Coercive Capacity and Economic Development: Subnational Evidence

• In Event: Poster Session: Comparative Politics in Developing Countries I

Abstract

A large theoretical literature has emphasized the role that state coercive capacity plays in economic development. By limiting external and internal challenges, governments which wield a monopoly of force over a given territory have longer time horizons and can extract higher rents, encouraging investments in productivity-enhancing public goods. On the other hand, high coercive capacity may also limit economic development by limiting government accountability, resulting in high degrees of productivity-inhibiting taxation/rent-extraction.
The empirical literature on state capacity has found limited evidence that coercive capacity is either a productivity-enhancing or inhibiting. We argue that these ambiguous results may be due to both the way coercive capacity is typically measured as well as the use of countries as the unit of observation. First, measurement error in these coercive capacity proxies may attenuate the estimated relationship between coercive capacity and economic development. Specifically, military expenditure and employment likely do not capture all government expenditure on coercive capacity and in particular exclude police expenditure/employment. In addition, the total population is typically used as the denominator in the military employment share rather than the adult population, which will bias this measure of coercive capacity downward for countries with high fertility. Second, small sample sizes of cross-country analysis may lead to underpowered tests of the effect of coercive capacity on development.
To address these issues, we utilize geo-coded Afrobarometer survey data to generate subnational estimates of coercive capacity for 33 countries in Africa, a continent characterized by non-democratic institutions and low/uneven levels of state capacity and economic development. Specifically, we use questions concerning respondent occupation to determine whether each respondent to a member of the “coercive sector” (military/police/security services). After assigning respondents to an arbitrary PRIOGRID cell, we use the fraction of respondents in the cell who are employed in the coercive sector as our cell-level estimate of coercive capacity.
We then examine the within-country correlates of coercive capacity as well as the within-country relationship between coercive capacity and proxies for public goods, rent extraction, and economic development, which are all derived from the 2016 round of the Afrobarometer Survey. To account for the non-random within-country assignment of coercive capacity, we control for a host of exogenous control variables (e.g. distance from the coast, rugged terrain), political geographic variables (e.g. distance from the capital, distance from the border, the historic political power of local ethnic groups), and controls for historic urbanization (e.g. population density in 1950, distance and size of the nearest city in 1950, urban potential in 1950). Our results imply that higher local coercive capacity is associated with higher levels of economic development, as proxied by asset ownership and levels of urbanization. Coercive capacity is also associated with higher levels of public goods, proxied by the presence of government facilities such as a school, hospital, electricity, piped water as well as average years of education for respondents in the region. We also find suggestive evidence that rent-seeking, proxied by respondent experience with bribery, is higher in locations with higher coercive capacity, consistent with theory suggesting that state capacity facilitated government investment in public goods by allowing the government to internalize the benefits of the resulting increase in economic activity. Coefficient stability tests indicate that these relationships are unlikely to be due to unobserved correlates of coercive capacity and the outcome variables. In addition, we find that distance from the capital city is robustly correlated with coercive capacity, thus potentially explaining high levels of urban concentration in Africa.

Author

• Sidney Carl Turner, Fors Marsh Group LLC