Economic Effects of Domestic and Neighbouring Countries Cultural Diversity. Erkan Gören - PDF

Description
Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören This Version: March 2013 Abstract This paper investigates the economic growth impact of cultural diversity, both domestically

Please download to get full document.

View again

of 21
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Information
Category:

Gadgets

Publish on:

Views: 15 | Pages: 21

Extension: PDF | Download: 0

Share
Transcript
Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören This Version: March 2013 Abstract This paper investigates the economic growth impact of cultural diversity, both domestically and in neighbouring countries, in a balanced panel of 94 countries covering the period 1970 to The measures of cultural diversity used in this article were derived from a recently developed computer algorithm intended primarily to measure linguistic distances in an automated fashion. The empirical analysis suggests that the degree of cultural diversity in contiguous neighbouring countries has substantial positive effects on domestic per capita income growth, even controlling for a broad set of regional, institutional, religious and other proximate factors of economic growth. The conclusion is that culturally homogeneous countries gain a strategic advantage over their culturally diverse neighbours. Keywords: cultural diversity; ethnic diversity; economic growth JEL: O11; O5 I would like to thank Jürgen Bitzer and the seminar participants at the Carl von Ossietzky University Oldenburg 2012 for useful comments and suggestions. All remaining errors are my own. Carl von Ossietzky University Oldenburg, Department of Economics, Campus Haarentor, Building A5, Oldenburg, Germany, Tel.: , 1 Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören 2 1 Introduction In recent years, a vast literature has emerged debating the impacts of ethnic heterogeneity on socioeconomic factors such as economic development and public policy choices. Since the influential contribution of Easterly & Levine (1997), much attention has been devoted to the negative relationship between ethnic diversity and contemporary per capita income growth (e.g. Alesina et al. (2003); Alesina & La Ferrara (2005); Garcia-Montalvo & Reynal-Querol (2005a)). More recent contributions have investigated deeply rooted factors of cross-country income differences, arguing that genetic differences across countries, as a proxy for barriers of technological diffusion, may account for different paths of economic development (e.g. Spolaore & Wacziarg (2009) and Ashraf & Galor (2012)). The discussion of the impacts of ethnic heterogeneity is not restricted to economic development alone. Researchers have also investigated the determinants of civil conflicts, where measures of ethnic polarisation are strong predictive variables explaining the incidence and duration of civil wars (e.g. Vanhanen (1999); Doyle & Sambanis (2000); Collier (2001); Reynal-Querol (2002); Fearon & Laitin (2003); Garcia-Montalvo & Reynal- Querol (2005b) and Esteban & Ray (2008)). Furthermore, the literature has also found a negative empirical relationship between ethnic heterogeneity and the provision of public goods across nations (e.g. Mauro (1995); Alesina & Perotti (1996); Hall & Jones (1999); La Porta et al. (1999); Annett (2001); Alesina et al. (2003) and Alesina & Zhuravskaya(2012)) and also in ethnic communities within nations (e.g. Alesina et al. (1999) and Miguel (2004)). The impacts of ethnic heterogeneity on redistribution have attracted recent attention among some scholars (e.g. Alesina et al. (2001); Luttmer (2001); Vigdor (2004); Desmet et al. (2009, 2012)). Since measures of ethnic diversity appear to have a strong predictive power in crosscountry studies, they have become standard explanatory variables in development economics. While the effects of ethnic diversity on socioeconomic outcomes (e.g. economic growth, public goods, civil conflicts, etc.) are generally acknowledged, the concrete mechanisms by which ethnic diversity functions in contemporary societies have been debated extensively. For example, Annett (2001) found that ethnically heterogeneous societies are more prone to political instability, and that political pow- Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören 3 ers may therefore increase unproductive government expenditures in order to reduce the risk of being overthrown. Alesina & Perotti (1996) reported that political instability is detrimental to investment decisions. But investments in physical capital are conducive to per capita income growth. Especially in the context of cross-country growth regressions, the majority of the literature has relied on reduced-form regressions, leaving unresolved the question of how precisely ethnicity affects growth. 1 Since the measures used for ethnic heterogeneity are sometimes linked to a specific notion of cultural diversity (e.g. Fearon (2003)), one may argue that differences in cultural attitudes across countries are responsible for cross-country differences in economic outcomes. Guiso et al. (2006) relates culture to the values, norms and beliefs in a society that change only very slowly over time and that have an explicit effect on cross-country differences in economic outcomes. These cultural dimensions (e.g. interpersonal trust) are interpreted as channels by which culture affects various economic outcomes. 2 Some of the variables that may proxy for cultural attitudes are religious affiliations, ethnic background, common language, and genetic differences. 3 The contribution of this paper is twofold. First, it seeks to redefine the concept and measurement of cultural diversity. The most frequently used approach in the economics literature to date has been to rely on established definitions of ethnic groups, although some researchers have carefully constructed their own ethnic classifications. 4 Furthermore, many researchers have generally assumed that cultural distance is constant and identical across groups due to the difficulty of defining and measuring this concept. 5 This approach has two serious shortcomings. On the one 1 See the article by Garcia-Montalvo & Reynal-Querol (2005a), which identifies three main channels (civil wars, government consumption and investments) by which ethnicity indirectly affects growth. 2 See, for instance, Gorodnichenko & Roland (2011) for an extensive analysis of cultural impacts from various sources on output per capita. 3 The main idea behind genetic distance is that it captures barriers to the diffusion of development and hence the adaptation of complex technological and institutional innovations (Spolaore & Wacziarg (2009)). Furthermore, societies that are more closely related genetically may have a stronger basis of trust and may thus exchange information more intensively. 4 See, for instance, the classification of groups in Alesina et al. (2003) and Fearon (2003). 5 One argument for treating cultural distance across groups as constant and equal was put forward by Garcia-Montalvo & Reynal-Querol (2005b), who argued that [...] the dynamics of Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören 4 hand, the widely used measure of ethnic fractionalisation is highly sensitive to the number of groups, and on the other hand, it does not take into account the asymmetric notion of alienation perceived across groups. Hence, the first contribution of this paper is to provide a set of ethnic diversity measures adjusted for the cultural resemblance across groups using a lexicostatistical setup. These newly constructed measures will be referred to as cultural diversity. The notion underlying this definition comes from Greenberg (1956), who argued that from two geographical regions with the same population share, the one with the lowest resemblance across groups should exhibit a higher measure of diversity. Based on this definition, a number of researchers have developed useful measures of cultural similarity across groups. For example, Fearon (2003) and Desmet et al. (2009) developed the concept of phylogenetic trees to capture the genealogy and hence the relationships among languages as a proxy for resemblances among ethnic groups. In contrast to this heuristic approach, the present article proposes a computerised lexicostatistical method to derive measures of resemblance across language pairs. To the best of my knowledge, to date only the contribution of Desmet et al. (2005) has used a lexicostatistical (but non-computerised) approach in constructing lexical percentages of cognates for basic meanings across languages. In their contribution, however, the analysis is limited to a restricted number of Indo-European languages. Therefore, the wide variety of Asian, African and indigenous Latin American languages is not considered because of the lack of data availability. The lexicostatistical approach has a clear advantage over the use of language trees. For example, Spolaore & Wacziarg (2009) mentioned that language trees sometimes rely on arbitrary classifications of languages into groups, so that the discrete number of common nodes across languages may not capture such cultural distances appropriately. The use of lexicostatistical percentages will at least weaken such shortcomings. Despite the advantage of the lexicostatistical approach, the lack of data availability has discouraged researchers from using lexicostatistical percentages to construct measures of cultural distances. 6 the we versus you distinction is more powerful than the antagonism generated by the distance between them. 6 For instance, Desmet et al. (2009) used distance measures based on tree diagrams to construct different indices of diversity because the data in Dyen et al. (1992) had only covered Indo-European Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören 5 Hence, this paper should also be viewed as an attempt to close this research gap. Second, this article seeks to explain the effects of cultural diversity in contiguous neighbouring countries on domestic per capita income growth. The empirical analysis is based on a balanced panel of developed and emerging countries for the period 1970 to Although the direct effects of ethnic diversity on per capita income growth are well documented, until now, to the best of my knowledge, no study has investigated potential spatial externalities of cultural diversity from contiguous neighbouring countries. This paper argues that possible detrimental effects of cultural diversity on economic growth may be conditional on the particular location of the country in question. Arguments in favour of this thesis come from the literature on international investment decisions, in which it is argued that international investors may see language and cultural differences as a possible source of information disadvantages, which contributes to the well-known home bias phenomenon (see, e.g. the contributions of Grinblatt & Keloharju (2001); Chan et al. (2005); di Giovanni (2005); Bhattacharya & Groznik (2008); and Siegel et al. (2011) among others.). It follows, then, that international investors favour culturally similar countries to overcome possible adjustment costs in the foreign market. Furthermore, if potential international investors have a choice among a range of countries in a particular location, they may choose countries that are culturally homogeneous, since these countries have higher economic development, better public policies and a lower incidence of civil wars. Hence, having an ethnically homogeneous society and well-established political institutions in a highly culturally diverse environment (e.g. Botswana) may be favourable for a country s own economic development. This statement will be tested in the empirical part of this paper. The empirical analysis reveals substantial positive effects of cultural diversity from nearby countries on domestic per capita income growth that are not captured by a broad number of regional, institutional, religious or other proximate factors of economic growth. Hence, the empirical results are in line with the statement that culturally homogeneous countries surrounded by languages and neglected the wide variety of African, Asian and American languages. The number of observations in the empirical analysis in Spolaore & Wacziarg (2009) was reduced substantially once the authors controlled for linguistic distance in their regressions, also using the more detailed data in Dyen et al. (1992). Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören 6 culturally diverse countries gain a strategic advantage that is conducive to domestic per capita income growth. The paper is organised as follows. Section 2 reviews the methods used in the economics literature to measure cultural similarity using language tree branches, and discusses the advantage of lexicostatistical methods over the use of language trees. The computerised lexicostatistical approach chosen here to derive similarity measures between pairs of languages is presented in Section 3. Section 4 discusses the data set and estimation methodology. The empirical results are presented in Section 5. Finally, Section 6 concludes and highlights some important points for further research. 2 Measurement Issues Affecting Cultural Distance This section introduces Greenberg s (1956) measure of cultural diversity and highlights some important shortcomings of recently developed approaches to measure the cultural similarity between ethnic groups. Greenberg s (1956) index of cultural diversity, GI, has the following form: 7 GI = 1 G G p i p j r ij, (1) i=1 j=1 where p i and p j refer to the respective proportions of groups i and j in the particular country s population. The resemblance between groups i and j is captured by the factor r ij, which is restricted between 0 and 1 (where a value of 1 means full cultural similarity between the two groups). The idea behind this functional form is that 7 It is well known that Greenberg s (1956) measure of diversity is closely related to the generalised polarisation index introduced in the seminal paper of Esteban & Ray (1994). The latter define the concept of polarisation as the sum of interpersonal antagonisms, where the main distinguishable feature is income differences between the different groups. They also limit the class of allowable functions by imposing three important axioms. It is well known that for particular parameter values, this general polarisation measure can generate different measures of diversity: namely, ethnic fractionalisation, polarisation and measures of peripheral heterogeneity. These results are available from the author upon request. Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören 7 from two geographical regions with the same population share, the one with the lowest resemblance across groups should exhibit the highest measure of diversity. It can be shown that if the resemblance between all groups is 0, then GI exactly equals the widely used Herfindahl-based ethnic fractionalisation measure. An important feature of the Greenberg index is that one could also use very disaggregated data of population shares and let the resemblance factors decide the degree to which different ethnic groups are merged. A crucial point in assigning cultural distances to groups is the specification of the resemblance factors r ij. Fearon (2003) used the concept of linguistic tree diagrams to investigate this issue. He argued that the number of shared tree branches of two languages could be used as a proxy for the cultural similarity between the groups. By assigning each pair of groups a corresponding number of shared tree branches and dividing this value by the maximum number of language classifications, one obtains the cultural similarity between this language pair. However, as Fearon (2003) mentioned, this function should be an increasing and concave function of the resemblance ratio, indicating that early divergence in a language tree corresponds to earlier cultural divergence rather than later divergence. This property is achieved by raising the resemblance ratio to the power of δ, requiring that the exponent is bounded between 0 and 1. Higher values of δ assign more cultural difference to more minor differences in linguistic structure. Nevertheless, the use of language trees to construct cultural resemblance factors has three main shortcomings. First, the main disadvantage of language trees is that all living languages are considered to be equidistant from their corresponding proto-languages within and also across languages. Second, not all languages have the same number of language branches, which means that one tends to overestimate the cultural distance between languages from the same language family. Therefore, tree diagrams give no information about the point in time when a given pair of languages diverged. 8 The last shortcoming of this approach lies in the functional form of the resemblance function. Fearon (2003) argued that this function ensures that earlier divergence between two lan- 8 See, for instance Serva & Petroni (2008) for a brief overview of the use of lexicostatistical percentages to derive separation times of language families. This research field is also known as glottochronology, and was developed primarily by Swadesh (1952). Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören 8 guages should indicate greater cultural distance than later divergence. Nevertheless, the degree to which two languages differ in their cultural distance depends crucially on the parameter δ, which captures this notion precisely. Therefore, the choice of the parameter δ has a non-negligible impact on the measures of cultural diversity. 9 While Fearon (2003) used a value of one-half, Desmet et al. (2005) argued that for their measures of linguistic diversity, values of δ in the range of [0.04, 0.10] continue to give similar results in terms of the statistical significance of the linguistic effects on redistribution. This discussion shows that in using language trees to measure cultural distances, one has to rely on arbitrary parameter choices that are crucial in measuring the cultural distance across ethnic groups. 10 Another related research line argues in favour of lexicostatistical methods in measuring the cultural similarity across groups. In particular, Desmet et al. (2005) have used the lexicostatistical measures in Dyen et al. (1992) as proxies for cultural distance between language pairs. The lexicostatistical method is used mainly for language classifications and is based on four main steps: 11 First, a list of basic meanings is collected that are so fundamental that every language contains them (e.g. meanings like I, we, water, fire, etc.). Based on these words, a linguist is needed to provide a carefully considered judgement of whether two words in a particular language pair share the same basic meaning and ancestry in both languages. After the judgement process is completed for any two particular languages, step 3 deals with the computation of lexicostatistical percentages between them. Finally, step 4 deals with the subgrouping of different languages using an appropriate statistical approach (i.e. clustering methods) based on the computed percentages of cognates in the previous step. Dyen et al. (1992) focussed on 200 different basic words and meanings for 84 Indo-European languages, and computed the number of cognates between given pairs of languages. The term cognate was used if the two languages have an unbroken history from the same ancestor. The result is a symmetric 9 Fearon (2003, Footnote 26) acknowledges that depending on this parameter value, one obtains different correlation measures between cultural and ethnic fractionalisation, where higher values for δ indicate a higher correlation between the two measures. 10 Alesina & La Ferrara (2005) noted that this approach is of a more heuristic nature. 11 For a brief overview of this method, see Appendix 3 in Dyen et al. (1992). Economic Effects of Domestic and Neighbouring Countries Cultural Diversity Erkan Gören 9 matrix of percentage co
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks