key: cord-0038380-vox7ws35
authors: Klasen, Stephan; Waibel, Hermann
title: Vulnerability to Poverty in South-East Asia: Drivers, Measurement, Responses, and Policy Issues
date: 2014-03-03
journal: World Dev
DOI: 10.1016/j.worlddev.2014.01.007
sha: 3afae08e52a576dbeb3cd4d16b1ea23d7d823244
doc_id: 38380
cord_uid: vox7ws35

nan

During the last 10-15 years, many developing countries did very well on many fronts. Average income growth rates have been quite high, in fact substantially higher than in industrialized countries, and absolute income poverty has come down substantially (IMF, 2012) . The Millennium Development Goal (MDG) of halving absolute income poverty has already been reached by 2011, 4 years ahead of schedule (World Bank, 2013) . While much of this success in reducing poverty is related to particularly high growth rates in some populous Asian economies (including China, India, Indonesia, and Vietnam), substantial rates of per capita growth and associated poverty reduction has been experienced in the majority of countries from all regions, including Sub-Saharan Africa. In addition, education levels improved almost everywhere and mortality has fallen in many countries (World Bank, 2013) .

At the same time, the last few years have been characterized by a very high level of volatility with important repercussions for poverty in developing countries. This high volatility is, first, linked to sharply increasing and highly volatile commodity prices, including fuels, minerals, and food. While some developing countries have also benefited from these high commodity prices, many poor households have suffered as they are net purchasers of food and are also negatively affected by rising energy prices (Dorward, 2012) ; food and energy prices rose dramatically in 2007 and 2008, then fell and have risen again in 2011-12 to very high levels (from which they have now receded somewhat). A second source of volatility has been high volatility in world output growth, with repercussions for developing countries which are deeply linked to the world economy. Most notably, the sharp downturn in 2009 as a result of the global financial crisis has affected developing countries substantially, although the effect has been heterogeneous and was more muted than anticipated due to strong policy responses in industrialized and some large developing countries, as well as continued high growth in key emerging economies (IMF, 2012) . However, the continued uncertainty of global financial markets caused by high levels of debt, macroeconomic instability and unconventional monetary policy of countries in Europe and the USA poses threats for developing countries as it affects capital flows to developing countries as well as their exports; moreover, there is rising uncertainty about growth prospects of key emerging economies, including particularly China, India, and Brazil. Lastly, there are increasing dangers of global non-economic shocks that could affect seriously developing countries. Among them are natural catastrophes such as floods and droughts (some already resulting from climate change), with prospects of further extreme events likely in the decades ahead (IPCC, 2011). In addition, there is the recurrent threat of global health scares such as Severe Acute Respiratory Syndrome (SARS) or Avian flu.

These more macro-level sources of vulnerabilities filter down to household levels depending on their exposure to the markets that are negatively affected by these global shocks. Such aggregate events come on top of persistent exposure to more micro-level shocks facing households in many developing countries (Dercon, 2005) . These shocks include demographic shocks such as illness and death of household members, economic shocks such as unemployment, adverse price or output developments, some of which are in turn related to local or more widespread floods, droughts, pest outbreaks, and other natural disasters. The susceptibility to shocks is particularly high in rural areas where the exposure to shocks, especially those related to adverse weather conditions, pests and diseases, as well as volatility in input-and output prices is large. At the same time, the ability of rural households to counteract these shocks is particularly limited due to imperfect credit and insurance markets. Some of these micro-level shocks are purely idiosyncratic, affecting only individual households; others affect entire villages and regions, and some, of course, are directly linked to the macro volatilities discussed above. As a result of these risks at different levels, one cannot be sure that many households, whose average income is now above the poverty line, have left poverty for good. Also, one might suspect that the movements in and out of poverty intensify due to the increasing volatilities just described.

In response to these developments, it is critical to assess to what extent households are affected by these risks, what impacts they have if they occur, and how they can shield themselves against them. These are precisely the questions that the research agenda on measuring vulnerability to poverty and analyzing its determinants is trying to address. Starting with some initial work in the late 1990s, including the World Bank's WDR on poverty (World Bank, 1999) , this research program has defined vulnerability as an ex ante concept that is trying to assess the propensity of households to be negatively affected by shocks. In contrast to the work on income poverty measurement, which determines actual poverty levels ex post and seeks to explain trends and determinants in poverty, vulnerability research tries to examine the ex ante poverty risk of households. This is critical from both a research as well as a policy perspective. From a research perspective, this approach thereby integrates shocks and risks into poverty research and combines both strands to make ex ante assessments (and its drivers) of future poverty risks. From a policy perspective, it is clear that successful poverty reduction policies should target those people at risk of future poverty, rather than simply try to help those who were identified to be poor in the past. In fact, as static poverty rates have been falling in most developing countries (including Thailand and Vietnam, the two countries we use as case studies for our research program on vulnerability to poverty) while risks and shocks persist (and might be on the rise), the relative importance of this ex ante view of vulnerability to poverty is clearly of particular relevance for anti-poverty policies.

Despite the importance of this issue, available research on vulnerability, its determinants, and policy implications is still rather limited. While there has been a considerable amount of conceptual work on the measurement of vulnerability to poverty, there is a great lack of empirical work implementing these concepts, studying the determinants of vulnerability from both a theoretical and empirical perspective, and deriving policy conclusions. This is largely related to the high demands on specialized data required to analyze vulnerability, as well as the complexity of the task of assessing ex ante risks. As shown, for example, in the paper of Povel in this special issue, empirical implementation of a forward-looking vulnerability concept requires an accurate prediction of the likelihood of different types of shocks to materialize for different household types, the severity of such shocks, and their impact on household wellbeing. This requires panel household surveys that include detailed modules on shocks, risks, and their impacts on household well-being.

The papers in this special issue extend both the conceptual as well as the theoretical and empirical research frontier on analyzing vulnerability to poverty, with applications to the emerging economies of Thailand and Vietnam. The papers are the results of a multiyear interdisciplinary collaborative research program on vulnerability to poverty in these two countries, based also on three-wave (and continuing) household panel data covering 4000 households in six provinces of the two countries generated for this program (see Klasen & Waibel, 2013) . This special issue is bringing together selected papers from this research program that are of particular interest for scholars and policy-makers concerned about measuring and analyzing vulnerability to poverty, examining its determinants and household responses, and devising options for policy to reduce it.

The first paper concerns the conceptualization of vulnerability. While most researchers define vulnerability as the risk to be poor in future (e.g., Calvo & Dercon, 2005; Chaudhuri, Jalan, & Suryahadi, 2002; Hoddinott & Quisumbing, 2003) , the first paper by Povel argues that this approach has some shortcomings as it considers those that are expected to be chronically poor as "vulnerable" while those who might face very large shocks that nevertheless place them above the poverty line as not vulnerable. Also, these definitions are very sensitive to the (essentially arbitrary) setting of the poverty line. To address these issues, Povel proposes a new way to conceptualize vulnerability to poverty as the likelihood to face a large downside risk of falling substantially below the present level of well-being, thereby incorporating insights from behavioral economics on the importance of the current level of well-being for future welfare assessments. He then applies this measure empirically to Thailand and Vietnam, thereby also developing and applying a methodology of how, using the experience of shocks, such risks can be quantified to produce estimates of ex ante vulnerability of households. The paper nicely shows the empirical difficulties of quantifying the likelihood and impact of shocks on future well-being (and thus on vulnerability) and shows how the custom-made panel data from Thailand and Vietnam are able to overcome many of these difficulties. It also shows that his measure provides important complemen-tary insights to the existing measures in the literature. Empirically he finds that vulnerability to downside risk is higher in Vietnam than in Thailand, largely related to the higher severity of shocks there.

The next three papers examine measurement and determinants of vulnerability to poverty. The paper by Phung et al. examines measurement issues in household surveys designed to measure vulnerability to poverty. Given the complexity of survey instruments trying to measure shocks, risks, and vulnerability to poverty, measurement error is a key concern and can seriously affect measurement of poverty and vulnerability. The paper shows that interviewer, setting, and respondent characteristics affect measurement but that they can be adequately addressed to avoid biases for analyses. The authors show that controlling for these effects can enable researchers to avoid biased assessments of consumption, poverty, and risk, key issues for accurate assessment of vulnerability of households. They also show how appropriate choice of questionnaire design, timing of interviews, and choice of interviewers and respondents can be used to reduce measurement error which is of relevance for improving the accuracy of household surveys more generally.

Turning to determinants of vulnerability, the paper by Klasen et al. examines whether and which types of female-headed households are more prone to shocks and vulnerability to poverty in the two countries using panel fixed-effects models. It thereby asks the question whether there is evidence for a feminization of vulnerability. It finds that, in comparison with male-headed households de facto female-headed ones (where the male head is absent) are found to be consumption richer in Thailand, but prone to more severe shocks in both countries. Furthermore, the results suggest that in Thailand single female-headed households are less vulnerable to poverty than households headed by men. However, in Vietnam these households are particularly consumption poor and vulnerable to poverty. The authors show, however, that this is mostly due to their greater poverty rather than their higher risk exposure. Thus it is their chronically poor incomes rather than their higher risks that lead to their high poverty risk. In contrast, these same single de jure female-headed households are less vulnerable in Thailand, linked to their higher incomes and lower shock severity. This relates to the paper by Povel and shows that vulnerability measures that focus on downside risk only (as suggested by Povel) and those that focus on the likelihood of being poor (as used in this paper and in much of the literature) can yield different findings that provide important complementary insights. The findings suggest that differentiation by subgroups of headship is important for policy development and targeting as well as future research on drivers of poverty and vulnerability.

The study by Gloede et al. examines how the experience of shocks affects risk aversion of households and, as a result, can lead to household behavior that trap households in a state of high poverty risk. Poor and vulnerable households that are affected by shocks respond to the exposure to shocks with greater risk aversion, even when controlling for a large set of sociodemographic variables. Therefore, shocks can perpetuate vulnerability to poverty via their effect on risk attitudes which then lead to risk minimization strategies that reduce the likelihood to escape poverty. These results point to another important feed-back mechanism between shocks and poverty traps via the effect on risk attitudes. They extend this general finding to various categories of shocks and find differences in the responses to these shocks between Thailand and Vietnam.

The third set of papers examines household responses to reduce vulnerability focusing particularly on migration as a 2 WORLD DEVELOPMENT response to reduce poverty risk. The paper by Loc et al. examines the impact of rural-urban migration on household welfare in Vietnam as one response to escape poverty and vulnerability. It investigates to what extent shocks motivate rural household members to move to urban areas, whether migrants better off in terms of working conditions and quality of life and what the effect of migration is on rural household's welfare and vulnerability to poverty. The analysis uses panel data of 2200 households from rural Vietnam covering the period 2007-2010, and a tracking survey of 299 migrants from 2010. It shows that migration, especially migration for employment, is a livelihood support strategy for households exposed to agricultural and economic shocks. Migration for education is more likely observed among households with higher human capital and those who are financially better off. Migrants perceive themselves to be better off at the place of destination, but income losses from shocks of their rural households may reduce their employment quality. The results from difference-in-difference specifications with propensity score matching techniques suggest that migration has positive income growth effects, and that these effects are more pronounced in provinces with fewer job opportunities. These effects help not only migrant households moving out of poverty, but it also reduces poverty and vulnerability in rural areas. The paper by Junge et al. examines internal return migrants and the impact they have on reducing vulnerability and as well as their impact on regional economic development. This paper draws a distinction between local return migrants, who move back to their home village, and regional return migrants, who return to other places within their home provinces in Thailand and Vietnam. They first analyze determinants of internal return migration and find that while local return migrants are less educated than continuing migrants, this negative human capital selection cannot be found for regional return migrants. Secondly, after returning local return migrants often engage in agriculture, while regional return migrants continue in non-farm employment, thereby diversifying the household income and consequently reducing their vulnerability and bringing potential for future economic development.

The last paper by Kemper et al. examines the impact of a particular policy on vulnerability to poverty. In particular, the paper analyzes the impact of land rights on reducing vul-nerability to poverty in Vietnam, investigating whether increasing security of tenure via land titling can improve the ability of households to smooth consumption and thus manage shocks. It does so by using impact evaluation techniques controlling for endogeneity of land rights. In particular, it identifies the channel of impact through which improved property rights affect consumption volatility. It finds that land certification increases consumption levels for rural households, but also consumption volatility for a subgroup of households identified by an instrument. They show that the certification program affects consumption outcomes of this subgroup through, presumably more risky, credit-based agricultural investment at the intensive margin.

While each paper provides interesting and unique insights, some general messages cut across the papers from this special issue. First, the papers in the volume show first that it is indeed possible to empirically study vulnerability to poverty despite the high demands made on the data. For such analyses, however, custom-made panel data that carefully assess risks, shocks, and their impacts on well-being are required and great care must be exercised to reduce measurement error and increase the reliability of the results. Second, the determinants of vulnerability to poverty depend on country contexts. As the papers on gender and risk attitudes show, processes that generate and sustain vulnerability are quite distinct in these two countries, requiring country-specific analyses and their careful interpretation. Third, factors affecting mean income levels and shocks can be quite different. As the example of land titling in Vietnam or the situation of different categories of female-headed households in the two countries show, factors associated with rising incomes might go alongside higher volatility of incomes, suggesting real trade-offs. Fourth, internal outmigration as well as return migration can have a substantial impact on vulnerability of households. The results here suggest that migration pays off for the migrants and regional return migrants can help diversify income sources and thereby reduce vulnerability to poverty. Lastly, there remains much scope to further refine the concepts of vulnerability and adapt them to particular policy contexts. For some contexts, it might be relevant to focus on vulnerability as poverty risk, while in other contexts the threat of negative shocks might be particularly important for policy-makers who want to sustainably improve well-being in developing countries.

Measuring individual vulnerability: Department of Economics discussion paper no. 229

Assessing household vulnerability to poverty fro cross-sectional data: A methodology and estimates from Indonesia: Discussion paper 0102-42

Risk, insurance, and poverty: A review

The short and medium-term impacts of rises in stable food prices

Methods for microeconometric risk and vulnerability assessments: Social protection discussion paper 0324

World economic outlook

Vulnerability to poverty: Theory, measurement and determinants with case studies from Thailand and Vietnam

World development report 1999-2000: Attacking poverty

Global monitoring report