Jurnal Ekonomi & Studi Pembangunan Volume 23 No 1, April 2022 Article Type: Research Paper New Evidence on Ethnic Diversity and Social Capital in Indonesia Muhammad Ryan Sanjaya Abstract: This study analyzes the relationship between the various measures of ethnic diversity and social capital in Indonesia, particularly trust and tolerance (towards other religions and other ethnic groups), using a nationally representative survey. The 2009 National Socioeconomic Survey asked almost 300 thousand individuals on social capital from the 2010 population census data used to construct measures of ethnic and religious diversity. The study’s main contribution to the literature is the inclusion of ethnic similarities in the construction of these diversity variables using linguistic tree data taken from Ethnologue. Using the multiple linear regression method, the study found that all measures of diversity are negatively associated with trust but are positively associated with tolerance, even after controlling for individual-, household- and district-level characteristics. Anticipating the possibility of endogeneity in the diversity variables, this study uses geographical variables to instrument these variables. Nevertheless, the use of the instrumental variable regression method does not change the main result. Keywords: Ethnic diversity; Social capital; Trust; Tolerance; Indonesia JEL Classification: D19; O53; Z13 Introduction The economics literature on the role of ethnic diversity is vast. Some early studies, such as Easterly and Levine (1997), tried to link ethnic diversity with economic growth in sub-Saharan Africa wherein the two showed an inverse relationship. The following literature on this topic has included additional controls such as interacting the measure of ethnic diversity with democracy (Alesina & La Ferrara, 2005) and controlling for migration (Bove & Elia, 2017), to linking it with other socioeconomic variables such as trust (Algan & Cahuc, 2014) and conflict (Esteban et al., 2012), to find the sources of such diversity (Ahlerup & Olsson, 2012; Ashraf & Galor, 2013; Michalopoulos, 2012; Moslehpour et al., 2021). Some studies linked ethnic diversity with socioeconomic variables in Indonesia, such as how it can affect public goods provision (Alesina et al., 2019; Siburian, 2019), social capital (Mavridis, 2015), and conflict (Bazzi et al., 2019). AFFILIATION: Department of Economics, Faculty of Economics and Business, Universitas Gadjah Mada, Special Region of Yogyakarta, Indonesia CORRESPONDENCE: m.ryan.sanjaya@ugm.ac.id THIS ARTICLE IS AVALILABLE IN: http://journal.umy.ac.id/index.php/esp DOI: 10.18196/jesp.v23i1.13863 CITATION: Sanjaya, M. R. (2022). New Evidence on Ethnic Diversity and Social Capital in Indonesia. Jurnal Ekonomi & Studi Pembangunan, 23(1), 35-50. ARTICLE HISTORY Received: 02 Feb 2022 Revised: 13 Mar 2022 28 Mar 2022 Accepted: 29 Mar 2022 https://scholar.google.co.id/citations?hl=en&user=-UBt1poAAAAJ https://economics.feb.ugm.ac.id/ https://economics.feb.ugm.ac.id/ https://economics.feb.ugm.ac.id/ https://economics.feb.ugm.ac.id/ mailto:m.ryan.sanjaya@ugm.ac.id http://journal.umy.ac.id/index.php/esp https://journal.umy.ac.id/index.php/esp/article/view/13863 https://crossmark.crossref.org/dialog/?doi=10.18196/jesp.v23i1.13863&domain=pdf https://creativecommons.org/licenses/by-sa/4.0/ Sanjaya New Evidence on Ethnic Diversity and Social Capital in Indonesia Jurnal Ekonomi & Studi Pembangunan, 2022 | 36 Due to the limitations set, this study will only discuss how ethnic diversity is related to social capital such as trust and tolerance. Social capital is an important feature of a society that would help members of different ethnic, religious, and other groups effectively achieve the goals of the community. It is only with the trust towards others and respecting differences that society would function, without the need or minimize the necessity of having formal institutions. Therefore, it is important to understand how the many different members of a society could work together. Indonesia, in particular, is blessed with a sheer amount of ethnocultural diversity due to its unique geographical locations. The 2010 census data shows the country’s population speaks 1,204 different languages and there were 964 ethnic groups. Such diversity is a real challenge to not only the government in ensuring that the policies are inclusive but also to grassroots communities that deal with members of different groups daily. The association between social capital, such as trust, and ethnic diversity are not particularly clear. There are three often cited theories on how social capital may be related. The first theory, often dubbed as constrict or “hunkering down” theory, predicts that diversity makes people trust less towards everyone, which implies lower (within- group) trust and (outgroup) tolerance (Putnam, 2007). The second is “conflict theory” which predicts that the more diverse a community is, the higher the trust and lower tolerance (Blalock, 1967). The reason for such prediction is that the perception of threat (from an outsider) makes people increase the bonding among the same group while, at the same time, increasing their prejudice towards the outsider. Finally, the “contact theory” is more optimistic as diversity is expected to increase both trust and tolerance (Allport, 1954). The rationale behind the theory is that contact with out of the group members would bridge differences and therefore increase the social capital. Empirically, some studies show how the two seem to have a negative correlation. For example, in a cross-country study, trust was found to be lower in a society that is heterogeneous (Alesina & La Ferrara, 2002). Whereas in another cross-country study, no robust relationship was found for these variables per se (Finseraas & Jakobsson, 2012), but when multidimensional social structures are introduced, the negative association emerges. Other variables are also found to be associated with trusts, such as the positive association between trust and economic performance (Algan & Cahuc, 2013), the negative correlation between trust and income inequality (Bjørnskov, 2008), and that females tend to trust more (Falk et al., 2018). A whole chapter in the handbook of economic growth was also dedicated to research on not only the relationships between social capital and economic performance but also with institutions (Algan & Cahuc, 2014). The importance of the latter is also expounded in a recent review paper that relates culture—which includes generalized trust—with institutions (Alesina & Giuliano, 2015). More recently, a review study shows the prediction from the contact theory seems to have less support than the alternatives as most empirical studies indicate a negative relationship between social trust and ethnic diversity (Dinesen & Sønderskov, 2017). The literature is also not limited to cross-country but also within-country variations, including those in the context of developing countries such as Indonesia. With hundreds of local languages spoken by 273.5 million people, Indonesia is unique in terms of its large Sanjaya New Evidence on Ethnic Diversity and Social Capital in Indonesia Jurnal Ekonomi & Studi Pembangunan, 2022 | 37 ethnic diversity resulting in the socio-economic implications of having such diversity. Anecdotal evidence shows the country’s experience with ethnic and religious violence, particularly during the transition period of 1998 to 2005, which serves as an indication that diversity matters. Previous studies have shown the conflicting effects of ethnic diversity on conflict: ethnic clustering has some positive correlations with local conflict (Barron et al., 2009), but different measures seem to have opposite effects on conflict (Bazzi et al., 2019). Meanwhile, Gaduh (2012) provides some evidence that trust (tolerance towards members of other religions) is negatively (positively) associated with religious heterogeneity. More recently, using individual-level data from the Indonesian Family Life Survey (IFLS), Mavridis (2015) shows that ethnic diversity is negatively correlated with trust but is positively correlated with ethnic tolerance. It is rather unfortunate that there is no further elaboration on why the result turns out that way, except for the positive outlook that diversity is a good thing to have in a society. The author used data from the fourth wave (2007) of the IFLS that surveyed more than 40 thousand individuals, where trust is elicited through the lost wallet questions (i.e., the likelihood that a lost wallet is returned) and tolerance is simply elicited by asking whether the respondent can trust people of the same ethnicity and whether worship place for people of a different religion is acceptable. One important aspect of the literature is regarding how ethnic diversity is measured, where most of the literature uses fractionalization and polarisation as proxies. Ethnic fractionalization refers to the probability of two strangers belonging to different ethnic groups, whereas ethnic polarization shows how divided or polarized the society is. In the simplest formulas used to construct these variables, most authors use population data by ethnic group, assuming that one ethnic group is distinct from another group. Consequently, such formulas ignore the possibility of two ethnic groups that are similar, but categorically different. Accordingly, researchers have begun to find a better approximation for ethnic diversity that could accommodate such a possibility and found the answer by using linguistic similarity (Desmet et al., 2009; Esteban & Ray, 2011). The implication of using such measures of ethnic diversity is vast, especially in countries like Indonesia, where, arguably, the different ethnic groups may be similar if they reside nearby. The literature on social capital can be categorized into two in their construction of the variable that is either using survey or experiment. The latter is often used to establish causality using a small sample (e.g., Chuah et al., 2013; Chen & Sriphon, 2022), and typically utilizes the trust game (Berg et al., 1995). However, due to the cost of conducting proper economic experiments and the relatively low generalizability of the results, most studies on social capital rely on the survey method. Most cross-country studies use survey methods taken from the World Values Survey and the Gallup World Poll, but national surveys are also very common as used in studies on Indonesia cited above. This study is unique as it extends Mavridis (2015) in three regards. First, instead of using IFLS that only covers around half of the total districts, we use data from a national survey to provide a better representation of Indonesia. Second, this study weighs ethnic diversity with ethnic similarities using an objective measure from Ethnologue. This is important as Sanjaya New Evidence on Ethnic Diversity and Social Capital in Indonesia Jurnal Ekonomi & Studi Pembangunan, 2022 | 38 there are gradations of ethnic differences: some ethnic groups are ethno-linguistically close to each other (e.g., Minang & Malay), while others are distant (e.g., Javanese & Dani in Papua). Third, this study provides a causal identification using an instrumental variable regression method. Research Method This study uses the 2009 National Socioeconomic Survey (Susenas) to measure trust and tolerance in Indonesia. The 2009 Susenas was gathered by the Statistics Indonesia (Badan Pusat Statistik) and is representative of Indonesia as it covers all districts of the country. The social capital module of the survey was not conducted every year and 2009 was the closest year to the 2010 census that provides the data for ethnic diversity. We also use the same survey to obtain individual and household characteristics used as controls in the regressions. The Susenas survey asked 291,532 individuals whether they trusted their neighbors, village figures, government, and village apparatus. Therefore, the survey reflects respondents' direct, particularized trust with their neighborhood rather than a generalized trust which measures respondents’ belief over a stranger. This study calculates trust as the mean response to these questions. Similarly, the tolerance questions were also particularized, i.e., whether they were content with having neighbors of different ethnicity or religion, and their opinion regarding a hypothetical plan of building a worship place of different religions in the neighborhood. This study uses two distributional measures, fractionalization and polarization, as proxies for ethnic diversity. Specifically, ethnolinguistic grouping is used as an appropriate proxy for ethnicity in Indonesia (Mancini, 2008) which helps estimate linguistic similarities. Ethnolinguistic groupings were based on the language spoken at home from the 2010 census (Minnesota Population Center, 2015), which was matched with linguistic similarities calculated from Ethnologue’s website (Simons & Fennig, 2017). IPUMS- International sampled (geographically stratified and systematic) the original census data with an expansion factor of 10. As of the writing of this article, IPUMS-International has only the 2010 population census as the latest available data. For each district, this study calculates the Greenberg-Gini index (ethnolinguistic fractionalization weighted by inter-group distance dmn) (Esteban & Ray, 2011) as: 𝐺 = ∑ ∑ 𝑠𝑚𝑠𝑛𝑑𝑚𝑛 𝑁 𝑛=1 𝑁 𝑚=1 . Here sm and sn are the share of group m and n, respectively. Thus, 0