Do LGBT-supportive Corporate Policies Enhance Firm Performance?

Prior research provides evidence that lesbian, gay, bisexual, and transgender (LGBT)-supportive corporate policies are related to important human resource functions, such as enhanced recruitment and retention. In addition, prior research indicates that investors view the adoption of such policies positively. We examine the firm-performance mechanisms underlying favorable stock-market reactions based on an integration of perspectives from corporate social responsibility and the business case for diversity. Specifically, we estimate a hierarchical linear model (HLM) to account for the nested nature of our data (firms nested within states) and find that (1) the presence of LGBT-supportive policies is associated with higher firm value, productivity, and profitability; (2) the firm-value and profitability benefits associated with LGBT-supportive policies are larger for companies engaged in research and development (RD and (3) the firm-value and profitability benefits of LGBT-supportive policies persist in the presence of state antidiscrimination laws. In supplemental analyses, we find that firms implementing (discontinuing) LGBT-supportive policies experience increases (decreases) in firm value, productivity, and profitability. We are among the first to link LGBT-supportive policies specifically to financial performance outcomes as well as to develop and test a multilevel model of these relationships. Our results have important implications for theory and research on LGBT issues in organizations, human resource managers, and policymakers.


Introduction
Significant advances have been made in recent history in the United States for the lesbian, gay, bisexual and transgender (LGBT) community. In 2011, "Don't Ask, Don't Tell" was repealed, effectively allowing gay men and lesbians to serve openly in the military. In 2013, the U.S. Supreme Court handed down two key rulings related to same-sex marriage. The Defense of Marriage Act (DOMA), which defined marriage as a union between a man and a woman, was repealed in a 5-4 decision (United States v. Windsor, June 26, 2013). In another 5-4 decision (Hollingsworth v. Perry, June 26, 2013), the court dismissed an appeal from proponents of California's Proposition 8, which had banned same-sex marriages in that state, on jurisdictional grounds-returning the case to the Ninth Circuit and effectively allowing same-sex marriages in that state to resume. There is also increased public and political support for the passage of federal non-discrimination employment legislation known as the Employment Nondiscrimination Act of 2013, and the U.S. Senate recently voted to approve the Act in a 64-32 vote.
Historically, public opinion of homosexuality has been negative. In 1965, 70% of respondents in a public opinion poll indicated that homosexuals were harmful to American life (Herek, 2002). Since that time, public opinion of homosexuality has changed significantly, becoming more positive (Hicks & Lee, 2006). Today, while approximately 40% of Americans feel that homosexuality is "always wrong" (Bowman, Rugg, & Marisco, 2013), most Americans (63%) feel that discrimination against gay men and lesbians is somewhat serious or very serious (Gallup, 2014), and 67% favor expanding federal hate crime laws to include sexual orientation (Gallup, 2014). Organizational support for LGBT workers also has risen noticeably in recent years. Whereas 61% of the Fortune 500 had non-discrimination policies in 2002, 88% did in (Human Rights Campaign, 2014.

Management and organization researchers have become increasingly interested in
organizational support for LGBT workers (e.g., Anteby & Anderson, 2014;Creed, Scully, & Austin, 2002;Day & Schoenrade, 1997;Griffith & Hebl, 2002;Kaplan, 2006;King & Cortina, 2010;Pichler & Ruggs, in press;Theodorakopolous & Budhwar, 2015;Trau & Härtel, 2007). Recent research at the organizational level has focused on potential financial benefits to organizations for adopting LGBT-supportive policies and practices. For instance, Johnston and Malina (2008), Wang and Schwarz (2010), and Li and Nagar (2013) examine company-level data to determine whether investors appreciate LGBT-supportive policies, and each of these studies documents at least some support for the hypothesized positive association between these policies and stock prices. However, these studies rely on stock market returns data to provide evidence of this positive association and call for additional research examining the fundamental performance improvements yielding their results.
We address these calls by developing and testing a cross-level theoretical model of firmperformance outcomes associated with LGBT-supportive policies (Figure 1), which is the overarching purpose of our study. We do so using a more generalizable dataset than has been used in previous research, offering a number of important and unique benefits, which we describe in the methods section. We therefore provide a robust and valid test of the proposition that organizations adopting LGBT-supportive policies perform better than non-adopters. To our knowledge, ours is the most sophisticated test of this proposition to date. In so doing, we address calls for more research on the economic imperative organizations face in terms of adopting LGBT-supportive policies, as well as the role of state laws as related to the protection of LGBT workers (e.g., King & Cortina, 2010;Pichler & Ruggs, in press). Results of our research have important implications for organizations, policymakers, and human resources management.

Integrating Corporate Social Responsibility and the Business Case for LGBT-Supportive Policies
Corporate social responsibility (CSR) theory. Howard Bowen, a pioneer of social responsibility theory, described CSR as a reference "to the obligations of businessmen to pursue those policies, to make those decisions, or to follow those lines of actions which are desirable in terms of the objectives and values of society" (Bowen 1953: 6). Although there are a variety of theoretical perspectives of CSR (Carroll, 1994;Garriga & Melé, 2013;McWilliams & Siegel, 2001), the prevailing framework is that of corporate social performance (CSP; Carroll, 1979;Preston, 1978), which is closely tied to stakeholder theory (Donaldson & Preston, 1995). The key proposition of stakeholder theory as it applies to CSR is that firms have varied stakeholders who are affected by and affect the firm in terms of, for instance, its performance (Freeman, 1984).
Thus, the CSP model posits that firms respond to social issues and social responsibility, in part, due to an economic imperative (e.g., Clarkson, 1995;Garriga & Melé, 2013;Wartick & Cochran, 1985). Simply put, firms attempt to behave in a socially responsive way so as to engage stakeholders in a manner that improves performance.
Numerous studies document a positive association between CSR, broadly defined, and firm performance (Blazovich & Smith, 2011;Chung, Eneroth, & Schneeweis, 2003;Rushton, 2002;Waddock & Graves, 1997). Findings suggest that firm performance is positively related to several components of CSR including environmental reputation (Clarkson, Li, & Richardson, 2004), the implementation of labor-friendly policies (Faleye & Trahan, 2011), and employee satisfaction (Edmans, 2011). The CSR literature is continually evolving, reflecting the shifting beliefs of what society views as good or responsible behavior (Carroll, 1999). It follows that CSR, perhaps especially social responsiveness, is largely dictated by society's current objectives and values. Given recent changes and trends in public opinion, public policy, and organizational adoption of LGBT-supportive policies, it seems particularly timely to better understand the legitimacy of such rationales for adopting these policies. CSR is theorized to benefit firms in terms of financial performance, even for a relatively controversial socially responsible practice.
While public opinion has changed in the recent past, negative attitudes towards gay men and lesbians are still more normative and socially accepted than other forms of prejudice (Ragins, 2004;Pichler, 2007). The adoption of LGBT-supportive policies is also relatively controversial (Creed et al., 2002;Kaplan, 2006;Chuang, Church & Ophir, 2011). In fact, firms often face backlash from key stakeholders, such as customers and investors, when adopting LGBT-supportive policies (e.g., Kaplan, 2006). A recent example is the One Million Moms boycott of Mattel for an article related to one of their product lines, American Girl dolls, featuring two gay fathers (Starr, 2015). Hence, firms adopting LGBT-supportive policies face competing challenges when it comes to CSP issues in the sense that they may be attempting to fulfill a social purpose that could threaten their fundamental responsibility to create and share wealth with key stakeholder groups (Clarkson, 1995). Thus, policy adoption in this case represents a unique test of CSR theory in the sense that adoption may not be universally seen as socially responsive, or may be less so than other diversity-related policies, such as work-family benefits (see Kossek & Pichler, 2007).
In the case of this study, the adoption of LGBT-supportive policies, as compared to other types of diversity management policies or practices, may not necessarily be related to firm performance, in part, because stakeholders may not react positively to these policies. Consistent with scholarship that suggests commitment to diversity and inclusiveness is an important aspect of CSR (Colgan, 2011;Snider, Hill, & Martin, 2003), we suggest that LGBT-supportive policies are an increasingly important, although more stigmatized, type of diversity-enhancing policy.
They are also increasingly important as related to outside stakeholder perceptions, namely to consumer perceptions. We explain these points further below by integrating the CSP model of CSR with the business case for diversity, which we see as an important contribution to CSR theory as applied to controversial policies. In so doing, we shed light on why firms should benefit from LGBT-supportive policies despite their relatively controversial nature.
The business case for diversity. Diversity experts have suggested that organizational support for diversity can have performance-enhancing benefits for organizations, and research generally supports this proposition (Richard, 2000;Theodorakopoulos & Budhwar, 2015). This logic is often used by organizations when explaining the potential benefits of CSR programs and socially responsible behavior, such as the adoption of diversity-enhancing policies. The literature on the business case for diversity is a useful complement to the CSR perspective outlined above.
In fact, CSR is increasingly important in a diverse and global business environment (Berkley & Watson, 2009). Research suggests that diversity at the organizational level and organizational support for diversity in terms of certain policies and practices are related to enhanced firm performance (Kossek & Pichler, 2007). In this connection, there is a growing body of research suggesting that organizational policies that support LGBT workers specifically may be a source of competitive advantage (Bell, Özbilgin, Beauregard, & Sürgevil, 2011;Day & Greene, 2008;King & Cortina, 2010) and should also be related to firm performance.
Estimates indicate that there are approximately nine million LGBT individuals living in the United States (Renna, 2011), comprising up to 12% of the workforce (Day & Greene, 2008)-a relatively large minority group. From a stakeholder perspective, as more firms implement LGBT-supportive policies, a company's decision not to adopt such rules may send a signal to potential employees that the firm is not socially responsible when it comes to antidiscrimination and support for diversity. Indeed, support for a variety of forms of diversity and inclusion seems increasingly "in vogue" (Theodorakopoulos & Budhwar, 2015). Recent research suggests that LGBT-supportive policies and practices are increasingly important to workers regardless of their sexual orientation. For instance, a recent poll found that 6% of heterosexual respondents indicated the availability of domestic-partner benefits is the most important factor when considering a new job (Badgett, 2006). Cordes (2012) contends that creating inclusive work environments through the adoption of LGBT-supportive policies helps establish companies as "employers of choice" for all employees, not just gay employees. Badgett (2006) suggests employees view companies with LGBT-supportive policies as more open and supportive.
In addition to more human resources-related benefits, gay consumers, perhaps especially gay men, are more likely to have higher levels of disposable income than their heterosexual counterparts (Colgan, Creegan, McKearney, & Wright, 2007;Iwata, 2006;Paul, McElroy, & Leatherberry, 2011;Valenti, 2012), which seems important from a stakeholder perspective. A recent survey (Experian Marketing Services, 2012) finds that the household income of married/partnered homosexual men (women) exceeds that of their heterosexual counterparts by $21,500 ($7,200). The buying power of the gay market, sometimes referred to as the "pink dollar," was estimated to reach $835 billion by 2014 (Paul et al., 2011). Hence, companies ignore the gay consumer at the peril of their own bottom line. In 1994, only 19 Fortune 500 companies targeted advertising campaigns at gay consumers, a paltry figure compared to the 175 Fortune 500 companies that did so in 2005. Gay consumers are distinct in that they are brand loyal (Iwata, 2006;Valenti, 2012) and prefer to buy products from companies with LGBT-supportive corporate policies (Clermont, 2006;Valenti, 2012). Conversely, companies with anti-gay actions or policies face a growing risk of losing business from both homosexual and heterosexual consumers. For example, Proctor and Gamble, AT&T, American Express, and Xerox pulled advertising from Dr. Laura Schlessinger's radio talk show to distance themselves from the host's anti-gay rhetoric (Valenti, 2012). Thus, it seems that support for diversity and CSR are closely tied when it comes to organizational rationale for adopting LGBT-supportive policies.
That said, to date there is no direct evidence supporting this proposition. Emerging management and organization research indicates that markets may react positively to LGBTsupportive policies (Proposition 1, Figure 1). Employing an event-study design, Johnston and Malina (2008) found a significant and positive abnormal return on the initial day of announcing the adoption of LGBT-supportive policies, based on scores on the HRC Corporate Equality Index (CEI), but no significant abnormal return when the window is extended to three days, suggesting an overall neutral market reaction to the announcement. Wang and Schwarz (2010) found that changes in firms' CEI scores are positively associated with stock-price changes during the subsequent year. Li and Nagar (2013) find excess annual returns of 14% in the year following the decision to adopt same-sex domestic-partner benefits. The authors suggest the excess returns may be due to an increase in profitability, documenting an increase in return on assets (ROA) following the adoption of these benefits. Firm performance (e.g., profitability and productivity) are related to stock market returns (Proposition 2, Figure 1; Bao & Bao, 1989;Belkaoui, 1999;Cho & Pucik, 2005;Varaiya, Kerin & Weeks, 1987). Thus we propose, consistent with Li and Nagar (2013), that positive stock market reactions to LGBT-supportive policies may be due, in part, to a link between supportive policies and firm performance.
-----Insert Figure 1 here -----Although this emerging literature would suggest that LGBT-supportive policies may be related to firm performance, these studies are limited in their scope and generally do not use firm performance outcome variables. Hence, while the stock returns-based results offered by Johnston and Malina (2008), Wang and Schwarz (2010), and Li and Nagar (2013) suggest that investors believe LGBT-supportive policies to be value-adding, perhaps due to the factors we outlined and explained above, our study investigates the actual existence of underlying value-adding performance (Hypothesis 1, Figure 1). Consistent with previous research on the outcomes of firm-level policies (e.g., Huselid, 1995) and assertions from scholars in the LGBT literature (Li & Nagar, 2013), we utilize multiple indicators of firm performance to assess the potential gains of implementing LGBT-supportive policies: firm value, productivity, and profitability. Based on the literature reviewed above, we expect that LGBT-supportive policies will be positively related to these performance metrics.

HYPOTHESIS 1.
LGBT-supportive corporate policies are positively associated with firm value, productivity, and profitability.

Engagement in R&D Activities as a Moderator of LGBT-Supportiveness and Firm Performance Relationships: A Resource-Based View
Wang & Schwarz (2010) state, "Future research is needed to explore how…boundary conditions influence the relationship between GLBT nondiscrimination policies and…organizational-level work-related outcomes" (pg. 212). We examine potential moderators of relationships between LGBT-supportive policies and firm performance outcomes related to characteristics of the firm or its strategy, consistent with the previous research on resource-based view of the firm (Richard, 2000). We leverage the resource-based view of the firm, research on demand for highly skilled labor, as well as the literatures reviewed above to develop a hypothesis about engagement in R&D activities as a firm-level moderator of the relationship between LGBT-supportive policies and firm performance (Hypothesis 2, Figure 1).
The resource-based view of the firm (Barney, Wright, & Ketchen, 2001;Barney & Wright, 1997) would suggest that support for LGBT workers should be more strongly related to firm performance among firms with relatively more demand for highly skilled labor, which we measure as engagement in R&D activities (Faleye & Trahan, 2011;Berman, Bound, & Griliches, 1994). The premise of this theory is resources that are valuable, rare, non-substitutable, and inimitable can provide competitive advantage for firms (Barney, 1992(Barney, , 2001Wernerfelt, 1984).
Human resources, or the way that they are managed, may be a particularly important source of competitive advantage for specific types of firms (Barney & Wright, 1997). Human resources are particularly valuable for firms that rely more on knowledge-based work, for instance (Becker & Gerhart, 1996). Specific human resource management policies, such as LGBT-supportive policies, may be a source of competitive advantage when tied to a firm's unique characteristics, competencies or strengths. Thus, we propose that social responsiveness, in the form of the adoption of controversial yet employee-supportive policies, may be more beneficial for firms with that engage in R&D activities. This seems highly consistent with what Garriga and Melé (2013) refer to as "social investments in a competitive context" (pg. 54) as a strategy for gaining competitive advantage. We believe this is an important integration of theoretical propositions from the CSP model of CSR with the resource-based view of the firm.
Firms that require highly skilled workers place more value on employee recruitment and retention (Faleye & Trahan, 2011). The CSR and diversity management literatures suggest that firms adopt LGBT-supportive policies to enhance recruitment and retention (e.g., Huffman, Watrous-Rodriguez, & King, 2008;Trau, 2015). Florida and Gates (2004) found that, while locating in a geographic region with high overall diversity positively correlates with financial performance for high-technology firms, the concentration of the gay population in the area is the foremost indicator of high-technology firms' success. The authors argue that the concentration of gay people in a geographic area reflects the "frontier" or "fringe" culture upon which the hightechnology field thrives. It seems important for such firms to be able to leverage this population through effective human resources (HR) policies, specifically LGBT-supportive policies. In addition, we expect based on the CSR and diversity management literatures that highly skilled heterosexual workers will be more attracted to and more interested in staying with firms that adopt LGBT-supportive policies (e.g., Badgett, 2006;Cordes, 2012). Thus, engagement in R&D activities may moderate relationships between LGBT-supportive polices and firm performance.
Consistent with previous research (e.g., Faleye & Trahan, 2011;Berman et al., 1994), we use research and development (R&D) expenditures as a way to operationalize demand for highly skilled labor. Not only are these expenditures important to the success of high-tech firms specifically (e.g., Florida & Gates, 2004), but also changes toward high-skill, non-production labor are closely tied to R&D expenditures among firms in general (e.g., Berman et al., 1994).
Moreover, firms rely on highly skilled workers to capture benefits from risky investments in firm-specific R&D expenditures (Faleye & Trahan, 2011). When firms compete based on firm-specific skills or knowledge (exemplified by R&D activities), low turnover and long-term employee commitment are critical to sustaining a competitive advantage (Batt, 2002), which we expect can be sustained through effective HR and diversity management in the form of LGBTsupportive policies (e.g., Badgett, 2006;Metcalf & Rolfe, 2011). Firms are most likely to benefit from CSR efforts that are tied to their business (Porter & Kramer, 2006), and we therefore expect that firms requiring highly specialized labor benefit more from LGBT-supportive policies because these policies should help firms attract and retain such labor.

HYPOTHESIS 2. The positive associations between
LGBT-supportive corporate policies and firm value, productivity, and profitability increase with engagement in R&D activities.

State-Level Policy as a Cross-Level Moderator of LGBT-Supportiveness and Firm Performance Relationships: Leveraging Perceived Organizational Support Theory
There is currently no federal legislation preventing discrimination on the basis of sexual orientation, making it effectively legal to use sexual orientation as a basis for making employment decisions. States and municipalities are free to adopt their own anti-discrimination laws, and such anti-discrimination laws vary significantly at state and local levels, as do attitudes towards gay men and lesbians (Pichler, 2007). Twenty-one states and the District of Columbia prohibit employment discrimination based on sexual orientation (Human Rights Campaign, 2014). The presence or absence of an anti-discrimination law at the state level may influence both the adoption of LGBT-supportive policies (Chuang et al., 2011) as well as the extent to which adoption is related to firm performance, which is yet untested in the literature. Thus, we leverage perceived organizational support theory (Eisenberger, Huntington, Hutchison, & Sowa, 1986) to develop a hypothesis about the presence or absence of a state-level anti-discrimination law as a cross-level moderator of the relationship between firm-level LGBT-supportive policies and firm performance outcomes (Hypothesis 3, Figure 1).
Perceived organizational support theory proposes that employees personify their organizations and feel more supported when their organizations demonstrate genuine concern for their well-being. These care qualities can be demonstrated in a variety of ways, such as through compensation policies or practices or supervisory supportive behavior (Rhoades & Eisenberger, 2002;Kossek, Pichler, Bodner, & Hammer, 2011). Perceived organizational support theory (Eisenberger et al., 1986;Rhoades & Eisenberger, 2002) would suggest that firms with LGBTsupportive policies, especially anti-discrimination policies, in states where sexual orientation discrimination in employment decisions is prohibited by law may not garner as much benefit from these firm-level policies as adopter firms in states without such legislation. Presumably, voluntary adoption becomes interpreted as a signal of care and support (e.g., Eisenberger et al., 1986). We expect that when employers adopt LGBT-supportive policies where employment discrimination on the basis of sexual orientation is already prohibited, the positive benefits we described above, in terms of social responsiveness among employees and consumers, will be muted.
Previous research suggests this interaction of firm-level policies and state-level laws is important to investigate. For instance, Ragins and Cornwell (2001) report that both protective legislation (state/local anti-discrimination laws) and organizational policies and practices (e.g., anti-discrimination policies, same-sex domestic partner benefits, etc.) reduce perceived workplace discrimination among LGBT employees. However, the beneficial effect of firm policies in terms of discrimination perceptions is almost three times as large as that of state/local laws. Research also has shown that anti-discrimination legislation is effective when it comes to mitigating sexual orientation discrimination (Barron & Hebl, 2010). Although no research to our knowledge has investigated the role of firm-level LGBT-supportive policies as well as state-level laws in the same model as they relate to firm performance outcomes, the literature on sexual orientation discrimination suggests that both firm policies and state laws may affect outcomes that are related to social responsiveness, such as employee attraction and workplace discrimination (see Pichler, 2007).
We propose that, when the adoption of LGBT-supportive policies is not voluntary but state mandated, adoption is less likely to be perceived as socially responsive by stakeholders and thus less likely to lead to enhanced firm performance outcomes.
HYPOTHESIS 3. The positive associations between LGBT-supportive corporate policies and firm value, productivity, and profitability decrease in the presence of state-level antidiscrimination laws.

Data
Our initial sample contains 26,243 firm-year observations with non-missing information observations that we lose due to missing governance and/or compensation data are different from the 4,619 observations that we retain in ways that affect the generalizability of our results.

Explanatory variable. The indicator variable
LGBT policy is the explanatory variable of interest in our primary analysis. We set this variable equal to the "Gay & Lesbian Policies" rating from the MSCI ESG STATS database, which codes this variable as 1 if the company has "notably progressive" LGBT-supportive corporate policies in year t, and 0 otherwise. MSCI ESG STATS uses multiple criteria to define "notably progressive" policies, including commendation by the HRC as an industry or corporate leader on such policies, earning an HRC CEI score of 60 or above, or the presence of both an anti-discrimination policy and the extension of benefits to same-sex domestic partners.
Response variables. We model four response variables that capture the underlying theoretical constructs of firm value, factor productivity, employee productivity, and profitability. (2011), we define these four response variables as follows:

Following Faleye and Trahan
firm value = Tobin's Q: (total assets -total common equity + fiscal year closing price × common shares outstanding) / total assets; factor productivity = the residual from the Cobb-Douglas production function, 1 which we estimate within each industry-year group: ln(net sales) = α + β1 ln(net property, plant, and equipment) + β2 ln(employees) + ε; employee productivity = ln(net sales / employees); profitability = return on assets: operating income after depreciation / total assets.
Moderating variables. We also examine the moderating effect of engagement in R&D activities on the associations between LGBT policy and our four response variables. Following Faleye and Trahan (2011) We also interact this variable with our explanatory variable of interest (LGBT policy × R&D).
The coefficient for the LGBT policy × R&D interaction tests whether, among firms with engagement in R&D activities (R&D=1), firms with LGBT-supportive policies experience higher firm value, productivity, and profitability relative to companies without LGBT-supportive policies. A positive coefficient for this interaction would provide support for Hypothesis 2.
We also examine whether the presence of an anti-discrimination law at the state level influences the performance outcomes associated with implementing LGBT-supportive policies at the firm level. To examine this issue empirically, we include the following variable in our model: state law = an indicator variable coded 1 if the state in which the firm is located prohibits employment discrimination based on sexual orientation in year t, and 0 otherwise.
We also interact this variable with our explanatory variable of interest (LGBT policy × state law).
The coefficient for the LGBT policy × state law interaction tests whether the associations between LGBT-supportive policies and firm-performance outcomes differ among firms headquartered in states with anti-discrimination laws (state law=1). A negative coefficient for this interaction would provide support for Hypothesis 3.
Control variables. Prior research (e.g., Faleye & Trahan, 2011) Yermack, 1996) demonstrates that these variables influence firm value, productivity, and profitability and thus represent necessary controls when examining the association between LGBT-supportive corporate policies and our four response variables. We include indicator variables to capture industry-and year-specific influences on our four response variables. An Appendix provides definitions of all variables used in this study.

Analytical Approach
We use a hierarchical linear model (HLM) to test our hypotheses. HLM accounts for the nested nature of our data: in examining the association between firm-level LGBT-supportive policies and firm-performance outcomes, we must consider that firms are headquartered within states, some of which have enacted state-level anti-discrimination policies. Specifically, we use Proc Mixed in SAS with restricted maximum likelihood estimation (Singer, 1998;Suzuki & Sheu, 1999;Bell, Ene, Smiley, & Schoenberger, 2013). This procedure provides simultaneous analysis of within-and between-group variance in a regression framework, thus allowing for the examination of higher-level units (i.e., states) on lower-level outcomes (i.e., firm performance) while maintaining the appropriate level of analysis (Hofmann, 1997)-in this case, firm performance variables. This procedure takes into account the non-independence of our nested data (i.e., firms within states). Before testing our multilevel model, we followed procedures outlined in the multilevel literature (Raudenbush & Bryk, 2002;Hofmann, 1997;Hofmann, Griffin, & Gavin, 2000), namely testing a null model with no predictor variables to check for systematic between-group variance in our Level 2 variable (i.e., state-level anti-discrimination policies).
We estimate this model four times, once for each of our four response variables. Thus, in our primary analysis, we nest firms within states and utilize a "levels" specification, regressing the levels of firm value, factor productivity, employee productivity, and profitability on the presence or absence of LGBT-supportive corporate policies, as well as interactions of this LGBT policy variable with measures of both engagement in R&D activities (R&D) and state antidiscrimination laws (state law), while controlling for other factors known to affect our response variables.

Descriptive Statistics
Table 1 presents descriptive statistics. Table 2 shows the percentage of sample firms with LGBT-supportive corporate policies by year. We present these annual percentages for observations without (N=26,243) and with (N=4,619) control variables but discuss only those for the reduced sample. In 1996, the first year in our sample period, this percentage is 4.26% (four of 94 firms). By 2009, the last year in our sample, this percentage climbs to 20.58% (135 of 656 firms). While this increase reflects changing societal views of homosexuality, it also may reflect boards and managers acting in the best interest of shareholders by enacting policies that enhance firm value, productivity, and profitability as our hypotheses contend.
-----Insert Tables 1 and 2 here ----- Table 3 presents bivariate correlations between each of the response and explanatory variables from our primary analysis. The correlation coefficients show that LGBT policy is positively and significantly associated with firm value, factor productivity, employee productivity, and profitability. These correlation coefficients provide preliminary evidence, consistent with Hypothesis 1, that the presence of LGBT-supportive corporate policies is associated with higher firm value, productivity, and profitability. Most of the control variables from our primary analysis are significantly correlated with both the response variables (firm value, factor productivity, employee productivity, and profitability) and the explanatory variable of interest (LGBT policy), emphasizing the importance of including them as controls in our model.

Results of Hierarchical Linear Model
Before we estimate our four regression models, we assess the appropriateness of HLM (untabulated). First, in all four models, the covariance parameter estimate for the intercept is significant (p<0.01), indicating significant variation in each of our four response variables among the states. Second, in the profitability model, the covariance parameter estimate for state law is significant (p<0.05), indicating that the effect of state anti-discrimination laws on profitability varies significantly among the states. Taken together, these results justify our use of HLM rather than simple OLS regression to test our hypotheses.  -----Insert Table 4 here ----- In the firm value and profitability models, the coefficients for the LGBT policy × state law interaction (0.07 and 0.01, respectively) are non-significant, indicating that state antidiscrimination laws do not influence the association between LGBT-supportive corporate policies and either firm value or profitability. However, in the factor productivity and employee productivity models, the coefficients for the LGBT policy main effect (0.16 and 0.13, respectively) are positive and significant (p<0.01), but the coefficients for the LGBT policy × state law interaction (-0.14 and -0.16, respectively) are negative and significant (p<0.01), and the sums of the main effect and interaction coefficients (0.16 + -0.14 = 0.02 and 0.13 + -0.16 = -0.03, respectively) are non-significant, indicating that the positive association between LGBTsupportive corporate policies and productivity exists only in states that permit employment discrimination based on sexual orientation. These results suggest that LGBT-supportive corporate policies do not enhance productivity in the presence of state anti-discrimination laws.
In the factor productivity and employee productivity models, the coefficients for diversity are positive and significant, indicating that policies promoting other types of workforce diversity also improve productivity. Consistent with Faleye and Trahan (2011), the coefficients for employee relations are positive and significant in all four models. The coefficients for certain other explanatory variables (e.g., size, leverage, investment opportunity, board size, and CEO ownership) are significant in Table 4, indicating the importance of controlling for these factors in our primary analysis. 2

Practical Significance
In the firm value model, for firms without R&D activities (R&D=0), mean firm value increases from 2.12 (intercept) to 2.14 (intercept + LGBT policy), indicating a 0.94% higher firm value for companies with LGBT-supportive policies, but this difference is not statistically significant. For firms with R&D activities (R&D=1), mean firm value increases from 2.37 (intercept + R&D) to 2.87 (intercept + LGBT policy + R&D + LGBT policy×R&D), indicating a 21.10% higher firm value for companies with LGBT-supportive policies, and this difference is statistically significant (p<0.01).
In the employee productivity model, for firms without R&D activities (R&D=0), mean employee productivity increases from 5.28 (intercept) to 5.42 (intercept + LGBT policy), indicating a 2.65% higher employee productivity for companies with LGBT-supportive policies, and this difference is statistically significant (p<0.01). For firms with R&D activities (R&D=1), mean employee productivity increases from 5.28 (intercept + R&D) to 5.46 (intercept + LGBT policy + R&D + LGBT policy×R&D), indicating a 3.41% higher employee productivity for companies with LGBT-supportive policies, and this difference is statistically significant (p<0.01). Results are similar in the factor productivity model.

Supplemental Analyses
To address potential endogeneity (i.e., whether implementing LGBT-supportive corporate policies may result in enhanced firm value, productivity, and profitability, or better performing firms may be more likely to implement such policies), we conduct two supplemental analyses: (1) estimating a "changes" specification of the model from our primary analysis and (2) causality tests based on temporal precedence. While we do not tabulate the results of these supplemental analyses, they are available by request from the corresponding author.

"Changes" Specification
Hypothesis 1 predicts an association between LGBT-supportive corporate policies and firm performance. This association may result from financially strong firms adopting LGBTsupportive policies. However, our literature review and theory development suggests that this relationship is such that adoption of policies may cause a change in financial performance.
Therefore, we provide a more robust test of the proposition that adoptions of LGBT-supportive policies are related to firm performance. We predict that LGBT-supportive policies promote better performance by testing the association of changes in firm performance with implementations and discontinuations of LGBT-supportive policies.
We first address this endogeneity concern using a "changes" specification. Specifically, we regress the change in firm value, factor productivity, employee productivity, and profitability from year t-1 to year t on the implementation, continuation, or discontinuation of LGBTsupportive employment policies. Thus, we code the explanatory variable of interest (∆ LGBT policy) as follows: 1 for firms that implement a policy (i.e., LGBT policy=0 in year t-1 and 1 in year t), 0 for firms that neither implement (i.e., LGBT policy=0 in years t-1 and t) nor discontinue LGBT policy=1 in years t-1 and t) a policy, and -1 for firms that discontinue (i.e., LGBT policy=1 in year t-1 and 0 in year t) a policy in the current year. In addition to ∆ LGBT policy, we measure all other explanatory variables (except R&D, state law, industry and year fixed effects) as changes from year t-1 to year t. For parsimony, we omit interactions of ∆ LGBT policy with R&D and state law and instead focus on the ∆ LGBT policy main effect. The coefficient for ∆ LGBT policy tests whether firms that implement (discontinue) LGBT-supportive employment policies experience increases (decreases) in firm value, productivity, and profitability relative to companies that neither implement nor discontinue such policies. Results reveal that the coefficient for ∆ LGBT policy is positive and significant for all four response variables. Thus, we provide evidence that firms implementing (discontinuing) LGBT-supportive employment policies experience increases (decreases) in firm value, productivity, and profitability relative to companies that neither implement nor discontinue such policies.
A total of 127 firm-year observations either implement (∆ LGBT policy = 1) or discontinue (∆ LGBT policy = -1) LGBT-supportive policies during our sample period. In an additional supplemental analysis (untabulated), we separate observations that implement (N=105) from those that discontinue (N=22) such policies. The coefficient for implementing observations is marginally significant (p<0.1) in the firm value, factor productivity, and employee productivity models and significant (p<0.05) in the profitability model. The coefficient for discontinuing observations is non-significant (p>0.1) in the firm value and profitability models, marginally significant (p<0.1) in the factor productivity model, and significant (p<0.05) in the employee productivity model.

Causality Tests Based on Temporal Precedence
The results of the "changes" specification provide evidence that adoptions (terminations) of LGBT-supportive corporate policies are associated with improvements (impairments) in firm value, productivity, and profitability. This is important in the context of this study because firms may be less likely to behave in a socially responsible way when they are performing less well financially (Campbell, 2007). To examine the direction of causality in these results, we rely on temporal precedence (Granger, 1969). First, we estimate a model to discern if the contemporaneous performance improvements associated with adoptions of LGBT-supportive policies persist in the year after adoption. In the firm value, factor productivity, and employee productivity models, results indicate that the positive outcomes associated with adopting LGBTsupportive corporate policies not only persist but also grow in the year after adoption.
Second, we estimate the logistic model from Li and Nagar (2013) to discern if past performance outcomes influence firms' decisions to adopt LGBT-supportive corporate policies.
Consistent with our primary analyses, the contemporaneous measures of firm value, factor productivity, employee productivity, and profitability are positively and significantly associated with adoptions of LGBT-supportive policies. However, neither the one-year nor the two-yearlagged measure of these variables is associated with these adoption decisions. Taken together, the results of these causality tests provide evidence that past firm value, productivity, and profitability do not influence adoptions of LGBT-supportive corporate policies, but adoptions of these policies do result in future improvements in firm value and productivity.

Discussion
The over-arching purpose of this study was to develop and test a cross-level theoretical model of firm performance outcomes associated with LGBT-supportive policies. The CSR literature has shown that socially responsible behavior is related to firm performance. The diversity management literature and the literature on LGBT workers complement the CSR perspective to support a rationale for positive links between LGBT-supportive policies and firm performance, at least conceptually, given that these links are not straightforward due to the relatively controversial nature of LGBT-supportive policies. Ours is the first study to date to directly test this relationship, not only with a longitudinal approach, but also with rigorous tests to address potential endogeneity concerns and tests of theoretically relevant moderators. Given that ours is the first study to test this relationship, this is an important first step towards understanding firm performance outcomes of LGBT-supportive policies and boundary conditions of those relationships.

Implications for Theory
Previous research has found that LGBT-supportive policies are associated with positive stock market performance, but the reasons underlying investors' positive assessments of these policies were heretofore uninvestigated. We therefore addressed calls from previous scholars (e.g., Wang & Schwarz, 2010) to develop and test logic for why these relationships may exist. In order to do so, we integrated key propositions from the CSP model of CSR and the business case for diversity. We also leveraged the resource-based view of the firm and perceived organizational support theory to develop logic for moderators of relationships between LGBTsupportive policies and firm performance.
Often, firms expect backlash from anti-gay groups and organizations (Clermont, 2006); for example, when the American Family Association (AFA) targeted several companies (e.g., Disney, Ford, Kraft Foods, Procter & Gamble, and PepsiCo.) because of their LGBT-supportive policies (Iwata, 2006). Although companies maintaining their LGBT-supportive policies in the face of protests may actually garner both favor and firmer commitment from "pink dollar" consumers (Baker, Strub, & Henning, 1995), it is not clear that LGBT-supportive policies will, on average, be related to improved financial performance. In fact, some firms may also perceive a financial burden in connection with the enrollment of gay employees' dependents in various employee-benefit programs. Even though enrollment statistics nationwide suggest actual increased enrollment due to domestic partners reaches only a fraction of a percent in most cases (Cordes, 2012), stakeholders may not react as positively to the adoption of LGBT-supportive policies as compared to other diversity-supportive policies.
Although firms may and do experience backlash due to LGBT-supportive policy adoption, our underlying argument is that these policies are increasingly important because attitudes towards gay men and lesbians are changing. In other words, the CSP model is helpful but insufficient when it comes to developing a theory of firm performance outcomes of LGBTsupportive policies and should be complemented by the business case for diversity. The business case for diversity suggests that when firms adopt policies that encourage all employees to bring their whole selves to the workplace, they tend to be more productive and may benefit by becoming an employer of choice (see Kossek & Pichler, 2007). Thus, organizations can potentially benefit from LGBT-supportive policies in terms of financial performance even when they experience backlash.
As previous scholars in this literature have suggested (e.g., Wang & Schwarz, 2010), it is important to understand not only if firms that adopt LGBT-supportive policies perform better than others, but why and under what conditions. We were interested in why some firms benefit more than others when it comes to adopting policies that are not necessarily broadly seen as socially responsive, an important insight into CSR theory. The resource-based view posits that resources, including human resources, can be valuable if they are uniquely matched to an organization's characteristics in a way that is difficult to imitate. The strategic CSR literature also suggests that when CSR practices are tied to the ways in which firms compete, these practices should be more strongly related to financial performance (Porter & Kramer, 2006).
Hence, we proposed that firms with engagement in R&D activities are more likely to perform better when adopting LGBT-supportive policies because these firms have a greater need to be seen as an employer of choice.
Another condition under which firms might benefit more from adopting controversial policies is when they are perceived as truly socially responsive. Here, we leveraged perceived organizational support theory (Eisenberger et al., 1986), which posits that employers are more likely to be perceived as genuinely supportive when they voluntarily adopt policies. Thus, the potential performance-enhancing benefits firms garner -in this case those related to being perceived as an employer of choice and socially responsive in the eyes of consumers -may be diminished. We see this as an important insight into CSR theory.

Implications for Research and Human Resources Management
In the case of LGBT-supportive policies, there is currently no federal legislation requiring or even promoting adoption; it is at the discretion of the employer organization. In general, our findings may be of particular interest to corporations considering the enactment of LGBT-supportive policies but are on the fence about it in some way. Because fiscal demonstrations are oftentimes the best motivator for companies questioning the expansion of diversity initiatives (Fassinger, 2008), and few such demonstrations exist (Metcalf & Rolfe, 2011), our study helps fill the gap in the existing literature on this dimension. Our results, combined with those from previous studies (Johnston & Malina, 2008;Li & Nagar, 2013;Wang & Schwarz, 2010), suggest that firms with LGBT-supportive policies benefit on key factors of financial performance, which, in turn, increase the investor perception of the firm.
We believe our results support the notion that LGBT-supportive policies are important, in part, because they maximize a firm's ability to attract highly skilled labor in tight labor markets.
In other words, when firms manage social responsiveness in the form of diversity-enhancing policies and labor demand effectively, this represents an opportunity for increasing competitive advantage. This integration of perspectives from the CSR and RBV literatures is consistent with anecdotal evidence that organizational attraction and recruitment are important reasons to adopt such policies. We recommend firms with R&D activities consider leveraging these policies.
We also tested the role of state-level anti-discrimination laws as a moderator of relationships between firm-level policies and firm performance measures, and in this sense we are among the first to develop and test a multilevel model of support for LGBT workers. We developed rationale for cross-level moderation based on an integration of propositions from the CSP model of CSR and perceived organization support theory. Results here are mixed in the sense that firm-level policies are related to some, but not all, measures of operational performance when testing state laws as a boundary condition. First, this suggests that state laws are important (e.g., Barron & Hebl, 2010;Ragins & Cornwell, 2001). This also suggests that state laws may be a boundary condition that limits the benefits to firms adopting LGBTsupportive policies. This could suggest that firms who are first-movers might garner the most benefits in terms of adoption, an important consideration for firms that have been reticent to adopt supportive policies.
Although we feel that ours is a meticulously designed study, we encourage others to continue investigating the types of questions that are at the heart of this research. We hope to promote more interest in and around the topic of rationales for adoption of LGBT-supportive policies and LGBT workers more generally (King & Cortina, 2010;Pichler & Ruggs, in press), a relatively under-studied group in the management and organization literature (Pichler, 2007