Title: Consumption vulnerability to risk in rural Pakistan
Abstract: Abstract Abstract As one of the dimensions of vulnerability, this paper empirically investigates the inability of rural dwellers to cope with negative income shocks. A variable coefficient regression model is applied to a two-period household panel dataset collected in the North-West Frontier Province, Pakistan, an area with high incidence of income poverty and low human development. The empirical model allows for a different ability to smooth consumption, approximated by a linear function of households' attributes, and controls for the endogeneity of observed changes in income, using qualitative information on subjective risk assessment. Estimation results show that the ability to cope with negative income shocks is lower for households that are aged, landless and do not receive remittances regularly. Acknowledgement The author is grateful to two anonymous referees and to Bob Baulch, Nicholas Minot, Christopher Udry, Robert Townsend, T. Paul Schultz, Koji Yamazaki, Yasu Sawada and other seminar participants at the NEUDC conference, the Pakistan Institute of Development Economics, Yale University, the Foundation for Advanced Studies on International Development, Hitotsubashi University and Tsukuba University for useful comments on earlier versions of this paper. The author acknowledges financial support from the Ministry of Foreign Affairs, Japan, and Seimei-Kai. The views reflected in this paper, however, are solely the author's own and should not be taken to represent any of the sponsors. Notes 1. See Ligon and Schechter (2002 Ligon, E. and Schechter, L. 2002. Measuring vulnerability: the director's cut, WIDER DP 2002/86, Helsinki: United Nations University, World Institute of Development Economics Research. [Google Scholar]) and Kamanou and Morduch (2004 Kamanou, G. and Morduch, J. 2004. "Measuring vulnerability to poverty". In Insurance Against Poverty, Edited by: Dercon, S. Oxford University Press. [Crossref] , [Google Scholar]) for an overview of different means of quantifying vulnerability. 2. See Harrower and Hoddinott (2004 Harrower, S. and Hoddinott, J. 2004. Consumption smoothing and vulnerability in the Zone Lacustre, Mali, IFPRI FCND Discussion Paper No.175, Washington, DC: International Food Policy Research Institute. [Google Scholar]) for another attempt to distinguish the effects of negative and positive income shocks and to incorporate heterogeneity in the ability to smooth consumption. 3. Since the dataset used in this paper is a two-period panel dataset and all empirical variables are in differences or initial levels, time subscripts are dropped below. 4. Equation (2) nests a model of village level consumption smoothing such as the one estimated by Townsend (1994 Townsend, R. 1994. Risk and insurance in village India. Econometrica, 62(3): 539–591. [Crossref], [Web of Science ®] , [Google Scholar]). The nested model corresponds to the case that b 1,k = b 2,k = 0 for all k > 0 and b 1,0 = b 2,0. 5. The empirical model used by Harrower and Hoddinott (2004 Harrower, S. and Hoddinott, J. 2004. Consumption smoothing and vulnerability in the Zone Lacustre, Mali, IFPRI FCND Discussion Paper No.175, Washington, DC: International Food Policy Research Institute. [Google Scholar]) is the closest to the one used in this paper, both allowing for different effects of income shocks depending on two factors: the sign of Δxi and household characteristics. However, Harrower and Hoddinott (2004 Harrower, S. and Hoddinott, J. 2004. Consumption smoothing and vulnerability in the Zone Lacustre, Mali, IFPRI FCND Discussion Paper No.175, Washington, DC: International Food Policy Research Institute. [Google Scholar]) examined the different impacts due to the two factors separately, while the model in this paper incorporates them simultaneously. Further, they employed only dummy variables in vector Zi . In other studies, variable coefficient models differentiating positive and negative anticipated shocks are adopted to test for credit constraints on consumption smoothing (for example, see Garcia et al. (1997 Garcia, R., Lusardi, A. and Ng, S. 1997. Excess sensitivity and asymmetries in consumption: an empirical investigation. Journal of Money, Credit, and Banking, 29(2): 154–176. [Crossref] , [Google Scholar]) and Jacoby and Skoufias (1997 Jacoby, H. and Skoufias, E. 1997. Risk, financial markets, and human capital in a developing country. Review of Economic Studies, 64(2): 311–335. [Crossref] , [Google Scholar])). The model of this paper attempts to differentiate positive and negative unanticipated shocks because this asymmetry is an important aspect of vulnerability. Kochar (1995 Kochar, A. 1995. Explaining household vulnerability to idiosyncratic shocks. American Economic Review, 85(2): 159–164. [Web of Science ®] , [Google Scholar], 1999 Kochar, A. 1999. Smoothing consumption by smoothing income: hours-of-work responses to idiosyncratic agricultural shocks in rural India. Review of Economics and Statistics, 81(1): 50–61. [Crossref], [Web of Science ®] , [Google Scholar]) also distinguished the impacts of negative and positive unanticipated income shocks explicitly. As a study using sources of income fluctuations rather than just income, Dercon and Krishnan (2003 Dercon, S. and Krishnan, P. 2003. Risk sharing and public transfers. Economic Journal, 113: C85–C94. [Crossref] , [Google Scholar]) incorporated both negative shocks (such as illness) and positive shocks (such as village aid receipt). 6. For instance, see Jalan and Ravallion (1998 Ravallion, M. 1998. Expected poverty under risk-induced welfare variability. Economic Journal, 98: 1171–1182. [Crossref] , [Google Scholar], 2000 Jalan, J. and Ravallion, M. 2000. Is transient poverty different? Evidence for rural China. Journal of Development Studies, 36(6): 82–99. [Taylor & Francis Online], [Web of Science ®] , [Google Scholar]) and McCulloch and Baulch (2000 McCulloch, N. and Baulch, B. 2000. Simulating the impact of policy upon chronic and transitory poverty in Rural Pakistan. Journal of Development Studies, 36(6): 100–130. [Taylor & Francis Online], [Web of Science ®] , [Google Scholar]), who regressed the household level measure of transient poverty, a la Ravallion (1988 Jalan, J. and Ravallion, M. 1998. Transient poverty in postreform rural China. Journal of Comparative Economics, 26(2): 338–357. [Crossref], [Web of Science ®] , [Google Scholar]), on household attributes. 7. See Kurosaki and Hussain (1999 Kurosaki, T. and Hussain, A. 1999. Poverty, risk, and human capital in the Rural North-West Frontier Province, Pakistan, IER Discussion Paper Series B No.24, March 1999, Hitotsubashi University. [Google Scholar]) and Kurosaki and Khan (2001 Kurosaki, T. and Khan, H. 2001. Human capital and elimination of rural poverty: a case study of the North-West Frontier Province, Pakistan, IER Discussion Paper Series B No.25, January 2001, Hitotsubashi University. [Google Scholar]) for details of these surveys. The reference period for each survey is fiscal years 1995/96 and 1998/99 respectively (Pakistan's fiscal year is the period from July 1 to June 30). 8. In the survey, a household is defined as a unit of co-residence and shared consumption. A typical joint family in the region, where married sons live together with the household head who owns their family land along with their wives and children, is treated as one household, as long as they share a kitchen. When the household head dies or becomes aged, the land may be distributed among sons, who start to live separately on that occasion. In our survey when we encountered such cases, each family of each son was counted as one household. 9. The most frequent reason for attrition was migration. Some households had migrated out from the village and others had sent all their adult males to work in foreign countries or in Pakistani cities. As shown in Appendix 2, attrition occurred more for households living in Village A and whose heads were more educated. Education and risky environments are thus associated with higher propensity to migrate. 10. The adult equivalence scale currently adopted by the Government of Pakistan was used in calculating the size of a household, where individuals aged more than 17 years old are given the weight of 1.0 and all other individuals are given the weight of 0.8. Pakistan Rupees (Rs.) in the empirical section of this paper are all in 1996 values. 11. The Government of Pakistan decided the official poverty line in August 2002 based on the Pakistan Integrated Household Survey, 1998/99. The poverty line corresponds to 2,350 kcal per adult per day of food intake. 12. Nevertheless, the hypothesis of an independent Markov process was rejected at 1 per cent level (χ2 test of independence in two-way contingency tables yielded a test statistic of 51.2, whereas the 1 per cent critical value for a χ2 variable with 16 degrees of freedom is 32.0). Therefore, there exists a tendency to remain in the same status although the transition to other statuses is frequent. 13. The first stage regression results for other explanatory variables in equation (2) are available from the author on request. 14. See Appendix 1 for the definition of these variables. 15. F-test applied to estimation results in Table 6 rejected the constrained version at 1 per cent level. The Hausman test rejected the exogeneity of explanatory variables in Table 6 and the instrumental variables listed in Table 5 passed the over-identification test (both at 5 per cent level). 16. Results based on different specifications are available from the author on request. 17. Instead of the adult equivalence scale currently adopted by the Government of Pakistan (see note 10), the unweighted number of household members and the Government's old formula (1.0 for those aged more than 16, 0.85 for those aged between 11 and 16, 0.75 for those aged between 6 and 10 and 0.45 for individuals aged less than 6) were tried. 18. Instead of regressing all of the explanatory variables in equation (2) on instrumental variables used in Table 5 in the first stage, Di and Δxi in equation (2) were replaced by the fitted values of Di and Δxi (obtained from regressing these two variables on instrumental variables used in Table 5) in the second stage. 19. They can be identified through a comprehensive measure of vulnerability if the underlying household model of income/consumption dynamics is specified thoroughly. See Ligon and Schechter (2002 Ligon, E. and Schechter, L. 2002. Measuring vulnerability: the director's cut, WIDER DP 2002/86, Helsinki: United Nations University, World Institute of Development Economics Research. [Google Scholar], 2003 Ligon, E. and Schechter, L. 2003. Measuring vulnerability. Economic Journal, 113: C95–C102. [Crossref], [Web of Science ®] , [Google Scholar]). 20. Models by Elbers and Gunning (2003 Elbers, C. and Gunning, J. W. 7–9 April 2003. Estimating vulnerability, paper presented at The Staying Poor: Chronic Poverty and Development Policy Conference, Chronic Poverty Research Centre, 7–9 April, University of Manchester. [Google Scholar]) and Zimmerman and Carter (2003 Zimmerman, F. J. and Carter, M. R. 2003. Asset smoothing, consumption smoothing and the reproduction of inequality under risk and subsistence constraints. Journal of Development Economics, 71(2): 233–632. [Crossref], [Web of Science ®] , [Google Scholar]) provide basic modelling ideas that are applicable to the context of this paper.
Publication Year: 2006
Publication Date: 2006-01-01
Language: en
Type: article
Indexed In: ['crossref']
Access and Citation
Cited By Count: 77
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot