Linking LGBT inclusion and national innovative capacity

This paper investigates the relationship between the social inclusion of lesbian, gay, bisexual and transgender (LGBT) people, and technological innovation. It empirically establishes that LGBT inclusion helps foster human capital skills, thus strengthening national innovative capacity. Exploiting cross-country data, this research provides suggestive evidence that social tolerance towards homosexuality is positively correlated with the economic complexity index, a novel measure of cross-country differences in innovative capabilities. Individual-level analyses, based on data from the World Values Survey, reveal that surveyed respondents who self-report tolerance towards homosexual acts tend to have positive attitudes towards technological innovation, ceteris paribus. This lends credence to the international evidence. Further analyses indicate that the link between LGBT inclusion and innovation is partially mediated through the accumulation of human capital. The main findings suggest that the social exclusion of LGBT people, at least to some extent, impedes long-run economic development by hindering innovative activities.


Introduction
The existence of large and persistent disparities in economic prosperity across the world remains one of the most perplexing issues facing economists. Some early contributions to this line of inquiry hold that the exclusion of women and/or marginalized groups within an economy matters for international differences in income per capita. Specifically, many studies indicate that gender discrepancies in various dimensions of empowerment and wellbeing, particularly education, health and employment opportunities, exert a negative influence on economic growth and development (see, e.g., Knowles et al., 2002;Duflo, 2012). More recently, Badgett et al. (2019) postulate that the social inclusion of lesbian, gay, bisexual, and transgender (LGBT) people positively affects income levels. Their empirical analysis is based on regressing GDP per head on an internationally comparable index of LGBT inclusion, which reflects legal rights and protections afforded to homosexual individuals across countries.
Much of the existing literature has focused on estimating the effects of gender differences in different aspects of well-being on the worldwide distribution of GDP per capita. By contrast, the extent to which the social exclusion of homosexual people matters for economic performance is still open to question. Badgett et al. (2019), to my knowledge, is the only study investigating the relationship between discrimination against the LGBT community and comparative development across the globe. The interest of their paper, in particular, lies exclusively in the effects of LGBT inclusion on income per capita. Nevertheless, little is known about the mechanisms through which the social exclusion of LGBT people helps shape global income differences. If innovation, as proposed by this research, is a key channel through which homosexuality-supportive policies transmit to economic growth, a better understanding of this link helps formulate relevant policies. To speak to those debates, the current study attempts to examine the impacts of the social inclusion of LGBT people on national innovative capacity, based on cross-country and individual-level analyses.
More specifically, the main inquiry of this paper is motivated by three arguments. First, there has been significant progress in social tolerance towards homosexuality in many Western societies (Badgett et al., 2019). However, substantial discrimination against LGBT people remains widespread in much of the developing world. Specifically, homosexual acts are still illegal, and may result in severe punishments, including the death penalty, in many parts of Asia, Africa and the Middle East (Bailey et al., 2016). Moreover, a recent study by Ayoub and Kollman (2020) reveals that there exist significant differences in the recognition of LGBT rights across European countries. Second, the fifth Sustainable Development Goal emphasizes the importance of espousing equality in gender roles. Therefore, reducing discrimination against LGBT people is essential for achieving this goal. Furthermore, the extent to which we can create an LGBT-supportive environment across the world arguably depends on our understanding of the contribution of the social inclusion of homosexual people to economic performance. This paper puts forward the idea that promoting human rights and protections of the LGBT community helps spur technological innovation and economic growth. As such, this study advocates for the social inclusion of LGBT people, particularly in developing economies where homophobia typically prevails. Third, technological innovation is widely perceived as a key driver of productivity (or income) levels, and health outcomes (Vu, 2020). This points to the desirability of exploring the institutional and/or social environment that enhances national innovative capacity.
The main objective of this paper is to explore the link between the social inclusion of LGBT people and cross-country differences in innovation. I posit that reducing discrimination against homosexual behaviors enhances human capital accumulation, thus strengthening national innovative capacity. Furthermore, creating an LGBT-friendly environment would attract an inflow of talents because it signals the acceptance of diversity, creativity and open-mindedness. It follows from this line of reasoning that improvements in the social tolerance towards LGBT people can contribute to economic prosperity through bolstering innovation. The proposed hypothesis is tested by performing the empirical analysis at both the macro-and micro-level. More specifically, I use the LGBT inclusion index developed by Badgett et al. (2019) to estimate cross-country OLS regressions. In line with the central hypothesis, the study finds that LGBT inclusion is positively correlated with the economic complexity index, a novel measure of country-level innovative capabilities. The paper also carries out individual-level analyses, exploiting data from the World Values Survey. The results demonstrate that self-reported tolerance towards homosexuality is positively correlated with surveyed respondents' attitudes towards science and technology, and new ideas, creativity, taking risks, adventure and changes. This lends support to the cross-country evidence. I also find that LGBT inclusion affects innovation by reinforcing the quality of human capital.
This research offers a fresh perspective to the following strands of research. Specifically, this paper builds upon the literature investigating the economic impacts of wide discrepancies in gender roles throughout the world (Duflo, 2012;Knowles et al., 2002). I add evidence to this debate that the social inclusion of marginalized groups of a population, particularly LGBT people, exerts a positive influence on national innovative capacity, which is arguably an important driver of long-term growth. Furthermore, this research relates to several firm-level studies documenting a relationship between discrimination against LGBT people and firm performance, focusing in a few high-income societies (Brenner et al., 2010;Button, 2001;Griffith & Hebl, 2002;Pichler et al., 2018). Accordingly, discrimination against homosexual people at the work place is associated with lower levels of job satisfaction and organizational commitment among LGBT employees. By contrast, LGBT-supportive policies in the work environment reduce job anxiety among gay and lesbian employees, thereby promoting firm productivity. The current research goes beyond previous studies by establishing the link between LGBT inclusion and innovation at the country level covering a larger number of world economies. This provides a generalized understanding of the contribution of LGBT inclusion to economic performance across the world.
A final distinguishing feature of the current study stems from adopting the economic complexity index as a new measure of innovative capacity. The findings of the current research also contribute to an emerging strand of literature examining the role of economic complexity in explaining cross-country comparative development (Hartmann et al., 2017;Hidalgo & Hausmann, 2009;Lee & Vu, 2020;Vu, 2020). Importantly, very few studies pay attention to the social and institutional environment that shapes the level of economic complexity. This paper documents suggestive evidence that reducing the social exclusion of LGBT people is linked to greater country-level innovative capabilities captured by the economic complexity index.

Why does LGBT inclusion matter for innovation?
The central hypothesis is that the social inclusion of LGBT people helps promote innovation through enhancing the quality of human capital. Conventional wisdom in development economics holds that human capital is a key conduit of innovation, which critically affects productivity and income differences across the globe (Gennaioli et al., 2012). Specifically, countries endowed with better human capital tend to innovate more, because technological innovation arguably depends on people's curiosity, imagination, risk-taking and cooperation. By this logic, the social inclusion of LGBT people may promote national innovative capacity by strengthening human capital accumulation within an economy.
It is noteworthy that the quality of human capital reflects knowledge, skills and health capital, all of which contribute to national innovation and long-term growth through shaping individuals' productivity (Becker et al., 1990). This reveals that any barriers to the accumulation of human capital may hinder national innovative capacity. As put forward by Badgett et al. (2019), the social exclusion of LGBT individuals, measured by the absence of legal rights and protections afforded to LGBT people, prevents them from acquiring better human capital skills. A key explanation for this argument is that discrimination against homosexuality lowers educational attainment, and leads LGBT people to drop out of schools (Badgett et al., 2019;Kosciw et al., 2013). This is consistent with numerous studies documenting that gender discrimination is detrimental to educational attainment, thus hindering economic development (Duflo, 2012;Knowles et al., 2002). Additionally, there exists ample evidence at the subnational level that LGBT people tend to experience considerable discrimination in schools by their teachers and fellow students. For example, a survey conducted in European countries reveals that approximately 90% of LGBT individuals had witnessed negative comments or conduct in schools. 1 Khan et al. (2005) indicate that 50% of males who have sex with males in Bangladesh and India reported that they had been assaulted or harassed by teachers or other students. This significantly increases the probability of dropping out of schools due to the negative impacts of school harassment and assault (Khan et al., 2005).
The social inclusion of the LGBT community may also foster human capital accumulation through improving health outcomes and providing employment-related opportunities (Badgett et al., 2019). 2 Several studies reveal that LGBT people typically suffer from violence in Indonesia (Arivia & Boangmanalu, 2016) and India (Khan et al., 2005). This may translate into significant loss of human health, thus impeding national innovative capacity. Indeed, the LGBT community tends to suffer from health disparities, including HIV, depression, anxiety and suicidality, compared with their heterosexual counterparts (Badgett et al., 2019). Health issues reduce the productivity of LGBT people, which hinders national innovative capacity. It is well established that discrimination in employment is associated with reduced monetary profit when discriminatory employers refuse to recruit socially excluded workers who are more or at least as productive as others (Badgett et al., 2019;Becker, 1971). Hence, marginalized groups who face discrimination at the work place eventually end up with jobs in which they are less likely to exploit their full capabilities and skills. An early study by Bergmann (1971) reveals that people facing discrimination in employment opportunities may end up with unproductive occupations or even become unemployed. This suggests that LGBT inclusion can foster the accumulation of human capital through education and health improvements, and growing job-related opportunities, leading to productivity improvements. Therefore, societies characterized by a prevalence of discrimination against homosexuality are less productive and innovative. By contrast, countries that promote the social inclusion of LGBT workers are more likely to experience improvements in national innovative capacity.
According to an early view proposed by Mokyr (1990), diversity and tolerance constitute the fundamental drivers of the innovation process. Innovative activities, in particular, tend to proliferate in places with less discrimination against nonconformists. The underlying intuition is that technological advances depend on the social acceptability of unconventional people because innovators are mostly eccentric individuals (Mokyr, 1990). This implies that social tolerance afforded to homosexual people may spur creativity, knowledge, skills and innovation. There also exists some empirical evidence supporting the argument that the social inclusion of LGBT people fosters technological change and economic outcomes. Florida (2002), for instance, documents a positive link between the share of bohemians and the quality of human capital across regions in the US. Constructing a novel measure of the bohemian population at the regional level, the author finds that the correlation between the bohemian index and the talent index is 0.553. For example, areas with the highest values of the bohemian index (e.g., San Francisco, Seattle, and Boston) are likely to enjoy better human capital. By contrast, regions characterized by the lowest bohemian index (e.g., Buffalo, San Antonio and Cleveland) tend to suffer from lower levels of human capital accumulation. Moreover, social tolerance towards marginalized individuals, including homosexual people, acts as a catalyst for the development of technologicalbased industries (Florida, 2002). The results reveal that the most high-tech regions (e.g., San Francisco, Boston, Seattle, and New York) are also in the top ten bohemian regions (Florida, 2002). Furthermore, Florida et al. (2008) indicate that the social inclusion of gays and lesbians is conducive to human capital skills and regional development in the US.
The positive association between LGBT inclusion and innovation also builds upon numerous studies demonstrating that treating these marginalized individuals equally at the work place helps improve firm performance (Griffith & Hebl, 2002;Pichler et al., 2018). This viewpoint, in particular, asserts that reducing any discrimination against LGBT people in the work environment could nurture business outcomes because it helps improve productivity of these workers. Furthermore, LGBT-friendly policies implemented at the firm level would reduce the negative consequences associated with any discrimination against homosexuality, such as health care and absenteeism costs (Badgett et al., 2019). According to Button (2001), equal treatment in the work place would benefit LGBT employees by improving their mental health and enhancing their commitment to companies. Griffith and Hebl (2002) also find that LGBT-supportive policies are associated with lower levels of job anxiety, and help improve job satisfaction among gay and lesbian employees. Furthermore, tolerance towards homosexuality at the workplace motivates LGBT individuals to disclose their sexual orientation, which eventually enhances mental health and productivity of LGBT employees (Ragins et al., 2007). Exploiting a sample of 534 LGBT employees in the US, Ragins et al. (2007) document that homosexual employees reported less fear, and were more inclined to disclose their sexual identity when working with LGBT-supportive groups. Hence, non-discrimination towards homosexuality enhances firm productivity.
LGBT inclusion also plays an important role in strengthening the relationship between these marginalized groups and their co-workers and employers (Brenner et al., 2010). This is particularly important for the development of technologyintensive industries that typically require coordination in the work place. Additionally, better cooperation at the firm level may improve the utilization of the existing human capital and bolster innovative capabilities. For this reason, if reducing discrimination against homosexual people fosters firm performance, LGBT-friendly policies would strengthen national innovative capacity.
Another argument for why an LGBT-friendly environment matters for national innovative capacity lies in the assertion that tolerance towards homosexuality signals low barriers to entry of high-quality human capital. The basic idea is that places with greater social diversity and tolerance are more likely to attract an inflow of talents (Florida, 2003). The social inclusion of the LGBT community creates an open business environment that nurtures diversity and creativity, which is of importance for immigration and innovation (Badgett et al., 2019;Florida, 2003). This is particularly relevant for the development of technology-intensive industries and economic prosperity when examining the effect of LGBT inclusion on innovation from a cross-country framework. Noland (2005), for instance, demonstrates that the social tolerance of homosexuality exerts a positive influence on foreign direct investment. Therefore, I argue that innovative activities proliferate in countries with tolerance and inclusiveness of the LGBT community.

The model
To explore the relationship between LGBT inclusion and national innovative capacity, I set up the following model: where ECI denotes the economic complexity index, a novel measure of innovative capabilities for country i ( i = 1, 2, … , 116 ).
LGBT stands for the LGBT inclusion index developed by Badgett et al. (2019). captures the estimated effects of LGBT inclusion on national innovative capacity. X corresponds to a set of main control variables, including trade openness (Trade), financial development (Finance), government size (Gov_size) and the log of population size (Pop_size). is the error terms. See also Table 1 and the online Appendix for a more detailed description and summary statistics of all variables, and data sources.
It is important to discuss the motivations of estimating the cross-sectional models to investigate the link between LGBT inclusion and innovation. First, the main interest of the current study lies in the impacts of the social inclusion of LGBT people on the cross-country variation in innovative capabilities. Estimating cross-country regressions, therefore, is relevant for capturing the long-term relationship between ECI and LGBT, which tends to be relatively stable over the years. Second, the LGBT index, constructed by Badgett et al. (2019), exhibits little variation within a country over time. This is consistent with the argument that discrimination against homosexuality appears to be an enduring feature of many societies throughout the world.
The choice of main control variables is partially motivated by Vu (2019) who examines the determinants of economic complexity -a novel measure of national innovative capacity. A potential confounder is trade openness, which promotes innovation through enhancing the cross-border dissemination of knowledge and skills (Sweet & Maggio, LGBT This index reflects cross-country differences in the social inclusion of LGBT people, captured by legal rights and protections afforded to homosexual individuals. Source: the Global Index on Legal Recognition of Homosexual Orientation developed by Badgett et al. (2019) 1.36 1.15

ECI
The economic complexity index reflects the availability of productive (innovative) capabilities that allow an economy to produce more sophisticated (high-productivity) products. This is a novel measure of cross-country differences in national innovative capacity. . Moreover, financial development is an important driver of technological innovation, as suggested by Hsu et al. (2014). Government size also exerts a positive influence on national innovative capacity through shaping the provision of public goods, including education and health services, well-functioning legal systems and social order (Sweet & Maggio, 2015;Vu, 2019). As explained below, using ECI to capture cross-country differences in innovation exploits the idea that innovative economies can produce and export a diverse range of sophisticated (high-productivity) products. A key concern relates to the possibility that the diversity of a country's export structures is purely attributed to increasing returns to scale in production and exports, possibly driven by population size (Sweet & Maggio, 2015). Therefore, I include these variables in the main analysis. Table 1 contains a description of key variables included in the baseline model specification.

Measuring international variation in LGBT inclusion
To my knowledge, there exists no internationally comparable measure of well-being disparities (e.g., earning, education or health discrepancies) between homosexual and heterosexual individuals (Badgett et al., 2019). Moreover, measuring the size of the LGBT community across countries appears to be challenging due to the absence of sexual orientation questions in most demographic or household surveys, particularly in developing economies. In particular, demographic questions about sexual identity or orientation are particularly sensitive in many societies characterized by substantial and persistent discrimination against homosexuality, thus being excluded from public surveys. Obtaining a comparable measure of the size of the LGBT community across countries also rests upon the consistency of questions about sexual identity or orientation of surveyed respondents. 3 Following Badgett et al. (2019), this paper exploits a global dataset of legal rights and protections afforded to homosexual individuals to measure international differences in the social inclusion of LGBT people (Table 1). More specifically, Badgett et al. (2019) introduce the Global Index on Legal Recognition of Homosexual Orientation (LGBT). The construction of this index relies on a threestage procedure. The first stage involves identifying the types of laws related to the social inclusion of LGBT people. In the second stage, these authors attempt to collect internationally comparable data on these laws. The final stage requires assigning numerical values to these laws, thus providing a proxy for the social inclusion of homosexual individuals. As put forward by Badgett et al. (2019), the LGBT index utilizes eight categories of legal rights that have been implemented to reduce discrimination against LGBT people across the globe. They include (i) Legality of consensual homosexual acts between adults; (ii) Equal age limits for consensual homosexual and heterosexual acts; (iii) Explicit legal prohibition of sexual orientation discrimination in employment; (iv) Explicit legal prohibition of sexual orientation discrimination regarding goods and/or services; (v) Legal recognition of the non-registered cohabitation of same-sex couples; (vi) Availability of registered partnership for same-sex couples; (vii) Possibility of second-parent and/or joint adoption by same-sex partners; and (viii) Legal option of marriage for same-sex couples. See Badgett et al. (2019) for more details. To estimate the cross-sectional models, I calculate a simple average of this index between 1966 and 2011. Figure 1 depicts the cross-country variation in LGBT inclusion, captured by legal rights and protections afforded to homosexual individuals.

Measuring international variation in innovation
Conventional proxies for innovation include the number of patents and R&D expenditure. The drawbacks of using these measures to capture the cross-country variation in innovative capabilities are well documented (Sweet & Maggio, 2015). For instance, economists have typically made use of R&D expenditure as an input-based measure of innovative capabilities (Kleinknecht et al., 2002). However, the extent to which R&D spending translates into real innovative capacity critically depends on the institutional environment that shapes the efficiency of utilizing this input (Sweet & Maggio, 2015). Hence, we can hardly infer anything about the output side of the innovation process from R&D expenditure (e.g., commercially oriented innovation such as the introduction of new products, services or processes).
The number of patents has been popularly used as an (intermediate) output-based indicator of innovative capabilities (Kleinknecht et al., 2002). Nevertheless, a "culture of patenting" varies significantly between countries in terms of the criteria of patent applications and the degree of protection of patented innovations (Sweet & Maggio, 2015). This may lead to measurement errors in cross-country studies if innovative activities are not reflected in the number of patents due to lack of a "culture of patenting", which tends to prevail in the developing world (Sweet & Maggio, 2015). Furthermore, technological progress can be driven by non-patented or unpatented inventions (Kleinknecht et al., 2002). Importantly, innovation is defined as an incremental process, obtained via the accumulation of both tangible (explicit) and intangible (tacit) knowledge (Nelson, 2005). The number of patents reflects only the "explicit" side of innovation but it says nothing about "tacit" knowledge (Sweet & Maggio, 2015). Moreover, some patents are not relevant for creating commercially valuable products (Kleinknecht et al., 2002). Thus, the number of patents reflects inventions rather than innovation. It follows from these arguments that using conventional measures of innovation is subject to potential measurement errors bias. Fig. 1 The worldwide distribution of LGBT inclusion. This figure depicts international variation in the social inclusion of LGBT people, with darker areas corresponding to societies with better legal rights and protections afforded to homosexual individuals. See also Table 1 Against this background, this paper attempts to capture cross-country differences in innovation by using the economic complexity index (ECI), developed by Hidalgo and Hausmann (2009). Innovative activities generally take the form of creating new products, services, and processes. National innovative capacity critically hinges on the stock of "tacit" and "explicit" knowledge available within a country. For this reason, innovation can be directly inferred from the availability of productive capabilities within an economy, and its ability to assimilate and exploit existing productive knowledge. Building upon these ideas, Hidalgo and Hausmann (2009) develop ECI in which the accumulation of productive capabilities can be measured by examining the types of products a country produces (and exports).
More specifically, the construction of ECI relies on the "diversity" and "ubiquity" of a country's export bundles. First, diversity captures the number of products a country can produce and export with revealed comparative advantage. The central idea holds that a country is endowed with a larger set of productive capabilities if it can produce and export a diverse range of products. Moreover, product diversification reflects the ability to assimilate and utilize innovative capabilities to create commercially valuable products, which is a key driver of technological progress and economic growth. Second, ubiquity reflects product sophistication based on the extent to which a country's products are popularly produced and exported by many other economies. Low-ubiquity products (e.g., smartphones, machinery, chemicals and metals), which require many hard-to-find innovative capabilities, are generally produced in only a few economies possessing such capabilities (Felipe et al., 2012). This is because the production of sophisticated products is only viable in places where prerequisite technologies and knowledge are available. By contrast, ubiquitous products (e.g., agricultural, wood, raw materials and commodities, and textiles) can be easily produced in many parts of the world as they require much fewer productive capabilities (Felipe et al., 2012). For example, Japan, Germany and the US, among others, are the most complex economies in the world because they can produce a diverse range of low-ubiquity products, such as medical imaging and machinery (Felipe et al., 2012). Meanwhile, Cambodia, Papua New Guinea, and Nigeria are relatively  Table 1 technologically backward because they mainly produce ubiquitous commodities (Felipe et al., 2012). Figure 2 depicts cross-country differences in innovative capabilities, captured by ECI.
Recent studies employ ECI to capture the variation in innovative capabilities across the world. They establish that ECI is a key determinant of a wide range of economic outcomes. For instance, there exists evidence that countries with higher values of ECI experience higher rates of economic growth (Hidalgo & Hausmann, 2009;Zhu & Li, 2017), less income inequality (Hartmann et al., 2017;Lee & Vu, 2020), better health outcomes (Vu, 2020), inclusive institutions (Vu, 2019), and increased labor share (Arif, 2021). Table 2 contains an overview of related empirical studies examining causes and consequences of economic complexity. Given that ECI directly matters for economic prosperity, this paper investigates the contribution of LGBT inclusion to national innovative capacity captured by ECI. Figure 3 illustrates the partial effects of LGBT inclusion on innovation. Consistent with the main hypothesis, the social inclusion of LGBT people is positively correlated with national innovative capabilities, holding other things constant. This suggests that countries with better legal rights and protections afforded to homosexual people are endowed with higher levels of economic complexity, which reflects an enhanced capacity to innovate. Table 3 presents OLS estimates of the effects of LGBT on ECI. In column (1), I report the unconditional estimates. Accordingly, the coefficient on LGBT is positive and statistically significant at the 1% level, which lends support to the central hypothesis articulated in Sect. 2. From columns (2) to (5), I gradually incorporate a set of main control variables in the regression. As discussed previously, these factors are key drivers of national innovative capacity. Controlling for these country-level characteristics helps mitigate plausible concerns about omitted variables bias. The results demonstrate that the estimated coefficients on LGBT remain positive and statistically significant at the 1% level after ruling out the impacts of potentially confounding factors (Table 3)

. A one-unit increase in
LGBT is associated with a roughly 0.32-unit increase in ECI, which equates to approximately one third of a standard deviation of ECI (column 5, Table 3). This reveals that the social inclusion of homosexual people exerts an economically significant influence on national innovative capacity.
Overall, the main findings suggest that improvements in legal rights and protections afforded to LGBT people play an important role in fostering national innovative capacity. The aforementioned arguments reveal that the social inclusion of homosexual people helps spur technological innovation through fostering the quality of human capital. By contrast, countries characterized by substantial discrimination against the LGBT community may suffer from reduced human capital accumulation, leading to less innovation. These results complement the findings of Badgett et al. (2019) by improving our understanding of a potential channel through which the social inclusion of LGBT people transmits to global income differences. Furthermore, the current research contributes to an emerging line of research examining the contribution of economic complexity to economic performance Table 2 Empirical studies using the economic complexity index This table provides a brief review of selected empirical studies exploring causes and consequences of economic complexity, which is a novel measure of national innovative capacity. See also the main text Study Key findings Hidalgo and Hausmann (2009) This paper develops the method of reflections to construct ECI, and documents evidence of the positive effects of economic complexity on economic growth across countries Zhu and Li (2017) Exploiting a world sample of countries, this paper revisits the link between economic complexity and economic growth. Their findings are consistent with Hidalgo and Hausmann (2009). Furthermore, this study reveals that the effects of economic complexity on economic growth are quantitatively larger in countries with better human capital Hartmann et al. (2017) This paper examines the relationship between economic complexity and income inequality across countries. It empirically establishes that complex economies tend enjoy an equal distribution of income, holding everything else constant Vu (2019) This paper investigates the extent to which institutional quality helps foster economic complexity. The author provides suggestive evidence of a positive relationship between economic complexity and the quality of institutions across countries. This is closely related to the current research that attempts to uncover the determinants of economic complexity, which is a novel measure of national innovative capabilities Lee and Vu (2020) This paper argues that the distributional impacts of economic complexity are reinforced by human capital accumulation Vu (2020) This study goes beyond the existing literature by investigating the association between economic complexity and national health status. Using cross-country data, the author documents robust evidence of the positive impacts of economic complexity on different measures of population health. Accordingly, complex economies characterized by a diverse range of employment opportunities are more likely to enjoy better health outcomes Arif (2021) A recent empirical analysis by Arif (2021) reveals that economic complexity helps improve labour share via increasing the bargaining power of workers    (Table 2). In particular, I put forward that achieving higher levels of economic complexity can be facilitated by LGBT inclusion. 4 Trade openness is expected to facilitate the cross-border diffusion of knowledge and technologies, which may foster technological innovation. This paper, however, finds that the effects of trade liberalization on country-level innovative capacity are statistically insignificant at conventionally accepted levels (Table 3). Therefore, I do not find evidence supporting the argument that trade openness is an important driver of innovation. By contrast, the impacts of financial development, government size, and population size on ECI are positive and statistically significant at conventionally accepted levels (Table 3). These findings are consistent with previous studies (e.g., Hsu et al., 2014;Sweet & Maggio, 2015;Vu, 2019).

Controlling for other effects
To avoid obtaining spurious estimates, I replicate the main analysis by accounting for numerous potential confounders, and report the results in Table 4. The online Appendix contains a detailed description of these additional controls.
First, I control for the effects of the diversity of birthplaces of immigrants. This is motivated by recent studies linking birthplace diversity and economic development (Alesina et al., 2016;Bahar et al., 2020). In particular, Bahar et al. (2020) find that an index of population diversity is positively correlated with ECI. It is argued that countries with greater social tolerance towards LGBT people would attract immigrants, thus enhancing population diversity. Nevertheless, my findings are robust to accounting for these effects (Table 4). In contrast to Bahar et al. (2020), the effects of birthplace diversity on ECI are imprecisely estimated at conventionally accepted levels of statistical significance. This suggests that social tolerance towards LGBT people plays a more prominent role in shaping national innovative capabilities. Second, I incorporate dummy variables for legal origins in the regression. The basic idea is that common-law countries endowed with greater protection of private property rights can be more innovative, relative to their civil-law counterparts. The baseline estimates, however, are insensitive to accounting for legal traditions. Third, I control for the effects of land suitability and resource endowments because these factors may affect the development of technology-intensive industries, following Vu (2019). The inclusion of these controls in the benchmark model fails to alter the main results. Finally, my results may yield a spurious relationship between the social inclusion of LGBT people and innovation if I fail to control for the quality of institutions and income levels. For this reason, I rule out the impacts of democracy, institutional quality, and income per capita, as demonstrated in Table 4. Accordingly, the estimated coefficients of LGBT are still positive and statistically significant at the 1% level in all cases. When all of these additional control variables are included in the regression, the contribution of social tolerance towards homosexuality to national innovative capacity remains precisely estimated at conventional levels of statistical significance (Table 4). Overall, I find that the baseline findings are robust to controlling for numerous potentially confounding factors.

Additional robustness checks
To provide a valid basis for causal inference, I perform numerous other sensitivity checks. The results of additional robustness analyses are provided in the online Appendix to conserve space.
First, I rule out the possibility that the positive relationship between LGBT inclusion and innovation across the globe is exclusively driven by the inclusion of specific groups of countries. To this end, I exclude countries located in the same continent because they may share common cultures, histories and geographic characteristics. Moreover, countries whose values of the LGBT index equal zero are removed from the sample. The inclusion of continent dummies also accounts for unobserved time-invariant heterogeneity across world regions that may simultaneously affect innovation and the social inclusion of homosexual people. As shown in the online Appendix Table A3, the baseline results remain intact. Hence, the main findings are unlikely to be purely attributed to the inclusion of specific categories of countries or unobserved region-specific factors.
Second, another concern relates to the role of cultural factors in shaping technological innovation and social tolerance towards homosexual acts. For example, it is widely established that individualistic cultures characterized by inclusive institutions are key driving forces of innovation (Gorodnichenko & Roland, 2017). Collectivistic societies, relative to their individualistic counterparts, are less likely to reward individuals deviating from norms and standards, and tend to emphasize conformity. Hence, cultures plausibly shape the social inclusion of the LGBT community and innovation. To address this issue, I allow several proxies for the cross-country variation in different cultural dimensions to enter the benchmark model specification. These additional controls include social trust, the cultural dimension of individualism/collectivism, and the fractions of the population practicing major religions. The results reported in the online Appendix Table A4 indicate that my findings are not attributed to these cultural factors. Third, I check for potential bias induced by constructing a simple average of LGBT across the period 1966-2011. It is important to re-emphasize that the social inclusion of LGBT people exhibits little variation over the years within a country. Hence, the use of an average of LGBT is relevant in this context. However, one could well argue that the baseline estimates are driven by the period chosen to compute LGBT. Therefore, I replicate the benchmark results by using the main variable of interest measured across different periods.
The results indicate that the coefficients on LGBT retain their signs and significance levels (the online Appendix Table A5). Furthermore, the use of alternative measures of national innovative capacity fails to alter the core findings (the online Appendix Table A6).
Finally, the paper checks for robustness to spatial dependence. The underlying idea is that innovative capabilities and social tolerance towards homosexuality may transcend national borders through the international diffusion of knowledge and technologies, and socio-economic interactions between world economies. The presence of such relationships between countries leads to inconsistent estimates. To mitigate this concern, I calculate the standard errors that correct for spatial dependence (the online Appendix Table A7). 5 It is also evident from the online Appendix Table A8 that my findings are unlikely to be exclusively driven by potential outliers.

Individual-level evidence
The cross-country evidence is suggestive of a positive link between LGBT inclusion and national innovative capabilities. Although I attempt to control for numerous confounding factors, a key threat to identification relates to the effects of unobserved country-specific factors. Given that the findings are derived from a cross-country empirical framework, it is impossible to rule out this concern using country fixed effects (FEs). This motivates an analysis at the subnational level. Unfortunately, there exists no comprehensive dataset of both LGBT inclusion and innovation at the regional level across the globe. Moreover, the construction of such data would be very challenging. Therefore, I employ data from the World Values Survey to explore the relationship between surveyed respondents' attitudes towards homosexuality and technological innovation (Table 5). For this purpose, I specify the following model:  The bold value illustrates the findings for the main variable of interest This table reports empirical estimates of the relationship between LGBT inclusion and innovation at the individual level, using data from the World Values Survey. Robust standard errors in parentheses. Homosexuality is surveyed respondents' perceptions on whether homosexual behaviors are justifiable, recorded on a categorical scale from 1 ("never justifiable") to 10 ("always justifiable"). The dependent variables are different measures of attitudes towards science and technology, new ideas, taking risks and changes. Higher values correspond to positive attitudes towards innovative activities. See also the main text and the online Appendix for a more detailed description of variables *** p < 0.01, **p < 0.05, *p < 0.1 Table 5 (continued) in which the dependent variables are different measures of attitudes towards innovative activities of individual i in country c, surveyed in wave v of the World Values Survey. The main variable of interest is Homosexuality, which captures self-reported tolerance towards homosexual behaviors. X stands for a vector of individual-level controls, including age, age squared, income, gender, the degree of trust in others and educational attainment.
The paper uses data conducted from face-to-face interviews in more than 60 countries, spanning over six waves of the World Values Survey from 1981 to 2014. The main proxy for LGBT inclusion is derived from a question in which surveyed respondents are asked about the extent to which they think homosexuality is justifiable (Homosexuality). 6 The answers to this question were recorded on a categorical scale from one ("never justifiable") to 10 ("always justifiable"). Therefore, this question reflects individual-level perceptions on social tolerance towards homosexual behaviors. Higher values of the dependent variable correspond to greater tolerance towards homosexual behaviors within a society. I adopt seven questions reflecting people's attitudes towards science and technology, and new ideas, taking risks and changes as the outcome variables. Furthermore, individual-level controls are incorporated in all the regressions, including age, age squared, income levels, a binary variable for male, social trust, and educational attainment. 7 Country dummies are also added to all the regressions to control for unobserved country-specific factors. I further include religion and wave FEs in the model specification. The online Appendix contains a detailed description of these variables. 8 Table 5 reports the estimation results from the individual-level analysis. The dependent variable used in column (1) is whether survey participants agree that we depend too much on science versus faith (E220). The answers are coded from one to ten, with higher values corresponding to negative views about science and technology. For ease of interpretation, I recode this variable by multiplying it by minus one, so that higher values represent positive attitudes towards technological advances. The second question is whether surveyed respondents agree that science and technology are changing our life too fast (E219). Higher values imply negative views about technological progress. I also recode this question so that higher values denote positive views about technological changes (column 2). The next question is whether respondents think our world is better off because of science and technology (column 3). As evident in columns (1) to (3), Homosexuality enters all the regressions with a statistically significant and positive coefficient. This suggests that social tolerance towards homosexual acts is positively associated with individual-level attitudes towards science and technology.
The remaining columns of Table 5 present empirical estimates of the relationship between Homosexuality and surveyed respondents' views about new ideas, taking risks and changes. In column (4), the dependent variable is whether survey participants agree that new ideas are better than old ones (E046). Furthermore, I use the question about attitudes towards the importance of new ideas and creativity (A189). The answers are also coded from one to ten with higher values corresponding to greater disagreements with this viewpoint. I also recode this variable by multiplying it by minus one to make it easy to interpret the findings. In column (6), I use the question in which surveyed respondents are asked whether they welcome or worry about changes (E047). An increase in this variable is associated with positive views about changes. The final dependent variable adopted in column (7) is whether survey participants disagree about the importance of taking risks and adventure (A195). For ease of interpretation, this variable is also re-calculated by multiplying it by minus one. The estimated coefficients of Homosexuality are statistically significant at the 1% level when different dependent variables are used (except in column 6). 9 Taken altogether, my findings suggest that the social inclusion of homosexual people is associated with positive attitudes towards new ideas, creativity, changes, adventure and taking risks across individuals in the World Values Survey.
Overall, the individual-level analysis reveals that people who self-report tolerance towards homosexuality tend to have positive attitudes towards technological progress. The effects of Homosexuality on innovation remain precisely estimated at conventional levels of statistical significance after accounting for a wide range of possible confounders, including individuals' characteristics, unobserved country-and wave-specific factors. The inclusion of religion dummies helps address a concern that my findings just proxy for other cultural and religious factors. Therefore, the individual evidence lends further support to the baseline findings.

Potential channels of transmission
The central hypothesis rests upon the premise that the social inclusion of LGBT people helps improve the quality of human capital (Sect. 2). Furthermore, LGBT-supportive policies arguably signal low barriers to inflows of talents, thus enhancing national innovative capacity. The cross-country OLS estimates lend empirical support to the main hypothesis. This section provides some further evidence on a key mechanism underlying the relationship between LGBT inclusion and innovation.
To this end, the study replicates the benchmark estimates by controlling for different measures of human capital (Panel A, Table 6). It is evident from Table 6 that the effects of LGBT inclusion on innovation remain precisely estimated at the 1% level of significance. Importantly, the magnitude of the estimated coefficients of LGBT reduces significantly when the proposed channel of influence is incorporated in the regression. For instance, the results in column (1) of Panel A indicate that the baseline estimates decrease to nearly a half when I control for the human capital index. This suggests that the contribution of LGBT inclusion to national innovative capacity is, at least partially, mediated by human capital accumulation. Next, different measures of human capital accumulation are regressed on LGBT (Panel B, Table 6). Following Kraay (2019), I employ the World Bank's human capital index in column (1). This indicator captures the expected human capital that a child born today may obtain by the age of 18, considering any risks associated with poor health and education prevailing in his/her country (Kraay, 2019). In column (2), I estimate the effects of the social inclusion of homosexual people on years of schooling. Hanushek and Woessmann (2012) demonstrate that a measure of cognitive skills performs better than the years of schooling when it comes to explaining comparative development across countries. For this reason, I adopt an index of cognitive abilities as the dependent variable in column (3). In the last column, I use an index of national IQs (intelligence) obtained from Lynn and Meisenberg (2010). This metric captures the cross-country variation in cognitive attainment, which is highly correlated with educational attainment (Lynn & Meisenberg, 2010). Using different proxies for human capital skills, I find that the social inclusion of LGBT individuals is positively associated with human capital accumulation. This is consistent with the proposition that LGBT inclusion affects innovation through strengthening the quality of human capital.
The impacts of LGBT inclusion on human capital accumulation also remain robust to controlling for various confounding factors (the online Appendix Table A9). This empirical exercise is similar to that reported in Table 4. It is important to note that LGBT-supportive policies may affect the innovation process through other channels such as national creativity. However, a major challenge with exploring other potential mechanisms stems from the availability of comparable data across the world. In this regard, a potential avenue for future research is to examine other channels of transmission that would help advance our understanding of the relationship between social tolerance towards LGBT individuals and innovation.

Conclusion
It is widely acknowledged that gender disparities in many aspects of empowerment and well-being, including education, health and employment opportunities, are detrimental to economic growth and development (Duflo, 2012;Knowles et al., 2002). Nevertheless, the extent to which discrimination against LGBT people affects economic performance has received scant attention among economists. This is surprising given that substantial and persistent discrimination against LGBT and/or other marginalized groups still exists in many parts of the world. To the extent that fostering social inclusiveness of the LGBT community contributes to enhancing social justice and economic development, we need to obtain a better understanding of the influence of LGBT inclusion on economic performance.
This paper builds upon a recent study by Badgett et al. (2019) that develops an index of LGBT inclusion for a large number of countries. It empirically establishes a positive relationship between LGBT inclusion and income per capita across countries. This paper goes beyond the existing literature by postulating that legal rights and protections afforded to homosexual people play an important role in fostering national innovative capacity, which is a key driver of sustained growth. To test this proposition, I carry out empirical analysis at the country and individual levels. I also employ the economic complexity index as a novel measure of innovation to address several concerns about using conventional innovation metrics. The baseline results, obtained from estimating cross-country OLS models, lend credence to a positive link between LGBT inclusion and innovation. Additionally, the main findings withstand a wide range of robustness analyses.
To rule out a major concern that the cross-country evidence is confounded by unobserved country-specific factors, I perform individual-level analyses using data from the World Values Survey. The subnational evidence reveals that surveyed respondents who self-report tolerance towards homosexual acts have positive attitudes towards science and technology, new ideas and creativity, adventure and taking risks, and changes. The results are insensitive to accounting for a variety of individual-and country-level characteristics. Having established a positive link between the social inclusion of LGBT people and national innovative capacity, this paper examines a key mechanism underlying this relationship. It provides suggestive evidence that LGBT inclusion exerts a positive influence on human capital accumulation, which acts as a catalyst for innovation.
To conclude, this research documents strong and robust positive effects of LGBT inclusion on innovation. Nevertheless, my findings by no means suggest that crosscountry differences in innovative capacity are exclusively attributed to tolerance towards