Poverty Dynamics of Households in Rural China
| Author | Jing You,Katsushi S. Imai |
| DOI | http://doi.org/10.1111/obes.12044 |
| Published date | 01 December 2014 |
| Date | 01 December 2014 |
898
©2013 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 76, 6 (2014) 0305–9049
doi: 10.1111/obes.12044
Poverty Dynamics of Households in Rural China*
Katsushi S. Imai† and Jing You‡
†Economics, School of Social Sciences, University of Manchester, UK
(e-mail: Katsushi.Imai@manchester.ac.uk)
‡School of Agricultural Economics and Rural Development, Renmin University of China,
No.59 Zhongguancun Street, Beijing 100872, China (e-mail: jing.you@ruc.edu.cn)
Abstract
The objective of our study is to identify patterns and causes of households’ transitions into
and out of poverty using the long household panel data on rural China in 1989–2009. We
propose a discrete-time multi-spell duration model that not only corrects for unobserved
heterogeneity, but also addresses the endogeneity due to dynamic selection associated
with household’s livelihood strategies. The household choosing farming or out-migration
as a main livelihood strategy was more likely to escape from persistent poverty than those
taking local non-agricultural employment. The present study emphasizes the central role
of agriculture in helping the chronically poor escape from poverty.
I. Introduction
While rural poverty has continued to decline in China due to spectacular economic growth
over the last three to four decades, much of the reduction is concentrated in two relatively
brief periods: 1979–84 and 1995–97 (Ghosh, 2010). Substantial reduction in rural poverty
was achievedin 1979–84 as a result of de-collectivization of agricultural production and the
introduction of the Household Responsibility System which dramatically raised agricul-
tural productivity (Lin, 1992). Reduction of rural poverty was accelerated again in 1995–
97 by significant increases in procurement prices of agricultural products which pushed
up income growth of rural households (Benjamin, Brandt and Giles, 2005). Since then,
although the speed of poverty reduction has slowed down (Chen and Ravallion, 2010), it
has been increasingly difficult for the government or international agencies to direct their
poverty alleviation policies or aid programmes to those who remain poor in rural China as
*Corresponding author: JingYou.The authors acknowledgeuseful comments and advice from Armando Barrientos,
Obbey Elamin, RaghavGaiha, Masashi Hoshino, Kunal Sen, Xiaobing Wang and participants inWorkshop on Poverty
and Inequality in China and India, the University of Manchester in March 2012, RIEB Seminar at Kobe University
in April 2012, Econometrics Seminar in Manchester in May 2012, and GRIPS/TWID Conference on ‘Risks, Social
Networks, and Development’in Tokyoin December 2012. The second author would like to expressthe deepest thanks
to Kristian Bernt Karlson and Francesco Devicienti for their help in programming. The research is supported by the
Ministry of Education of China (MOE) Project of Humanities and Social Science (Grant No. 13YJCZH231), the
Fundamental Research Funds for the Central Universities, and the Research Funds of Renmin University of China
(Grant No. 13XNK014). The authors greatly appreciate useful comments from three anonymous referees and John
Knight, an editor of the journal. Only the authors are responsible for any errors.
JEL Classification numbers: C33, C41, I32, O15.
Poverty dynamics of households in rural China 899
they tend to live in remote areas (World Bank, 2009). While some rural households seem
to be chronically poor, considerable mobility in and out of poverty has been reported in
rural China (e.g. McCulloch and Calandrino, 2003; Gustafsson and Ding, 2009) and thus
‘transient poverty’ is a non-negligible part of total poverty (Jalan and Ravallion, 1998;
Duclos, Araar and Giles, 2010). An effort to help the poor more efficiently thus calls for
understanding of the pattern and causes of households’ poverty transitions as households
manage their livelihoods in response to the changing environment. Incorporating a time
dimension into the analysis of household poverty is crucial not only for understanding the
evolution of households’ poverty status and underlying causes, but also for designing and
implementing effective anti-poverty programmes.
The empirical literature of ‘poverty dynamics’ has addressed transitions of poverty
status for a household or an individual over long periods in both developed and developing
countries, usually using household panel data. A typical approach to analyzing poverty
dynamics is to include lagged poverty status as an additional independent variable to
capture the dynamics of poverty (Cappellari and Jenkins, 2002; Antman and McKenzie,
2007). However, considering poverty status only in the previous year may oversimplify
the dynamics across many years and fail to recognize the cumulative nature of household
poverty.
To address this concern, the present study analyses poverty dynamics by using the du-
ration model which takes account of how long the household has been or has not been in
poverty as well as when it moved in or out of poverty. One of the significant advantages of
the duration model is to track an individual’s unique history and experience. That is, our
model captures the story in which, for example, a farming household experienced poverty
for four years due to the low productivity, but escaped from poverty as one of the household
members had access to non-farm employment and stayed ‘non-poor’ for four years and
then slipped into poverty and remained poor for the next six years due to the illness of a
household head. While some studies have analyzed poverty in developed countries using
the duration model (Canto, 2002; Devicienti, 2002, 2011; Maes, 2013), there have been
few works on developing countries.1More specifically, to explore the pattern of poverty
dynamics, we incorporate unobserved heterogeneity in a discrete-time duration model and
apply a fully non-parametric approach to the long household panel data on rural China.This
methodology aims to minimize possible misspecifications to offer better estimates. Our
framework controls for (i) unobserved heterogeneity that can be correlated across multiple
poverty transitions of each household and (ii) the dynamic selection underlying multi-path
transitions, neither of which has been done in the context of developing countries.This en-
ables us not only to understand trajectories of household poverty status, but also to identify
the underlying socio-economic factors which influence the changes in poverty status.
It is noted that the above empirical literature on ‘poverty dynamics’ is closely associated
with the parallel empirical literature on ‘poverty traps’. For example, Carter and Barrett
(2006) proposed an asset-based approach to distinguish a structural component of poverty
that is systematically poor over the years from ‘poverty that passes naturally with time
due to systemic growth process’ (p. 178). They also suggested the need for controlling
1Exceptions include Baulch and McCulloch (2002) for Pakistan, Bigsten and Shimeles (2008) for Ethiopia and
Glauben, Herzfeld and Wang(2012) and You(2011) for China.
©2013 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial