Does Lesbian and Gay Friendliness Pay Off? A New Look at LGBT Policies and Firm Performance
Veda Fatmy1, John Kihn2, Jukka Sihvonen3, Sami Vähämaa4 University of Vaasa, Department of Accounting and Finance
January 16, 2018 Abstract This paper examines the relationship between corporate LGBT friendly policies and performance. Using data on U.S. firms from 2003 to 2016, we find that LGBT friendliness is positively associated with firm performance. Specifically, we find strong evidence that firms with more LGBT-friendly policies are more profitable and have higher stock market valuations. Our findings also indicate that LGBT-friendliness is positively associated with firm and board size. Furthermore, we document that that the positive effect of progressive LGBT policies on firm performance is larger for firms headquartered in less religious or decisively Democratic states, while being weaker or non-existent for firms headquartered in more religious or decisively Republican states. Overall, our empirical findings indicate that diversity considerations and progressive employee policies may create value for the firm.
JEL classification: D22, G30, G31, G39, J15, J70, J83, M12, M50 Keywords: LGBT policies, firm performance, sexual minorities, employee policies
We would like to thank Audur Arna Arnardottir, Kirsten Cook, Markku Kaustia, Pontus Troberg, and conference and seminar participants at the 2017 Diversity Section Meeting of the American Accounting Association, the 9th Annual Workshop of the Nordic Corporate Governance Network, the Hanken School of Economics Workshop on Corporate Governance, the Finnish Graduate School of Finance Summer Workshop, and the University of Vaasa for valuable comments and suggestions. 1 University of Vaasa, School of Accounting and Finance, P.O. Box 700, FI-65101 Vaasa, Finland; Tel.: +358 XXXXX; E-mail address: vfatmy @uva.fi 2 University of Vaasa, School of Accounting and Finance, P.O. Box 700, FI-65101 Vaasa, Finland; Tel.: +358 272 5423; E-mail address:
[email protected] 3 University of Vaasa, School of Accounting and Finance, P.O. Box 700, FI-65101 Vaasa, Finland; Tel.: +358 29 449 8506; E-mail address:
[email protected] 4 University of Vaasa, School of Accounting and Finance, P.O. Box 700, FI-65101 Vaasa, Finland; Tel.: +358 29 449 8455; E-mail address:
[email protected]
2
1. Introduction Corporate social advocacy can be a particularly tricky business venture. When done right, it can increase firm reputation and shareholder returns. However, it can just as easily be misconstrued as an unnecessarily divisive gimmick from an institution that is supposed to maintain a degree of incorporeal neutrality. Of particular interest in recent years is the escalating social and legal contention over LGBT equality in the United States. With LGBT equality currently being such a polarizing subject, firm policy can garner strong reactions from stakeholders. In this paper, we try to determine the independent effect of LGBT friendly policies on firm value and profitability, and the differences in the effect that arise from state demographics.
Ordinarily, when the public deems a company’s actions to be ‘unethical’ or ‘unjust’, it can be damaging to its reputation and lead to reductions in sales and value. This is true concerning news of mistreatment of employees, environmental damage, and even executive scandals (Waddock, 2003; Collins, Ericksen and Allen, 2005; Yadav, Han and Rho, 2015; Matsumura, Prakash and Vera-Munoz, 2014; Cline, Walking and Yore, 2016). However, a company’s LGBT policy – its decision to discriminate or be inclusive – is met with inconsistent results. Disney, Starbucks, Uber and Adidas have all successfully embraced LGBT culture and incorporated non-discrimination policies with positive outcomes. Moreover, Barilla’s initial statements defending the exclusion of same-sex couples from their advertisements were met with outrage (CNN, 2013). However, when Target declared its support for the transgender community by inviting guests to use ‘the restroom … corresponding with their gender identity’ in 2016, it evoked indignation in its customer base. One can argue that because of the significant
3
differences in stakeholder reaction, demographics play an important role in predicting the effects of corporate LGBT friendliness.
Of course, if it were simply a matter of aligning corporate values with those of target consumers and investors, our research would offer contributions of little value or practical implications. However, as prior literature confirms, corporate LGBT policies affect not just consumer and investor relations but also internal operations. In their 2013 publication, ‘The Business Impact of LGBT-supportive Workplace Policies’, Badgett, Durso, Kastanis and Mallory assimilate the results of various studies highlighting the effects of LBGT discrimination on human resources, employee productivity and recruiting and retaining talent. The report substantiates the hypothesis that firms who are more LGBT inclusive tend to have higher employee productivity and a larger number of LGBT persons in their executive positions. These companies are in a better position to maximize the utilization of the talent of sexual minorities. Moreover, it can be argued (and validated)5 that companies wish to maintain a corporate culture of inclusion and non-discrimination in order to attract qualified, talented employees from all around the country.
There is good reason to presume that inclusivity and equal opportunity for sexual minorities are necessary conditions for the firm to fully benefit from their productivity and commitment. There is theoretical basis for this claim in Appelbaum et al.’s (2000) Ability-
5
In a letter addressed to the Governor and House Speaker of Mississippi, CEOs from Levi Strauss & Co, GE, Hyatt
Hotels Corp, Choice Hotels Intl., Dow, Whole Foods, Hewlett Packard and Pepsico express regret at the HB 1523, stating that it makes it ‘challenging for businesses….to recruit and retain the nation’s best and brightest workers and attract talented students’ (Human Rights Campaign, 2016).
4
Motivation-Opportunity theory, which suggests that an employee’s discretionary effort may be achieved through a combination of three factors: their own knowledge, skills and abilities, their motivation to perform (determined intrinsically and extrinsically), and their access to opportunities to participate (Marin-Garcia and Tomas, 2016). The viability of HPWSs (High Performance Work Systems) such as the AMO framework have been backed empirically numerous times in recent literature, including but not limited to studies by Becker & Huselid, 1998; CIPD, 2006; Datta et al., 2005; Flood, Guthrie, Liu, & MacCurtain, 2005; Guthrie, 2001; Huselid, 1995 and MacDuffie, 1995. They have been linked repeatedly to better performance measures, productivity and innovation. Barring evidence to the contrary, HPWSs should work just as effectively for a diverse workforce as a heterogeneous one. This extrapolation is backed by Moss-Kanter 1983, Cox and Blake 1991 and Schwartz 1989, who find a significant link between managing diversity equitably and enhanced innovation, workforce talent and turnover, respectively. Armstrong et al. (2010) are among the first to study the effects of Diversity and Equality Management Systems (DEMS) and firm performance. Moreover, in order to attract and retain top talent in a competitive economy, employers must also compete in terms of the incentives they can provide to new recruits. These include employment benefits such as healthcare and spousal benefits, and may even extend to factors such as a flexible and nurturing work environment. Generation Y, or Millennials in particular are more likely to seek ‘personally rewarding’ employment by considering the level of social contact and recognition they can obtain in the workplace (Earle, 2003).
These reasons motivate our interest in the effects of a firm’s LGBT friendly policies on its performance. Other researchers have attempted to statistically corroborate the mechanisms
5
through which these effects take place (for instance, Shan et al. 2017), and while our data presents us a fresh opportunity to dissect and trace the workings of the CEI, we find it feasible to limit the scope of this paper to the aforementioned variables of interest. The study needs to not only be a thorough one – accounting for multiple methods in which a firm’s stance on LGBT anti-discrimination manifests itself, but also a recent one – covering the later part of the 21st century wherein sexual minorities and the transgender community welcomed higher levels of social and legal acceptance. We hypothesize a significant positive relationship between corporate LGBT friendliness, and firm value and profitability. We test this relationship using the Corporate Equality Index (CEI) as a proxy for LGBT friendliness. The CEI is published annually by the Human Rights Campaign and assigns a score ranging from 0 to 100 for a company’s record of LGBT non-discrimination in the preceding year. In addition, we hypothesize that local demographics – specifically political and religious – tend to significantly influence the relationship between LGBT friendliness and firm performance.
Our paper is not the first to study the relationship between corporate LGBT policy and firm performance. In fact, Pichler, Blazovich, Cook, Huston and Strawser published their paper ‘Do LGBT-supportive corporate policies enhance firm performance?’ earlier this year, in which they found the standalone effect of good LGBT policies to be insignificant toward firm value and profitability in the absence of R&D, but find the relationship between good LGBT policies and factor and employee productivities to be strongly significant. They use ‘Gay and Lesbian Policy’ ratings from the MSCI ESG database, which is a dummy variable assuming 1 for a sufficient level of LGBT policy implementation. The MSCI rating is at an obvious disadvantage
6
when compared to the Corporate Equality Index, since it not only simplifies various aspects of LGBT policy and corporate culture, but also has a lower variance.
However, our research is the first in our knowledge to successfully document a significant effect of state demographics on the relationship between firm LGBT friendliness and performance. Pichler et al. (2017) choose instead to rely on state politics in a bid to demonstrate the effects of voluntariness in adopting non-discriminating policies. While voluntariness is an interesting aspect in and of itself, it is not necessarily representative of the company’s motivations in their policy choice. A company going above and beyond its legal requirements does not represent objective voluntariness without controlling for stakeholder expectations. For instance, while a fashion company may come under scrutiny for lagging in its LGBT acceptance, similar rigidity in an oil and gas company would not provoke many. Like industry, differences in values and cultural norms across geographical boundaries lend significant context to corporate LGBT policy decisions. The inclusion of political and religious state demographics in our study is therefore a fresh perspective on the financial effects of corporate LGBT policy.
Prior research on the subject includes Wang and Schwarz’s paper ‘Stock Price reactions to GLBT non-discrimination policies’. Published in 2010, the study pits the CEI score against stock returns while controlling for company characteristics relevant to an LGBT-friendly work environment. Their findings suggest that companies with higher CEI scores tend to outperform lower scoring firms. These results are then corroborated by Shan, Fu and Zheng (2017) in similar research. Moreover, Zheng et al. (2017) find evidence of higher productivity through the labor market for firms that have LGBT friendly policies. In addition, Johnson and Malina (2008) document an initial increase in firm value for high scoring firms on the CEI announcement date.
7
Our use of the CEI is therefore by no means without precedent, however, it covers the most extensive period, and benefits from the substantial increase in the number of firms featured on the CEI by the Human Rights Campaign in recent years.
The purpose of this paper is twofold. First, we establish that a discrete multi-component measure of LGBT friendliness has a positive independent effect on firm performance. For this purpose, we conduct a multivariate cross-sectional analysis with period fixed effects for approximately 600 US firms from 2003 till 2016, in which the CEI is regressed against Tobin’s Q and ROA while controlling for firm characteristics, the overall firm Corporate Social Responsibility score, and board characteristics. Our empirical findings strongly support our claim that LGBT friendliness is positively and significantly related to both firm value and profitability. Next, we show that this effect is dependent on the demographics of firm headquarter state, as a proxy of stakeholder concentration and firm values. For this, we test our model using data for states with low (high) religiosity and Democrat (Republican) leanings separately. We find the aforementioned relationship to be highly significant (not significant) within the respective subsamples. We also conduct two-stage least squares regression to eliminate concerns about endogeneity in our main regression model.
The paper is organized into 4 sections. Section 2 explains the data and variables used. Section 3 details the methodology, the results of the empirical analysis and robustness checks. Section 4 presents the main findings of the study and their implications for practical implementation and further research.
8
2. Data and variables The sample used in our empirical analysis consists of 585 publicly traded firms in the U.S. This sample is limited to firms that have a CEI score for at least 5 of the 13 years, and available data for all control variables in the corresponding years. In addition, firms that trade as penny stocks were excluded from the sample. The final sample includes firms from all major SIC industry classifications that are publicly traded on NASDAQ, NYSE, AMEX or other regional U.S. stock exchanges. Moreover, these firms are headquartered across 42 different states. These states are consequently used as a metric for the location where the firm conducts a majority of its business (hires the most employees/sells the most products and/or services).
2.1 LGBT friendliness A firm's CEI score, published by the Human Rights Campaign ("HRC") since 2002, is used as the main independent variable in our study. This choice is motivated by the assessment of the CEI as the most comprehensive measure of a company’s LGBT friendly employment policies and public advocacy. The CEI is released by the HRC in the fall of each year, and represents an assessment of the company’s LGBT friendliness based on questionnaires and news articles during the year leading up to the date of publication. Since 2007, the HRC has been publishing forward-looking CEI reports (i.e., publishing them under the title of the upcoming year). However, we have edited the data in our sample so that each CEI score corresponds to its year of publication.
9
One of the most important objectives of the CEI is to provide reliable and up-to-date data for ‘LGBT and allied consumer and employees’ that are searching for employment or ‘buying for workplace equality’ (HRC 2017). That is not to say that our use of the CEI data in research is a precedent for this resource. Only recently, Wang and Schwartz (2012) published their research on the relationship between LGBT non-discrimination and stock returns, which uses a firm’s CEI score as their main independent variable. The CEI report ranks all participating firms by the number of LGBT friendly policies in place and the measures they take to actively avoid LGBT-discrimination in public. The firms are awarded scores consisting of discrete values from 0 to 100, where 0 represents the least LGBT friendly workplace, and 100 represents the highest possible score. Unlike the Gay and Lesbian Policy dummy, the CEI is able to provide a sufficiently varied measure across all firms, moreover, one that is exclusively dedicated to policies and actions regarding sexual minorities. Table 1 presents a detailed summary of the criteria included in the CEI and the score awarded for each criteria.
(insert Table 1 about here)
2.2 Firm performance
Our two dependent variables that measure firm performance through value and profitability are Tobin’s Q and Return on Assets ("ROA"). Tobin’s Q is widely used to estimate firm value in studies pertaining to firm and board characteristics and investment. First introduced by Nicholar Kaldor in 1966 and then later by James Tobin and William Brainard in
10
1968, Tobin’s Q is essentially the ratio of the market value of the firm’s assets and the 'replacement costs’ of those assets (Tobin 1977). Given that the replacement costs of firm assets is difficult, if not impossible, to estimate, many studies have estimated Tobin’s Q by using the book value of assets in place of total asset value (Wernerfelt & Montgomery 1988, Servaes 1991, Stulz & Lang 1994, Weston & Allayannis 2001, and Pedersen et al. 2006, etc.). In this paper, Tobin’s Q is calculated using the formula given below:
Thus, the sum of market capitalization and total book value of liabilities is the numerator, and total book value of assets is the denominator. Similarly, ROA is used as a measure of firm profitability, and has ample precedents for its use in prior literature (e.g., Carroll et al. 1985, Kumar & Sharma 2011, Zhou et al. 2008, Crampton & Patten 2008, Dooley & Lerner 1994). The following formula is used to calculate ROA:
, where the numerator represents the net income available to common shareholders and the denominator represents the sum of total current assets, long term receivables, investments in unconsolidated subsidiaries, net property plant and equipment, and other assets (i.e., as measured by book value). The values for these variables are collected from Datastream for each publicly listed firm included in the CEI (i) for the time period 2003 through 2016 (t).
11
2.3 Control variables
Our choice of control variables is motivated by prior literature. Firm specific controls such as size, growth and risk are important when testing for firm performance because key performance variables such as net income and the market value of assets are dependent on firm size, the riskiness of its stock, its debt structure, etc. In addition, Wang and Schwartz (2010) and Pichler (2017) control for overall employment quality by including measures of employee benefits or diversity. This is important to ensure that the effect on value and profitability is not driven by overall better worker policies. In addition, Corporate Social Responsibility has shown to be on aggregate positively related to firm value and growth (Gregory, Tharyan & Whittaker, 2013). To this effect, we control for CSR by including the ESG score in our study. The ESG (Environmental, Social and Governance) score is a compound CSR rating collected annually by Thompson Reuters. Moreover, Yermack (1996) and Cheng et al. (2008) find a positive correlation between small board size and firm performance. Knyazeva and Knyazeva (2013) also find board independence to be positively correlated with profitability and operating performance. Therefore, the aforementioned board characteristics are included as controls in our study. Specifically, we control for size (measured by the natural log of total assets), leverage (measured by the equity-to-assets ratio), growth (measured by change in sales), risk (measured by annualized firm beta), board size and board independence. In addition, in regressing the CEI score against Tobin’s Q, we also control for profitability as represented by ROA. We use two dummy variables – one for religion, and the other for politics – to split our sample and perform simultaneous and comparative empirical tests. The data for percentage of
12
highly religious people per state is taken from a Gallup survey conducted in 2013, and is assumed to be approximately constant over the sample period. The data for state politics is derived from election results starting from 2000 up to 2012. The state political majority is determined by the result of the election and is carried backward over the four years leading up to the election. This is a conservative approach to assigning state politics, and is meant to alleviate concerns that political changes might drive firm policies and stakeholder sentiment instead of vice versa. There are three possible states for the ‘politics’ variable: 1 for states that voted Democrats, 0 for states that voted Republicans and 0.5 for states that were within a 5% margin in the election results.
3. Empirical analysis
3.1 Descriptive statistics and correlations
In this section, we first present descriptive statistics for the main dependent and independent variables and control variables in Table 2. All variables have been winsorized at the 1% level to remove extreme values. In addition, all variables are presented in their natural numeric form without logarithms or other transformations.
(insert Table 2 about here)
In contrast to its slight negative skew, the CEI has a high sample mean. This may represent a voluntary response bias in the dependent variable, because firms that are likely to respond to the
13
CEI questionnaires are those that have either already accomplished a comfortable level of LGBT friendliness or are motivated to do so. In extreme cases, firms that wish to express opposition to LGBT non-discrimination may also respond. However, this bias should not affect our sample of dependent and control variables. This is further corroborated by the sample mean and the spread of our main dependent variables, Tobin’s Q, and ROA. Both variables display sufficient heterogeneity and there is sufficient variance between the minimum and maximum values. Our control variables also display a similar heterogeneity. However, the values for board size and independence have a similar distribution to the CEI score, which is indicative of a significant positive relationship between strong board characteristics and LGBT friendliness. Table 3 presents the pairwise correlation coefficients for the variables used in our study. Tobin’s Q is significantly and positively correlated with the CEI score and presents a strong case for our first hypothesis. Profitability itself, however, has no significant correlation to LGBT friendliness. Meanwhile, LGBT friendliness is also significantly and positively correlated with size and board characteristics, while being significantly and negatively correlated with both the growth rate and the risk.
(insert Table 3 about here)
The correlations of the CEI with firm characteristics have interesting implications. At a cursory glance, it would appear that larger firms with low growth and beta are more likely to have better LGBT-policies. This is to be expected: large, well-established firms are known to provide significantly better employee benefits. This relationship, however, is also known to be
14
significantly different across industries (Burke & Morton 1990). The pairwise correlation coefficients for most of our variables are quite low (below 0.31, with the exception of the correlation of ROA and Tobin’s Q) which mitigates any concerns about multicollinearity.
3.2 Univariate analysis
In the first test, we split our sample into two groups: (1) firms that are more LGBT friendly – i.e. have a CEI score of 100, and (2) less LGBT friendly firms, i.e. firms that have a CEI score less than 45. We then conduct a univariate analysis in which we compare the means and medians of our dependent and control variables across the two subsamples. This is accomplished by carrying out two-tailed t-tests and Wilcoxon rank-sum tests, with the null hypothesis being that there are no significant differences between the means and medians of the two subsamples for each variable.
(insert Table 4 about here)
The mean and median of each dependent variable, along with the results of the univariate analysis are presented in Table 4. By comparing the two subsamples, it is apparent that there are significant differences for the means and medians of profitability and almost all control variables. The means and medians for Tobin’s Q and ROA are significantly higher in the subsample of more LGBT friendly firms and these differences are significant at the 1% level. It is also worth noting that the difference in the means tests suggests that more LGBT-friendly
15
firms tend to be larger, have lower debt, growth and beta and bigger and more independent boards than less LGBT friendly firms. They are also more likely to have high Corporate Social Responsibility scores (i.e. perform better on all three measures; environmental and social responsibility and corporate governance). These results provide further evidence for a statistically significant positive relationship between LGBT friendliness and firm performance. They do not, however, shed light on the nature of the relationship (the size, and the direction). For this purpose, we perform two subsequent empirical tests. Firstly, we conduct a multivariate least square regression analysis using industry and time fixed effects. Later, in order to study the effects of political and religious differences across states, we repeat this regression for separate subsamples. Secondly, to determine the direction of the relationship between LGBT friendliness and firm performance, we conduct a two-stage least squares regression. The following sections present the method and results of these regressions in greater detail. 3.3 Regression results
In order to test the relationship between LGBT friendliness and firm performance, we perform a multivariate analysis, controlling for firm characteristics that directly impact value and profitability. The main regression model is presented below as Equation (1)
Performance j,t = +
1LGBTj,t + 10(Industry
2-9 (Firm-specific
fixed-effects)t +
controls) j,t
11(Year
fixed-effects)t +
(1) j,t
16
where Performancej,t represents our main dependent variables, Tobin’s Q and ROA, for firm j at time t. The main explanatory variable, LGBTj,t, is the CEI score for firm j at time t, where t corresponds to the year of publication. Our firm-specific controls include firm size, leverage (equity-to-assets ratio), growth (change in sales), risk ( eta), ESG (Thompson Reuters’ company score for Corporate Social Responsibility), board size, and board independence. For regressions performed using Tobin’s Q as the dependent variable, profitability (ROA) is also included as an explanatory variable. Year and industry fixed effects are added to control for trends in the aggregate time series and control for variation across industries. First, we regress our dependent variables, Tobin’s Q and ROA, against firm specific controls excluding board characteristics, using OLS method and white period standard errors and coefficients. Year and industry fixed effects are not included in these tests. The results of these regressions are presented in Table 5 in the columns ‘Model 1’ and ‘Model 4’ respectively. The results show that the CEI score is statistically significant at the 1% level and positively related to both dependent variables. More specifically, a 1% increase in the CEI score would be matched by a 0.2% increase in Tobin’s Q and a 1.5% increase in ROA while controlling for relevant firm characteristics. For the next regression, we include industry dummy variables and year fixed effects. Fixed effects can explain a lot of variation in the data; however, in this case, the coefficient of the CEI score remains statistically significant at the 1% level for Tobin’s Q and at the 5% level for ROA. These results can be seen in the columns ‘Model 2’ and ‘Model 5’ of Table 5.
17
Lastly, the main regression is performed using all control variables (including board size and board independence) and industry and year fixed effects. The results are displayed in the columns ‘Model 3’ and ‘Model 6’ of Table 5. Once again, the coefficient of the CEI score remains statistically significant at the 1% level for Tobin’s Q and at the 5% level for ROA. The magnitude of the effect on ROA and Tobin’s Q drops per percentage increase in CEI with the addition of fixed effects.
(insert Table 5 about here)
These results provide sufficient evidence to support our first hypothesis, namely that LGBT friendliness is significantly and positively related to firm performance. Firstly, our findings so far are also surprisingly different from those of Pichler et al. (2017). For one, LGBT friendliness has been shown to be statistically significant in predicting Tobin’s Q and ROA in all three scenarios (with and without year and industry fixed effects and/or board characteristics controls). Secondly, unlike the MSCI dummy variable used in Pichler et al.’s (2017) study, the CEI score is significantly and positively related to ROA. Lastly, it is encouraging to find that the relationship remains significant and robust while controlling for overall firm Corporate Social Responsibility. This suggests that LGBT policies are a strong enough indicator on their own of firm performance. It can also serve as evidence that the impact of sexual minorities on the economy is surprisingly intricate and unsurprisingly may exceed the surveyed workforce figure of 9 million (Mallory, 2011).
18
For our second hypothesis, we once again use the model presented in Equation (1). However, this time, we split our sample into ‘Conservative’ and ‘Liberal’ states. The ‘Conservative’ sample comprises of firms that are headquartered in Republican and more religious states, and the ‘Liberal’ sample comprises of firms headquartered in Democratic and less religious states. Here, more religious states are those in which more than 32% of the population (the sample median) identifies as highly religious according to Gallup data on state demographics. The remaining are classified as less religious states. Since both conditions for politics and religiosity must be met for states to be classified as either liberal or conservative, the sum of observations in the resulting samples is smaller than the number of observations in the entire sample. Our strict selection for the ‘Liberal’ and ‘Conservative’ subsamples means that nearly half our observations are omitted from this regression.
(Insert Table 6 here)
The results of the effects of politics and religion on the relationship between LGBT friendliness and firm performance can be seen in Table 6. Columns 1 & 2 display the results of the regression of Tobin’s Q for firms in Liberal and Conservative states respectively. The coefficient of the CEI score remains positive and statistically significant at the 1% level in the Liberal subsample, but this significance is lost in the Conservative subsample. Columns 3 & 4 display the results of the regression of ROA for the Liberal and Conservative subsamples respectively. Once again, only in the Liberal subsample does the CEI coefficient retain a statistical significance.
19
These results confirm our expectations about the role demographics play in the CEIperformance relationship. Specifically, the political and religious leanings of state populations impact some of the effect of firm LGBT friendliness on performance. It may be argued that headquarter state is somewhat imperfect representation of target demographics for most large companies that span not only state but national borders. However, it is important to remember that most large companies have already garnered a thorough understanding of their target consumer base and their corporate culture. In such cases, barring industry constraints, they would likely locate their headquarters in a state that represents this demographic and culture. Recall footnote 1, in which we learn that several corporations forewent the opportunity to expand operations in conservative areas. Moreover, industry trends highlight that corporate LGBT culture is usually aligned with the culture of the location constraint (Sadree, Lees, 2001). Large IT firms, usually headquartered in Silicon Valley, tend to be more LGBT-friendly, matching the liberality of California, while large oil and gas firms headquartered in the midsouth, are usually less LGBT-friendly, similar to the conservativeness of Texas. 3.4 Endogeneity
The results of our empirical tests indicate that higher LGBT friendliness has a significant effect on both firm value and profitability, and that this effect is stronger in more liberal states. However, before we can confirm whether these results corroborate our first and second hypothesis, it is important to test for endogeneity or reverse causality. While we control for several firm specific variables that have been shown to affect firm performance, it is possible that some missing variables (like other social performance indicators as suggested by Pichler et al. (2017)) may have a significant explanatory effect on our dependent variables. In addition,
20
one could argue that the nature of the relationship between Tobin’s Q and firm performance is converse; that is, firms with higher performance are more inclined to provide LGBT friendly policies. Therefore, we test the results for the previous regressions for endogeneity. Specifically, we perform a two-stage least squares regression using the dummy variables for political leanings as our instrumental variables. These instrumental variables are meant to be strongly correlated with the LGBT friendliness of corresponding firms, while being uncorrelated with their performance. In the first stage, the instrumental variables and the primary control variables are regressed against the CEI score. This instrumented CEI score is then used in the second stage against our initial dependent variables, Tobin’s Q and ROA respectively. The results of the two-stage least square regression are presented in Table 7.
(Insert Table 7 here)
The results of the first stage regression in Table 7 are highly revealing. While the coefficient for the instrumental variable ‘Democrat’ is positive and strongly significant, the coefficient for the other instrumental variable ‘Republican’ is not significant. This is in keeping with the results of our regression of the conservative subsample, and suggests that not only is the relationship between LGBT friendliness and CEI score stronger in Democrat states, but that Democrat states tend to have firms with higher CEI scores. In the second stage, we can see that both the regressions for Tobin’s Q and ROA survive the tests for endogeneity. The coefficient of the instrumented CEI score is positive and significant when tested against both variables.
21
Therefore the two-stage least squares regressions indicate that not only is firm LGBT friendliness dependent on state politics, but that state politics play a vital role in the relationship between LGBT friendliness and firm performance. Moreover, we further test our instrumental variables for strength and over-identification. The partial F-stat test has a value of 43.34, well above the recommended restriction of 10. Since we use more than one instrumental variable, we conduct the J test (Sargan-Hansen test 1975), which relies on the number of instruments and the R-squared of the original instrumental regression. The computed statistic has a value of 0.33, which approximates the recommended range for this test. 3.5 Robustness checks
We perform several additional tests in order to strengthen the reliability of our findings. While our split sample analysis for the role of state religiosity and politics (Table 6) does provide greater insight, concerns might arise due to the reduced size of the subsamples. Therefore, we first create a dummy variable for politics and religion separately. The politics dummy variable assumes the value of 1 when the state is Democrat and 0 otherwise. The religion dummy variable assumes the value of 1 when the number of highly religious people per state is greater than 32%, and 0 otherwise. We then interact the dummy variables for politics and religion with the CEI score, and perform multivariate regressions using the model presented in equation (1) and with the addition of our interacted variables. In both cases, the results of the regression corroborate our previous findings (i.e., the interaction of state politics and LGBT friendliness is positively and significantly related to both Tobin’s Q and ROA), while the
22
interaction of state religiosity and LGBT friendliness is negatively and significantly related to the two dependent variables. Next, we consider the variations across the sample for CEI scores. While we have shown both political and religious subsamples to be relevant to the hypothesized relationship, we have not seen how the relationship varies across divisions in the sample of CEI scores in a multivariate setting. We therefore create two dummy variables, namely ‘CEI low’ and ‘CEI high’, and regress these on our main dependent variables using the model presented in equation (1). Once again, our results corroborate our previous findings. ‘CEI high’, the dummy variable that is 1 when CEI = 100, is significantly and positively related to both Tobin’s Q and ROA. On the other hand, ‘CEI low’, the dummy variable that is 1 when CEI < 45, is significantly and negatively related to Tobin’s Q, while its relation with ROA is not significant. In order to test which percentile of CEI score demonstrates the most significant positive relation to firm performance, we repeat the multivariate regression from the model in equation (1) using four subsamples. In the first scenario, we limit the CEI score by the constraint 0
CEI
25, in the second scenario the constraints place the CEI score between 25 and 50, in the third scenario between 50 and 75, and in the fourth between 75 and 100. Interestingly, and contrary to expectations, the largest effect on firm performance does not proceed from the subsample with the highest CEI scores. Instead, the largest effect comes from the subsample where CEI is constrained between 50 and 75. This suggests that firms that are on satisfactory levels of LGBT friendliness and still have room for progress stand to gain the most in terms of firm value and profitability from an increase in their CEI score.
23
In addition, we test for the effects of firm CEI scores that are above or below the industry average. Since the average CEI score across industries varies significantly, this is to ensure that firms’ CEI scores have been standardized over the entire sample. This is achieved by first regressing the CEI score against industry classifications. The resulting coefficients are then used to calculate residuals for each firm-year observation. These residuals are used in the original model with the exception of industry fixed effects. The coefficient for LGBTfriendliness remains positive and significant at 1% for Tobin’s Q and for ROA. These findings suggest that higher CEI scores are related to better firm performance, even when they are subjective to the industry average. Lastly, in order to strengthen the argument for the independence of the effect LGBT friendliness has on firm performance, we control for a further CSR related exogenous variable. We divide the sample of firms into separate subsamples based on their Social score (retrieved from the Thompson Reuters ESG database). The Social score is a continuous variable ranging from 0 to 100. We run regressions from equation (1) on our subsample and find that firms with a Social score of above 60 exhibit the strongest positive relationship between LGBT friendliness and firm performance.
4. Conclusions
The purpose of this paper is to determine the effect, if any, that corporate LGBT friendliness has on firm performance. Specifically, we study the impact on two key variables, Tobin’s Q (value), and ROA (profitability). Our research is motivated by the recent rise in corporate and political advocacy for the LGBT community in the U.S., and the study by Pichler
24
et al. (2017) that is the first study to investigate a link between LGBT friendly policy and firm performance. Deviating from Pichler et al.'s (2017) methods, we use the CEI as a proxy for LGBT friendliness, and present our reasoning behind the choice of CEI score over that of the MSCI index. Our first hypothesis states that firms that are more LGBT friendly (have a higher CEI score) have higher firm value and profitability. Results of the empirical analysis strongly support this hypothesis. When regressing CEI score against both Tobin’s Q and ROA using firm and industry fixed effects, and controlling for firm specific and board characteristics, we find that the coefficient for the CEI score is positive and significant. To be precise, a 1% increase in the CEI score predicts an approximately 0.15% increase in value and approximately a 1.2% increase profitability. These results are more significant than those achieved by Pichler et al. (2017) using the MSCI ESG index, and they strengthen the evidence for financial implications of a firm’s LGBT policy and the argument for implementing LGBT friendly policies in the workplace. Additionally, we also test for the effect of the political and religious climate of the headquarter state on the relationship between LGBT friendliness and firm performance. For this purpose, we divide the data into two samples, namely ‘Liberal’ and ‘Conservative’. The ‘Liberal’ sample comprises of firms that are headquartered in Democratic and less religious states, and the ‘Conservative’ sample comprises of firms that are headquartered in Republican and more religious states. We first perform a difference in means test for our performance variables across the two samples to ensure that firm value and profitability do not display significant correlation with these demographic conditions. The test confirms that not only is the difference in mean across the two samples not significant, but the means themselves are nearly identical. We then
25
regress firm performance and CEI using the model specified in equation (1). To this effect, our second hypothesis states that for firms headquartered in Liberal states, the relationship between LGBT friendliness and firm value is stronger than for firms headquartered in Conservative states. The results of the second set of regressions support our hypothesis. While the significance of the effect of LGBT friendliness on performance is wiped away in conservative states, it is retained in Liberal states. Specifically, the effect of CEI on Tobin’s Q is significant at the 1% level and the effect of CEI on ROA is significant at the 5% level in liberal states. The lower significance of the CEI’s effect on performance in conservative states might be because the stakeholders of firms in these states (specifically their clients and consumers) are less likely to react adversely to news of LGBT discrimination. As an example, consider the controversy surrounding the anti-LGBT remarks of the president and COO of Chick-fil-A, a restaurant headquartered in Atlanta, Georgia. LGBT rights activists across the country condemned the statement of President Dan Cathy who insinuated that couples in same sex marriages were ‘prideful and arrogant’ and had the ‘audacity’ to challenge God’s commandments on the institution of marriage. Thomas Menino, the mayor of Boston, went so far as to state that he would not allow the company to open franchises in the city unless they were LGBT friendly (Greg 2012). In contrast to the Boston mayor's reaction, conservative states had a very different response. Former Arkansas Governor Mike Huckabee proposed a ‘Chick-fil-A’ appreciation day as a backlash against calls for boycott, and the restaurant sales increased by 29.9% within two months of the controversy (Holpuch 2012). Our research presents evidence suggesting that firms in less religious and Democrat states can suffer from significantly lower value and profitability if they do not provide sufficient
26
LGBT anti-discrimination policies to their employees and maintain controls for supporting an LGBT friendly environment. Moreover, the findings also imply that firms in more religious and Republican states face less motivation not only socially but financially to improve their LGBT policies. It is important to note that the results are robust and significant while controlling for the ESG ratings, as this suggests that the effect of LGBT friendly policies is independent of overall corporate social responsibility. These results also indicate that a demographic representation like the ones we have used may have a more significant or discernible influence than state legislation on the relationship between LGBT friendliness and firm performance. This could be because state legislation regarding LGBT employment policy has little variance across the country, and can be quite polarizing in the reaction it garners from state residents. Further research can investigate the relationship between LGBT friendliness and firm performance to a deeper extent by identifying the individual elements of the CEI that have a strong impact on firm performance. It would also be interesting to study the stock price reactions to news regarding LGBT policies, such as changes in the CEI scores by the Human Rights Campaign. Either way, with the rising conflict over conservatives’ measures, the avenues of research on the impact of this crucial policy are rife with possibilities.
27
References
Allayannis, George, Weston, James P., 2001. The Use of Foreign Currency Derivatives and Firm Market Value. The Review of Financial Studies 14:1, 243-276. Albuquerque, R., Durnev,A., & Koskinen Y., 2012. Corporate Social Responsibility and asset pricing in industry equilibrium. Geneva Summit on Sustainable Finance. Aupperle, Kenneth E., Carroll, Archie B., Hatfield, John D., 1985. An Empirical Examination of the Relationship between Corporate Social Responsibility and Profitability. The Academy of Management Journal 28:2, 446-463. Badgett, M.V. Lee, Durso, Laura E., Kastanis, Angeliki, Mallory, Christi, 2013. The Business Impact of LGBT-supportive Workplace Policies. The Williams Institute. May 2013. Available
online:
http://williamsinstitute.law.ucla.edu/wp-
content/uploads/Business-Impact-LGBT-Policies-Full-May-2013.pdf Bebchuck L, Fired J., 2004. Pay without Performance. Cambridge, Massachusetts and London. Brainard, WC., Tobin, J., 1968. Econometric Models: Their problems and usefulness pitfalls in Financial Model building. American Economic Review 58:2, 99-122. Burke, Thomas P., Morton, John D., How Firm Size and Industry Affect Employee Benefits. Monthly Labor Review 113, 35-57. Cheng, Shijun, 2008. Board Size and variability of Corporate Performance. Journal of Financial Economics 87:1, 157-176. Cline, Brandon N., Walking, Ralph A. and Yore, Adam S., 2016. The Consequences of Managerial Indiscretions: Sex, Lies and Firm Value. Journal of Financial Economics (Forthcoming)
28
Collins, Christopher, Ericksen, Jeff and Allen, Matthews, 2013. Human Resource Management Practices, Worforce Alignment, and Firm Performance. Human Resource Management 52:2, 153-173 Crampton, W., Patten, D. J., 2008. Social Responsiveness, Profitability and Catastrophic Events: Evidence on the Corporate Philanthropic Response to 9/11. Journal of Business Ethics 81:4, 863-873. Deloitte 2013. Resetting Trends: Global Human Capital Trends 2013. Dooley, R.S., Lerner, L.D., 1994. Pollution, profits and stakeholders: The constraining effect of economic performance on CEO concern with stakeholder expectations. Journal of Business Ethics 13:9, 701-711. Edmans, Alex, 2012. The link between Job Satisfaction and Firm Value, with Implications for Corporate Social Responsibility. Academy of Management 26:4, 1-19. Golnaz Sadri, Brian Lees, (2001) "Developing corporate culture as a competitive advantage", Journal
of
Management
Development,
Vol.
20
Issue:
10,
pp.853-859,
https://doi.org/10.1108/02621710110410851 Gompers, Paul, Ishii, Joy, Andrew Metric, 2003. Corporate Governance and Equity Prices. The Quarterly Journal of Economics 118:1, 107-156. Greenberg Quinlan Rosner, 2015. Public Opinions on LGBT Equality and Non-Discrimination. Published
15.09.2015.
Available
online:
http://www.gqrr.com/articles/2015/9/16/newhrc-polls-public-opinions-on-lgbt-equalityand-non-discrimination Gregory, Alan, Tharyan, Rajesh & Julie Whittaker, 2013. Corporate Social Responsibility and Firm Value: Disaggregating the Effects on Cash Flow, Risk and Growth. Journal of Business Ethics 124:4, 633-657.
29
Holpuch, Amanda, (2012). Chick-fil-A appreciation day brings huge crowds to a fast-food chain. London, UK. Guardian. August 1, 2012. Johnston, D. and Malina M. A., 2008. Managing Sexual Orientation Diversity: The impact on firm value. Group and Organization Management 33, 602-625. Kaldor, Nicholas, 1966. Marginal Productivity and the macroeconomic theories of distribution: Comment on Samuelson and Modigliani. The Review of Economic Studies 33:4, 309319. Knyazeva, Anzhela, Knyazeva, Diana, 2013. Ownership change, institutional development and performance. Journal of Banking and Finance 37:7, 2605-2627. Lang, Larry H.P., Stulz, Rene M., 1994. Tobin’s Q, Corporate Diversification and Firm Performance. Journal of Political Economy 102:6, 1248-1280. Matsumura, Ella Mae, Prakash, Rakhna and Vera-Munoz, Sandra C., 2014. Firm Value effects of Carbon Emissions and Carbon Disclosures. The Accounting Review 89:2, 695-724. Mele, Christopher, 2016. “In North Carolina and Mississippi, Backlash Grows over Rights Law”. The New York Times. April 12, 2016. Newport, Frank, 2014. Mississippi is the Most Religious U. S. State. Available online: gallup.com/poll Peters, Mark, Silverman, Rachel Emma, 2016. “Big Business Speaks Up on Social Issues”. The Wall Street Journal. April 17, 2016. Pichler, S., Blazovich J., Cook, K., Huston, J., Strawser, W., 2017. Do LGBT-supportive Corporate Policies Enhance Firm Performance? Human Resource Management. Forthcoming. Servaes, Henri, 1991. Tobin’s Q and the Gains from Takeovers. The Journal of Finance 46:1, 409419.
30
Shan, Liwei, Fu, Shine and Zheng, Lu, 2017. Corporate Sexual Equality and Firm Performance. Strategic Management Journal 38:9, 1812-1826. Thomsen, Steen, Pederson, Torben, and Hans Kurt Kvist, 2006. Blockholder ownsership: Effects on firm value in market and control-based governance systems. Journal of Corporate Finance 12:2, 246-269. The Human Rights Campaign 2003-2015. The Corporate Equality Index: Rating Workplaces on LGBT
Equality.
Available
online:
http://www.hrc.org/campaigns/corporate-
equalityindex Waddock, Sandra, 2003. Stakeholder performance implications of Corporate Responsibility. Journal of Business Performance. 5:2-3, 114-124. Wang, P., Schwartz, J., 2010. Stock Price Reactions to GLBT nondiscrimination policies. Human Resource Management 49:2, 195-216. Wenerfelt, B., Montgomery, CA., 1988. Tobin’s Q and the importance of focus in firm performance. The American Economic Review 78:1, 246-250. Yadav, Prayag Lal, Han, Seung Hun and Rho, Jae Jeung, 2015. Impact of Environmental Performance on Firm Value for Sustainable Investment: Evidence from Large US Firms. Business Strategy and the Environment 25:6, 402-420. Yermack, David. Higher Market Valuation of companies with a small board of directors. Journal of Financial Economics 40:2, 185-211.
31
Appendix Table 1. Criteria for CEI Scores (Human Rights Campaign 2017).
Criteria 1
Equal Employment Opportunity policies a) Sexual Orientation for all operations b) Gender Identity for all operations
15 points 15 points
c) Contractor/Vendor standards include sexual orientation and gender identity 5 points Criteria 2
Criteria 3
Employment benefits a) Equivalent Spousal and Partner benefits b) Other "soft" benefits c) Transgender inclusive health insurance coverage Organizational LGBT competency a) Competency training, resources and accountability measures b) Employee group or Diversity council
Criteria 4
Criteria 5
10 points 10 points 10 points
Public commitment LGBT-specific efforts (recruitment, philanthropy etc.)
10 points 10 points
15 points
Deductions for large-scale anti-LGBT blemish 25 point reduction for recent cases of LGBT discrimination 100 points
32
Table 1. Descriptive statistics. Variable
Mean
Median
P1
P25
P75
P99
Std. dev.
No. of obs.
69.02
85.00
-25.00
29.00
79.00
100.00
32.90
5295
Tobin's Q
1.947
1.486
0.094
1.1310
2.1804
7.6352
1.943
7709
Profitability
3.736
4.778
-22.8132
1.6452
9.0969
90.173
0.778
7763
77645894
19342500
168400
48178000
2570000000
236000000
7972
Leverage
0.345
0.358
-29.0641
0.2971
1.3357
14.5512
0.495
8008
Growth
1.107
0.054
-3.4620
-0.0225
0.1032
11.9126
7.142
7438
Risk
1.042
1.023
-1.5000
0.8116
1.2691
3.3593
0.435
7448
ESG
46.615
43.120
0.0000
18.6983
68.8342
95.6600
16.904
6113
Board size
10.351
11.000
7.0000
10.0000
13.0000
33.0000
3.748
6635
Board independence
73.961
81.820
0.0000
56.2500
87.9100
100.0000
24.918
6553
LGBT friendliness: CEI score Firm performance:
Control variables: Size
7457000
This table reports the descriptive statistics for the entire sample of each variable. The CEI Score is reported as published by HRC in their annual CEI report, and ranges from 0 to 100. The dependent variables, Tobin’s Q and ROA are calculated as follows: , , , . The control variables include Size and , = , = ,
,
(the natural log of Total Assets), Leverage (the equity-to-assets ratio), Growth (change in annualized Sales), Risk (annualized firm beta), ESG is the firm score for Corporate Social Responsibility, Board size (number of people on the board of directors) and Board independence (percentage of independent board members). The time period for these variable observations is 2003 – 2016.
33
Table 2. Correlations. (1)
(2)
(3)
(4)
(5)
(1) CEI score
1.0000
(2) Tobin's Q
0.0801
*
1.0000
(3) Profitability
0.0756
*
0.5765
*
1.0000
(4) Size
*
-0.3564
*
-0.2099
*
1.0000
*
0.1435
*
0.2959
*
-0.2152
0.1343
*
0.1703
*
0.0123
(7) Risk
0.2631 0.0436 0.0336 0.0772
-0.1431
*
-0.2184
*
0.1015
*
(8) ESG
0.1966
*
0.0387
*
0.0538
*
0.0599
*
0.0300
(9) Board size
0.1875
*
-0.1441
*
-0.0587
*
0.4667
*
-0.1078
(10) Board independence
0.0101
(5) Leverage (6) Growth
*
0.0313
0.0007
0.0396
*
(6)
(7)
(8)
(9)
(10)
1.0000 0.0997 -0.1141
-0.0360
1.0000 *
*
0.0096
1.0000
-0.0774
-0.0583
*
1.0000
0.0113
-0.0638
*
0.0586
*
1.0000
0.0853
*
-0.0406
-0.0536
*
-0.0124
1.0000
This table reports the pairwise correlation coefficients for all variables in this study. The CEI Score is reported as published by HRC in their annual CEI report, and ranges from 0 to 100. The dependent variables, Tobin’s Q and ROA are calculated as follows: , , , . The control variables include Size (the natural log of Total and , = , = ,
,
Assets), Leverage (the equity-to-assets ratio), Growth (change in annualized Sales), Risk (annualized firm beta), ESG (company score for Corporate Social Responsibility), Board size (number of people on the board of directors) and Board independence (percentage of independent board members). All variables have been winsorized at 1% and 99%. * denotes significance at the 1% level.
34
Table 3. Univariate tests. Less LGBT friendly Mean Tobin's Q Profitability
Median
More LGBT friendly Mean
Median
Diff. in means
Diff. in medians
0.430 0.348 0.525 0.413 -0.095 *** -0.065 *** 4.788 4.703 5.600 4.851 -0.812 *** -0.148 16.131 16.007 17.274 17.198 -1.143 *** -1.191 *** Size 0.317 0.345 0.318 0.315 -0.001 0.030 *** Leverage 0.055 0.044 0.043 0.034 0.012 * 0.010 ** Growth 1.066 1.043 1.036 1.017 0.030 * 0.026 * Risk 43.415 40.700 51.961 46.520 -8.546 *** -5.820 *** ESG 10.065 10.000 11.562 11.000 -1.497 *** -1.000 *** Board size 74.985 81.820 78.949 83.333 -3.964 -1.513 *** Board independence *** The table reports the results of two-tailed t-tests and Wilcoxon rank-sum tests for the null hypothesis that there is no difference in the means and medians between less LGBT friendly and more LGBT friendly firms. The less LGBT friendly subsample consists of firm-year observations with CEI scores below 45, while the more LGBT friendly subsample includes firm-year observations with CEI scores of 100. Firm performance is measured by Tobin’s Q and Profitability. Tobin’s Q equals the sum of market capitalization and total liabilities divided by the book value of total assets and Profitability is the return on assets. Size is measured as the logarithm of total assets, Leverage is the equity-to-assets ratio, Profitability is the return on assets, Growth is the percentage change in sales from year t–1 to year t, Risk is the market model beta coefficient, ESG is the company score for Corporate Social Responsibility, Board size is the logarithm of the number of board members, and Board independence is percentage of independent board members. ***, **, and * denote statistical significance at the 0.01, 0.05, and 0.10 levels, respectively.
35
Table 4. Regression results. Tobin's Q Constant CEI score Size Leverage Profitability Growth Risk
Model 1
Model 2
1.834 *** (23.18) 0.002 *** (11.08) -0.096 *** (-16.69) -0.003 (-1.22) 0.040 *** (14.14) 0.180 *** (5.44) -0.074 *** (-2.67)
1.728 *** (12.20) 0.002 *** (4.88) -0.088 *** (-10.05) -0.003 (-1.48) 0.039 *** (15.30) 0.156 ** (2.18) -0.043 (-1.55)
Corporate governance
Model 3 1.771 *** (10.84) 0.002 *** (4.79) -0.091 *** (-8.80) -0.002 (-0.80) 0.040 *** (15.18) 0.160 ** (2.22) -0.045 (-1.63)
Model 4
Model 5
Board independence
25.949 *** (8.66) 0.018 *** (2.65) -1.001 *** (-5.24) -0.133 ** (-2.51)
25.181 *** (7.66) 0.014 ** (2.06) -1.171 *** (-5.45) -0.135 ** (-2.46)
8.828 *** (10.67) -4.442 *** (-13.86)
8.323 *** (6.87) -3.991 *** (-6.60)
8.410 *** (7.04) -3.803 *** (-6.12) 0.033 ** (2.09) 0.074 (0.68)
0.000 (-0.38) No No
Yes Yes
Yes Yes
Model 6
27.651 *** (15.78) 0.020 *** (5.07) -1.116 *** (-9.64) -0.138 *** (-3.13)
0.000 (0.33) 0.001 (0.17)
Board size
Industry fixed-effects Period fixed-effects
Profitability
0.002 (0.12) No No
Yes Yes
Yes Yes
No. of cross-sections 485 485 448 485 485 448 No. periods 12 12 12 12 12 12 No. of observations 2,993 2,993 2,870 2,993 2,993 2,870 Adjusted R2 0.58 0.60 0.61 0.18 0.20 0.21 F-stat. 245.80 *** 211.20 *** 184.82 *** 43.35 *** 38.03 *** 35.03 *** The table reports the estimates of six alternative versions of Equation (1). The dependent variable is Tobin’s Q in Models 1-3 and Profitability (return on assets) in Models 4-6. CEI score measures the firm’s LGBT-friendliness. The control variables used in the regressions are defined as follows: Size is measured as the logarithm of total assets, Leverage is the debt-to-equity ratio, Profitability is the return on assets, Growth is the percentage change in sales from year t–1 to year t, Risk is the market model beta coefficient, Corporate governance measures the strength of corporate governance, Board size is the logarithm of the number of board members, and Board independence is percentage of independent board members. The t-statistics (in parentheses) are based on robust standard errors which are adjusted for heteroskedasticity and are clustered by firm. ***, **, and * denote statistical significance at the 0.01, 0.05, and 0.10 levels, respectively.
36
Table 5. The effects of politics and religion. Tobin's Q Less religious Constant
1.562*** (15.97)
CEI score Size
Profitability
(11.21)
1.951*** (9.10)
Republican 1.397*** (11.29)
Less religious 29.476*** (27.65)
More religious 14.191*** (4.89)
Democratic 27.184*** (16.77)
Republican 19.247*** (7.14)
0.002***
0.002***
0.001***
0.018***
-0.001
0.013***
0.004
(7.17)
(4.82)
(4.35)
(3.96)
(7.22)
(-0.16)
(2.69)
(0.61)
-0.079***
-0.103***
-0.108***
-0.070***
-1.479***
-0.452**
-1.289***
-0.989***
(-8.39)
(-7.24)
0.002
-0.007**
0.000
-0.003
-0.144***
-0.089
-0.134*
-0.213*
(0.78)
(-2.21)
(0.13)
(-0.52)
(-4.38)
(-0.91)
(-1.81)
(-1.87)
0.048*** (15.95)
Growth
1.966***
Democratic
0.001***
(-20.31) Leverage
More religious
Profitability
(-16.68)
0.029*** (9.99)
0.039*** (11.21)
(-15.49)
(-2.33)
(-13.98)
(-5.59)
0.039*** (9.73)
0.149***
0.199***
0.218**
0.070
8.014***
8.382***
9.679***
6.560***
(3.07)
(3.31)
(2.02)
(1.07)
(9.32)
(7.61)
(8.64)
(3.08)
-0.010
-0.098**
-0.014
-4.206***
-3.952***
-2.418***
(-0.27)
(-2.21)
(-0.34)
(-1.95)
(-6.06)
(-5.62)
(-4.93)
0.001
0.000
0.000
0.002
0.027***
0.059***
0.034**
0.045***
(1.18)
(0.10)
(0.03)
(1.38)
(2.64)
(3.19)
(2.44)
(4.00)
0.000
0.003
-0.004
0.009**
0.116
-0.009
0.026
0.402***
(-0.12)
(1.03)
(-0.46)
(2.40)
(1.56)
(-0.08)
(0.50)
(5.63)
-0.001***
0.001
0.001
-0.001
0.007
-0.002
0.013
-0.022*
(-2.96)
(0.81)
(0.37)
(-1.47)
(0.69)
(-0.10)
(0.84)
(-1.66)
Industry fixed-effects
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Period fixed-effects
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Risk Corporate governance Board size Board independence
-0.06*
-3.61*** (-9.90)
37
Table 5. Continued. Tobin's Q No. of cross-sections No. periods No. of observations Adjusted R2 F-stat.
Profitability
Less religious
More religious
Democratic
Republican
Less religious
More religious
Democratic
Republican
285
163
245
183
285
163
245
183
12
12
12
12
12
12
12
12
1,863
1,007
1,492
771
1863
1007
1,492
771
0.67
0.53
0.63
0.54
0.259
0.1694
0.23
0.17
158.45***
47.50***
107.57***
39.24***
29.359***
9.92***
20.04***
7.70***
The table reports the estimates of eight alternative versions of Equation (1) using subsamples of firms located in less and more religious states and in decisively Democratic and Republican states. State-level religiosity is measured by the percentage of highly religious people as reported by the 2013 Gallup Survey. State-level political majority is derived from the outcome of the Presidential elections between 2000 and 2012. A state is classified as Democratic (Republican) if the Democratic (Republican) candidate won the presidential election with a margin of at least five percent. The dependent variable is Tobin’s Q in Models 1-4 and Profitability (return on assets) in Models 5-8. CEI score measures the firm’s LGBT-friendliness. The control variables used in the regressions are defined as follows: Size is measured as the logarithm of total assets, Leverage is the debt-to-equity ratio, Profitability is the return on assets, Growth is the percentage change in sales from year t–1 to year t, Risk is the market model beta coefficient, Corporate governance measures the strength of corporate governance, Board size is the logarithm of the number of board members, and Board independence is percentage of independent board members. The t-statistics (in parentheses) are based on robust standard errors which are adjusted for heteroskedasticity and are clustered by firm. ***, **, and * denote statistical significance at the 0.01, 0.05, and 0.10 levels, respectively.
Table 6. Instrumental variable regressions. First-stage regression Variable
CEI score
Constant
-32.733 *** (-4.07)
Second-stage regressions Tobin's Q 1.968 *** (9.52)
Profitability 27.861 *** (6.92)
Instrumental variables: Democratic
12.473 ***
Republican
(-2.62) *
(8.64) (-1.66) Independent variables: Instrumented CEI score Size Leverage
0.006 ***
0.089 ***
(3.05)
(2.61)
4.174 ***
-0.113 ***
-1.518 ***
(8.76)
(-7.33)
(-5.14)
-0.247
-0.001
-0.115 **
(-1.50)
(-0.34)
(-2.02)
Profitability
0.038 *** (13.81)
Growth Risk Corporate governance Board size Board independence
-9.757 ***
0.212 ***
9.076 ***
(-2.58)
(2.77)
(7.07)
-7.268 ***
-0.017
-3.219 ***
(-4.39)
(-0.49)
(-4.86)
0.339 ***
-0.001
0.005
(7.85)
(-1.04)
(0.24)
0.728 **
-0.002
0.012
(2.46)
(-0.36)
(0.10)
0.029
0.000
0.003
(0.51)
(-0.32)
(0.12)
Industry fixed-effects
Yes
Yes
Yes
Period fixed-effects
Yes
Yes
Yes
No. of observations
2,874
2,870
2,870
0.23
0.53
0.10
2
Adjusted R F-stat.
36.07 ***
180.40 ***
35.83 ***
The table reports the estimates of two-stage instrumental variable regressions. CEI score measures the firm’s LGBT-friendliness. The instrumental variables for the CEI score in the first-stage regressions are the state-level Democratic and Republican dummy variables. A state is classified as Democratic (Republican) if the Democratic
(Republican) candidate won the presidential election with a margin of at least five percent. The dependent variables in the second-stage regressions are Tobin’s Q and Profitability (return on assets). The control variables used in the regressions are defined as follows: Size is measured as the logarithm of total assets, Leverage is the debt-to-equity ratio, Profitability is the return on assets, Growth is the percentage change in sales from year t–1 to year t, Risk is the market model beta coefficient, Corporate governance measures the strength of corporate governance, Board size is the logarithm of the number of board members, and Board independence is percentage of independent board members. The t-statistics (in parentheses) are based on robust standard errors which are adjusted for heteroskedasticity and are clustered by firm. ***, **, and * denote statistical significance at the 0.01, 0.05, and 0.10 levels, respectively.
