Can Social Protection Tackle Chronic Poverty?

Abstract

Recent developments in social protection have shifted its focus on to risk and vulnerability. These contribute to poverty directly, but also indirectly through the response of poor households to risk. The extent to which social protection interventions could address chronic poverty is unclear. A hard and fast distinction between transient and chronic poverty suggests a bifurcation in anti-poverty policy, with social protection addressing the former, and asset transfer policies the latter. To the extent that factors behind chronic poverty extend beyond the direct and indirect impact of risk on households, social protection can at best constitute a partial response. The paper discusses these issues and concludes that ‘broad’ social protection can have an important role in interrupting risk and vulnerability among the chronic poor.

Les développements récents sur la protection sociale se sont concentrés sur les notions de risque et de vulnérabilité. Ces derniers contribuent directement aux phénomènes de pauvreté, et indirectement également, à travers la réponse des ménages pauvres au risque. L'impact des modifications de la protection sociale sur la pauvreté chronique reste peu clair. La distinction radicale entre pauvretés chronique et transitoire suggère une modification des politiques anti-pauvreté, la protection sociale étant affectée à la pauvreté transitoire, et les politiques de transfert d'actifs à la pauvreté chronique. Dans la mesure où les facteurs déterminants de la pauvreté chronique dépassent les questions de l'impact direct et indirect du risque sur les ménages, la protection sociale peut au mieux constituer une réponse partielle. L'article propose une analyse de ces problèmes, et conclut qu'une protection sociale ⟨élargie⟩ peut avoir un rôle important en matière de réduction du risque et de la vulnérabilité associée à la pauvreté chronique.

Notes

Armando Barrientos is Senior Lecturer in Public Economics and Development and David Hulme is Professor of Development Studies at the Institute for Development Policy and Management, University of Manchester, UK; Andrew Shepherd is Senior Research Fellow, Overseas Development Institute, London, UK. The authors would like to thank participants at the CPRC ‘Staying Poor: Chronic Poverty and Development’ Conference held in Manchester, 7–9 April 2003, and Andrew McKay for helpful comments on an earlier version of this paper.

 1. See the selection of papers in a Special Issue on ‘Chronic Poverty and Development Policy ’ in World Development [Hulme and Shepherd, 2003 b]. Also see Chapter 4 of The Chronic Poverty Report 2004–05 [ Chronic Poverty Research Centre, 2004 ].

 2. The World Bank's ‘Social Protection Sector Strategy’ paper finds important similarities between social protection and the sustainable livelihoods approach of the UNDP (United Nations Development Programme) and DFID (Department for International Development) [ World Bank, 2001 ].

 3. For a discussion of the transformation of social protection within the World Bank, see World Bank [2001], and for a restatement of social protection within the ILO, see ILO [2001].

 4. See for example the discussions on targeting [ Atkinson, 1995 ; Sen, 1995 ; van de Walle, 1998 ].

 5. Our focus is on donor definitions here because they have been the main actors in promoting the concept. To varying degrees, the governments of developing countries adopt a donor definition, merge elements of different definitions, or use the term as a broad label lacking a clear specification.

 6. Jalan and Ravallion, for example, ‘define transient poverty as the contribution of consumption variability over time to expected consumption poverty. The non-transient component is the poverty that remains when inter-temporal variability in consumption has been smoothed out’ [2001: 83].

 7. However, it does not exclude risk and vulnerability as a causal factor in observed chronic poverty.

 8. This applies even where the chronically poor are defined as those poor in some, or most, observations. In the context of a panel with six observation points, with consumption collapsed to a [0,1] range and with the chronically poor defined as those with at least three poverty spells, households with consumption patterns [0,0,0,0,0,0], [0,0,0,1,1,1], or [0,1,0,1,0,1] are observationally equivalent.

 9. They decompose the poverty gap for each household into a time-mean (chronic) and a time-variant (transient) component.

10. In the context of asset acquisition and diversification as a means of reducing vulnerability to poverty (transient poverty), Dercon [2001] notes that asset prices are normally covariant with income such that raising the asset base of the poor may not reduce their vulnerability, it may actually raise it. Jalan and Ravallion [2001] acknowledge that facilitating capital investment among the chronic poor may actually increase their exposure to risk.

11. Methodological issues surrounding the poverty line [ Ravallion, 1996 ] and the dimensions of poverty [ Bourguignon and Chakravarty, 2002 ] are also relevant here.

12. The issue of measurement error is relevant here [ Baulch and Hoddinott, 2000 ; McKay and Lawson, 2003 ].

13. In the context of a panel with six observation points, the chronically poor defined as those with time-mean income at or below a poverty line of 0.5, and with consumption collapsed to a [0,1] range, households with consumption patterns [1,1,1,0,0,0], [0,0,0,1,1,1], or [0,1,0,1,0,1] are observationally equivalent.

14. In the third measure of chronic poverty, these issues do not arise because risk and vulnerability affect both chronic and transient poverty. Suryahadi and Sumarto [2001] carried out a study of changes in poverty associated with the 1997 Indonesian crisis. Using a version of the predicted consumption approach (CP3) in Table 1, they find that both transient and chronic poverty rose after the crisis, but proportionally the rise in chronic poverty was steeper. A rise in vulnerability to poverty directly affects chronic poverty.

15. A number of studies, especially those focusing on the multidimensionality of poverty, have suggested that the poor and non-poor are fuzzy sets [ Chiappero Martinetti, 2000 ; Lelli, 2000 ].

16. An additional analytical complexity arising when attempting to measure chronic poverty concerns the likelihood that those experiencing chronic poverty will have higher mortality rates because they die from causes that are eminently preventable among the non-poor. As a consequence, a preventable death (the ultimate poverty experience?) may appear, in the analysis of poverty incidence in panel datasets, as improved average household standards of living for the survivors.

17. ‘Drivers’ in the language of Hulme and Shepherd [2003a] and Hulme, Moore and Shepherd [2001].

18. This was the argument of the Culture of Poverty school of thought in the 1950s and 1960s.

19. It is important to consider more closely the data used. Chronic poverty is characterised by deprivation persisting over time, and in Figure 1 the persistent poor are those with adult equivalent incomes at or below 60 per cent of median income in at least three consecutive years in the period 1993–1997. The data were reported in Fouarge [2003] and correspond to waves 1 to 5 of the ECHP Survey. Social protection is proxied by social security and welfare spending as a proportion of GDP. This is a measure of transfer payments to compensate for loss of income or inadequate earnings capacity. The value used in Figure 1 is an average of five-year averages for the period 1972–1997. The data were reported in Besley et al. [2003] and come from IMF Government Finance Statistics. The data used can be said to reasonably capture the time dimension involved in the relationship postulated.

Additional information

Notes on contributors

Andrew Shepherd

Armando Barrientos is Senior Lecturer in Public Economics and Development and David Hulme is Professor of Development Studies at the Institute for Development Policy and Management, University of Manchester, UK; Andrew Shepherd is Senior Research Fellow, Overseas Development Institute, London, UK. The authors would like to thank participants at the CPRC ‘Staying Poor: Chronic Poverty and Development’ Conference held in Manchester, 7–9 April 2003, and Andrew McKay for helpful comments on an earlier version of this paper.