Middle Class and Economic Development in Nigeria: Evidence from Micro Data

Abstract

In this study, we examine the extent to which middle class households can compel governments to provide public goods nearer their neighbourhoods. Measuring district middle class as the percentage of households in 2-4 quintiles of the asset distribution spectrum, we find in OLS and probit regressions that higher percentage of district middle class is associated with closer proximity to public goods such as water, post-office, point-of-public-transportation, food market, all season road, and availability of government organised refuse disposal services. Results show that a percentage point increase in district middle class is associated with lesser travel distance to the nearest water source by 0.54, post-office by 1.853, point of public transportation by 0.720, food market by 0.824 and all season local roads by 0.753. Similarly, a percentage point increase in district middle class is associated with greater access to government provided refuse disposal services by 0.103. The middle class coefficient estimates are economically large, and statistically significant at 1 percent. Their presence in the district has favourable effects on the proximity to primary schools, secondary schools, clinics and hospitals, greater use of electricity and community care services. Their presence is associated with greater level of local security, reducing the probability that an individual experiences burglary, armed robbery, personal theft, community violence, physical harm and increasing the probability that an individual could feel much safer walking at night in the neighbourhoods. Their impacts on the quality of services offered in primary and secondary schools located in their neighbourhoods are large and statistically significant. A number of tests performed on OLS and probit regressions do not suggests that middle class households might be moving into locations nearer to where public goods are situated. Middle class coefficients remain relatively stable when both dependent and independent variables are re-specified. Middle class coefficients are robust to corrections for historical and contemporary variables and possible omission of historic and current variables. The coefficient estimates are also robust to the removal of influential observations. The use of outlier robust regression technique such Least Absolute Deviation (LAD) produces coefficient estimates similar to those of OLS and probit regressions. To address bias due to endogeneity, measurement error, and model specification simultaneously, we use the 2SLS as complementary identification strategy. Results from 2SLS regressions show that district middle class causes the provision of public closer to middle class neighbourhoods. In all specifications, 2SLS coefficients are considerable higher than OLS and probit estimates, and are precisely estimated. 2SLS instruments pass both tests of relevance and validity. Finally, we find out district middle class use community development association (CDA) membership, greater access to print and electronic media information, higher ability pay levies/taxes for public goods, greater ability to manage conflicts and the promotion of political competition as the mechanisms connecting middle class households to greater public goods proximity.