Integrating corporate social responsibility and profitability into best practice selection: the case of large Taiwanese firms

Abstract

Corporate social responsibility (CSR) is rarely incorporated into best practice selection, despite the fact that most large multi-national corporations receive heavy pressure from stakeholders to engage in CSR. In contrast with CSR, profitability is always recognized as a major element when selecting best practices (or best firms). Without incorporating CSR into best practice selection, firms with high profitability but low CSR can be identified as best practices. If requested to emulate high-profitability yet low-CSR ‘best practices’ to improve the organizational performance, firms may increase profitability but also lower their social responsibility, eventually causing an abnormality. Data envelopment analysis (DEA) is a powerful benchmarking tool for situations where multiple inputs and outputs need to be assessed to identify best practices and improve productivity in firms. However, simply adding the CSR measure to DEA outputs does not change the selection of best firms. Specifically, when CSR and profitability measures are pooled as DEA outputs, the conventional DEA identifies the firms with low CSR as best practices. To overcome this drawback, a new benchmarking method is proposed to select best practices with an integrated perspective of CSR and profitability. The proposed method is applied to an empirical case including 66 large firms in Taiwan and opens up a new path for future benchmarking research.

Appendices

Appendix A: conventional DEA model

Consider a group of n firms, with each firm j (j = 1, …, n), using m inputs x _ij (i = 1, …, m) and producing s outputs y _rj (r = 1, …, s). Model (1) shows the input-oriented DEA with the constant returns to scale (CRS) assumption (Charnes et al. 1978).

$$\theta_{k}^{*} = min\theta - \varepsilon \left( {\mathop \sum \limits_{i = 1}^{m} s_{i}^{ - } + \mathop \sum \limits_{r = 1}^{s} s_{r}^{ + } } \right)$$

subject to

$$\theta x_{ik} = \mathop \sum \limits_{j = 1}^{n} \lambda_{j} x_{ij} + s_{i}^{ - } ,{\quad} i = 1,2, \ldots ,m$$

(1)

$$y_{rk} = \mathop \sum \limits_{j = 1}^{n} \lambda_{j} y_{rj} - s_{r}^{ + } ,{\quad} r = 1,2, \ldots ,s$$

$$\lambda_{j} ,s_{i}^{ - } ,s_{r}^{ + } \ge 0, {\quad}j = 1,2, \ldots ,n, i = 1,2, \ldots ,m, r = 1,2, \ldots ,s,$$

where $s_{i}^{ - }$ is the input excess of i-th input and $s_{r}^{ + }$ is the output shortfall of r-th output. The efficiency score θ ^* is expressed as a number between 0 and 1. An evaluated firm with an efficiency score equal to 1 is deemed efficient relative to other firms. The reference set for an inefficient firm _k is defines as E _k . E _k  = {j|λ ^k _j  > 0} (j ∈ {1, …, n}). An optimal solution can be expressed as $\theta^{*} x_{k} = \sum\nolimits_{{j \in E_{k} }} {x_{j} \lambda_{j}^{*} + s^{{ -^{*} }} }$ and $y_{k} = \sum\nolimits_{{j \in E_{k} }} {y_{j} \lambda_{j}^{*} - s^{{ +^{*} }} }$. These can be further interpreted as $x_{k} \ge \theta^{*} x_{k} - s^{{ -^{*} }} = \sum\nolimits_{{j \in E_{k} }} {x_{j} \lambda_{j}^{*} }$ and $y_{k} \le y_{k} + s^{{ +^{*} }} = \sum\nolimits_{{j \in E_{k} }} {y_{j} \lambda_{j}^{*} }$. These relations suggest that the efficiency of (x _k , y _k ) for firm _k can be improved if the input values are reduced radially by the ration θ* and the input excesses recorded in $s^{{ -^{*} }}$ are eliminated. Similarly efficiency can be attained if the output values are augmented by the output shortfalls in $s^{{ +^{*} }}$.

Appendix B: stratification DEA model

Assume that there are n firms which produce s outputs by using m inputs. The set of all firms is defined as J ¹ and the set of efficient firms in J ¹ is defined as E ¹. Then the sequences of J ^z and E ^z are defined recursively as J ^z+1 = J ^z − E ^z. Model (2) presents the stratification DEA model with the input-oriented CRS assumption (Seiford and Zhu 2003).

$$\theta_{k}^{z} = min\theta$$

subject to

$$\sum\limits_{{j \in J^{z} }} {\lambda_{j} x_{ij} \le \theta x_{ik} }$$

(2)

$$\mathop \sum \limits_{{j \in J^{z} }} \lambda_{j} y_{rj} \ge y_{rk}$$

$$\lambda \ge 0,$$

x _ij and y _rj are i-th input and r-th output of firm k. When z = 1, Model (2) becomes Model (1).