interest Income to Asset Ratio; LIQAST = Liquid Asset to Total Asset Ratio; DDTD = Demand Deposits to Total Deposits; and RELNPL = Difference Between Bank NPL and Average Industry NPL.

Notes: 1. ^a indicates significance at 1%; ^b indicates significance at 5%; and ^c indicates significance at 10%.

2. λ is a likelihood ratio static calculated as − 2 [log(likelihood (H₀)) – log (likelihood (H₁)]. It has an appropriate chi-squared distribution with degree of freedom equal to the number of independent constraints under the H₀ hypothesis. *indicates λ is statistically significant, since λ values are greater than the appropriate mixed chi-square static presented in Table 1 of Kodde and Palm (1986).

Note: Std. Dev. = Standard Deviation; Min = Minimum; and Max = Maximum.

Note: Reported are mean values of profit efficiency of state owned banks (SOBs), domestic private banks (DPBs) and foreign banks (FBs). The values in parentheses are t-statistics for testing the equality of means. ^{a, b} , and ^c indicate significance at 1, 5, and 10%, respectively.

Note: Values in parentheses are percentage of banks.

Notes: 1. ^a indicates significance at 1%.

2. SIZE = Log of Assets; EQAST = Equity Capital to Asset Ratio; LOAST = Loans and Advances to Asset Ratio; LODEP = Loans and Advances to Deposits Ratio; NIIAST = Non-Interest Income to Asset Ratio; LIQAST = Liquid Asset to Total Asset Ratio; DDTD = Demand Deposits to Total Deposits; RELNPL = Difference Between Bank NPL and Average Industry NPL; GDP = Annual Growth Rate of Gross Domestic Product; INF = Annual Rate of Inflation; HHI = Herfindhal Herschman Index; STATE = Dummy Variable for State Owned Banks; and FOREIGN = Dummy Variable for Foreign Banks.

On the other hand, standard profit function assumes the existence of perfect competition in both input and output markets, requiring information on the prices of the output, which in most cases are not available. Further, it treats small and large banks as same and they should have same variable outputs when facing the same input and output prices, and other appropriate variables that may lead to scale bias since small banks cannot reach the same output levels (Vander Vennet, 2002; Kasman & Yildirim, 2006).

In fact, Das et al. (2005) measured standard profit efficiency for Indian commercial banks. They used investments, performing loan assets, and non-interest income. For the first two outputs, the respective prices are average interest earned per rupee unit of investments and average interest earned per rupee unit of performing loan assets. However, for the third output, the study has taken the amount of noninterest income itself.

Another competitive approach to define bank inputs and output in the empirical literature is production approach, which assumes financial institutions as providers of services for account holders. It considers deposits as an output, because they involve the creation of value added associated with liquidity and safe-keeping services provided to depositors (Yildirim & Philippatos 2007). Considering banks as financial intermediaries between savers and investors, the study prefers the intermediation approach and measures output in rupees (1999–2000 prices).

Berger and DeYoung (1997) for the U.S. banks and Williams (2004) for the European banks found mixed evidence on the exogeneity of nonperforming assets.

According to DeYoung and Hasan (1998), this variable is superior to more standard economic condition variables, such as gross domestic product (GDP) growth rate or unemployment rate, because it stresses the portion of economic conditions that are most relevant to the banks. Following studies such as Berger and Mester (1997), DeYoung and Hasan (1998), Akhigbe and McNulty (2003, 2005) this variable is included in the profit function.

The other widely used functional form is Fourier-flexible form, which combines a standard translog functional form with the nonparametric Fourier-functional. Despite its added flexibility, Berger and Mester (1997) found the difference in mean efficiency estimates between the translog and the Fourier to be negligible. In addition, Swank (1996) in a comparative analysis between translog and Fourier-flexible found supportive evidence for stability of the multi-product translog specification. Further, Fourier form requires larger number of parameters (Hasan & Marton 2003). Therefore, for these reasons, and given the limed date set, the present study avoids its adaptation and prefers the translog, which is widely used in empirical studies.

Where θ indicates the absolute value of the minimum value of profits over all banks in the period. This transformation is necessary to estimate the profit frontier without excluding all the banks with negative values from the sample. Such that, the dependent variable is ln(1) = 0 for the bank with the lowest profit, and positive for all other banks (Maudos et al. 2002; Bonaccorsi di Patti & Hardy, 2005; Fitzpatrick & McQuinn, 2008).

Standard symmetric assumptions are imposed on the second order parameters as ξ_ij = ξ_ji and τ_ij = τ_ji in accordance to economic theory. Since the study estimates alternative profit function that does not contain output prices, the estimation procedure does not restrict profits to be homogeneous of degree one in prices. But, studies that have mainly done comparative analysis between cost and profit efficiency applied the principle of homogeneity in input prices to keep the functional forms equivalent. Berger and Mester (1997) argued that this is not necessary for alternative profit frontier, since contemporary banking markets being perfectly competitive is far from reality, thereby assuming that banks maximize their profit by choosing input quantities and output prices, given both input prices and output quantities. In fact, Deyoung and Hasan (1998), Maudos et al. (2002), and Fitzpatrick and McQuinn (2008) followed this method.

Some of the variables included in the profit function also appear in Equation (6) one way or another way. For example, indicate these variables by X. Including X in Equation (6) will not cause bias or inefficiency in the estimation. Because the Profeff is constructed from both Π (which is observed) and u (which is estimated). X will affect Profeff only through its impact on Π, because by definition X is orthogonal to u (DeYoung & Hasan, 1998).

Ariff and Can (2008) found a positive relationship between profit efficiency and liquidity risk for Chinese commercial banks.

Akhigbe and McNulty (2005), Yildirim and Phillippatos, (2007), and Ariff and Can (2008) found a positive relationship between off-balance sheet activities and profit efficiency.

Hasan and Marton (2003) found that concentration in liquid assets reduced profit inefficiency for Hungarian banks, while Iannotta et al. (2007) found that liquid assets reduced profitability of banks for the European banking industry.

DeYoung and Hasan (1998) revealed mixed results on the effects of demand deposits to total deposits on the profit efficiency of de nova commercial banks for the United States.

DeYoung and Hasan (1998), Akhigbe and McNulty (2005), Yildirim and Phillippatos (2007), Ariff and Can (2008), and Fuentes and Vargara (2007) found that a higher nonperforming loans ratio will increase profit inefficiency. However, Casu and Girardone (2004) found a positive association between nonperforming loans and profit efficiency.

The lower the index is, the higher the competition is. The index ranges from close to zero to 10,000. It classifies a market with a result of less than 1,000 to be a competitive marketplace, a result of 1,000–1,800 to be a moderately concentrated marketplace, and a result of 1,800 or greater to be a highly concentrated marketplace (Brown & Frederick, 1988).

In banking literature, the approaches to estimate market competition are classified into two-structural and nonstructural approaches. The structural approaches are (i) Lerner index, (ii) Boone indicator, and (iii) Herfindhal-Hirschman Index (HHI). The nonstructural approaches are (i) Iwata model, (ii) Bresnahan and Lau (BL) model, (iii) H-statistic or Panzar and Rosse (PR) model, and (iv) X-inefficiency. However, to examine the impact of structural effects on firm performance, that is, testing structure conduct performance (SCP) hypothesis, studies typically used the HHI or n-firm concentration ratio (CRn) as an exogenous indicator of market power or an inverse indicator of the intensity of competition. Following studies such as DeYoung and Hasan (1998), Maudos et al. (2002), Fuentes and Vergara (2003), and Akhigbe & McNulty (2003, 2005), and Kasman and Yildirim (2006), the present study included HHI to examine effects of market structure on the performance of Indian banks.

Kasman and Yildirim (2006) and Pasiouras et al. (2007) found that inflation contributes to lower profits, whereas Koutsomanoli-Filippaki et al. (2009) found the opposite.

For instance, in 2007–08, State Bank of India raised Rs. 16,376 crore through a rights issue in March 2008, in which the government contributed over Rs. 10,000 crore to subscribe to its share. Besides, the government has also provided over Rs. 3,800 crore to Punjab and Sind Bank, Vijaya Bank, Central Bank of India, and UCO Bank, as part of a plan to spend Rs. 20,000 crore on recapitalization of state-owned banks. The move is aimed at strengthening these banks' capital base to enable them to lend more.

Statistical Tables Relating to Banks in India, Reserve Bank of India, various issues.