CULTURE, INSTITUTIONS AND THE WEALTH OF NATIONS

CULTURE, INSTITUTIONS AND THE WEALTH OF NATIONS

                    Yuriy Gorodnichenko                                   Gerard Roland

University of California, Berkeley and NBER                                      University of California, Berkeley, NBER and CEPR

Abstract: We explore the link between the individualism-collectivism dimension of culture and innovation and long-run growth. We argue that a more individualist culture leads to more innovation and to higher growth because of the social status rewards associated with innovation in that culture. For our baseline estimates, we use data on the frequency of particular genes associated with collectivist cultures, as well as a measure of distance in terms of frequencies of blood types, and historic prevalence of pathogens to instrument individualism scores. The relationship between individualism and innovation/growth remains strong even after controlling for institutions and other potentially confounding factors. We also provide evidence consistent with two-way causality between culture and institutions.

Acknowledgments: We thank Vladimir Asriyan, Dominick Bartelme, Insook Lee, and Donna Kim for excellent research assistance. We are grateful to three referees for extensive comments. We benefited from discussions with Philippe Aghion, Yann Algan, John Bonin, Olivier Coibion, Lewis Davis, Lena Edlund, Fred Finan, David Laitin, Amir Licht, Edward Miguel, Gerard Padro-I-Miquel, David Romer, Guido Tabellini, Daniel Treisman, Luigi Zingales, and seminar participants at UC Berkeley, the Booth School of Business, Michigan, the Harvard-MIT Development seminar, IMT Lucca, the University of Sienna, Stanford, Princeton, UCLA Anderson School of Business, Sciences Po Paris, the European University Institute, Facultés Universitaires Notre Dame de la Paix in Namur, Solvay Business School of Economics and Management, the “Macroeconomic Across Time and Space” conference, the Berkeley Center for Political Economy conference on Endogenous Institutions, the NBER Workshop in Political Economy and the National Academy of Sciences Sackler Colloquium on “Dynamics of Social, Political, and Economic Institutions”, conference participants in the First Annual conference of Belgian economists in 2010 and the 2011 AEA meetings. We are especially grateful to Romain Wacziarg for his comments and suggestions, to Laura Fejerman from UCSF for clarifications on genetic questions, and to Diego Comin and Marti Mestieri for sharing their results on the margins of technology diffusion. Gorodnichenko thanks the NSF, the Sloan Foundation, and the

Hellman family fund for financial support.                                                               

1. Introduction

One of the central questions in economics of growth and development is why disparities in income and development across countries are large and persistent. Despite decades of research, this question continues to puzzle the profession as the bulk of the difference is attributed to variation in productivity, a residual component not accounted by observed factors. It is widely perceived that the key conduit of economic growth and productivity enhancements is innovation that brings new goods and services to the economy as well as new ways to produce existing goods and services.  In this paper, we argue that individualist culture plays a key role in stimulating innovations and hence in explaining long-run economic growth, alongside with other important factors such as institutions and human capital.   

The idea that culture is a central ingredient of economic development goes back to at least Max Weber who, in his classical work “The Protestant Ethic and the Spirit of Capitalism,” argued that the protestant ethic of Calvinism was a very powerful force behind the development of capitalism in its early phases.  Weber saw culture as the driving force behind differences in economic development. His theory was in direct opposition to Karl Marx’s thesis that culture is determined by the level of economic development and by the economic interests of the various social classes. Although Landes (1998) and others have argued that culture played a fundamental role in explaining the wealth of nations, and the literature on the economic effects of culture is growing fast, there has so far been little systematic work examining theoretically and empirically the effect of culture on long-run growth and development. 

To be clear, we define culture as the set of values and beliefs people have about how the world (both nature and society) works as well as the norms of behavior derived from that set of values. This definition highlights that culture affects not only social norms but also economic behavior such as the propensity to save or to innovate and many other economic decisions such as fertility choices, investment in education, charitable contributions or the willingness to contribute to public goods. Culture is directly related to institutions, broadly defined, in the sense that culture, like formal political or legal institutions as defined by North (1990), imposes constraints on individual behavior. 

In our analysis in this paper, we focus on only one dimension of culture that may be relevant for long-run growth: individualism versus collectivism.^[1] Individualism is a cultural trait that emphasizes personal freedom and achievement. Individualist culture therefore awards social status to personal accomplishments such as important discoveries, innovations, great artistic or humanitarian achievements and all actions that make an individual stand out. In contrast, collectivism emphasizes embeddedness of individuals in a larger group. It encourages conformity to a group, loyalty and respect to one’s superiors, and discourages individuals from dissenting and standing out.^[2] Although one may obviously contemplate other cultural aspects, the individualism-collectivism distinction is considered by cross-cultural psychologists to be the main dimension of cultural variation (see Heine, 2007) and it has potentially important economic effects. For example, Greif (1994, 2006) uses this distinction in his path-breaking work showing strong effects of culture on economic outcomes.

Several main elements of the difference between individualism and collectivism play a role in our theory. Because individualism emphasizes personal freedom and achievement, it awards social status to personal accomplishments such as important innovations. On the other hand, individualism can make collective action more difficult because individuals pursue their own interest without internalizing collective interests. Collectivism makes collective action easier in the sense that individuals internalize group interests to a greater degree. However, it also encourages conformity and discourages individuals from dissenting and standing out. This framework implies that individualism should encourage innovation more, everything else equal, but collectivism should have an advantage in coordinating production processes and in various forms of collective action.³ Despite this trade-off, we argue that the advantage of individualism has a dynamic effect, whereas the advantage of collectivism has only a static effect. As a result, the advantage of individualism in innovation dominates over longer horizons, thus giving individualistic culture an edge in long-run economic growth.

We bring the argument to the data by testing the effect of individualism versus collectivism on long-

run growth. Ideally, we would like to have a reliable measure of individualism from centuries ago to see how cultural differences of the past affected long-run growth. Unfortunately, our measure of individualism is from the second half of the twentieth century and exists only as a cross-sectional variable. In principle, this is not necessarily damning for our research if culture changes slowly. Nevertheless, this mistiming in the measurement of culture raises several concerns. In particular, our measure of culture might be endogenous to economic outcomes. Therefore, finding a convincing causal effect of culture on long-run growth would require a valid instrumental variable (IV). It is extremely difficult to find foolproof instrumental variables for cross-country regressions. We have nevertheless come up with several instrumental variables that are jointly strongly suggestive of a possible causal link from individualism to long-run growth. For the first set of instrumental variables, we use information on prevalence of certain genes in a population (the frequency of the S-allele in the serotonin transporter gene 5HTTLPR making people more prone to depression when confronted with stressful events and the frequency of the G allele in polymorphism A118G in -opoid receptor gene creating a stronger psychological pain from social exclusion) as well as historical pathogen prevalence in a particular geographical area. According to recent advances in genetics and psychology, the genetic variables appear to directly affect personality traits and can explain the prevalence of collectivist culture in certain populations. In a similar spirit, strong prevalence of pathogens can incline populations to adopt a collectivist culture.  Chiao and Blizinsky (2009), Way and Liebermann (2010) and others argue that communities with a higher frequency of these two genes and with a higher pathogen prevalence developed social norms to adapt to this genetic and epidemiological environment. These data are good candidates for instrumental variables, and they can be argued to satisfy the exclusion restriction. Specifically, the two genetic variables are not plausibly correlated to income per capita through any other channel than collectivism. One might think that historical pathogen prevalence affected income per capita via health, but health variables that may or may not affect income (e.g., life expectancy) are not significantly correlated with historical pathogen prevalence. Unfortunately, cross-country coverage is limited to approximately 40 countries for the two genetic variables, which are the cleanest instrumental variables one can currently obtain in this kind of work. 

Another instrumental variable that is more widely available worldwide is a measure of genetic distance between the population in a given country and the population in the United Kingdom, which is the second most individualistic country in our sample.^[3] A large literature studying values of descendants of immigrants as a function of the country of origin (see Fernandez, 2010, for a survey) documents that parental transmission of culture is a fundamental determinant of the cultural values of individuals. Obviously, parents transmit their cultural values as well as their genes to their offspring. Populations that interbreed a lot should be genetically and culturally close because a similar parental transmission mechanism is at work in both cases. Therefore, measures of genetic distance can be seen as a proxy measure of differences in cultural values. In this case, when we use genetic distance as an instrumental variable, we do not postulate a causal relationship between genes and cultural attributes such as individualism. We simply exploit the correlation between genetic distance and cultural differences across populations as both genes and culture are transmitted from parents to offspring. Since there are no identified direct genetic causes for why some countries became wealthier than others, genetic distance is very likely to satisfy the exclusion restriction. Furthermore, we use only “neutral” genetic markers that have no direct effect on fitness (i.e., ability to think, run, work, etc.) and thus economic or cultural outcomes.  These neutral genetic markers are very unlikely to be affected by economic outcomes, and thus we can exclude reverse causality in our instrumental variable estimates. We use genetic distance based on frequencies of blood types, which is the genetic information available for the largest number of countries.^[4] A potential drawback of genetic distance is that there could be channels other than individualism through which genetic distance can be indirectly related to long-run growth (e.g., another cultural dimension). Because it is more difficult to argue a priori that genetic distance satisfies the exclusion restriction, we combine this variable with the other instrumental variables mentioned above and apply standard statistical tests for the exclusion restriction. Our measure of genetic distance successfully passes these tests, and one can thus feel comfortable using it as instrumental variable for a set of countries larger than one can cover with the genetic variables mentioned above. 

Conditional on the quality of our instrumental variables, our econometric results suggest a statistically and economically significant effect of individualism on income per worker. According to some of our estimates, a one standard deviation increase in the individualism score nearly doubles income per worker Our results are robust to the introduction of different types of controls and different measures of long-run growth as well as to using dyadic regressions or alternative instrumental variables based on linguistic properties of individualist cultures. Although our estimates are based on cross-country variation, these estimates are also remarkably consistent with regional variation within countries like Italy where there exists considerable cultural variation across regions. In addition, the effects of individualism on total factor productivity and innovation are also very strong, thus suggesting that individualism pushes the technological frontier and thus that the effects we estimate capture more than simple technological diffusion. 

To rule out alternative explanations for differences in economic development and to isolate the effect of individualism on economic development from the alternative channels, we employ a battery of checks and tests. First, we explore how our results vary across subsamples of countries that were differentially exposed to these alternative channels. For example, we report results estimated on a sample of African, Asian or European countries to exclude the possibility that our results are influenced by including countries in Americas and Oceania where colonization by European settlers was particularly important. We find that our results are remarkably consistent across subsamples based on continents (e.g., Asia vs. Europe) or levels of development (e.g., OECD vs. non-OECD economies). We also take into account migrations that have taken place between countries over the last 500 years, exploiting the Putterman and Weil (2010) data, and our results hold if we restrict our sample to countries having roughly the same ethnic composition as 500 years ago.

Second, we introduce controls for alternative determinants of economic development and examine

how our estimates of individualism’s effects vary with the inclusion of these additional controls. For example, we may find strong effects of individualism on economic outcomes because individualism can be correlated with the quality of institutions (e.g., Hall and Jones, 1999, Acemoglu et al., 2001), human capital (e.g., Barro and Lee 2001), legal origin (e.g., La Porta et al. 2008), ethnic fractionalization (e.g., Fearon 2003), speed of technology diffusion (e.g., Spolaore and Wacziarg, 2009, Fogli and Veldkamp, 2012), and remoteness from Europe (e.g., Redding and Venables 2004)—key variables that have been shown to be correlated with economic outcomes we want to explain. We document that controlling for these additional factors does not change our conclusions that individualism explains a significant fraction of variation in economic development. Furthermore, we find that individualism and income per capita continue to be strongly related even in dyadic regressions where we can control for country fixed effects thereby ruling out explanations based on a large class of potentially relevant but omitted variables. Thus, individualism has an effect on economic development that is independent of institutions and of other commonly suggested factors, and our estimates are not driven by any omitted variable bias we could think of.   

In light of these findings, we also examine the interactions between individualism and institutions— measured by the average protection against expropriation risk as in Acemoglu et al. (2001)—using our instrumental variables for individualism and the Acemoglu et al. (2001) settler mortality instrument for institutions. We cannot exclude a two-way interaction, culture affecting institutions and institutions affecting culture. However, when using settler mortality data constructed by Albouy (2012), we find that the link from institutions to culture is much weaker and loses robust significance. This result is consistent with Roland (2004) who argues that culture tends to change more slowly than political or legal institutions and, therefore, might have an important effect on the choice of political and legal institutions itself.   

Third, we examine within-country variation of occupational choices across ethnic groups so that we can further minimize the effects of potentially omitted factors in our cross-country regressions. In particular, our theory predicts that persons from ethnic groups that are characterized as more individualistic should enroll in research oriented occupations, which require independent thinking and deviation from conventional ways of doing things, more frequently than persons raised in the traditions of more collectivist cultures. Using U.S. Census data, we find support for this prediction: people from more individualistic cultures are more likely to become scientists and researchers. 

In short, we examine many other potential channels suggested in the previous literature via which genetic distance might indirectly affect economic outcomes. We find that individualism is still positively related to innovation and long-run growth after controlling for these other potential explanations. While we cannot rule out the possibility of an omitted variable driving both individualism and economic development, one may find it increasingly difficult to propose a plausible, quantitatively important alternative that we did not attempt to control for in our empirical analysis. Together with the evidence, based on cross-cultural psychology, on the effects of the distribution of genetic endowments on collectivist culture, these results show that individualism is empirically relevant for understanding economic development and should be included in theories of economic growth.

Our findings contribute to the nascent literature emphasizing the effects of culture on economic outcomes (see Spolaore and Wacziarg (2013) for a review).  Greif (1994, 2006) modeled the effects of individualist versus collectivist beliefs on contract formation, social stratification and the expansion of markets in the late Medieval trade in the Mediterranean. Bisin and Verdier (2000, 2001) examined the dynamics of intergenerational transmission of cultural preferences taking into account family choices of cultural transmission and effects of social environment. Fernandez (2013) modeled cultural change as Bayesian learning in the context of changes in attitudes towards labor force participation. Tabellini (2008b, 2010) studied how the cultural transmission of values of cooperation can affect the form of institutions, which in turn reinforces norms of cooperative behavior. Ashraf and Galor (2007) model the trade-off between non-conformism and conformism at different stages of development and provide a theory of why China was richer in the Malthusian stage of development but lagged behind in the industrialization stage. Doepke and Zilibotti (2008) developed a model to explain the cultural transmission of the values of the preindustrial middle class (thriftiness, hard work) in the industrialization process as well as their eventual social success and the demise of the landed aristocracy while Corneo and Jeanne (2010) argue that cultural transmission can result in poverty traps. In subsequent work, Doepke and Zilibotti (2013) show how in entrepreneurial societies, innovation and risk-taking create incentives for cultural transmission of values of thrift and risk-taking, which in turn sustain a high level of entrepreneurship and innovation. Fernandez, Fogli and Olivetti (2004), Fernandez and Fogli (2009) and Giuliano (2007) examined the effects of culture on fertility choices, family living arrangements and labor supply decisions. Barro and McCleary (2003) argue that economic growth is affected by religious beliefs (e.g., existence of hell and heaven). Knack and Keefer (1997) considered the effect of social capital on economic performance.  Aghion et al. (2010) found a negative correlation between trust and the level of regulation in societies. Guiso et al. (2003, 2009) examined the effect of trust on economic attitudes and international trade patterns, and Giuliano et al. (2014) investigated the link between geography, genetic distance, transportation costs and economic variables. Tabellini (2008a) and Licht et al. (2007) provide evidence consistent with a causal link from culture to institutions and Jellema (2009) provides evidence consistent with a causal link from cultural practices to a society’s basic achievements (such as the presence of writing, the wheel or money) documented for different cultures in Murdock’s (1967) Ethnographic Atlas. In line with Roland (2004), Murrell and Schmidt (2011) show that in seventeenth century England cultural change preceded the important institutional changes brought about by the Glorious Revolution of 1688.

The rest of the paper is organized as follows. Section 2 presents our argument for how individualism

and economic growth can be related. In section 3, we discuss the data used in our empirical analysis. Section 4 contains our empirical analysis of how individualism can affect economic development. Sections 5 and 6 examine the interplay between individualism, institutions and other factors. In Section 7, we investigate occupational choices of various ethnic groups in the USA. Section 8 makes concluding remarks. 

2. The economic argument

In this section, we synthesize how individualism/collectivism can affect long-term growth and development via innovation and production. Our discussion is intentionally narrative to formulate the argument in general terms (we relegate to Appendix A a simple endogenous growth model, which we find useful in making our argument precise, and in particular in differentiating static and dynamic effects of culture).  Hence, we focus on general themes documented by previous studies in sociology, economics, history, case studies in the business management literature, etc. 

While there is a general agreement that technological innovations are the central conduits of economic growth and development, a central question is how innovation is stimulated. Obviously, monetary rewards from patents, market power, etc. provide strong incentives for innovation. However, there are other important dimensions such as social status that can also compensate innovators for their efforts. Our main hypothesis is that individualistic societies permit and encourage more innovation than collectivist societies by providing a higher social status for individuals making important discoveries. There is ample evidence (e.g. Merton 1973) that social reward with heightened status is the most significant part of the total reward for scientists. Indeed, many probably have dreamed of becoming the first to discover a new element, a new law or a new technology.  By stimulating more innovations, individualism gives a dynamic advantage that can lead to higher economic growth. In contrast, collectivist societies emphasize the role of collective effort and give less social status reward to innovation. They reward conformity more and discourage individuals from dissenting (see e.g. Bond and Smith, 1996).

High status rewards can counteract the disincentive effects of high tax rates because while income and wealth can be expropriated, social status cannot. Thus even if a country has bad institutions with high expropriation risk, there can still be incentives to innovate if there is a high enough status reward to innovation. Clark (2007) argues against the view that institutions are important for long-run growth by pointing to the fact that institutions in England around the time of the Industrial Revolution were no better than in many developing countries today, whose institutional weaknesses are precisely cited as the main cause of their underdevelopment. Bringing individualist culture in the picture, which has been shown by historians to exist in England at least since the thirteenth century (Macfarlane, 1978), the negative effect of predatory institutions on long-run growth can be offset by the social status reward to innovation that is present under an individualist culture. 

The comparative advantage of collectivist societies is hypothesized instead to be on the production side, which almost always involves combining inputs and hence requires coordination of workers/units. 

Such coordination is easier to achieve in collectivist cultures that value harmony, conformity and team effort. For example, Liker (2003) documents that teamwork and consensus building are among defining features of the Japanese way to run business. Efforts to copy the Japanese organization inside U.S. automobile factories however failed in their attempts to catch up with the efficiency of Japanese automobile firms, since American carmakers could replicate lean production but could not imitate Toyota’s culture. Because enhancing an assembly line by improved coordination is likely to run into diminishing returns, the production advantage of collectivism is static and may be interpreted as a level effect.

In addition, while the vast majority of fundamental innovations were made in the U.S. and Western Europe (see for example Harrison, 2004), which have a highly individualistic culture, collectivist countries may be good at incremental innovations. For example, the color TV was invented by RCA, an American firm, but Japan ended up making the best TV sets. Sony invented the walkman which was a great consumer success starting in the 1980s. However, the key invention of the compact cassette was made by Philips, a European firm. Similarly, Sony introduced the VCR but the technology was invented by Ampex, an American firm which was unable to make its VCR affordable to households. One can argue that incremental innovations have diminishing returns (i.e., one can relatively easily improve a cassette player in terms of design and functionality but one needs a radical innovation to create a CD player) and gains from incremental innovations are limited in the long run and, hence, the technological frontier is likely to be pushed by the individualistic societies. Collectivist societies may be able to close some of the gap in technology via the international diffusion of technology, an element that we do not incorporate in our model. However, it is important to acknowledge that this diffusion is a gradual process: growth theories analyzing the diffusion of development emphasize that the tacit and idiosyncratic nature of technological knowledge make it impossible to transplant new technologies costlessly and immediately to other countries. In practice, investments are needed to master an existing technology and adapt it to local conditions (see Aghion and Howitt, 2009, Jones, 2002, Evenson and Westphal, 1995, Grossman and Helpman, 1991). If diffusion of technology from leaders to laggards is gradual, one should thus observe a stationary distribution of income differences with leaders (more individualist and hence more innovative countries) being richer since they are technologically a few steps ahead of laggards.^[5],[6]  

This reasoning can shed new light on episodes of “reversal of fortune”. In the Malthusian stage

when labor is allocated almost exclusively to production of final goods (food, clothes, etc.) and virtually no labor is allocated to innovation, collectivist societies, which enjoy a greater level of coordination, may be richer than individualistic societies. This prediction is consistent with, for example, China being richer, more urbanized and more densely populated than much of Western Europe in 1500. However, as the economy exits the Malthusian stage (e.g. after the Black Plague), the collectivism-individualism difference across cultures starts to play a new and different role. Since individualistic societies grow faster than collectivist societies outside the Malthusian stage, countries with an individualistic culture eventually become richer and thus one may observe a “reversal of fortune”, i.e. those countries catch up and become more affluent than collectivist countries that initially had a higher level of development. 

In summary, there is a trade-off between the benefits and costs of individualism and collectivism. Our overview suggests that the benefits of individualism affect the output growth rate while the costs of individualism affect the level of output. In the long-run, the latter effect, which is dynamic, should thus dominate the former effect, which is static. Hence, despite the short-run trade-off, countries with a more individualistic culture should unambiguously grow faster and eventually enjoy a higher level of output. In what follows, our objective is to explore empirically whether cultural attributes such as individualism/collectivism are strong predictors of incomes, productivity and innovation. 

3. Data

A key question for our empirical analysis is how to measure individualism. A well-known measure of individualism (and other cultural dimensions) at the country level was developed by Hofstede (2001) who initially used surveys of IBM employees in about 30 countries. To avoid cultural biases in the way questions are framed, the translation of the survey into local languages was done by a team of English and local language speakers. With new waves of surveys and replication studies, Hofstede’s measure of individualism has been expanded to 96 countries.^[7] In a nutshell, the individualism score measures the extent to which it is believed that individuals are supposed to take care of themselves as opposed to being strongly integrated and loyal to a cohesive group. Individuals in countries with a high level of the index value personal freedom and status, while individuals in countries with a low level of the index value harmony and conformity. Hofstede’s index as well as the measures of individualism from other studies use a broad array of survey questions to establish cultural values. Factor analysis is used to summarize data and construct indices. In Hofstede’s analysis, the index of individualism is the first factor in work goal questions about the value of personal time, freedom, interesting and fulfilling work, etc. This component loads positively on valuing individual freedom, opportunity, achievement, advancement, recognition and negatively on valuing harmony, cooperation, relations with superiors.⁹ This index measures quite well the notion of individualism given above. Similarly, the emphasis on harmony, cooperation and good relations with superiors fits well with the notion of collectivism given above and strongly suggests greater capacity at coordination within the group but also a stronger sense of conformity and a fear of sticking out. Although Hofstede’s data were initially collected mostly with the purpose of understanding differences in IBM’s corporate culture, the main advantage of Hofstede’s measure of individualism is that it has been validated in a number of studies.^[8] For example, across various studies and measures of individualism (see Hofstede (2001) for a review) the United Kingdom, the USA and Netherlands are consistently among the most individualistic countries, while Pakistan, Nigeria and Peru are among the most collectivist.  Figure 1 represents a world map of Hofstede's individualism scores. 

The causality between individualism and economic outcomes can a priori flow in both directions. For example, as we have argued above, more individualist countries may be wealthier because individualism fosters innovation. On the other hand, one might reason that a more affluent economy can support a more individualist culture. Indeed, there is a long tradition in social sciences starting with Marx claiming that economic development affects a country’s culture.

To address this potential endogeneity problem, we use a number of instrumental variables. We first use genetic and epidemiological data which the recent literature in cross-cultural psychology has directly linked to collectivism. A first set of data is from Chiao and Blizinsky (2009) who document a strong correlation between collectivism and the presence of a short (S) allele in the polymorphism 5‐HTTLPR of the serotonin transporter gene SLC6A4. This allele is known in psychology to put individuals at greater risk for depression when exposed to life stressors. The mechanism linking individual genetic traits and culture is that a collectivist culture protects individuals from these stressors by embedding them more strongly in communities with strong social links thus providing strong psychological support networks. These data are complemented with data assembled in Inglehart et al. (2014) for a total of 43 countries. We also use data from Way and Liebermann (2010) showing that collectivism is also strongly correlated with the G allele in polymorphism A118G in the -opoid receptor gene that leads to higher stress in case of social rejection. Way and Liebermann (2010) also reason that collectivist culture can be seen as providing psychological protection from social rejection. These data are complemented by various other sources (see Appendix F) for a total of 34 countries. Finally, we use epidemiological data put together by Murray and Schaller (2010) for 96 countries on pathogen prevalence, complementing earlier work by Fincher et al. (2008).^[9] Given a strong correlation between pathogen prevalence and collectivism, Fincher et al. and Murray and Schaller argue that stronger pathogen prevalence pushed communities to adopt more collectivist values emphasizing tradition, putting stronger limits on individual behavior, and showing less openness towards foreigners. Collectivism is thus understood as a defense mechanism created to cope with greater pathogen prevalence. 

 We then combine each of these instrumental variables with a measure of genetic distance between people in different countries and perform statistical tests of overidentification to check whether genetic distance meets the exclusion restriction. To the extent that culture is transmitted mainly from parents to children, so are genes. Thus, genetic markers can be used as a proxy for cultural markers and this instrumental variable should be seen as a proxy measure of cultural transmission. To be clear, we do not postulate a causal effect between genetic distance and cultural distance. Instead, we exploit the correlation between cultural and genetic transmission from patents to offspring. Since economic development is unlikely to affect genetic pools in a matter of a few centuries, one can reasonably expect that genetic distance is a good IV for differences in cultural attributes. These genetic data originate from Cavalli-Sforza et al. (1994) which provides measured genetic markers for roughly 2,000 groups of population across the globe. These data contain allele frequencies (alleles are variants taken by a gene) for various ethnic groups. Since we want to eliminate the feedback from economic outcomes to genetic variation, we focus on neutral genetic markers that are not related to evolutionary fitness, and thus economic performance. Furthermore, as discussed in CavalliSforza et al. (1994), genetic variation for countries not affected by massive colonization since 1500s was largely determined during the Neolithic migration of early humans thousands of years ago.¹² We use the Mahalanobis distance between the frequency of blood types in a given country and the frequency of blood types in the UK, which is the second most individualistic country in our sample.¹³ The geographical distribution of the Mahalanobis distance measure is displayed in Figure 2.¹⁴ Using the frequency of blood types is attractive because, apart from being neutral genetic markers (i.e., different blood types do not cause a higher level of intelligence, output or individualism), the frequency of alleles determining blood types is the most widely available genetic information and thus we can construct the most comprehensive (in terms of

                                                            

¹² Note that the genetic and cultural data were collected predominantly in 1950s through the early 1970s.  On the other hand, our measures of economic outcomes are generally from the 21^st century. This difference in the timing of explanatory/instrumental variables (i.e., culture and genetic variables) and dependent variables (i.e., economic outcomes) helps us to alleviate certain types of endogeneity (e.g., recent strong migration of skilled workers).  ¹³ The advantage of using distance relative to the U.K. is that U.K.’s population is genetically more homogenous than the population in the U.S.A.—the most individualistic in the world—and that the U.K. is often described as the cradle of individualism and the Industrial revolution. Indeed, the share of indigenous (as of year 1500) population in modern U.K. is over 94 percent. Results are very similar when we use distance to the U.S.A. Note also that we get similar results when we use the distance to the most collectivist countries (Guatemala, Pakistan, Mozambique, Tanzania, etc.). ¹⁴ The Mahalanobis distance between a vector x and y picked from distributions X is

/

                                                                                  ,                            Σ

where Σ is the covariance matrix for X.  In our contexts, Σ var _,̅                 _,̅ where A and B denote blood types and c indexes countries. We obtain the Euclidian distance , when Σ is set to the identity matrix. Thus, the

Euclidian distance between country c and the USA is equal to        ,                       _,^̅              _,^̅                _,^̅              _,̅          ^/. 

country coverage) measure of genetic distance.^[10] Another key advantage of utilizing frequency of blood types is that we can exploit alternative sources of information (e.g., Red Cross) about frequency of blood types to corroborate our data from DNA studies. In a series of robustness checks, we also employ aggregate measures of genetic distance constructed in Cavalli-Sforza et al. (1994) and used in Spolaore and Wacziarg (2009).^[11] 

Since the genetic data are available at the level of ethnic groups while our analysis is done at the country level, we aggregate genetic information using ethnic shares of population from Fearon (2003).^[12] Specifically, if we define blood frequency f_bec for blood type b and ethnic group e in country c, then the

country level blood frequency for type b is calculated as ̅ ∑ where s_ec is the share of ethnic group e in the population of country c.^[13]  In a robustness check, we also employ an instrumental variable based on linguistic peculiarities of individualistic cultures. Specifically, in languages where the pronoun cannot be dropped in a sentence there is a greater differentiation between the individual (first person of the singular) and the community, whereas in languages where pronouns can be dropped there is less emphasis on such a differentiation. Kashima and Kashima (1998) and others document that prohibition of pronoun drop is strongly correlated with individualism.^[14] This instrumental variable was used in Licht et al. (2007), Tabellini (2008a) and other papers studying the effects of culture on socioeconomic outcomes. 

The sources of data on economic outcomes are standard. We take income per worker data in 2000 from the Penn World Tables (version 6.3). To control for differences in factor endowments, we use data on total factor productivity (TFP) from Hall and Jones (1999) and Jones and Romer (2010). These two measures have been widely used as measures of long-run growth in the literature. 

Since the main conduit of individualism’s effect on growth in our argument is innovation, we proxy for the intensity of innovations with the innovation performance index and the log patents per million population from Economist Intelligence Unit (2007, 2009; henceforth EIU). EIU constructs patents per million population  as the sum of patents granted to applicants (by residence) from the 82 economies by three major government patent offices—the European Patent Office, the Japanese Patent Office, and the US

Patent and Trademark Office. The data are averaged over 2002-2007. Although the use of patent data has a number of problems, this is the single best available measure for innovation outputs.  The innovation performance index incorporates information on patents and alternative indicators of innovation output such as royalty and license fee receipts as a percentage of GDP, high-technology manufacturing output per head, high-technology services output per head, the number of citations from scientific and technical journals, etc. As documented in EIU (2007, 2009), these measures are highly correlated with other proxies for innovation performance such as UNIDO estimates of the share of medium- and high-technology products in a country’s manufacturing output and its manufacturing exports, and the results of a survey question from the World Economic Forum’s Global Competitiveness Report that asked respondents to rate the extent to which companies were adept at, or able to absorb, new technology. Thus, these measures of innovation are likely to capture salient features of innovative activities across countries. 

The timing of data collection is different across variables (see Appendix G). For example, while frequencies of blood types were collected in the 1940s and 1950s, other genetic data that we use were collected recently. The first individualism scores were constructed in the 1960s for a limited number of countries and the coverage increased gradually since then. Data on historic pathogen prevalence refer to early-to-mid 20^th century, before the epidemiological revolution.  Outcome variables such as output per worker, patents per capita, total factor productivity are generally available for recent years.  Ideally, one would like to have measures of our cultural variable (individualism) and of instrumental variables such as pathogen prevalence before the Industrial Revolution to correctly estimate the effect of individualism on long-run growth centuries later. Unfortunately, such data are not available.  Nevertheless, we think that, the quantitative significance of this mistiming is not likely to be large. First, the central premise of our argument (as is also demonstrated by a large body of research) is that culture is slow-moving so that culture “today” in a given country is similar to what it was in the past, even after centuries of economic development. Second, to strengthen the argument and minimize possible endogeneity of individualism scores, we use genetics-based instrumental variables that are unlikely to change materially since the Malthusian stage. This significantly strengthens our research design. In addition, as we discuss below, the plausible correlation structure of errors is such that our estimates could understate the strength of the relationship between individualism and the outcome variables. Third, historic pathogen prevalence data were constructed for the era preceding the revolution in the medical treatment of malaria, typhus, tuberculosis and other major contagious diseases. Hence, this measure is likely to provide a good proxy of pathogen prevalence for earlier periods. 

4. Baseline econometric specification and results

Our argument predicts that more individualistic countries should be more affluent since individualism encourages innovation.  Consistent with this prediction, Figure 3 shows that countries with more individualistic cultures enjoy higher levels of income, TFP and rates of innovation. Also, innovation is strongly positively correlated with income and TFP (Figure 4). These raw correlations, some of which were reported earlier in Hofstede (2001), are informative but they do not control for other factors and cannot be interpreted as causal relationships.^[15] 

To address these concerns, we employ the following basic econometric specification:

                                                                                                                                                               (1)

where i indexes countries, Y_i measures an economic outcome (e.g., log income per worker), IND_i is a measure of individualism,  is a vector of control variables and e_i is the error term.^[16]  The vector  includes commonly used controls for geography such as countries’ longitude and latitude, a dummy variable for being landlocked, and a set of dummy variables for continents. In addition to this standard set of geographic controls, we include the percentages of population practicing major religions from Barro and McCleary (2003) to ensure that our results are not driven by differences in the composition of people following various religions. 

As discussed above, we use several instrumental variables to deal with reverse causality in equation

(1). Figure 5 shows that countries with more individualistic cultures are genetically less distant from the U.K. The converse applies to countries with collectivist cultures. At the same time, countries with individualist and collectivist cultures are genetically distant from each other. Note the strong negative correlation between genetic distance (computed relative to the U.K. which has the second most individualistic culture) and individualism. 

Table 1 presents the OLS and IV estimates for the basic specification (1) where the dependent variable is log income per worker. In the basic OLS regression (column (1)), the coefficient on individualism is positive and significant. Specifically, a one standard deviation increase in individualism (say from the score of Venezuela to Greece, or from that of Brazil to Luxemburg) leads to a 66 percent increase in the level of income, which is a large effect.  Taking the blood distance to the U.K. as instrument (column (2)) yields a somewhat larger estimate of the coefficient on individualism. In columns (3) and (4), the key instrument is the frequency of the short (S) allele in the polymorphism 5‐HTTLPR of the serotonin transporter gene SLC6A4, which makes people more prone to depression when facing stressful events. In columns (5) and (6), the key instrument is the G allele in polymorphism A118G in the -opoid receptor gene that leads to higher stress in case of social rejection. Finally, columns (7) and (8) use historical pathogen prevalence as an instrument. The first stages for all IV regressions (columns 2 to 8) are strong. By and large, the estimates are similar across the specifications. 

Note that when we include blood distance as a second instrumental variable (columns 4, 6 and 8), the estimated coefficient remains similar in magnitude to what one can obtain using instruments separately. Furthermore, the overidentifying restriction tests cannot reject the null of instrumental variables being correctly excluded at any standard significance level. The results of the overidentification test, together with the similar magnitudes of the coefficients, strongly suggest that blood distance picks up the link between genetic distance and cultural distance along the individualism-collectivism dimension. Spolaore and Wacziarg (2009) interpreted instead genetic distance as a proxy for barriers to the diffusion of knowledge. But how geographical distance—a prominent barrier to diffusion—affects individualism should not be systematically related to how e.g. a particular variation in the serotonin transporter gene SLC6A4 affects individualism. While our measure of blood distance might a priori reflect such barriers, the variation in SLC6A4 cannot be reasonably suspected of directly reflecting barriers to the diffusion of knowledge. If our measure of blood distance were to be interpreted as a measure of barriers in the diffusion of knowledge, then the coefficient on individualism in the second stage regression should be quite different when we use two instruments (blood distance and the other genetic/epidemiological variable) compared to when we use only one instrument (the other genetic/epidemiological variable). Indeed, if that were the case, these different instrumental variables would pick different aspects of the variation in individualism, thus leading to a different estimate and also to a rejection in the test of over-identifying restrictions. As we can see from Table 1, however, this is not the case. The results in Table 1 are thus consistent with both instrumental variables picking up approximately the same aspects of the variation in individualism, thus confirming our interpretation of blood distance as a proxy for cultural distance. These clarifications are important, because even if the instrumental variables used in columns 3 and 5 are much more directly related to individualism and collectivism, they are currently available only for respectively 43 and 34 countries. Given that our blood distance instrument covers many more countries and it passes the overidentification test in Table 1 despite its potentially lower plausibility as an instrument, for the rest of the paper we will use blood distance as an instrumental variable so that we can have additional robustness checks with more controls and subsamples as well as more statistical power to reach sharper conclusions. 

Table 2 performs some first robustness checks. In row (1), we use as instrument for culture the Mahalanobis distance of frequency of blood types A and B in a country relative to the USA. In row (2), we use the frequency of blood types A and B separately so that we do not need to construct a distance measure to any particular country. In row (3), instead of using the Cavalli-Sforza et al. (1994) data on blood types, we use the data from the Red Cross. Although the Red Cross data are available for a much smaller set of countries, it does not require us to use ethnic shares in population to aggregate genetic data to the country level. In rows (4) and (5), we use the genetic distance data used by Spolaore and Wacziarg (2009). Their data also come from Cavalli-Sforza et al. (1994). In contrast to our blood distance, Spolaore and Wacziarg (2009) take genetic distances calculated by Cavalli-Sforza et al. (1994) for a larger set of genes. However, with a larger set of genes, the distance can be computed for only 42 ethnic subgroups of the world population. Similar to our approach, Spolaore and Wacziarg (2009) aggregate ethnic data to the country level using shares of ethnic groups in country populations. Row (6) uses the prohibition of pronoun drop as an instrument whereas in row (7), it is used as an instrument together with blood distance. In all cases, results are similar to the results we obtained for the baseline specification of Table 1. 

As an additional robustness check, rows (8)-(10) report results for a series of dyadic regressions that reduce the influence of using the U.K. as the comparison point for genetic distance. In particular, we estimate the following specification: 

                      Δ Δ                                 ∑                                          ∑                                                                          (2)

where Δ ≡lnln is the log difference in income per worker in country i and country j, Δ

                           is the difference between individualism scores in country i and country j,                  is an

indicator variable equal to one if  and zero otherwise,  is a set of additional controls (if included). We instrument Δ with the blood distance between countries i and j. We find that the estimates of  continue to be highly significant and positive even after controlling for country fixed effects and geographical distance between countries. These last two results are important because country fixed effects control for other possible country-specific omitted variables, and controlling for geographical distance ensures that our results are not driven by factors related to diffusion of development.

In Table 3, we use different dependent variables, in line with our hypothesis that innovation is the channel through which cultural differences lead to differences in long-run growth. Panel A presents for comparison regressions with log income per worker as the dependent variable.  The first four columns are OLS and the next four are IV regressions. In columns (2), (4), (6) and (8) we introduce continental dummies and in columns (3), (4), (7) and (8), we introduce geographical controls for landlocked countries, absolute values of country longitude and latitude and controls for the percentages of population practicing major religions in a country to make sure the effect of culture is not driven purely by religion. Panel B uses TFP from Hall and Jones (1999) as the dependent variable. We know indeed from their work that the main factor behind differences in incomes is variation in the level of TFP across countries. Panel C uses newer TFP data from Jones and Romer (2010). We find strong and positive effects of individualism on productivity. A one standard deviation increase in the individualism score leads to a 31 to 66 percent increase in TFP.  Note that the effect on TFP is smaller than the effect on income. This should be expected since differences in income per worker are due to differences in factor accumulation on top of differences in TFP. 

Finally, we perform a more direct test of our theory by regressing measures of innovation on individualism (Table 3, Panels D and E). With and without controls, we see a strong robust effect of individualism, confirming the channel going from individualism to innovation and to income and productivity. This finding is consistent with experimental evidence (e.g., Goncalo and Staw, 2006) showing that groups populated by individualistic persons generate more creative solutions to problems than groups populated by collectivist persons. Importantly, this finding also highlights that although countries may achieve a larger level of total factor productivity via diffusion of existing knowledge and willingness of people in individualistic cultures to accept new goods/services as well as new ways of producing goods/services, individualism affects the creation of knowledge.^[17] In other words, individualism not only helps countries to approach to the technological frontier, it also pushes the frontier. 

To assess whether the magnitudes of individualism’s effect on economic outcomes are plausible, consider differences in economic outcomes in Italy’s South and North, which is a prime example of the importance of cultural effects. In his classic book, Putnam (1994) argues that the North of Italy is culturally similar to Switzerland and Germany (the individualism score for Switzerland is equal to 68) while the South of Italy is similar to Spain (the score is 51). Our baseline regression results (column (8) in Table 3, panels A and C) predict that the difference in income per capita and TFP between Italy’s North and South should be 0.0291749.3% and 0.0181730.6% respectively. According to Italy’s statistical office income per capita in Southern regions is about 50% smaller than income per capita in Northern regions. Using the methods developed in Hall and Jones (1999), Aiello and Scoppa (2000) estimate the difference in TFP across two regions to be 27%. Thus predictions made from our cross-country regressions are remarkably similar to within-Italy variation in incomes and productivity and validate our parameter estimates.^[18] 

Note that China is not at all an outlier in our estimations. Despite its very fast growth for the last

thirty years, China still remains relatively poor. Panel A of Figure 3 illustrates that China is approximately half a log point below the regression line so that China’s income per worker would have to grow by more than 50 percent before it is on the regression line.  Even if China’s income per worker were as high as that of Mexico (approximately halfway between triple and quadruple of the actually observed income per worker in China), China would continue to look like a fairly typical data point in Panel A of Figure 3. 

5. Exploring other channels

By focusing on the individualism/collectivism dimension, specification (1) does not include other potentially important determinants of economic development. To the extent these determinants are positively correlated with individualism, one may overstate the contribution of individualism to long-run growth. To address this concern about omitted variables, we explore in this section how controlling for these potentially important factors alters our conclusions. 

A first major potential objection is that our results reflect migration patterns from the colonization era in which the Americas and Oceania were settled by European immigrants. One may also be concerned that our results are driven by a set of countries that for historical reasons were disadvantaged in economic development. If our theory explains income differences at the global scale, it is reasonable to expect our theory to explain income differences within continents where countries may be more similar. These concerns are important because, for example, Albouy (2012) argues that the theory of institutions as the fundamental cause of economic development has weak or no empirical support when tested within continents. Table 4 reports regression estimates for each continent separately and for OECD economies.^[19] By and large, we confirm our basic finding that individualism leads to higher income per worker. Even if we focus on OECD countries or relatively more developed countries in Europe and the Americas, individualism can explain a large fraction of variation in income. Although the coefficient on individualism is somewhat smaller for the subsample of developed countries, it does not necessarily mean that culture is less important. It simply reflects the fact that variation in incomes and individualism is more compressed in these countries and thus, with less variation in our key variables, measurement errors can have a stronger attenuation bias. This observation can also explain why the estimated coefficients are the largest for Africa where countries are extremely diverse in the level of development and individualism. For example, Morocco and Bhutan have individualism scores similar to those for Argentina and Spain, whereas Mozambique, Ghana, and Burkina Faso have some of the lowest scores in the world. Column (5) gives results for Africa, Europe and Asia where there was no massive migration of European settlers. Note that the coefficient in the IV estimation is even larger than in the results from Table 1 where the Americas and Oceania were included. In summary, our results are not driven by a particular continent and the effect of individualism is significant also within continents.  

Another concern is related to migration flows that have happened over the centuries across countries in a continuous manner. For example, countries with bigger economic opportunities could have attracted migrants from places that also happened to have more individualistic cultures. To address this concern, we use the Putterman and Weil (2010) data on migration flows between 1500 and 2000.  In column (1) of Table 5, we first replicate baseline OLS and IV regressions for our full sample. Then we restrict gradually the sample to those countries whose share of indigenous population as of 1500 in today’s population is larger than respectively 80 percent (columns (3) and (4)), 90 percent (columns (5) and (6)), and 95 percent (columns (7) and (8)). We thereby eliminate countries that have witnessed large migration flows since 1500. We find that the coefficients remain highly significant as we restrict the sample and, if anything, the point estimates get larger.  In summary, the results of Table 4 and 5 rule out the idea that our results reflect only migration patterns (most importantly, European settlers in the colonization period of the last 500 years) or the effects of being European (i.e., differences in individualism are not about Europe vs. the rest of the world).

A second major objection could be that individualism proxies for other forces of economic development. For example, a popular alternative explanation of economic development is the quality of institutions (see e.g. Acemoglu et al., 2001). Because cultural attributes and institutions are correlated and it is possible that culture simply captures the effect embodied in institutions, one needs to establish whether individualism has an effect separate from the effect of institutions. To differentiate effects of institutions and individualism, we augment the baseline econometric specification (1) with the average protection against expropriation risk between 1985 and 2009, a measure of institutions used by Acemoglu et al. (2001) and the majority of previous papers studying the effects of institutions on socioeconomic outcomes:^[20]

                                                                                                                                                                     (3)

where INST_i is a measure of institutions in country i. Estimates of equation (3) (see Table 6) show that individualism remains significant even after including institutions in the OLS and IV specifications. Individualism thus has a robust effect that is separate from institutions. The size of the estimated coefficient remains substantial. A one standard deviation increase in the individualism score leads to a 53 to 79 percent increase in the level of income without instrument for institutions and  to a 86 to 99 percent increase in the level of income when the institutional variable is instrumented using the settler mortality variable as in Acemoglu et al. (2001).²⁶ 

Note that the size of the effect of individualism on income remains fairly robust to including institutions and other controls. We cannot say the same for the institutional variable which is rather sensitive to including controls and individualism in the regression. Furthermore, the coefficient on institutions does not increase in the IV estimation (panel B) once individualism is included but rather tends to decrease, which was not the case in Acemoglu et al. (2001). Finally, the estimated effect of institutions is particularly unstable when we apply the correction for settler mortality as in Albouy (2012) and include individualism in the regression (columns 8 and 9 in panel B). We observe similar results (not reported) when we use innovation or TFP (rather than income per worker) as the dependent variable. In summary, there is an important contribution of culture to economic development that is independent of institutions. One can state that culture explains income differences across countries at least as much as institutions. However, data for institutions and cultural variables are available for a limited number of countries: Panel B has 39 observations because of the imperfect overlap between coverage of settler mortality and individualism data. Future research should re-examine these results when more data become available.  

Table 7 reports estimates of the effect of individualism on our outcome variables when we control for a variety of additional factors that have been investigated in the empirical literature on growth and other channels that might link individualism or genetic distance to growth. For example, individualism may be correlated with trust, which is often interpreted either as a cultural norm that reduces transaction costs or as a measure of social capital, which reflects the density of social networks and a culture of participation and citizenship. Using generalized trust constructed from the World Values Survey, a variable that has been widely used in the social sciences literature, we find some positive correlation between log income per worker and trust, but this relationship is not robust. Once we regress log income per worker on both individualism and trust, trust ceases to be significant while individualism remains robustly significant and quantitatively important.^[21] In Gorodnichenko and Roland (2011), we look at a large number of alternative available measures of culture (including the other Hofstede indicators) and conclude that there is no significant or robust effect on growth from cultural dimensions that are independent from the individualism-collectivism dimension. When analyzing the effect of culture on growth, the individualism-collectivism dimension thus appears to be the most relevant and robust relevant cultural variable. Note that this also further validates our use of genetic distance as a valid instrument for individualism since other cultural channels are either nonrobust or correlated with individualism.

Likewise, ethnic fractionalization, which previous literature found to be associated with weaker institutions and hence lower levels of output, does not appear be a robust predictor of output, patents or productivity. Furthermore, we do not find a statistically significant relationship between ethnic fractionalization—which also proxies for diversity—and output or any material change in the estimates of the coefficients on individualism when we augment this specification with nonlinear terms in ethnic fractionalization (not reported) and, therefore, our results for individualism are different from and not confounded by the diversity effects emphasized by Ashraf and Galor (2013). 

One may argue that individualism is likely to arise only when the level of education is high, and thus that individualism may proxy for the quality of human capital instead of having an independent effect on economic outcomes. To rule out this alternative explanation, we control for the Barro-Lee measure of average years of schooling for people over the age of 15. This variable is only significant in regressions on log TFP (columns (5) and (6)) and its inclusion does not affect the significance of individualism.

Similar to our previous specifications, we also control for average protection against expropriation risks, the share of people with a European descent in 1900, and legal origins, three popular measures of institutional quality. While legal origins and the share of people with a European descent do not have a robust association with our economic outcomes after controlling for other factors, protection against expropriation risks has a strong and robust association with the outcomes. Including these measures as additional regressors, however, does not alter our conclusions about the strong effects of individualism on income, patents, and productivity. 

Genetic distance may reflect geographical distance and thus capture transport costs in international

trade (see e.g. Giuliano et al., 2006), as well as the speed of technology diffusion rather than differences in cultural attributes. To address this concern, we introduce the log of the population-weighted distance of a country from the UK, which proxies for transportation costs from the cradle of the Industrial revolution.  Although this distance variable is negatively correlated with the log of income per worker, when it is combined with the individualism score, it is not statistically significant while individualism remains robustly significant both in the OLS and IV specifications. 

We argue that individualism’s effect on growth works through a higher level of innovation. It is possible, however, that instead of creating new technologies and products, individualism leads to higher income and productivity only or mainly through faster absorption of existing technologies as argued by Fogli and Veldkamp (2012). In other words, diffusion of technologies may be faster in more individualistic societies, hence leading these societies to enjoy higher levels of productivity and income. We have already shown that individualism influences the intensity of creation of new technologies and goods as measured by patents. To further separate these two channels, we control for the extensive margin (the average time lag for a technology to appear in a country since the technology is invented) and the intensive margin (the speed at which a technology spreads in a country) of technology diffusion constructed by Comin and Mestieri (2013). Specifically, we average the values of a margin for each country across 25 technologies (e.g., internet, synthetic fiber, cars)^[22] and use these averages as additional regressors. If the diffusion channel matters more than the innovation channel, we should observe individualism becoming statistically and economically insignificant once we control for measures of the speed of technology diffusion. If the opposite is true, then including measures of the speed of technology diffusion should have no material effect on the estimated coefficients on individualism. We find that while these two margins of diffusion are strongly correlated with our outcome variables, the margins are not systematically correlated with the outcomes once we control for other country characteristics. Moreover, the coefficients on individualism are barely affected, suggesting that individualism matters more because of the innovation channel than because of the speed of diffusion channel.  Again, this is clear evidence consistent with the channel we posited in this paper between individualism and long-run growth.

The control function approach adopted in Table 7 is likely to bias the estimate of individualism’s effect downward. Indeed, many of the controls (trust, education, etc.) are potentially endogenous but we do not have credible instruments for all of these variables and the data sets for which all instruments could overlap would be considerably smaller, as was already the case in Table 6 when combining only instruments for culture and institutions. These potentially endogenous regressors are likely to be correlated with our instrumental variables and the error terms across first- and second-stage regressions are plausibly positively correlated. Therefore, by not instrumenting these potentially endogenous variables, our IV regressions in Table 7 are likely to attribute some of the effects of individualism to these other regressors (see Appendix B for a more formal derivation of this result).  Thus, one could interpret our estimates on the individualism coefficients as conservative and, if we find a significant positive effect of individualism on growth, the true effect is likely to be larger.

In summary, although genetic distance may be correlated with non-cultural factors or cultural

factors other than individualism, none of the popular alternatives alters our main result that individualism plays an important role in determining economic development. 

6.  Causal channels between culture and institutions.

Given that individualism plays a role that is independent of institutions, we naturally want to examine whether individualism affects institutions or vice versa. Arguments could go both ways. One can reason that culture shapes institutions. When institutions are put in place, they correspond to a view of how the world works and are thus based on culture. The political transformations that took place in the Western world between the eighteenth and twentieth century from absolute monarchy and autocracy to republican and democratic regimes can be seen as based on the values of the Enlightenment that go back to the Renaissance period and the rediscovery and reappropriation of the Greek culture of rationality and democracy. The French revolution led to the abolition of monarchy and profound institutional changes that were inspired by the ideals of the Enlightenment. In contrast, large-scale revolts in China throughout its history led at best to the replacement of one emperor/dynasty by another one (Finer, 1997) because the Chinese imperial system was in line with the Confucianist culture and its view of the “good emperor” as father figure with the associated moral duties towards the people. Within that culture, dissatisfaction of the population tended to be interpreted as resulting from having a “bad” emperor. Replacing the latter with a “good” emperor who would behave according to the Confucianist moral cannons was thus seen as the appropriate response. Culture can thus be argued to affect institutional choices of a society. 

However, one can also make a case in favor of an opposite causal channel. People lived for centuries under empires characterized by different institutional organizations, be it the Chinese imperial system, the Ottoman Empire or the Austro-Hungarian Empire. The administrative apparatus of empires (as well as of smaller political entities) made it possible to influence the world view of people living within its boundaries, usually by the spreading of religions such as Islam under the Ottoman Empire or Catholicism under the Austro-Hungarian Empire.^[23] For example, Confucianism became widespread in China in part because it was adopted as the official ideology of the empire as early as the Han dynasty. Institutions can thus be argued to have affected the spread of a specific culture, and thus also the degree of individualism and collectivism. 

We thus test for the existence of two causal channels: from culture to institutions and from

institutions to culture. For this test we employ two econometric specifications:

                                                                                                                                                                   (4)

                                                                                                                                                                  (5)

where INST is a measure of institutions (i.e., protection against expropriation risk as in Acemoglu et al. (2001)), IND is a measure of individualism, X is a vector of controls, and e and u are error terms. In equation (4), individualism is instrumented with the blood distance we constructed before. In equation (5), protection against expropriation risk is instrumented with settler mortality.  If we find that  is significant while  is not, culture can be interpreted as causing institutions. If  is significant while  is not, institutions can be interpreted as causing culture. Joint significance of  and  can be understood as causation flowing both ways. The validity of these results will of course depend on the validity of the instruments used.

The results for equation (4) are reported in Panel A of Table 8. The effect of individualism on the strength of economic institutions is positive and significant thus implying a flow of causality from individualistic culture to institutions.^[24] This finding corroborates Tabellini (2008a) and Licht et al. (2007) who found similar results using different measures for culture and institutions. We report results for equation (5) in Panel B of Table 8. They indicate that causality also flows from institutions to culture when we use as instrument settler mortality from Acemoglu et al. (2001). However, the effect of institutions on culture ceases to be significant once one introduces settler mortality from Albouy (2012) and the first stage fit becomes quite poor.  Hence, the effect of institutions on culture might be less robust than the other way round. One must however be careful in interpreting all these results since they are based only on 39 observations, the countries for which the data on culture and institutions and their instruments overlap.  In short, culture appears to have a causal effect on institutions and is itself influenced by institutions, although the latter direction of causation is less clear cut than the former.

7. Within-country evidence 

Cross-country analysis may fail to control fully for differences in institutional factors or other sources of cross-country differences. However, we can examine the effect of culture within a given country, thereby holding institutional factors constant. Furthermore, by exploring within country variation, we can rule out alternative explanations based on differences in diffusion costs, geography, etc. Specifically, our model predicts that more individualistic cultures should ceteris paribus stimulate persons to choose research-oriented occupations that require independent thought and deviation from traditional ways of doing things. For this analysis, the USA is a particularly attractive research object since this country has many ethnicities and occupational opportunities that are relatively open for peoples of all origins and cultures. This special feature of the USA has been exploited by the epidemiological approach to culture pioneered by Fernandez and coauthors (see Fernandez, 2010), Giuliano (2007), Algan and Cahuc (2010), and others.

We use ethnicity, age, gender, birth place, educational attainment from the 1 percent and 5 percent public micro data (IPUMS) of the U.S. Census in 1970 and 2000 respectively. For the 2000 census, ethnicity is based on the respondent’s self-reported country of ancestry. For the 1970 census, ethnicity is based on the respondent’s response about the father’s birth place. Our sample includes only employed males who are aged between 25 and 60 and have non-missing information on ancestors (country of origin). The reason why we constrain the sample only to individuals with non-missing ethnicity information is because we then focus only on individuals who associate themselves with a particular culture (which could be different from the American one) and are likely to observe the traditions of their original cultures. We exclude females, unemployed and other ages to minimize the various possible selection effects.

We consider several sub-samples. The first sample split is determined by whether an individual is born in the USA so that we can attenuate the effects of high-human-capital migration into the USA (intuitively, high-human-capital migration from countries with low level of individualism could create a sample of highly individualistic U.S. persons from these countries, and thus the difference between persons from individualistic cultures/countries and collectivist cultures/countries would not be reflected in the sample). The second sample split is based on educational attainment: all persons vs. persons with a bachelor (or higher) degree. The higher is the level of educational attainment, the smaller should be the effect of differences in initial conditions and abilities across ethnicities on the estimates. 

Our approach has two steps. In the first step, we estimate the following probit:

                                       ΦΣ                            error                                                                                 (6)

where i and k index individuals and ethnic groups, ROO is a dummy variable equal to one if an individual has a research oriented occupation and zero otherwise, D is a set of dummies of each ethnicity, and the vector X includes controls such as age, age squared, and a set of dummies for educational attainment.  The omitted category in the set of ethnic dummy variables is British since the U.K. is the second most individualistic country in our sample. 

In the second step we estimate the following specification by least squares:^[25]

                                                   error                                                                                                  (7)

where  is the set of estimated coefficients  in regression (6) and  is Hofstede’s individualism score. Our theory predicts that  should be positive. 

Table 9 presents estimates from regression (7). Note that the estimate of  is larger when we constrain the sample only to U.S. born persons and when we consider persons with a certain educational threshold. The estimates of  indicate that persons coming from individualistic cultures are more likely to take research-oriented occupations than persons from collectivist cultures. Obviously, these estimates do not prove that persons from individualist cultures are more successful at innovation than persons from collectivist cultures but they clearly suggest that there is a cultural component at work in the choice of such occupations. 

8. Concluding remarks

We consider the hypothesis that individualism/collectivism can influence economic outcomes such as innovation and long-run growth and test this hypothesis using cross-country and micro-level data. Our evidence documents a strong relationship between these cultural attributes and economic outcomes even after controlling for a broad range of alternative explanations. Although one should be cautious in interpreting our results as causal—we rely on non-experimental data and therefore cannot rule out omitted factors completely—our instrumental variable estimates as well as a large battery of checks and tests provide preponderance of evidence suggesting a plausible causal interpretation of this relationship.  

There are clearly many pitfalls that should be avoided in interpreting our results. By no means

should our (or other) research on economic effects of culture be seen as implying a “ranking” of cultures in the world or a call for cultural revolutions. On the contrary, this research is aimed to better understand the tradeoffs implied by different cultures which are deeply rooted in history and change very slowly. We must better understand the world we live in and the values and beliefs upon which people in different countries base their expectations, judgments and calculations. Identifying effects of culture on economic outcomes should be interpreted in a way that leads to better dialogue and communication across cultures.  

On a more practical side, this research can help pinpoint effective margins of development policy and aid programs to developing countries. Depending on the strengths of various cultures, different emphases may have to be put on a spectrum of available policy tools. For example, aid for programs providing public goods may be more effective in collectivist societies than in individualist societies. In the latter, aid programs counting on local initiatives might be more effective. Alternatively, organizational support may have to be stronger for infrastructure projects in individualist societies, whereas in collectivist societies one may have to make special effort to encourage creative initiatives.  

Research on the economic effects of culture is still in its infancy. We hope that our results showing

the importance of culture for long-run growth will help to spur research in this direction. 

References

Acemoglu, D., S. Johnson, and J. Robinson (2001) “The Colonial Origins of Comparative Development: An Empirical Investigation,” American Economic Review 91, 1369-1401.

Acemoglu, D., S. Johnson, and J. Robinson (2002) “Reversal of Fortune: Geography and. Institutions in the Making of the Modern World Income Distribution,” Quarterly Journal of Economics 117(4), 1231-94.

Aghion, Ph., and P. Howitt (1998) Endogenous Growth Theory MIT Press, Cambridge Mass. 

Aghion, Ph., and P. Howitt (2009) The Economics of Growth MIT Press, Cambridge Mass.

Aghion, Ph., Y. Algan, P. Cahuc, and A. Shleifer (2010) “Regulation and Distrust," Quarterly Journal of Economics 125(3), 1015-1049.

Aiello, F., and V.Scoppa (2000) “Uneven Regional Development in Italy: Explaining Differences in Productivity Levels,” Giornale degli Economisti 59(2), 270-298.

Albouy, David. 2012. “The Colonial Origins of Comparative Development: An Empirical Investigation: Comment,” American Economic Review 102(6), 3059-3076.

Algan, Y., and P. Cahuc (2010) "Inherited Trust and Growth," American Economic Review 100(5), 20602092.

Ashraf, Q., and O. Galor (2007) “Cultural Assimilation, Cultural Diffusion and the Origin of the Wealth of Nations” CEPR Working Paper No 6444, London UK.

Ashraf, Q., and O. Galor (2013) “The Out of Africa Hypothesis, Human Genetic Diversity and Comparative Development,” American Economic Review 103(1), 1-46. 

Barro, R.J., and J.W. Lee (2001) “International Data on Educational Attainment: Updates and Implications,” Oxford Economic Papers 53(3), 541-563.

Barro, R.J., and R.M. McCleary (2003) “Religion and Economic Growth Across Countries,” American Sociological Review 68(5), 760-781.

Bisin, A., and T. Verdier (2000) ““Beyond The Melting Pot”: Cultural Transmission, Marriage, And The Evolution Of Ethnic And Religious Traits,” Quarterly Journal of Economics 115, 955-988.

Bisin, A., and T. Verdier (2001) “The Economics of Cultural Transmission and the Dynamics of Preferences,”Journal of Economic Theory 97, 298-319.

Bond, M., and P. Smith (1996) “Culture and Conformity:  A Meta-analysis of studies using Asch’s (1952b, 1956) line Judgment Task,” Psychological Bulletin 119(1), 111-137.

Cavalli-Sforza, L. L., P. Menozzi, and A. Piazza (1994). The History and Geography of Human Genes. Princeton University Press.

Chiao, J.Y., and K.D. Blizinsky (2010) “Culture-gene coevolution of individualism-collectivism and the serotonin transporter gene,” Proceedings - Royal Society. Biological sciences 277(1681), 529-537.

Clark, G. (2007) A Farewell to Alms. A Brief Economic History of the World. Princeton Univ. Press.

Comin, D.  and M. Mestieri (2013) “If Technology has arrived everywhere, why has income diverged?” mimeo Harvard University.

Corneo, G., and O. Jeanne (2010) “Symbolic values, occupational choice, and economic development,” European Economic Review 54(2), 237-251.

De Mooij, M. (2003) “Convergence and Divergence in Consumer Behaviour: Implications for Global Advertising,” International Journal of Advertising 22(2), 183-202

Doepke, M., and F. Zilibotti (2008) “Occupational Choice and the Spirit of Capitalism,” Quarterly Journal of Economics 123(2), 747-793.

Doepke, M., and F. Zilibotti (2013) “Culture, Entrepreneurship and Growth” in P. Aghion and S. Durlauf (ed.) Handbook of Economic Growth vol. 2 Elsevier Amsterdam, 1-48. 

Easterly, W., M. Kremer, L. Prichett, and L. Summers (1993) “Good policy or good luck? Country growth performance and temporary shocks,” Journal of Monetary Economics 32(3), 459-483. 

Economist Intelligence Unit (2007) Innovation: Transforming the way business creates. www.eiu.com. 

Economist Intelligence Unit (2009) A new ranking of the world’s most innovative countries. www.eiu.com.

Evenson, R., and L. Westphal (1995) “Technological Change and Technology Strategy” in Handbook of Development Economics, vol. 3, 2209-2299, Elsevier Amsterdam.

Fearon, J. (2003) “Ethnic and Cultural Diversity by Country,” Journal of Economic Growth 8(2), 195-222. 

Fernandez, R. (2010) “Does Culture Matter?” in J. Benhabib, A. Bisin and M. Jackson (ed.) Handbook of Social Economics, Vol. 1A, North-Holland, 481-510.

Fernandez (2013) “Cultural Change as Learning: The evolution of female labor force participation over a century,” American Economic Review 103(1), 472-500.

Fernandez, R., A. Fogli and C. Olivetti (2004) “Mothers and Sons: Preference Formation and Female Labor Force Dynamics,” Quarterly Journal of Economics 119(4), 1249-1299.

Fernandez, R., and A. Fogli (2009) “Culture: An Empirical Investigation of Beliefs, Work and Fertility,” American Economic Journal: Macroeconomics 1(1), 146-177. 

Fincher, C.L., R. Thornhill, D.R. Murray, and M. Schaller (2008) “Pathogen prevalence predicts human cross-cultural variability in individualism/collectivism” Proceedings - Royal Society. Biological sciences 275(1640), 1279-1285.

Finer, S. (1997). The History of Government, volumes I-III, Oxford University Press, Oxford.

Fogli, A., and L. Veldkamp (2012) “Germs, Social Networks and Growth,” mimeo. 

Giuliano, P. (2007) “Living Arrangements in Western Europe: Does Cultural Origin Matter?” Journal of the European Economic Association 5, 927-952. 

Giuliano, P., A. Spilimbergo, and G. Tonon (2014) "Genetic distance, transportation costs, and trade," Journal of Economic Geography 14(1), 179-198.

Goncalo, J.A., and B.M. Staw (2006) “Individualism–collectivism and group creativity,” Organizational Behavior and Human Decision Processes 100, 96–109.

Gorodnichenko, Y. and G. Roland (2011) “Which Dimensions of Culture Matter for Long-Run Growth?” American Economic Review. Papers and Proceedings 101(3), 492-498.

Greif, A. (1994) “Cultural Beliefs and the Organization of Society: A Historical and Theoretical Reflection on Collectivist and Individualist Societies,” Journal of Political Economy 102, 912-950.

Greif, A. (2006). Institutions and the Path to the Modern Economy: Lessons from Medieval Trade. Cambridge University Press.

Grosjean, P. (2009) “The Contributions of Spatial Proximity and History to Cultural Integration: A Gravity Approach”, mimeo, UC Berkeley.

Grossman, G., and E. Helpman (1991) Innovation and Growth in the Global Economy. MIT Press Cambridge Mass.

Guiso, L., P. Sapienza, and L. Zingales (2003) “People’s Opium? Religion and Economic Attitudes.” Journal of Monetary Economics 50: 225-282.

Guiso, L., P. Sapienza, and L. Zingales (2009) “Cultural Biases in Economic Exchange,” Quarterly Journal of Economics 124(3), 1095-1131.

Hall, R. E., and C.I. Jones (1999) “Why Do Some Countries Produce So Much More Output Per Worker Than Others?” Quarterly Journal of Economics 114(1), 83-116.

Harrison, I. (2004) Book of Inventions, Cassel Illustrated, London.

Heine, S.J. (2007) Cultural psychology. W. W. Norton & Company.

Hofstede, G. (2001) Culture’s Consequences: Comparing Values, Behaviors, and Organizations Across Nations. 2^nd edition. Sage Publications. 

Hoppe, M.H. (1990) A comparative study of country elites: international differences in work-related values and learning and their implications for management training and development” Ph.D. thesis University of North Carolina at Chapel Hill. 

Huang, Y., J. Lin, and Y. Qian (2013) “Does ethnicity pay? Evidence from overseas Chinese FDI in China,” Review of Economics and Statistics 95(3), 868–883.

Inglehart, R.F., S. Borinskaya, A. Cotter, J. Harro, R.C. Inglehart, E. Ponarin, and C. Welzel, 2014. "Genetic Factors, Cultural Predispositions, Happiness and Gender Equality," Journal of Research in Gender Studies 4(1), 32–100.

Jellema, J. (2009) “Cultural Variation, Genetic Heterozygosity and Economic Development,” mimeo.

Jellema, J., and G. Roland (2011) “Institutional Clusters and Economic Performance,” Journal of Economic Behavior and Organizations 79(1-2), 108-132. 

Jones, C. I. (2002) Introduction to Economic Growth. W.W. Norton New York. 

Jones, C. I., and P. Romer (2010) “The New Kaldor Facts: Ideas, Institutions, Population, and Human Capital,” American Economic Journal: Macroeconomics 2(1), 224-245. 

Kashima, E., and Y. Kashima (1998). “Culture and language: The case of cultural dimensions and personal pronoun use,” Journal of Cross-Cultural Psychology 29, 461-486.

Klenow, P. and A. Rodriguez-Clare (1997) “Economic growth: A review essay,” Journal of Monetary Economics 40(3), 597-617. 

Knack, S., and P. Keefer (1997) “Does Social Capital Have an Economic Payoff? A Cross-Country Investigation,” Quarterly Journal of Economics 112(4), 1251-1288.

La Porta, R., F. Lopez-de-Silanes, A. Shleifer, and R. Vishny. 1998. “Law and Finance.” Journal of Political Economy 106(6), 1113-1155. 

La Porta, R., F. Lopez-de-Silanes, and A. Shleifer (2008) “The Economic Consequences of Legal Origins,” Journal of Economic Literature 46(2), 285–332.

Landes, D. S. (1998) The Wealth and Poverty of Nations. Norton Publishers, New York.

Licht, A. N., C. Goldschmidt, and Shalom H. Schwartz (2007) “Culture Rules: The Foundations of the Rule of Law and Other Norms of Governance,” Journal of Comparative Economics 35(4), 659-688. Liker, J. (2003) The Toyota Way. McGraw-Hill.

Macfarlane, A. (1978) The Origins of English Individualism: Family, Property and Social Transition. Blackwell, Oxford.

Maddison, A. (2003) The World Economy. Historical Statistics OECD, Paris.

Merrit, A. (2000) “Culture in the cockpit: Do Hofstede’s Dimensions replicate?” Journal of Cross-Cultural Psychology 31(3), 283-301.

Merton, R. K. (1973) The Sociology of Science: Theoretical and Empirical Investigations. University of Chicago Press. 

Mourant, A.E., Ada C. Kopec, and Kazimiera Domaniewska-Sobczak (1976) The Distribution of the Human Blood Groups and Other Polymorphisms. Oxford University Press.

Murdock, G. (1967) Ethnographic Atlas. Pittsburgh, PA: University of Pittsburgh Press.

Murrell, P., and M. Schmidt (2011) “The Coevolution of Culture and Institutions in Seventeenth Century England”, mimeo, University of Maryland.

Murray, Damian R., and Mark Schaller (2010) "Historical Prevalence of Infectious Diseases Within 230 Geopolitical Regions: A Tool for Investigating Origins of Culture," Journal of Cross-Cultural Psychology 41(1), 99–108.

North, D. (1990) Institutions, Institutional Change and Economic Performance. Cambridge University Press. Platteau, Jean-Philippe (2000) Institutions, Social Norms and Economic Development, Harwood Academic Publishers Newark New Jersey.

Pritchard, J.K., and M. Przeworski (2001) "Linkage disequilibrium in Humans: Models and Data". American Journal of Human Genetics 69 (1), 1-14.

Putnam, R.D. (1994) Making democracy work: Civic traditions in modern Italy. Princeton Univ. Press.

Putterman, L., and D.N. Weil (2010) "Post-1500 Population Flows and the Long-Run Determinants of Economic Growth and Inequality," Quarterly Journal of Economics 125(4), 1627-1682.

Redding, S., and A.J. Venables (2004) “Economic geography and international inequality,” Journal of International Economics 62(1), 53-82.

Roland, G. (2004) “Understanding Institutional Change: Fast-moving and Slow-moving Institutions.” Studies in Comparative International Development 38: 109-131.

Schwartz, S. H. (1994) “Beyond Individualism/Collectivism: New Cultural Dimensions of Values” in Uichol K. et al, eds., Individualism and Collectivism: Theory, Method, and Applications. Sage. 

Schwartz, S. H. (2006) “A Theory of Cultural Value Orientations: Explication and Applications,” Comparative Sociology 5(2-3), 137-182. 

Shane, S. (1995) “Uncertainty avoidance and the Preference for Innovation Championing Roles”, Journal of International Business Studies 26(1): 47-68. 

Spolaore, E., and R. Wacziarg (2009) “The Diffusion of Development,” Quarterly Journal of Economics 124(2), 469-529.

Spolaore, E., and  R. Wacziarg (2013) “How Deep Are the Roots of Economic Development?” Journal of Economic Literature 51(2), 325-369.

Tabellini, G. (2008a) “The Scope of Cooperation: Values and Incentives,” Quarterly Journal of Eocnomics 123(3), 905-950.

Tabellini, G. (2008b) “Institutions and Culture. Presidential Address European Economic Association,” Journal of the European Economic Association 6, 255-294. 

Tabellini, G. (2010) “Culture and Institutions: Economic Development in the Regions of Europe,” Journal of the European Economic Association 8(4), 677-716.

Tills, D., A. C. Kopec, and R. E. Tills (1983) The Distribution of the Human Blood Groups and Other Polymorphisms. Oxford University Press. 

Van Nimwegen, T. (2002) “Global Banking, Global values: The in-house reception of the corporate values of ABN-AMRO” Ph.D. dissertation Nyenrode University Delft, Netherlands.

Way, B.M. and M.D. Lieberman (2010) “Is there a genetic contribution to cultural differences? Collectivism, individualism and genetic markers of social sensitivity,” Social Cognitive & Affective Neuroscience 5(2-3), 203-211.

Weber, M. (2002) The Protestant Ethic and the Spirit of Capitalism, translated by Stephen Kalberg, Oxford University Press, Oxford U.K.