**3. Data and methodology**

#### **3.1. Data**

To investigate our hypotheses, we start by extracting all firm-level constituents of the MSCI AC World Index, which captures large and medium market capitalization stocks of both developed and emerging market countries, on a monthly basis for the time period of 2009 to 2014. We then extract firm-level characteristics from FactSet Research Systems (hereafter, FactSet) and merge this database with MSCI's ESG database. To be included in our dataset, we require firms to have non-missing ESG scores. We also drop firms from Taiwan for consistency across our analyses, as the World Bank does not report important country-level statistics for Taiwan.<sup>2</sup> Finally, we only retain firms that have enough available data to construct control variables. This procedure yields 134,823 monthly observations of 2542 companies across 44 countries and 128 industries.

<sup>2</sup> https://datahelpdesk.worldbank.org/knowledgebase/articles/114933-where-are-your-data-on-taiwan/, retrieved on 30 March 2015

To validate the significance of cross-country variation valuation exposure to CSR, we observe the results of our investigations under differing institutional and macroeconomic conditions in later tests. In this study, we use MSCI's market classification criteria, which segregate our sample of 44 countries into 23 developed markets and 21 emerging markets. **Table 1** provides the number of firms by country.

For our analyses, we exploit a firm-level measurement of how much CSR a firm undergoes to empirically test our hypotheses. The source of this data is MSCI's ESG database, which independently rates firms on their environmental, social, and governance (ESG) performance,


This table displays the number of firms by country for the time period of 2009 to 2014. The sample includes all firms extracted from the MSCI AC World Index between 2009 and 2014 with sufficient firm-level and CSR data.

**Table 1.** The list of firms in each country.

*Hypothesis 1: CSR creates value for the firm.*

80 Firm Value - Theory and Empirical Evidence

esize that:

Scholars have also put forth evidence that CSR is heterogeneous in nature such that the inherent dimensionality of CSR has implications for value creation (e.g., [2, 13]). Thus, we hypoth-

*Hypothesis 2: The CSR-valuation relation is heterogeneous in nature and CSR dimension is dependent, such that there is significant heterogeneity in valuation effects across different groups of stakeholders.* Khanna and Palepu [21] introduce the notion of institutional voids, which they define as the absence of institutions or intermediaries that are instrumental in supporting business operations in the context of a country's capital, labor, and product markets, its regulatory system, and its mechanisms of contract enforcement. For example, in an environment with underdeveloped financial institutions, the absence of mechanisms such as financial reportage, watchdog oversight, and analyst coverage works to increase informational asymmetry and decrease market efficiency. It follows that these financial markets will experience a decrease in investor willingness, negatively impacting capital access and forcing firms to seek alternative means (e.g., [50]). Similarly, an environment with underdeveloped economic institutions may force firms to find innovative ways to obtain skilled labor. Anecdotally, Khanna and Palepu [21] describe how Microsoft was compelled to collaborate with local firms and other stakeholders to aid the development of China's software industry and subsequently demonstrated how this has led to significant benefits for the firm. Lastly, an environment with underdeveloped governmental institutions might require firms to leverage their relationship with the government and reputation established by prior dealings, as they cannot rely on the robustness of the judicial system. Indeed, Khanna and Palepu [49] theorize that a key motivation behind a firm's engagement in CSR arises from a need to fill these institutional voids to

subsequently allow their business to thrive in these markets. Thus, we hypothesize that:

*governmental institutions will result in a greater (lesser) valuation effect.*

**3. Data and methodology**

**3.1. Data**

2

March 2015

*Hypothesis 3: The CSR-valuation relation is moderated by the institutional frameworks that firms operate in, such that the presence of greater (lesser) institutional voids in financial, economic, and* 

To investigate our hypotheses, we start by extracting all firm-level constituents of the MSCI AC World Index, which captures large and medium market capitalization stocks of both developed and emerging market countries, on a monthly basis for the time period of 2009 to 2014. We then extract firm-level characteristics from FactSet Research Systems (hereafter, FactSet) and merge this database with MSCI's ESG database. To be included in our dataset, we require firms to have non-missing ESG scores. We also drop firms from Taiwan for consistency across our analyses,

only retain firms that have enough available data to construct control variables. This procedure yields 134,823 monthly observations of 2542 companies across 44 countries and 128 industries.

https://datahelpdesk.worldbank.org/knowledgebase/articles/114933-where-are-your-data-on-taiwan/, retrieved on 30

Finally, we

as the World Bank does not report important country-level statistics for Taiwan.<sup>2</sup>

assigning them a numerical ESG index score (from 1 to 100, with 100 being the highest). MSCI's ESG constructs indices of sustainable investment value and risk factors of more than 6300 public corporations worldwide using a specialized list of 150 RiskMetrics adjusted for various markets, regional, ownership, or sector differences.<sup>3</sup> MSCI only considers CSR issues that have a material impact on the firm, implying that the index score parallels the firm's investment in CSR. Throughout the course of this study, we utilize MSCI's global rating, which compares each individual firm's ratings to all rated firms.

for our sample. We also plot the time series average of the three CSR component scores over

Can Corporate Social Responsibility Fill Institutional Voids?

http://dx.doi.org/10.5772/intechopen.76164

83

While firms in developed markets tend to have better environment scores, we note that firms in both markets consistently improve their average score year over year. This phenomenon is not present when we examine social and governance dimensions. For the social dimension, firms in both markets appear to converge toward the middle score of 50 over time. For the governance dimension, we see that firms in emerging markets tend to outperform firms in developed markets. When we observe the marginal month-on-month changes over time, we see that CSR ratings for firms in both markets tend to stay constant over time and appear to have similar patterns of change across all three dimensions. This indicates that on average, a firm's ESG score tends to stay constant, but there are also firms that experience large changes in ESG scores. This is consistent with the fact that firms tend to undergo periodic, substantial investments in CSR (i.e., rethinking energy source procedures, reconceiving manufacturing processes to be more sustainable, etc.) versus gradual improvements over time (e.g., [8, 51]).

time from **Figures 1**–**3**.

**Figure 1.** Average environment score.

**Figure 2.** Average social score.

**Figure 3.** Average governance score.

**Table 2** reports the average overall environment (E), social (S), and governance (G) scores and marginal month-on-month changes in CSR component scores by year and market classification


This table displays both the full sample and subsample (i.e., developed/emerging market) averages of overall environment, social, and governance scores and marginal month-on-month changes in CSR component scores by year from 2009 to 2014.

**Table 2.** The summary statistics of CSR component scores by year.

3 MSCI's RiskMetrics increased its coverage from 105 dimensions to 150 dimensions starting May 2013. for our sample. We also plot the time series average of the three CSR component scores over time from **Figures 1**–**3**.

While firms in developed markets tend to have better environment scores, we note that firms in both markets consistently improve their average score year over year. This phenomenon is not present when we examine social and governance dimensions. For the social dimension, firms in both markets appear to converge toward the middle score of 50 over time. For the governance dimension, we see that firms in emerging markets tend to outperform firms in developed markets. When we observe the marginal month-on-month changes over time, we see that CSR ratings for firms in both markets tend to stay constant over time and appear to have similar patterns of change across all three dimensions. This indicates that on average, a firm's ESG score tends to stay constant, but there are also firms that experience large changes in ESG scores. This is consistent with the fact that firms tend to undergo periodic, substantial investments in CSR (i.e., rethinking energy source procedures, reconceiving manufacturing processes to be more sustainable, etc.) versus gradual improvements over time (e.g., [8, 51]).

**Figure 1.** Average environment score.

assigning them a numerical ESG index score (from 1 to 100, with 100 being the highest). MSCI's ESG constructs indices of sustainable investment value and risk factors of more than 6300 public corporations worldwide using a specialized list of 150 RiskMetrics adjusted for

that have a material impact on the firm, implying that the index score parallels the firm's investment in CSR. Throughout the course of this study, we utilize MSCI's global rating,

**Table 2** reports the average overall environment (E), social (S), and governance (G) scores and marginal month-on-month changes in CSR component scores by year and market classification

**Year Obs. ESG E S G ∆E ∆S ∆G** 16,976 44.04 44.75 52.36 45.28 0.14 0.05 0.03 22,995 44.91 47.30 52.99 45.22 0.28 0.07 −0.03 23,626 45.17 48.58 52.47 45.75 0.07 −0.07 0.03 23,484 43.24 51.38 55.05 43.91 0.20 0.16 −0.30 24,351 40.17 57.79 49.74 42.58 0.90 −0.24 0.30 23,391 44.44 63.51 50.81 47.18 −0.83 −0.22 −0.11 **Total 134,823 43.62 52.62 52.21 44.96 0.13 −0.05 −0.01**

 13,246 45.30 48.07 54.95 44.88 0.15 0.02 −0.22 17,718 45.17 50.83 55.41 43.91 0.28 0.05 0.01 17,605 45.65 52.61 55.36 44.63 0.06 −0.09 0.04 17,411 43.66 54.35 59.15 43.27 0.17 0.38 −0.16 17,345 40.08 60.73 51.59 41.00 0.85 −0.48 −0.01 16,903 44.17 66.06 51.19 46.18 −0.71 −0.21 0.41 **Total 100,228 43.96 55.67 54.62 43.93 0.14 −0.06 0.02**

 3730 39.57 32.99 43.16 46.70 0.12 0.18 0.94 5277 44.04 35.48 44.86 49.64 0.28 0.14 −0.16 6021 43.78 36.79 44.01 49.02 0.11 −0.02 −0.03 6073 42.06 42.85 43.29 45.75 0.31 −0.48 −0.70 7006 40.40 50.49 45.15 46.50 1.00 0.35 1.06 6488 45.16 56.88 49.82 49.77 −1.13 −0.25 −1.48 **Total 34,595 42.64 43.79 45.24 47.92 0.12 −0.02 −0.11** This table displays both the full sample and subsample (i.e., developed/emerging market) averages of overall environment, social, and governance scores and marginal month-on-month changes in CSR component scores by year

MSCI's RiskMetrics increased its coverage from 105 dimensions to 150 dimensions starting May 2013.

**Table 2.** The summary statistics of CSR component scores by year.

MSCI only considers CSR issues

various markets, regional, ownership, or sector differences.<sup>3</sup>

82 Firm Value - Theory and Empirical Evidence

3

**Developed markets**

**Emerging markets**

from 2009 to 2014.

which compares each individual firm's ratings to all rated firms.

**Figure 2.** Average social score.

**Figure 3.** Average governance score.

#### **3.2. Tobin's Q in cross-sectional regressions**

To assess the CSR-firm value relation, we examine the impact of CSR on firm value, utilizing monthly Tobin's Q (TOBINW) in our analyses. We define Tobin's Q as the market value of equity minus the book value of equity plus the book value of total assets divided by total assets (e.g., [13]). To mitigate the effect of outliers on our observations, we winsorize Tobin's Q at the 2.5 and 97.5 percentiles. **Figure 4** shows that firms in both developed and emerging markets generally experience similar patterns of firm valuation over the time period of 2009 to 2014. Empirically, we estimate the following equations below:

$$\text{Tobin's } \mathbb{Q}\_{\downarrow t} = \boldsymbol{\beta}\_0 + \boldsymbol{\beta}\_1 \text{ CSR } \text{Overall}\_{\downarrow t-1} + \boldsymbol{\beta}\_2 \text{ X}\_{\downarrow t-1} + \boldsymbol{\varepsilon}\_{\downarrow t} \tag{1}$$

in valuation even after controlling for other firm characteristics. Here, the null hypothesis expects these coefficients to be zero, while the alternate hypothesis is that they are significant

Can Corporate Social Responsibility Fill Institutional Voids?

http://dx.doi.org/10.5772/intechopen.76164

85

Next, we explore how the CSR-valuation relation changes in the presence of different institutional voids related to financial, economic, and governmental institutions. To capture the complex and multidimensional nature of a country's institutional framework, we collect a variety of county-level measures to serve as proxies for the presence of institutional voids. We then utilize these measures to observe the sensitivity of the CSR-valuation relation to institu-

First, we collect measures related to economic development. These include the log of gross domestic product (GDP) per capita (GDPPC) from the Economist Intelligence Unit, Index of Economic Freedom (FREE), and the ratio of total investment to GDP (CINV) from the International Monetary Fund (IMF) to capture the rate of infrastructural development.

Second, we collect measures related to financial market development. This includes the ratio of bank deposits to GDP (GFDDB) from the International Financial Statistics and IMF, the ratio of the outstanding domestic private debt securities to GDP (GFDDP) from the Bank for International Settlements, and the ratio of stock market capitalization to GDP (GFDDS) from

Lastly, we collect measures related to governmental institution development. We follow Low, Tee, and Kew [18] in utilizing the World Bank Governance Indexes (WBGI). The World Bank constructs indices from 441 variables taken from 35 different sources produced by 33 organizations (Kaufmann, Kraay and Mastruzz [53]). WBGI measures six dimensions of country governance, which include voice and accountability (WGIVA), government effectiveness (WGIGE), regulatory quality (WGIRQ), rule of law (WGIRL), control of corruption (WGICC), and political stability (WGIPS). **Table 3** reports the summary statistics of the key variables as

To explore the moderating effect of institutional voids on the CSR-valuation relation, we construct a series of dummy variables. For each measure, we sort countries according to their performance and assign them a value of 1 if they place in the lower 50th percentile for that month. The only exception is the ratio of total investment to GDP, where we assign countries a value of 1 if they place in the upper 50th percentile for that month. For each measure of institutional voids, we rerun our regression estimates with the inclusion of the dummy term and the interaction term of the dummy and CSR. This models the marginal valuation effect of CSR

in the presence of institutional voids. Thus, we estimate the following equation:

*β<sup>d</sup> CSRi,d,t−<sup>1</sup>* + *β<sup>4</sup> IFVi,<sup>t</sup>* + ∑

here, *IFVi,t* is a dummy that takes a value of 1 if the country that firm i operates in scores in the lower 50th percentile for a given measure of institutional framework strength at time t and

*n*=*1 N*

*β<sup>n</sup> CSRi,n,t−<sup>1</sup>* × *IFVi*,*<sup>t</sup>* + *<sup>8</sup> Xi*,*t*−*<sup>1</sup>* + *<sup>i</sup>*,*<sup>t</sup>* (3)

tional voids in financial, economic, and governmental institutions (e.g., [21, 49]).

the Global Stock Markets Factbook and Standard and Poor's.

well as these institutional void measures.

*Tobin*" *s Qi*,*<sup>t</sup>* = *β<sup>0</sup>* + ∑

*d*=1 *D*

and greater than zero.

**3.3. Institutional void analysis**

$$Tobin\text{'s }Q\_{i,t} = \beta\_o + \sum\_{d=1}^{D} \beta\_d \text{ CSR}\_{i,d,t-1} + \theta\_4 X\_{i,t-1} + \varepsilon\_{i,t} \tag{2}$$

here, *Tobin's Qi,t* is firm *i*'s Tobin's Q at time *t*. *CSR Overalli,t-1* is the overall index measure of CSR for firm *i* at time *t-1*. *CSRi,d,t-1* is the individual dimension index measures of CSR for firm *i* relative to dimension *d* (i.e., environment, social, governance) at time *t-1*. *Xi,t-1* is a vector of firm-level controls obtained from FactSet at time *t-1*, which include return on assets (LROAW), leverage-to-equity ratio (LLEVW), capital expenditure-to-asset ratio (LCAPXW), cash-to-asset ratio (LCASHW), year-on-year sales growth (LSGRW), advertising expenditure-to-total asset ratio (LADW), log of total assets (LASSET), and a dummy variable if the firm paid out dividends (LDDUM). In particular, we take special care to collect data on advertising expenditure as prior research has suggested that the valuation effect of CSR is moderated by firm visibility (e.g., [3, 52]). In order to mitigate the effect of outliers on our observations, we winsorize firm-level characteristics defined as ratios, namely, LROAW, LLEVW, LCAPXW, LCASHW, LSGRW, and LADW, at the 2.5 and 97.5 percentiles. We also include year dummies to account for yearly sources of heterogeneity. *εi,t* is the stochastic error term, assumed to be independent and identically distributed random variables with zero mean and constant variance. Similarly, we also include industry and country dummies to account for industry and country sources of heterogeneity. We are interested in the coefficient β<sup>1</sup> for Eq. (1) and βd for Eq. (2), which measures whether a firm's CSR drives changes

**Figure 4.** Average Tobin's Q.

in valuation even after controlling for other firm characteristics. Here, the null hypothesis expects these coefficients to be zero, while the alternate hypothesis is that they are significant and greater than zero.
