Ethnic heterogeneity and public goods provision in Zambia: Further evidence of a subnational 'diversity dividend' [book]

Rachel M. Gisselquist, Stefan Leiderer, Miguel Niño-Zarazúa
2014 WIDER Working Paper  
Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen
more » ... von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Abstract: The hypothesis that ethnic diversity has a negative impact on public goods provision is widely accepted. Notably, most work on this issue fails to distinguish adequately between national versus subnational governance. We find that subnational empirical evidence in particular is inconclusive, and speak to this gap with new analysis at the Zambian district level. Results lend strong support to an emerging body of work challenging the 'diversity debit' hypothesis: we find no clear evidence of a negative impact but instead a robust positive association with key welfare outcomes. Contra the conventional wisdom, future work should explore mechanisms underlying the 'diversity dividend' now suggested in multiple subnational analyses. All sources are: Authors' analyses based on data and methods as described in the text. 4 diversity debit hypothesis for subnational analyses as follows: First, we distinguish between budgetary outcomes and other measures of government performance, including related welfare outcomes. In the centralized case, local governments and constituencies have little direct influence on budgets according to the mechanisms outlined above as they neither contribute a large share of revenues (taxes) nor directly decide budgetary allocations or policies. At the level of implementation, however, subnational governments and constituencies do have direct influence on how budgets are spent and policies carried out. Community characteristics (including diversity) may also directly affect related welfare outcomes by influencing how communities use and interact with government services. For instance, diversity may have an impact on preferences or social capital which in turn may influence parents' decisions to enroll their children in school, and thus educational enrollment rates. Empirical studies The preceding discussion suggests that in considering the empirical bases upon which the conventional wisdom rests, it is useful to consider empirical studies at the national and subnational levels separately. Broadly speaking, empirical support for the diversity debit hypothesis appears relatively robust at the national level. 2 Easterly and Levine (1997) show a relationship between ethnic heterogeneity and low schooling and insufficient infrastructure, as well as other policy-related outcomes such as political instability, underdeveloped financial systems, distorted foreign exchange markets, and high government deficits. Baldwin and Huber (2010) find support in an analysis of 46 countries considering an aggregate measure of public goods provision based on ten variables related to education, health, sanitation, infrastructure, and the regulatory framework for private sector activity. Jackson (2013) finds support for the diversity debit hypothesis in the areas of education, drinking water, and electricity across 18 African countries, while Gerring et al. (2015) do so across 36 developing countries in terms of human development outcomes, including child mortality, fertility, education, and wealth. At the subnational level, however, empirical support for the diversity debit hypothesis is less clear. The most cited study at the subnational level is Alesina et al. (1999) , which uses US census data (1990) and budget information from all cities, metropolitan areas, and urban counties with populations of at least 25,000. This analysis considers a range of dependent variables on spending, showing that ethnic fractionalization, measured in racial terms, is negatively associated in a statistically significant manner with the share of public spending on roads, education, welfare, and sewage and trash pickup. Subsequent work has highlighted several weaknesses with these findings, however. First, as the study itself shows, some results are not consistent with the under-provision of public goods, e.g. spending on health. 3 Second, the results appear less robust when controls for state level effects are included, which seems wise within the US context, as fiscal responsibilities and regulations differ across states, as does ethnic fractionalization (Gisselquist 2014). 2 Why this relationship holds however remains open for discussion. Alesina et al.'s (1999) model is widely cited, but it relies on the median voter theorem which is generally formulated on the basis of two party competition under plurality rule -i.e., a different institutional context to many of the countries under analysis. Interpretation is further complicated by the quality of data on public goods provision and government budgets that is available at the crossnational level. 3 There is also a positive relationship with spending on police. This arguably can be reconciled within the model: polarized preferences may also lead to higher levels of social conflict and thus greater demands for policing.
doi:10.35188/unu-wider/2014/883-4 fatcat:kmkkzevidjabpawkgyki3evbl4