The Political Construction of Caste in South India Vijayendra Rao vrao worldbank org Development Research Group The World Bank And Radu Ban rban worldbank org London School of Economics and Development Research Group The World Bank August 2007 We thank seminar participants at the World Bank s research department and Karla Hoff for helpful conversations and comments Jillian Waid and Babu Srinivas Dasari provided excellent research assistance This paper reflects the views of the authors and should not be attributed to the World Bank its member countries or any affiliated organization We are indebted to the Dutch government and the Research Support Budget of the Development Economics Vice Presidency of the World Bank for financial support Abstract Are social institutions endogenous Can measures of social diversity e g fractionalization be treated as exogenous variables in assessing their impact on economic and political outcomes The caste system which categorizes Hindus into endogamous and stratified social groups is considered to be the organizing institution of Indian society It is widely thought to have stayed stable for hundreds if not thousands of years so deeply resistant to change that it has been blamed for everything from formerly anemic Hindu rates of growth to persistent inequality traps This paper uses a natural experiment the 1956 reorganization of Indian states along linguistic lines to demonstrate that the number and nomenclature of castes has significantly changed in linguistically matched villages i e mistakes in the reorganization at the borders of these states This shows that the caste system is not stable but a pliable institution endogenous to political change 2 Introduction A prolific line of research has explored the impact of ethnic and social diversity on a variety of economic and political outcomes Alesina and La Ferrara 2005 Economists however have paid far less attention to the possibility that social structures may be endogenous1 though historians and anthropologists have made it an active area of research2 While economists now recognize that political institutions may be endogenous to economic change Acemoglu Johnson and Robinson 2005 Engerman and Sokoloff 2003 they have tended to treat social categories such as caste and race as fixed in time and exogenous The small but growing literature on the economics of identity choice Caselli and Coleman 2006 Bloch and Rao 2001 Akerlof and Kranton 2000 which examines how individuals chose a social identity within a given and fixed set of alternatives has made a move in this direction Our focus however is on a different question what if the alternatives the names and the number of choices in identity categories themselves were endogenous This would have fundamental implications for the analysis and measurement of fractionalization and polarization besides raising questions about whether such measures of diversity could be included on the right hand side of an OLS regression Our focus in this paper is on the Indian caste system which has long been the archetype of a rigid and unchanging social institution that traps individuals within a hierarchical hereditary structure which determines their economic and social status We examine how caste can be transformed by political change More specifically we analyze the impact of an exogenous shift in the boundaries of states on caste structures the number of castes and their names in villages affected by the change Thus we question the widespread assumption among social scientists that caste structures are fixed given and very slow to change This assumption best described as a trope3 has long historical roots Beginning with Alberuni a thousand years ago scholars have considered the caste system as the organizing institution of Indian society A voluminous literature on the subject has evolved since then with various 1 A point made by Alesina and La Ferrara 2005 in their thoughtful review of the economics literature A review of the anthropology and history literature on this would require another paper but for important work in different contexts see Sahlins 1991 for Spain and Bayly 1999 for India 3 A persistent and familiar idea or theme i e an intellectual clich 2 3 western scholars commenting on its exotic and exploitative rules and practices from the Abb Dubois 1806 in the eighteenth century4 to Max Weber 1966 and James Mill 1820 in the nineteenth and Louis Dumont 1980 in the twentieth This Orientalist5 trope of caste as a system of inherited institutional rigidity has become an integral part of the scholarly canon and is reflected in the writing of politicians including Gandhi and Nehru to contemporary anthropologists and economists The description of caste in this literature primarily drawn from ancient Sanskrit texts such as the Manusmriti can be briefly summarized as follows Hindu society has been divided for millennia into four hierarchical groups or varnas led by Brahmins the scholars and teachers followed by Kshatriyas rulers and warriors Vaishyas traders and merchants and Shudras artisans and laborers A fifth group considered Untouchable is so low as to be outside the domain of the ritual hierarchy and relegated to occupations such as scavenging In everyday practice castes manifest as jatis6 endogamous groups defined within regional and linguistic boundaries which are mapped onto varna categories and thus bound within those hierarchies This conception of caste reached its most sophisticated expression in Louis Dumont s 1980 influential description of Hindus as a species that he called Homo Heirarchicus7 To paraphrase Dumont s complex argument the caste system is perpetuated by an ideology where upper castes justify the hierarchy because they internalize the belief that their inherited high status is inevitable as the fruit of efforts in past lives while lower castes similarly internalize the justification of their low status This stands in opposition to western notions of individual equality and as an alternative to the individually rational Homo Economicus The fast expanding literature on the economics of caste is not immune to the trope To cite a few examples Akerlof 1976 in his classic work on caste and the rat race outlines a model of a stable and persistent caste equilibrium where obedience to the caste code results in the sub optimal allocation of labor More recently Freitas 2006 formalizes a model of caste system that attempts to understand why this