Editor’s note: This post is part of a series showcasing Barcelona GSE master projects by students in the Class of 2016. The project is a required component of every master program.
Andrea Bacilieri, Abhijeet Khanna, Irene Pañeda Fernandez, Jonathan Stern
This paper examines how ethnic heterogeneity may affect the ability of Malawian rural households to solve collective action problems. The collective action challenges are natural shocks -floods, droughts, and irregular rain and availability of common pool resources – an irrigation system, a forest, and common pasture land. We measure household welfare through maize harvest and annual consumption. We find that ethnic polarization and fractionalization are unambiguously bad for maize harvest but, under natural shocks, the size of this negative relationship is reduced. This may be due to the way natural shocks cross ethnic lines and facilitate the overcoming of ethnic differences. The bad effects of polarization remain unchanged in the presence of a shock, suggesting that this is a more intransigent problem. With respect to consumption, we find diminishing returns to increased polarization, becoming negative for high levels of polarization. Results are strongest in the presence of a communal forest. This may be due to the repeated and continuous nature of communal forest management, and the way that polarization may facilitate the formation of coherent bargaining factions.
In this paper we explore the effects of ethnic fractionalization and polarization in the presence of natural shocks and common pool resources. By greater fractionalization we mean a smaller probability that any two individuals come from the same ethnic group. Polarization is a related concept which also takes into account the size of the bargaining factions- polarization being highest with two equally sized groups.
Our hypotheses were that ethnic heterogeneity would worsen the impact of shocks, and affect detrimentally the economic benefit derived from common pool resources. We sought to test these hypotheses by constructing a novel dataset for Malawi which combines indices of ethnic fractionalization and polarization calculated at the Territorial Authority level using the 2008 census and the Malawi Integrated Household Panel Survey for the year 2013. We argue for the exogeneity of our heterogeneity indices based on the low level of change in the ethnic makeup of Malawi over the past four years and the low level of migration within the country.
In the first part of our analysis we regress the log of maize harvest on the presence of shocks such as drought, flood and irregular rain interacted with our ethnic heterogeneity indices and a set of agricultural, climate, household and community controls. We find that ethnic polarization and fractionalization are unambigiously bad for maize harvest. Counter to our expectations, we find that fractionalization appears to lessen the impact of a drought or irregular rain on harvest, although the net effect of increases in fractionalization remains bad for harvests. We posit tentatively that the reduction in the effect of fractionalization in the presence of a shock could be due to the way natural shocks may cross ethnic lines and facilitate the overcoming of ethnic diff erences. The bad effects of polarization remain unchanged in the presence of a shock, suggesting that this is a more intransigent problem, and potentially a cause of enduring local level conflict.
In the second part of our analysis we regress the log of consumption on the presence of common pool resources such as forests, irrigation systems and common pasture land. We find no signicant relationship between consumption and fractionalization after testing both linear and quadratic specications. For polarization we find a quadratic relationship with consumption, which is strongest in the presence of a communal forest. This suggests that a certain degree of polarization could help communal forest management, with diminishing returns to increased polarization, becoming negative for high levels of polarization. We posit that this may be due to the repeated and continuous nature of communal forest management, and the way that polarization may facilitate the formation of coherent bargaining factions.
Through an exploration of the correlations between our ethnic heterogeneity indices and a set of community characteristics we find that greater heterogeneity is negatively correlated with school quality and the availability of agricultural inputs. These results cast some doubt on the exogeneity of ethnic heterogeneity. However given that the ethnic indices are slow moving over time, these correlations may also suggest some of the mechanisms by which fractionalization and polarization aff ect economic development in rural Malawi. Further work might seek to explore further these mechanisms, and whether the empirical findings of this paper can be replicated in other countries and contexts.