Spatial Poverty in Benin: Human Capital, Isolation and Governance

Abstract

Benin has witnessed significant economic growth over the past decades, positioning itself as a rising economy in the west African subregion. This notwithstanding, persistent subnational differences in poverty were observed, suggesting a challenge of growth inclusiveness. This study provides evidence on the effects of human capital differences, isolation, and institutional quality on the determinants of household expenditure and factors that explain the southern-northern welfare gaps. This study employs a decomposition technique based on re-centred influence functions. I find significant spatial differences in consumption expenditure at the mean and across selected quantiles, with an increasing gap along welfare distribution and driven largely by differences in the returns to households’ endowments in social public infrastructure. Access to energy and human capital measured by the education level of the head of the household generally favours the rich in the leading region and contributed significantly to explaining the difference in the welfare gap. The difference in access to roads widens the gap for the poor, whereas governance quality at the local level reduces it. To reduce the poverty gap, a public policy to improve returns of household endowments in the social sector should be integrated with other initiatives to address the underlying determinants of inequalities.

Keywords:

• inequality
• spatial poverty traps
• human capital
• isolation
• institutions
• governance

Acknowledgements

I thank the editor and the two anonymous referees for their valuable comments that helped improve the manuscript. I also sincerely thank Iangola Andrianarison for her diligent proofreading. I am also grateful to the West African Economic and Monetary Union (WAEMU) for providing the household survey data used in this study. Any remaining errors and omissions are the responsibility of the author. The views expressed in this paper are those of the author and do not necessarily reflect those of the United Nations Development Programme (UNDP).

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The global bottom 50 per cent captures only 8.5 per cent of total income and owns 2 per cent of wealth measured at purchasing power parity (PPP), while the top 10 per cent owns 76 per cent of total household wealth and captures 52 per cent of total income. Note that top wealth holders are not necessarily top income holders. Income is measured after the operation of the pension and unemployment systems and before taxes and transfers.

2 In 2013, the International Monetary Fund (IMF) qualified sub-Saharan Africa as the second fastest-growing region in the world, trailing only developing Asia (IMF African Department, & International Monetary Fund, 2013). However, for the African Development Bank, Africa has become the fastest growing continent in the world (Adb, 2013). In his recent World Economics Outlook Database for October 2021, the IMF again emphasised many African countries as the fastest-growing economies in 2022

3 The poverty rate increased between 2011 and 2015, reaching 36.2 per cent in 2011.

4 Target 1.1: By 2030, eradicate extreme poverty for all people everywhere, currently measured as people living on less than $1.25 a day. The poorest 20 per cent of people globally includes everyone living below the $1.25-a-day extreme poverty line; Target 10.1: By 2030, progressively achieve and sustain income growth of the bottom 40 per cent of the population at a rate higher than the national average

5 Benin is divided into 12 departments (French: départements) and 77 communes.

6 To examine the relationships between isolation and poverty, it is highly conveyable to focus on a limited number of indicators. One convenient way of doing this is through statistical techniques for data reduction, such as factor analysis (Afifi et al., 2011), and by selecting the factors that best summarise the data and using each factor as a single dimension summary index of isolation. This technique has been widely used to define an index of household assets by combining information on a diverse range of assets that households may own and reducing it to a single asset index (Sahn & Stifel, 2000).

7 I have grouped the remainders in a single modality because they represent only 6 per cent of the answers. The answers are: once a day, several times a week, once a week, once a month, and other

8 (i)Political functions to transform the inhabitant into a citizen and strengthen the sense of belonging to a national community and of local pride; (ii) The functions of territorial administration to improve the quality of local governance in all areas of communal competence (own, shared, delegated); and (iii) Territorial development functions to lift communities out of poverty.

9 In stata, I use the moransi command to calculate Moran’s I statistic (Kondo, 2021), and run the getisord command for the Getis-Ord $G{*}_i(d)$ statistic (Kondo, 2016).

10 The dataset does not contain any information on geographic location at the household level, either latitude or longitude coordinate, so I consider the commune as a local unit. The geographical coordinates at the communal level are taken from the simple maps website, which includes a city’s latitude and, longitude. This is a subset of all 3,623 places in Benin (and only some of the fields) from the World Cities Database, available at https://simplemaps.com/data/bj-cities

11 First-order stochastic dominance shows that this classification is independent of the poverty line.

12 I follow the standard poverty theory by including age squared as an explanatory variable. The life-cycle hypothesis indicates a possible quadratic relationship between welfare and age: comparatively, poverty is higher at younger ages, decreases in one’s middle-age, and rises again beyond a certain age.

13 0.11 = 0.1356/12.231 × 10)(resp. 0.10 = 0.1309/12.557 × 10).

14 0.66 = 0.8098/12.231 × 10 (resp. 0.79 = 0.9959/12.557 × 10).

15 The option rwprobit() is used. This procedure is based on the parametric strategy described in S. Firpo and Pinto (2016). This can be considered the equivalent of teffects, ipwra, and is part of the identification strategy of the counterfactual distribution (Firpo et al., 2018).

16 The errors are relatively small and statistically insignificant, indicating that the RIF-regression model does relatively well at tracking down the composition effect estimated consistently using the reweighting procedure.