The Impact of External Business Environment on Corporate Social Responsibility: Theoretical Implications and Empirical Evidence

The literature evidently demonstrates the CEO-centric effect upon firms’ Corporate Social Responsibility (CSR) engagements. Given the CEO’s strong discretionary power over CSR decisions and self-serving motive divergent from shareholder value maximization, it would be interesting to investigate the strategic decisions made by the CEO on CSR policy facing varied dimensions of market conditions. From an agency theory perspective, the paper develops a theoretical framework to model and clarify the relationships between firm’s CSR provisions and the three dimensions of external market conditions: market complexity, munificence, and dynamism. Furthermore, I empirically measure the market dimensions and test the propositions implied by the theoretical work. Consistent with the model implications, I found CEOs tend to invest more in CSR in a competitive market, less in CSR when the market is munificent and more in CSR when the market is dynamic and unpredictable. The results are consistent with the extant literature and shed light on the value relevancy of CSR activities.


Introduction
It is widely documented that corporate social responsibility (CSR) activities are deeply value relevant (see, for example, Szymanski and Henard, 2001, Godfrey, Merrill, and Hansen, 2009, and Luo and Bhattacharya, 2009). As such, corporate social performance (CSP) becomes an important dimension of firm performance, How to cite this paper: Qiao, Y. K. (2020). The Impact of External Business Environment on Corporate Social Responsibility: Theoretical Implications and Empirical Evidence. Theoretical Economics Letters, 10, 500-522.
https://doi.org/10.4236/tel.2020.103032 Theoretical Economics Letters theses. Section 5 summarizes the statistical tests and performs analyses on the empirical results in accordance with the theory and hypotheses. Section 6 gives concluding remark on the significance of the study and a brief outlook on the future study.

Environmental Dimensions
Rather than focusing mainly on the production market competition as in classic finance and economics literature, the organizational management literature has rigorous development on the environmental dimensions of external business conditions facing the firm (see, for example, Aldrich, 1979;Dess and Beard, 1984; and Keats and Hitt, 1988). According to the literature, those dimensions are theoretically defined as munificence, instability and complexity. Munificence or environmental capacity refers to the abundance of environmental resources and resulting capacity to support growth. Instability or dynamism reflects volatility or difficult-to-predict discontinuities in the dominant industry of a firm.
Complexity or market competition represents the heterogeneity and concentration of environmental elements. Williamson (1965), Khandwalla (1973) and Starbuck (1976) suggest that the industries demonstrating high monopoly power are less complex than those demonstrating less monopoly power. The literature has documented that the environmental context of the dominant industry affects firm-level strategies (Hill and Hoskisson, 1987;Hitt and Ireland, 1985).

Strategic Provisions of CSR
Extensive research attentions are also paid to addressing the relation between product market competition and CSR expenditures both theoretically and empirically (see, for example, Bagnoli and Watts, 2003;Fisman, Heal, and Nair, 2006;Fernández-Kranz and Santaló, 2010;and Zhang, Zhu, Yue, and Zhu, 2010).
The theories are divided. The signaling model in Fisman, Heal, and Nair (2006) shows that CSR provisions in private goods serves as a vertical differentiation tool in product market. Thus, in the separating equilibrium, socially responsible firms provide more CSR provisions in face of higher product market competition. However, in a pooled equilibrium, Bagnoli and Watts (2003) show that higher market competition (e.g. Bertrand price competition) leads to less than efficient amount of public goods provided in the private products since facing highly competitive market condition and shrinking profit margin, firms are incapable of carrying the additional costs incurred by CSR. Attempting to reconcile the literature, the model in Fernández-Kranz and Santaló (2010) shows that the net influence of competition on CSR depends on the tradeoff between the "marginal effect" and "business stealing effect", and thus depends on how the competition and demand function is modeled. Interestingly, both of the two opposing papers illustrate that in competition for socially responsible customers, Y. K. Qiao DOI: 10.4236/tel.2020.103032 503 Theoretical Economics Letters firms tend to provide more public goods. It is also consistent with the evidence in Zhang, Zhu, Yu, and Zhu (2010) that corporate philanthropic giving is positively related to advertising intensity.

Value Relevancy of CSR
It is widely documented in the literature that CSR activities are value relevant.
Firms are motivated to engage in social responsibility activities mainly to enhance customer satisfaction, build up intangible assets and goodwill, and stabilize financial conditions. Specifically, whereby improving customer satisfaction, socially responsible firms are more likely to enjoy higher customer loyalty (see, for example, Oliver, 1980 andBolton andDrew, 1991), receive higher reviews and evaluations (Brown andDacin, 1997 andSzymanski andHenard, 2001) and eventually are able to charge premium price. Prior studies have documented that the reputation and the socially positive image earned by the firm through proactively engaging in CSR provisions could be transferred to intangible assets and goodwill, leading to premium valuation (see, for example, McWilliams andSiegel, 2001 andLuo andBhattacharya, 2009). In face of external profit shocks, socially responsible firms are more likely to achieve financial stabilization and higher market value, according to the studies of resource-based theories (see, for example, Barney, 1991 andWernerfelt, 1984). In addition, firm is less likely to face accusations, filing complains, investigation and legal litigation especially in market depression period (Godfrey, Merrill, and Hansen, 2009).

Agency Problem
Agency theory tells us that unless CEO is the sole owner of the firm, the incentive of CEO shall always deviate from shareholder value maximization (see, for example, Berle andMeans, 1932 andJensen andMeckling, 1979). Provided that CSR is value relevant, CEO is likely to strategically utilize CSR to maximize her own utility according to the external market conditions. Thus, the theoretical modeling in the following section is centered at the utility optimization of CEO herself.

The Model
Based on the empirical evidence in Brick and Qiao (2017) and Bernard, Godard, and Zouaoui (2018), and the model in Fernández-Kranz and Santaló (2010) ( ) Fisman, Heal, andNair, 2006 andFernández-Kranz andSantaló, 2010), and I assume the incremental value is decreasing (concave curvature) according to the classic economic setting of diminishing marginal effect. Further, consistent with the typical form of cost function, cost of CSR expenditures ( ) i C CSP is a monotonically increasing convex function with respect to level of CSR provisions, i.e., 0 C′ > and 0 C′′ > (WLOG, I can also assume the marginal cost is constant 0 C′′ = ). According to Brick and Qiao (2017) and Bernard, Godard, and Zouaoui (2018), CSR expenditures are largely subject to CEO's discretionary power. Thus, the strategic decision making of CEO is the central part of modeling CSR provisions and business environment. The total compensation (w) of CEO is structured as follows.
i i w a bπ ε in which a is the base salary, positively related to the current level of market munificence, b represents the performance sensitive bonus and i ε denotes the random compensation shocks due to the uncertainty of the external business environment. Here I adopt the arguments of weak rational expectation (see, for example, Muth, 1961, andLovell, 1986) in the way that i ε represents a complete innovation term to the CEO and thus is exogenous to other component of CEO compensation. As such, i ε is a pure white noise, and is normally and independently distributed. The only prior knowledge of CEO as of making strategic decisions on CSR engagements is about the distribution of i ε , but not about any potential realizations. The distribution of i ε is clearly a reflection of the current level of market dynamism. Specifically, i ε is assumed to follow a demeaned Gaussian distribution, ( ) 2 0, i N σ , in which 2 i σ , as the sole factor parametrizing the distribution function, is endogenous to the current level of market dynamism facing the company and is negatively moderated by CSR provisions (stabilization effect of CSR, see, for example, Barney, 1991, Wernerfelt, 1984, and Godfrey, Merrill, and Hansen, 2009. Assume the CEO has a constant absolute risk aversion (CARA) utility function as follows, in which i w is the total income of the CEO in face of the current environmental conditions and i λ is the absolute risk aversion of the CEO and is determined by the CEO's current risk propensity. I further assume that the risk preference of the CEO is changing along with the variation of CEO's wealth, firm performance and For simplicity and without loss of generality, let 1, b a F = = and The strategic decision made by CEO on CSR expenditures can then be represented using the following optimization program.
Applying moment generating function (MGF), the above program is transformed as follows, Therefore, the FOC condition is as follows, in which 2 V σ = , denotes the variance of CEO's income uncertainty to simplify the notation. The optimal level of CSR provisions chosen by the CEO in frim i solves the above first order condition for the partial equilibrium. According to the extant literature on environmental dimensions and the value relevancy of CSR, the resulting equilibria is dependent on the parameters of external business environment: complexity or competition ( κ ), munificence ( µ ) and dynamism (θ ), and their relationships with the socially responsible activities undertaken by the firm. To clearly address the partial effect of each orthogonal environmental dimensions, for each firm i, I explicitly parametrize the differential equation of FOC as following.
Albeit without specializing the functional structures, several important implications can be drawn from the above general framework, in which the endogenous relation between each component and external parameters of business condition is explicitly denoted in the reduced form of comparative statics. In accordance with assumptions based on theoretical implications and empirical evidence in the extant literature, the following analytical process leads us to testable propo- On the other hand, according to the argument in Fernández-Kranz and Santaló (2010), provisions of CSR has "business stealing effect" in the sense that due to the warm glow altruism valued by consumers (Andreoni, 1989(Andreoni, , 1990, will increase in CSR provisions and this business stealing effect will become more severe in the more competitive market with more necessity of product differentiation, Proposition 1.b: Ceteris paribus, the aggregate effect of market complexity on the CSR provision is more likely to be positive when the advertising intensity of the industry is high. , the second component in the FOC captures the effects of environmental munificence ( µ ) and dynamism ( θ ). Specifically, when the external business environment has been experiencing rapid growth for years, the wealth of the CEO is rising due to the accrued incentive pays over years and the risk averse propensity of the CEO could be cushioned by the abundant resources of slack money. Furthermore, the CEO is more likely to hold optimistic perspective for the future development of the business and the degree of over-confidence inherent in her characteristics will be exaggerated, resulting in hard-driving managerial style in the context of decision making and policy implementation. Thus, the CEO is willing to take more risk as to making investment decisions on aggressive projects and i λ is decreasing. Since 0 λ µ ∂ < ∂ and muni- Finally, due to the stabilization effect of CSR (Barney, 1991;Wernerfelt, 1984;Godfrey, Merrill, and Hansen, 2009), the volatility of CEO's income brought by the market uncertainty will decrease in CSR provisions 1 0 2  and the sensitivity of V to CSR provisions is higher in more dynamic industries in the sense that when the environmental uncertainty is naturally high, firms with good social reputation would avoid more negative shocks than the same firm but in a stable business environment, . Analogous to the business stealing effect, the stabilization effect is expected to be more effective in the highly dynamic environment, leading to more expenditure on CSR. Obviously, the CSR provisions are more demanding in dynamic industries facing frequent negative shocks.
Assumption 3: The stabilization of effect of CSR provisions become increasingly effective when the environmental dynamism is higher, Proposition 3: Ceteris paribus, the influence of environmental dynamism on the level of CSR provisions chosen by the CEO is positive, The complete derivation of the above propositions using comparative statics is fleshed out in the appendix. In sum, based on the above theoretical work and testable propositions, I develop the following hypotheses to guide the empirical tests. Note that according to the assumptions associated with Proposition 1, the linkage between market complexity and CSR provisions is moderated by the advertising intensity. In addition to type II error, any potential negative results of the following hypotheses could be attributed to the fact that the underlying assumptions associated with propositions are not empirically valid.
H1: Market complexity is positively (negatively) related to CSR provisions.
H2: Market dynamism is positively related to CSR provisions.
H3: Market munificence is negatively related to CSR provisions.
To empirically verify the above hypotheses proposed in accordance with the in which z is a vector of three dimensions of external market conditions: complexity, munificence, and dynamism; x is a vector of fundamental covariates; t λ

Environmental Dynamism and Munificence
Regressing the logarithm of industrial sales against year, I have a model in time-series manner of 5 years rolling window as follows, in which t Y is the logarithm transferred industrial sales and t is year. According to Keats and Hitt (1988), the volatility of sales over every five years, measured as the antilog of standard error of regression slope coefficient (STD), is used as proxy for industrial dynamism. Similarly, the average growth of industry sales over every 5 years, measured as the antilog of regression slope coefficient (COEFF), is used as proxy for industrial munificence.

Environmental Complexity
As suggested by Williamson (1965), Khandwalla (1973) and Starbuck (1976) HHI is computed by summing up the squared market shares of each firm. In accordance with Keats and Hitt (1988) the complexity represents the competition structure in the market, which is closely related to HHI. I use one minus HHI as a proxy for environmental complexity (see, for example, Fang, Palmatier,

Measures of Corporate Social Performance (CSP)
Another methodological concern is how to construct a comprehensive measure of CSR with sufficient validity and reliability. For the sake of capturing the overall performance of CSR in a broad scope, I measure the level of social-friendly activities conducted by the firm using data from KLD Research & Analytics, Inc.  Agle, Mitchell, and Sonnenfeld, 1999;Deckop, Merriman, andGupta, 2006 andBae, Kang, andWang, 2011. In particular, I compute the scaled KLD score using the following formula: in which it KLD is the CSP metrics for firm i in year t, k t S is the number of strength items in year t for dimension k, k t C is the number of concern items in year t for dimension k, t K is the number of dimensions in the year, and k its Strength and k itc Concern are scores of strength and concern items in in year t for dimension k, respectively.

Other Control Variables of CEO and Firm Characteristics
Besides the above explanatory variables with specific moderating effects of interest, I also include other firm-based control variables in the econometric specification of the multivariate regression analysis. I use Compustat data to obtain  , Udayasankar, 2008, Cordeiro and Tewari, 2015. In order to remove the firm's size effect on variables such as R & D and ADV for intensity, I further modify those variables by scaling them by total assets. AT is taken logarithm to account for the skewness of sampling distribution. The detailed explanation and data sources are summarized in Table 1.  is positive and significant at 1%, suggesting that firms facing turbulent market conditions are more likely to engage in socially responsible activities, which is in line with the evidence demonstrated in Barney (1991), Wernerfelt (1984), and Godfrey, Merrill, and Hansen (2009)  influence of external market conditions stays unchanged statistically and the adjusted R squared for the specification with CEO fixed effects are much higher than those without, confirming the CEO-centric influence on CSR, which is in turn the foundation for the model development.

Results and Analyses
The primary findings of the empirical test are also robust to heteroscedastic standard error (Table 3), as well as using only main covariates (Table 4 & Table   5), in which much more data are employed in the empirical test. Further note that when using less firm covariates, the model includes CEO-firm linked fixed effects to capture unobservable heterogeneity in more granular level. Proxy for the firm's willingness to engage in earnings management. The proxy is calculated by subtracting the average level of accruals used by the firm's industry from the level of accruals scaled by the firm's total assets, in which accruals are defined as the difference between net income and cash from operations.

MTR
Marginal tax rate is the amount of tax paid on one more dollar of income, which is the proxy for the cost of investment in CSR in terms of after-tax dollars.
Via the website of John Graham

HHI
The Herfindahl-Hirschman index is a proxy for market complexity (e.g., Keats and Hitt, 1988), which is calculated by the sum of squared companies' net sales over the squared main industry sales to which the companies belong.

COEFF
The coefficient of regressing logarithmic industry sales on the fiscal year, using data of every five years (e.g., Keats and Hitt, 1988). It is a proxy for market munificence, indicating the average growth rate of industrial sales over five years.
Compustat STD This is the standard error of the coefficient generated from the above regression, which is a proxy for market dynamism. Compustat

Concluding Remarks
Building upon studies on the CEO-centric influence upon corporate social responsibility (see, for example, Fernández-Kranz and Santaló, 2010, Kang, 2017, Brick and Qiao, 2017, and Bernard, Godard, and Zouaoui, 2018, the paper steps The theoretical implications and empirical findings of the paper are of great significance both for policy makers and practitioners. To promote a socially responsible economy, the regulatory authority should pay attention to the external business environment in which firms operate, which may have vital impact upon the CSR provisions. Moreover, provided that the firm's social policy and CSR initiatives are largely dictated by the CEO, appropriate internal monitoring mechanism should be in place to mitigate the concern of self-serving behaviors and non-pecuniary consumption through philanthropic activities. One potential de-

Appendix A: Comparative Statics
As developed in the previous section, the first order condition for the partial equilibrium of firm i in a certain business sector can be expressed in the following notation, As shown in Figure 1, the shift of curve ( ) ,CSP κ Κ is dependent on the aggregate influence of the exogenous shock on the environmental dimension of market complexity. The power of two opposite forces, marginal effect and business stealing effect, is subject to the specific modeling structure and assumptions and therefore cannot be determined in the current analytical framework.