Demand for different types of public goods: evidence from Nigeria

Abstract

Preferences of Nigerian households vary across different types of public goods. For example, some prefer roads while others favor education even after controlling for the existing supply of these goods. What explains this variation? We argue that the perceived distributional consequences of specific public goods differ conditional on the personal characteristics of households. In particular, households demand the type of public good that (a) increases the utility of assets they already own and (b) resonates with their past experiences involving the lack of particular public goods. We test our argument with data on 123,000 Nigerian households. We find strong evidence for our argument across six types of public goods.

Keywords:

• Public goods
• preferences
• developing countries
• Nigeria
• simultaneous estimation

JEL CLASSIFICATIONS:

• H31
• H41
• H5

Acknowledgements

This paper greatly benefited from the comments of Vito D’Orazio, Shelby Grossman, Asli Leblebicioglu, Clint Peinhardt, David Samuels, and Todd Sandler. Any remaining errors are our own.

Notes

The authors note that no financial interest or benefit that have arisen from the direct applications of our research.

Supplemental data for this article can be accessed at https://doi.org/10.1080/00346764.2018.1424930

1 The Vanguard, ‘Housing as Tool for Economic Development’, November 16, 2015.

2 Ethnic map polygons were derived from the GeoEPR 2014 dataset which geo-codes all politically relevant ethnic groups (Vogt et al. 2015).

3 Equal Times, ‘Nigeria Gears Up for Education Protests’, September 18, 2013.

4 Daily Trust, ‘The Giant in the Dark’, April 13, 2014; Vanguard, ‘Outrageous Electricity Bills – Consumers Groan as Discos Smile to Banks’, October 29, 2014.

5 We account for possible wealth effects by controlling for levels of income in the empirical analysis. We nevertheless obtain statistically significant results for asset ownership, suggesting that asset ownership is not just a proxy for wealth.

6 Gender is an exception, with most respondents being male.

7 We also estimate the model with standard errors clustered on the district level. The results are robust to this methodological change. See Section 2 of the Online Appendix.

8 For space considerations, tables presented in the article omit the set of control variables. The full tables are available in Section 1 of the Online Appendix.

9 As education is likely to be highly correlated with income, there is the possibility that education is simply a proxy for wealth. Our empirical strategy accounts for this with controls for the welfare quintile of the household to test if – when controlling for wealth – the level of education has an independent effect on the demand for public goods.