國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3
for banks to contemplate.
The thesis is organized as follows. In Chapter 2, we present a brief review of existing literature in regards to the assessment of bank’s efficiency and the relationship between CSR and CFP1. Chapter 3 introduces the stochastic cost frontier approach. Chapter 4 depicts data collection, defines variables, and presents descriptive statistics. We explain how to construct the CSR index in details. Chapter 5 conducts our empirical study and analyzes the results, while the last chapter concludes the thesis and claims some limitations of the research. The thesis follows suit.
2. Literature Review
When estimating the efficiency of banking industry, there are various approaches and wide spectrum of independent variables can be nominees to assess efficiency. Efficiency can be estimated either by production frontier, or cost frontier, or profit frontier. The production function can be specified as Cobb-Douglas, CES (constant elasticity of substitution), and translog forms. A cost or profit function can also be specified as translog and Fourier flexible forms (see Gallant, 1981 and 1982), which requires the use of input and output prices. Most frequently used model in banking industry is the translog cost function, which is known as being flexible and allows for an interpretation on the marginal cost of production for banks (Clark and Speaker, 1994; Weil, 2013). The Fourier flexible functional form combines a standard translog function with additional terms that are linear combinations of sine and cosine functions, also referred to as Fourier series (Huang and Wang; 2003, 2004). There are already tons of existing works that apply either a parametric or non-parametric approaches. The stochastic frontier approach belongs to a parametric approach and the data envelopment analysis (DEA) belongs to a non-parametric approach. The SFA as mentioned in the first chapter is the main method of this thesis, which was initially developed by Aigner et al. (1997) and Meeusen and van den Broeck (1997) endeavored in cross-section
1 Here we define CFP as corporate financial performance in terms of either financial ratio, return on assets, return on equity, or the efficiency on banking scope by using net interest margin and technical efficiency.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4
data when utilizing SFA to estimate efficiency. Later, panel data is recommended by Schmidt and Sickles (1984) to estimate technical efficiency (TE) for firms.
Nonetheless, earlier panel data models all counted on the assumption of time-invariant efficiency, which is likely inconsistent with the reality. Not until Cornwell et al (1990) and Battese and Coelli(1992) relieved this assumption. Most researchers utilize maximum likelihood (ML) to estimate the parameters of interest.
The variable of CSR is treated as one of the environmental variables. We then investigate the impact of CSR on banks’ efficiency. It is interested to know whether CSR positively or negatively affects the cost efficiency of banks. Since there is little study regarding to CSR and efficiency, here we will mention some studies in regards to CSR with financial performance covering different industry. There are several papers analyzing the relationship between CSR and financial performance, including return on assets (ROA), return on equity (ROE), technical efficiency, and so on. Unfortunately, mixed results are found. We will briefly review some of them.
First, some scholars claim that conducting CSR incurs extra costs for firms (Vance, 1975), since firms would have to employ additional resources that would be otherwise used to produce goods and services in pursuit of maximizing economic profits (or minimizing production costs). Anginer et al. (2008), Brammer et al.
(2005), and Nejati and Ghasemi (2012) argue that CSR is a misuse of allocating corporate resource, because the deployment of resources on CSR will put firms in a relatively economic disadvantage at the same time. Walley and Whitehead (1994) demonstrate a limited return on CSR engagement. Contrarily, two studies related to meta-analysis, i.e., Margolis and Walsh (2003) and Orlitzky, Schmidt, and Rynes (2003), reveal the positive relationship between CSR and CFP in a modest level.
Alexander and Buchholz (1978), Aupperle et al. (1985), and Shane and Spicer (1983) fail to find the association of CSR with CFP. Meanwhile, positive relationship between CSR, ROA, and ROE is documented by Tsoutsoura (2004) by using Kinder, Lyndenberg and Domini (KLD)2 so do Orlitzky et al.’s (2003) and Rettab et al (2009) conduct same positive result of CSR in financial performance, personnel commitment, and corporate integrity. Barnett and Salomon (2006) further indicate that CSR and financial performance share the curvilinear relation which gives us the other idea to define the association of CSR with efficiency.
2 Kinder, Lydenberg and Domini & Co. (KLD), a social choice investment advisory firm which has made available their social performance database.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
5
On the other hand, if we only look into the specific industry, banking industry, the results shown in those theories are much more consistent. When digging into the socially financial performance of banking sector, researchers solidly support the hypothesis that the link between social and financial performance is positive (W.
Gary ; Simpson;Theodor Kohers, 2004). Additionally, when it comes to the CSR and efficiency in a single country, Ohene-Asare and Asmild (2011) are the scholars who contribute to this field greatly. They examine the efficiency of Ghanaian banks with CSR as one of the environmental variables and conclude that banks that are socially responsible may have economic advantages on profitability and efficiency.
They no more view CSR as conceptual grounds. Keffas and Olulu-Briggs (2011) confirm the correlation between CSR and financial performance in the U.S., UK and Japan. In their studies, there are two groups, one declares the presence of CSR, and the other group of CSR is absent. The authors come to the conclusion that CSR not only benefits bank’s capital adequacy but also accumulates the competitive advantages in the long-run. The positive and strong relevance of CSR and firm performance is agreed by the findings of Alafi and Hasoneh (2012), which are based on housing banks in Jordan.
The foregoing inconsistent results may be interpreted by, e.g., Baron (2001) and Dam et al (2009), who affirm that different motives of corporation initiating CSR result in dissimilar relationship between CSR and financial performance.
Distinct motives include altruism, strategic choices, and green washing3. The motivation of altruism usually predicts negative correlation between CSR and CFP in terms of non-performing loans. Strategic choices support positively association between CSR and financial performance, including ROA, ROE, net interest income, and non-interest income. Greenwashing presents no significant effect of CSR on FP (Wu and Shen, 2014).
Despite of the motivation differences that contribute to the conflicting result of CSR impact, some researchers deem the effect of CSR may be influenced by other exogenous factors, e.g., the growth of an industry that will positively mitigate the connection between application of CSR and CFP (Russo and Fouts, 1997). This
3 The altruism motive indicates that companies conduct CSR activities for their own sake.Recent theories of CSR (see McWilliams and Siegel, 2000, 2001; Baron, 2001; Hillman and Keim, 2001;
Bagnoli and Watts, 2003) assert that firms especially engage in CSR for 'strategic' or 'prot-maximizing'.
According to Dam et al. (2009), if no clear cost difference are observed between responsible and irresponsible corporations, then these firms are merely greenwashing hence no effects will become evident in their earnings.
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
6
implies that non-systematic risk and macroeconomic condition should also be considered at the same time. According to Barros et al. (2007), the efficiency scores estimated from the common frontier4 model are apt to be influenced by individual countries conditions. This implies that the omission of cross-country differences, such as demographic and economic and political systems, might lead to biased efficiency measures. Allen and Rai (1996) and Altunbas et.al (2001) estimate global cost frontiers for all banks from different countries, assuming common production technology across nations. There is a growing consensus on the importance of environmental variables played in efficiency estimation. Those environmental factors are not controlled by managers and must be treated as exogenous variables.
See, for example, Battese and Coelli (1995), Dietsch and Lozano-Vivas (2000), and Kumbhakar and Wang (2007). In this thesis we consider several of such variables that are associated with efficiency scores. Chapter 4 describes those variables, as well as those relevant to cost function.
In spite of the discussions of CSR influence, there is another interest issue having been examined, i.e., the cushion effect of CSR engagement on stock price and bond price, when firms happen to face negative events. Yang and Shiu (2016) collect 399 sample points, consisting of 55 financial and 344 non-financial firms.
Their results suggest that the CSR engagement does help firms exempt from the decrease in bond or stock price, when negative events occur. However, the insurance-like effect only exists with continuous involvement in CSR and will vanish within a short time following the occurrences of a second or subsequent, negative event.
To sum up, one should not expect CSR engagement can benefit firms within a short time period, say, one or two years. Mullen (1997) and Yang (2015) state that CSR programs usually require lasting 3-5 years to gain the firms. Moreover, Long-term CSR engagement not only benefits the efficiency of firm itself, but also decreases the effect of competitor’s long-term CSR by increasing its own long-term CSR engagement. There are few studies investigating the direct impact of CSR on bank’s efficiency in the context of SFA.
For this study, since input prices are available from the Bankscope databank, we suggest the use of the cost frontier to estimate the cost efficiency of the sample
4 Cost efficiency and alternative profit efficiency are estimated separately from best practice frontiers using panel data methods. A common frontier is specified due to similarities in regulation and technology across markets
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
7
banks, which may be a preferable measure for regulators and business consultants who attempt to gauge the costs and benefits to society from distinct policies to the conventional inefficiency measure based on a production frontier. It is known that a production function is suitable for a single output case without using the information on input prices which are key variables affecting the decision on inputs hiring. Therefore, the applying of cost frontier is recommended due to diversifying financial products banks provide (Huang,Chiang and Chen; 2011). Different from previous studies, we apply the SFA (SFA) to investigate the relationship between cost efficiency and CSR index that is compiled from EIRIS database.5