BACKGROUND PAPER FOR THE

WORLD DEVELOPMENT REPORT 2013

Jobs and Transitions Out of

Poverty: A Literature Review

Gabriela Inchauste
Abstract

This paper summarizes the existing literature on the links between labor and poverty reduction.
The vast majority of studies find that more and better paid work is critical in lifting people out of
poverty. Studies analyzing poverty dynamics and economic mobility find that improved returns
to endowments, particularly the returns to human and physical capital, are critical for exiting
poverty. Although non-labor incomes and demographic changes can help to lift the poor out of
poverty, there is no substitute for providing an enabling environment that fosters better work and
pay for the poor.

JEL classification: Q12, R23, J22

Keywords: Poverty dynamics

The findings, interpretations, and conclusions expressed in this paper are entirely those of the
authors. They do not necessarily represent the views of the World Development Report 2013
team, the World Bank and its affiliated organizations, or those of the Executive Directors of the
World Bank or the governments they represent.
 -1-

I. Introduction

The most plentiful resource that poor households have is their own labor. What are the
conditions under which more or better paid work allow people to exit poverty? Can these
conditions be summarized into a set of events or circumstances? A better understanding of the
enabling environment that allows the poor to pull themselves upward through their own efforts is
critical for the design of interventions aimed at poverty reduction and prevention.

This paper aims to summarize the existing literature on the links between labor and poverty
reduction through two related questions. First, how different are the events that lead people out
of poverty from those that drag people into poverty and what is the role of employment? Second,
how important are the returns to work relative to demographics, or other shocks? The
methodologies used to answer these questions have been varied, ranging from models estimating
the determinants of transitions across poverty states using household panel surveys, to models
that look at changes in welfare over time, to qualitative surveys based on community-based
inquiries. In each case the questions revolve around issues of economic mobility and poverty
dynamics.

These questions are related to a wide literature which has dramatically grown in the last decade.
Baulch and Hoddinott (2000) began summarizing existing studies on economic mobility and
poverty dynamics in developing countries at a time when panel data were scarce, resulting in a
limited number of studies able to track individuals over time. Baulch and Hoddinott argued that
this lack of data led to a considerable knowledge gap in understanding the factors associated with
movements into and out of poverty (poverty dynamics), and why some households increase their
well-being over the long-run (economic mobility). Since then, the number of panels available in
developing countries has expanded considerably1 along with the economic literature.
Development economists have since distinguished between chronic and transient poverty2,
explored tests for nonlinear income dynamics as evidence of poverty traps 3; estimated the effects
of volatility and risk on individual welfare4; estimated reduced-form determinants of income
mobility5; and analyzed the determinants of transitions into and out of poverty.6

With the growing availability of panel data, some of the caveats to their use have also become
clearer, including their representativeness over long periods of time and the inevitable problems
of attrition and non-response.7 Moreover, even when panel data are available, different
1
A recent compilation of databases in developing and transition countries by Baulch (2011) adds up to 66 different
panel datasets for developing countries. See
http://www.chronicpoverty.org/uploads/publication_files/Annotated_Listing_of_Panel_Datasets_in_Developing_an
d_Transitional_Countries.pdf
2
See Jalan and Ravallion (2000), McKay and Lawson (2003), Porter and Quinn (2008), Foster (2009), Calvo and
Dercon (2009); and Duclos et al (2010). For a recent review of methods used to define chronic poverty see Baulch
(2011).
3
Carter and Barrett (2006), Antman and McKenzie (2007), and Quisumbing and Baulch (2009).
4
See Gottschalk and Spolaore (2002); Bigsten and Shimles (2004)
5
See Cunningham and Maloney (2000), Maloney, Cunningham, and Bosch (2004), World Bank (2004).
6
For recent overview of methods, see Addison et al (2009) and Baulch (2011). For a survey in the context of Latin
American countries see Fields et al (2007).
7
For a review of problems with panel data see for instance the Special Issue “Attrition in Longitudinal Surveys,” of
The Journal of Human Resources, Spring, 1998, Vol. 33, No. 2.
 -2-

methodologies are available. Studies interested in better understanding poverty dynamics have
investigated the determinants of transitions into and out of poverty by modeling discrete
transitions or duration models to distinguish between short and long poverty spells. Studies
interested in economic mobility, on the other hand, typically use a continuous welfare measure
such as income or expenditure to model changes in welfare. Each method has its advantages and
drawbacks. Discrete choice models use high frequency panels that allow us to see that gross
movements into and out of poverty are typically much larger than net changes in the poverty
headcount. Moreover they highlight the importance of studying the drivers underlying the
upward and downward movements separately, since the underlying causes are not symmetric.
However, these models are sometimes seen as problematic because they reduce a continuous
welfare variable into discrete categories by imposing what is essentially an arbitrary poverty line.
Modeling changes in expenditure or income, on the other hand, allow the use of a continuous
variable, and are typically used to look at longer term dynamics. However these models cannot
easily link changes in welfare to poverty transitions. Finally, although duration models are able
to distinguish between short and long spells into poverty, they typically require multiple rounds
of panel data that are not always available for developing countries. In practice, each modeling
approach has been widely used, often with adjustments to minimize their drawbacks.8

In addition to the wider availability and use of panel data, micro-decomposition methods9 have
sought to account for the contributions to poverty reduction stemming from labor versus non-
labor income, while pseudo panel methods have substantially improved and can now delve into
some issues of economic mobility.10 The big advantage of microdecomposition and pseudo panel
methodologies is that they can be used on a wider set of countries with limited data. Typically,
they use multiple cross-sections of representative household surveys so that as long as the
surveys are comparable, they can look into movements that span long periods of time and do not
have the problems of attrition seen in panels. However, these methods cannot be used to identify
the events that allow people to transition out of poverty, nor do they answer questions about the
drivers of economic mobility.

This review finds that regardless of the methodology employed, the literature points to the
conclusion that more and better paid work has been critical in lifting people out of poverty.
Whether additional sources of household labor income come from greater diversification, higher
earnings per hour, or a greater number of hours worked, labor is at the core of what counts for
poverty reduction. Households able to exit poverty have experienced either improved returns to
labor (through more education or other productive assets), or higher returns to their human or
physical capital endowments.

The rest of the paper is structured as follows. The next section describes the literature on poverty
dynamics that has flourished in settings where income can be measured reasonably well both at
the household and at the individual level, where panel data is frequent and abundant, and where
the contribution of labor income can be quantified directly. The paper then turns to the evidence

8
See Baulch (2011) for a review of the methodological issues that arise when analyzing poverty dynamics using
panel data including measurement error, attrition and tracking.
9
See Juhn, Murphy and Pierce (1993), Bourguignon, Ferreira and Lustig (2006) and Barros et al (2006). For recent
reviews of these methods see Fortin et al (2011) and Essama Nssah (2012).
10
See Lanjouw et al (2011) for new pseudo panel methods to analyze income mobility.
 -3-

on economic mobility in settings where the rural population is a larger fraction of the economy,
where the focus has been on human and physical capital endowments and the impacts of shocks
on welfare. Section 4 describes alternative methods in the analysis of poverty dynamics,
including community based enquiries, mixed methods, and non-income dimensions of poverty.
Section 5 concludes.

II. Poverty dynamics when income can be measured

This section summarizes a growing literature on poverty dynamics, where the objective is to
determine the conditions under which a household moves out of poverty. Early studies using
panel data in developed countries used simple tabulations of the number of times a household
was poor over a fixed time period and variance components models. This literature then moved
to duration models led by work pioneered by Bane and Ellwood (1986), who pointed out that
different events are likely to have different impacts on poverty transitions and on the length of
time that households remain in poverty. Since then this methodology has been applied to a host
of countries where panel data are available and labor income can be measured relatively well. As
discussed below, the vast majority of empirical work in this area across a wide range of countries
suggests that most of the observed changes in poverty can be attributed to labor events, defined
as changes in household members' labor market attachment and earnings.

Early Models

Although tabulations of the number of times poor over a period offer a simple and transparent
account of changes in individual or household status, there is no attention focused on the events
which lead people into and out of poverty. As pointed out by Bane and Ellwood (1986), it is very
difficult to trace processes whereby persons may gradually or suddenly escape from poverty.
Moreover, the method takes no account of whether the observed poverty spells are censored or
not. This can lead to misleading conclusions about the duration of poverty spells and the
prevalence of short versus long spells.

Lillard and Willis (1978) were the first to use a variance components model to summarize the
probabilities of making a poverty transition for persons with different characteristics. In these
models, „income‟ is the dependent variable and dynamics are introduced through the residual
error structure, with assumptions about permanent and transitory components.11 In these models,
deviations from permanent income tend to be treated as random and behaviorally equivalent.
Typically all “disturbances” in income lead to the same temporal path of income in the future.

Hazard Models

However, as Bane and Ellwood (1986) argued, all changes in family income are not likely to
lead to the same sort of long-run dynamics given household income has many different sources,
across multiple household members, and can therefore change because any one of the income

11
Lillard and Willis (1978) assumed that error components are normally distributed, but Ulrick (2008) shows that
the normality assumption is not essential for this exercise. For a recent review of these models, see Jenkins (2011),
chapter 6.
 -4-

sources, household size or composition change. In fact, demographic changes have a dual effect,
changing both the number of people contributing to the household, as well as the household
equivalence-scale rate. In contrast, they argued that poverty transitions are triggered by the
occurrence of significant life-course events such as changes in household members' labor market
attachment and earnings, changes in non-labor income, or changes in household demography.
Therefore, these events are likely to have different impacts on poverty transitions and the length
of time a household remains poor. This idea led Bane and Ellwood (1986) to analyze poverty
transitions on the basis of duration models with the objective of understanding the sort of adverse
events that lead people into poverty and whether duration depends on how the poverty spell
began, and how (if ever) families are able to escape poverty.

Using the Panel Survey of Income Dynamics (PSID), Bane and Ellwood identify continuous
periods during which income falls below the poverty line and then calculate exit probabilities (or
hazard rates) to generate distributions of spell lengths for new spells. A key feature of this
method is that beginning and ending events are classified into mutually exclusive categories,
with the events defined using a hierarchy of event „importance‟. In this hierarchy, significant
changes in family structure change are most important, followed by changes in the sources of
income and changes in the needs of the household. Since every poverty transition is associated
with a single event, then by construction the sum of the trigger events add up to all observed
poverty changes. It is therefore straightforward to assess the importance of different trigger
events from the aggregate point of view. This approach has since been replicated in Stevens
(1994), Jenkins (2000) and Jenkins and Rigg (2001) for Britain, Oxley et al (2000/1) for six
OECD countries, and Cantó for Spain (2003).12 As panel data have become available in
developing countries, the approach has been used by Woolard and Klassen (2005) for KwaZulu
Natal in South Africa, Bigsten and Shimless (2004) in Ethiopia, and Nielson et al (2008) in
Chile.

In each of these studies, labor market changes were the most common reason for a significant
change in well-being (Figure 1). For example, Woolard and Klasen (2005) find that changes in
household income associated with the head or other family members getting a job are the most
important reason for households getting out of poverty in South Africa. Similarly, Nielson et al
(2008) find that 93 percent of households exiting poverty were associated with changes in labor
earnings in Chile.

12
Bane and Ellwood's method and a closely related one have been used to study poverty transitions in Britain by
Jenkins (2000), and Jenkins and Rigg (2001). Using the formats established by the latter, the Department for Work
and Pension's annual Low Income Dynamics publication now also includes tables linking trigger events and poverty
exits and entries. For a recent review, see Jenkins, 2011.
 -5-

Figure 1. Poverty Exits Classified by Main Trigger Effect

(share of all poverty exits)
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%

USA (1970-1982)

USA (1989-1993)
Argentina (B.A.) (1991-2003)
Sweden (1991-1996)

Spain (1985-1995)

UK (1991-1999)
Canada (1990-95)

Germany (1991-96)

Chile (1996-2001)

Peru (2002-2006)
United Kingdom (1991-96)

Argentina (2003-2006)

Costa Rica (2006-2008)

Brazil (2003-2006)

Ecuador (2002-2006)
South Africa (Kwazulu-Natal)
(1993-1998)

Bane and Ellwood (1986) method Jenkins and Schluter (2003) method

Non-labor 2/ Labor 1/

1/ Labor events include changes in household members' labor market attachment and earnings.
2/ Includes both changes in non-labor earnings (e.g., rents, pensions, land assets) and demographic changes
(change in number of children or adults, change in the head of household).
Source: Oxley et al (2000/1); Cantó (2003); Woolard and Klassen (2005); Nielson et al (2008); Beccaria and
Maurizio (2009); Beccaria et al (2011).

One problem that has been recognized with this approach is that multiple events may occur
simultaneously, and although one could treat each joint occurrence as a separate event, there are
practical limits. In a closely related methodology, instead of defining a set of mutually exclusive
events using a predefined hierarchy, Jenkins and Schluter (2003) focus on a subset of the most
important events individually as well as jointly. To make this more explicit, suppose that there is
a set of mutually exclusive events j=1,…,J, which trigger exits from poverty. Then, among poor
households the probability of exit from poverty between one year and the next is given by the
sum of the probabilities for households that exit by each of these different events:

(1)

Each term on the right hand side can be written as the product of the probability of each event
and the probability of exit conditional on event occurrence:

(2)
 -6-

This decomposition of the probabilities of transitions allows an assessment as to whether the

importance of a given event results from its high probability of occurrence, or from its strong
impact on the household income. This method has since been used by Cantó for Spain (2003),
and Beccaria et al (2011) for five countries in Latin America.

Despite the difference in method with the original Bane and Ellwood (1986) approach, these
studies tend to arrive at similar conclusions, namely that events exclusively related to the labor
market account for over 50 percent of the triggers associated with poverty exits (Figure 1).
Maurizio et al (2011) find that the most frequent among these labor events are either wage
growth (Argentina and Ecuador) or a rise in the number of employed household members
(Brazil, Costa Rica, and Peru), although these new jobs tend to be informal or precarious jobs
among the poor.

A fundamental problem with these methods recognized in the literature is that „changes in both
income and household size are themselves driven by a number of inter-related decisions about
household labor supply, household formation and fertility, as well as government tax and transfer
decisions (Oxley, Dang, and Antolín 2000). There have been some attempts to deal with this
problem. For instance, Machado and Ribas (2010) construct time-varying covariates that
represent changes in the Brazilian labor market in order to estimate the effects of aggregate
demand on poverty duration. Moreover, they find that the movement of industrial and
commercial workers into the service sector in metropolitan areas has caused a significant
reduction in the length of time that households remain in poverty. However, this movement has
been slow. They show that increasing the average wage of formal sector workers causes a
reduction in the probability of exiting poverty, while increasing the average wage of informal
sector wages causes an increased probability of exiting poverty.

Other papers have attempted structural models to deal with this problem. For instance, there are a
few studies evaluating the causal effect of divorce on income (see, for example, Aassve et al.
2007 and Ananat and Michaels 2007). However, these evaluation methods are often focused on
particular events or subsamples of the population, whereas the intended scope of analysis seeking
to better understand the dynamics of poverty is much wider. As a result, despite the potential
endogeneity between trigger events and poverty transitions, accounting for poverty transitions
among the population as a whole are still considered useful (Jenkins, 2011).

III. Poverty dynamics in rural settings

While the approaches described above have been useful in separating the impact of labor-related
events on transitions into and out of poverty, most of these studies have relied on frequent rounds
of panel data and contexts in which labor income can be measured relatively easily. These
methods have been less widespread in contexts where the majority of the population lives and
works in the agricultural sector, where it is difficult to separate the returns to labor from the
returns to land or other assets. In these contexts, the movement away from farm activities and
into off-farm income has often been associated with economic mobility. This section reviews
recent research that has exploited panel data to further probe the impact of rural non-farm
activity in poverty reduction.
 -7-

Farm and Nonfarm distinctions

Rural households are highly diversified, with off-farm sources of income accounting for 50
percent of total rural income in most countries (Figure 2). Not surprisingly then, studies focusing
on rural populations have given greater emphasis to movements out of poverty on account of
movements from the traditional rural agricultural sector to the nonfarm sector.13 Indeed, nonfarm
sources of income are consistently associated with increasing income inequality. For instance,
Davis et al (2010) find that wealthier households in rural areas have a higher level of
participation in, and greater income share from nonfarm activities. Conversely, agricultural
sources of income are generally the most important for the poorest households, with a greater
share of households at the bottom of the income distribution specializing in on-farm activities.

Figure 2. Sources of Income and Diversification

A. Source of Earnings B. Income Diversification
(share of rural income generating activities in (share of rural households with a diversified
total income) income portfolio)
100 60
90
80 50
70
60 40
50
40 30
30
20 20
10
0 10
Nigeria, 2004
Indonesia, 2000

Pakistan, 2001

Nicaragua, 2001
Bangladesh, 2000

Guatemala, 2000
Nepal, 2003

Vietnam, 1998

Madagascar, 1993
Bulgaria, 2001

Ghana, 1998

Tajikistan, 2003
Malawi , 2004
Ecuador, 1995
Panama, 2003

Albania, 2005

0
Pakistan, 2001

Nicaragua, 2001
Vietnam, 1998

Nepal, 2003

Guatemala, 2000
Ghana, 1998

Bulgaria, 2001

Tajikistan, 2003
Malawi , 2004
Nigeria, 2004

Panama, 2003

Albania, 2005
Indonesia, 2000

Ecuador, 1995
Madagascar, 1993

Bangladesh, 2000
Agricultural total (%) Off-farm total (%)

Source: Davis et al (2010).

However, as noted by Lanjouw (2007), simple stories about growth in the nonfarm sector driving
down rural poverty do not survive close scrutiny. Although it is true that rural poverty has been
declining alongside growing nonfarm sectors in some countries, this does not mean that the
nonfarm sector was responsible. Indeed, the direction of causality could have been the reverse or
alternatively both poverty and nonfarm sector growth were driven by a third factor, such as
technological change.

In fact, the evidence suggests that there is wide heterogeneity in rural nonfarm activities, and the
extent to which the poor can participate in an expansion of that sector varies significantly across
countries. As Reardon et al (2000) describe, there is evidence that in some contexts poor people
may be unable to overcome entry barriers to nonfarm activities. It would be an error to assume

13
See Lanjouw and Lanjouw, (2001); and recent reviews in Haggblade et al (2007), and Haggblade et al. (2010).
 -8-

that one can address poverty and inequality in the nonfarm sector without addressing farm-side
problems and vice versa.14

In order to better understand the relationship between the nonfarm sector and poverty dynamics,
it is best to look at the dynamic interactions using panel data sets. For instance, careful analysis
looking at region-level panel data for India finds that poverty is associated in a complex way
with the nonfarm sector. Although expansion in the nonfarm sector is directly associated with
poverty decline, and therefore expansion of the nonfarm sector also puts pressure on wage rates
in agriculture, where most of the poor are concentrated it also contributes to poverty reduction
indirectly (Lanjouw and Murgai, 2009).

Similarly, based on a series of panel data studies for East Asia, Otsuka et al (2009) argue that the
development of the nonfarm sector has led to an increase in nonfarm income and a major
reduction in rural poverty. In particular, they postulate that in the early stage of development,
when farming was a dominant source of income, access to land and agricultural technology were
the major determinants of farm household income. They point to work done by Estudillo,
Sawada, and Otsuka (2009) and Takahashi and Otsuka (2009), who use intergenerational
household surveys of villages and provinces in the Philippines spanning 2 decades. After
studying the determinants of changes in household incomes and educational attainment, they
conclude that the spread of the Green Revolution and the implementation of land reforms
stimulated the development of a land-pawning market and investments in schooling of the next
generation. Otsuka et al (2009) groups similar work and findings found in Thailand15,
Bangladesh16, and Tamil Nadu, India17 and argue that this reflects the notion that as economies
develop, the availability of nonfarm jobs increases so the importance of both the "quality" and
the "quantity" of human capital is highlighted, leading to a shift of household income structure
away from the farm to the nonfarm sector, motivated by an increase in nonfarm wage earnings.
They conclude that the development of the domestic nonfarm labor market was the single most
important factor behind the observed poverty reduction in these rural villages. However, as
pointed out by Thorbecke (2009), this analysis implicitly assumes that a larger supply of
educated workers from farm households will automatically obtain nonfarm jobs, since no attempt
was made to explore the determinants, extent, and magnitude of effective demand for nonfarm
jobs.

IV. Understanding Economic Mobility

Beyond the distinction between farm and nonfarm activities, there is a wide literature on
economic mobility, aiming to understand why certain households are able to increase their well-
being over the long-run, and what prevents other households from doing the same. These efforts
have led to models that focus on a continuous welfare measure, such as income or consumption,
which estimate the impact of idiosyncratic and community level shocks. Other models have
directly modeled transitions into and out of poverty. Transition studies are useful to the extent

14
Similar points have been raised by Dercon and Krishnan (1996), Escobal, (2001), and Lanjouw (2007).
15
Cherdchuchai and Otsuka (2006) use panel data for six villages between 1987 and 2004 in Thailand.
16
Nargis and Hossain, (2006) and Hossain et al (2009) use a panel from 1988-2004 in Bangladesh.
17
Kajisa and Palanichamy (2009) use detailed daily records of farming households from 1971 to 2003 in Tamil
Nadu, India.
 -9-

that they enable us to see that gross movements in and out of poverty are typically much larger
than net changes in the poverty ratio. Moreover they point to the fact that it is important to study
the drivers underlying the upward and downward movements separately, since the underlying
causes for these transitions are not symmetric. The results from these studies suggest that escape
from poverty is often due to a combination of improved labor returns to endowments, the
possibility that households accumulate assets, and to some extent, good fortune.

The impact of shocks on changes in household welfare

Beyond the distinction between farm and non-farm activities, studies have more generally
focused on economic mobility by modeling the change in household income or consumption
over time (Fields 1999 and 2003, Dercon, 2004, Deininger and Okidi, 2003, and Gunning et al,
2000). In particular, Dercon (2004) was inspired by the standard empirical dynamic growth
models as in Mankiw et al. (1992), where household growth rates are negatively related to initial
level of income, and positively related to a number of variables determining initial efficiency and
the steady state, including investment rates in human and physical capital. Assuming a Cobb–
Douglas production function for output, dependent on capital, labor and human capital, and
constant returns to scale, he models the change in consumption as:

(3)

where represents the household level of capital per capita and is a vector of commune or
region levels of capital (infrastructure, institutions and so forth) (Dercon, 2004). In order to allow
for the effects of economic shocks on consumption expenditure growth, Dercon assumes the
existence of a multiplicative risk, which can be idiosyncratic or common to all households in a
particular commune or region. Dercon argues that it is possible to introduce risk into equations
(3) as a control for shocks without adding any further distributional assumptions about the shock.
This risk variable can also be used to analyze the persistence of shocks across time when
sufficient data is available for a long period of time. With this model, Dercon (2004) investigates
the impact of weather and health related shocks on economic growth.

This type of model has been useful in determining the impact of economic shocks on changes in
household welfare characteristics and changes in the returns to those characteristics.18 However,
as Justino et al (2008) argue, consumption and earnings functions which impose constant
parameters across the entire consumption distribution, limit their application to the analysis of
the impact of economic shocks on household poverty transitions when the determinants of
household welfare have different returns to the poor when compared to the non-poor. In addition,
the model introduces shocks that are multiplicative in a Cobb-Douglas and therefore additive in
equation (3) so that shocks are analyzed individually. However, this abstracts from the
possibility that households might experience multiple shocks simultaneously.

One way to link the welfare variable with poverty status is to include the poverty status in the
initial year as one of the independent variables of a fixed effects regression. For example, Fields
et al (2003) use panel data for Indonesia, South Africa, Spain, and Venezuela and regress the
change in the log of per capita adult equivalent income on initial conditions such as the level of
18
For a recent literature review on studies using this approach in Latin America, see Fields et al (2007).
 - 10 -

income, the household head‟s characteristics, and changes in the head‟s and household
characteristics in four countries. Their results consistently show that initial income and job
changes of the head are the most important variables in accounting for income changes.
Moreover, changes in labor earnings are more important than changes in other sources of
income. Similarly, Woolard and Klasen (2005) use the same approach for South Africa, Lawson
et al (2006) for Uganda, and Lohano (2011) for Pakistan. In each case, initial conditions matter.
For instance, Woolard and Klasen (2005) find that initial household size, poor initial assets, poor
initial education and poor initial participation in the labor market hinder advancement for the
poor. However, they also find that improvement in education, reductions in household size and
improved employment opportunities are the most promising ways to improve incomes.
Similarly, Lohano (2011) finds that households that escaped poverty experienced an increase in
their asset ownership as well as marked improvements in their human development indicators.

Another way to link the expenditure variable with poverty status is to use interquantile
regressions calibrated to the average expenditures of the chronically poor and never poor in all
survey rounds. For instance Baulch and Vu (2011) apply this approach to rural Vietnam between
2002 and 2006, while Quisumbing (2011) applies it to rural Bangladesh over a six to twelve year
period. The results from these studies show that escape from poverty is due to a combination of
improved labor returns to endowments and household accumulation of assets (Table 1).

Differentiating household poverty transitions

The interest in differentiating entries and exits from poverty dynamics directly has spurred
models that can estimate movements into and out of poverty between two time periods using
discrete outcome models, such as a multinomial logit model (MNL). In these models, the
probability of moving from poor to non-poor states is estimated. These models are popular partly
because a large number of panel waves are not needed, as in the case for Markov 19 and hazard
models. However, given the difficulty in distinguishing the sources of income, these models
typically do not focus on trigger events as the hazard models described earlier. Instead, the focus
is typically on human and physical capital endowments. These usually include educational status
of the head and other household members, demographic characteristics of the household, and
assets including landholding. In addition, learning from the literature on shocks, these models
typically include measures of shocks experienced by the household, which include both
community level shocks (such as natural disasters), and idiosyncratic shocks, such as asset and
crop losses, illness and/or death related shocks, as well as social spending shocks.

These studies find that improved returns to endowments, particularly the returns to labor are
critical for exiting poverty. Obtaining an education or other productive endowment can be seen
as triggers to escape from poverty (Table 1). However, an absence of bad shocks such as illness
or death, and the presence of good shocks (such as rain) are also needed to boost the possibility
to escape from poverty. In contrast, falling into chronic poverty is due to a combination of
negative shocks, and a lack of resilience in the form of low assets. For example, Dercon and

19
See Capellari and Jenkins, 2004; Stewart and Swaffield, 1999; Canto et al, 2007 for application of Markov models
in developed country settings where multiple rounds of panel data are available.
 - 11 -

Table 1. Determinants of the Probability of Moving from Poor to Non-Poor, MNL estimates
Bangladesh Vietnam, Vietnam, Uganda, 1992-
1996-2006 2002-06 2/ 1993-98 2000
[Quisumbing, [Baulch and [Glewwe et [Deininger and
2011] Vu, 2011] al, 2002] Okidi, 2003]
Household characteristics
Age of household head 0 +*
Education of head - * 1/ +* +* +*
Gender of the head 0
Ethnic minority -* -*
Head is white collar +*
Head is not working -*
Household size +* -* 0
Share of children +* -* -* -*
Share of elderly 0 -* 0 0

Assets
Total land owned - * 1/ 0
Value of assets -* +* +*
Value of livestock -*
Road quality +* +*
Mains electricity +* 0
Clean water +*
Crop share of income 0 +*
Productivity of rice +*

Shocks experienced
Rainfall
Floods 0 -*
Crop losses -*
Livestock death 0
Asset or house losses 0
Legal or political shocks 0 + * 3/
Death of income earner 0
Death of other household earner 0
Illness-related income loss 0 0 0 0
Illness-related expenses -*
Dowry or wedding expenses +*
Property division 0
Dowry, wedding and illness expenses -*

Region dummies ** ** **

A positive and significant correlation between the probability of going from poor to nonpoor is identified as “+*”, while
a negative correlation is noted as “-*”. When the effects were tested but were insignificant, we note “0”.
1/ This result is reversed when quantile regression estimates of log per capita expenditures are run instead of an MNL.
2/ Results from sequential logit model on the probability of escaping poverty.
3/ Refers to ethnic fractionalization in initial period.
 - 12 -

Porter (2011) finds that households that become chronically poor have been seriously affected by
drought and illness. Quisumbing (2011) shows that rural Bangladeshi households with less than
median assets are especially hard hit. Similarly, Lohano (2011) finds that the drought of 1999-
2002 in Pakistan was especially hard on landless households because there was a collapse of
employment opportunities at the same time as rising food prices.

Although the results reported above and in Table 1 describe the direction of the main findings in
some of these studies, it is difficult to decompose the impact of labor effects on the probability of
escaping poverty, unless this is done explicitly by the author. For instance, Justino et al (2008)
use a multinomial logit model to determine the probability that household i experiences one of
the j mutually exclusive outcomes (remaining poor, poor to non-poor, non-poor to poor, and
remaining non-poor). The poverty transition results can be used to provide a decomposition of
the overall probability of escaping or entering poverty by each of the explanatory variables, as
follows:
(4)

where S is the share of the overall probability of outcome Y, n is the total number of
observations in the panel, k is the total number of poverty dynamic outcomes, are the values
of the explanatory variables for household i in period j and are the coefficients obtained from
the multinomial logit regression models.

Using household-level panel data for rural households in Vietnam between1992-1998, they find
that household poverty dynamics were significantly affected by an improvement in human
capital, improvements in the labor market (in particular to export employment), and an
improvement in agricultural production related to the rice and coffee booms. The decomposition
of poverty dynamic effects implies that employment effects, including human capital effects,
labor market effects, and the change in the employment share of the export sector account for 62
percent of the probability of rural Vietnamese households escaping from poverty.

V. Qualitative and Mixed Methods

While panel data methods are considered the most reliable given the comparability of the data
collected at different points in time, its disadvantages are its costs and the significant delay in
analysis that it entails (Addison et al, 2009). Another way to obtain information about the past
when inter-temporal panels are not available with sufficient frequency is to ask people about
their past and utilize this information. In this approach, each individual is engaged in a semi-
structured discussion about their life course with the objective of understanding the trajectory of
their well-being, and the causes underlying it.

However, many researchers are highly suspicious of this method, given the problems of recall,
including the difficulty in remembering what conditions were like at the exact time that is of
interest, and the fact that people tend to develop selective memories.20 As a result, this method is

20
See Addison et al (2009).
 - 13 -

often used extensively to collect qualitative data, often to triangulate other data, and/or to
understand the ways in which people subjectively interpret change over time. Two alternatives
are presented in this section. The first is to make data more reliable by asking an entire
community, rather than a single individual, about changes in the poverty status of households
within that community. A second method is to employ mixed methods by combining qualitative
and quantitative methods in an integrated fashion.

Community Based Enquiries

One alternative have been group-based methods, or community-based inquiries, which could
make data more reliable as interviewees debate each other's recall and researchers can triangulate
data between groups. For instance, in a series of studies following a Stages of Progress
methodology, survey teams met with representative community groups in India, Kenya, Peru,
and Uganda (Krishna, 2004; Krishna et al, 2004; Krishna et al, 2006a; and Krishna et al, 2006b
respectively). In each case, the community collectively defined „poverty‟ given their social
context, rather than using standard international measures. For example, in the case of Kenya,
poor households were defined as those unable to acquire food, clothing, perform repairs to their
roof, allow for primary education of children, and those unable to acquire a chicken and a goat.
Given the agreed definition, then all households in the village are listed and the community
group is asked to describe the status of each household 25 years earlier compared to their status
today.

In each case, the authors identify households moving into and out of poverty. Poor health, large
family size together with an uneconomic subdivision of land, and expenses on death feasts are
the most important reasons for descent into poverty in Kenya, India, Peru and Uganda (Krishna
et al, 2004, 2006a and 2006b). In addition, in India, high expenses on marriages and high interest
payments on private debt are also important contributors to descent into poverty. For households
escaping from poverty, diversification of incomes sources by establishing links with the urban
economy was critical in Kenya, India, and Peru. Among those escaping poverty in Kenya, 73%
reported a household member obtaining a job. In India, diversification of income sources was a
principal factor for escaping poverty in 70% of cases. However these data are not amenable to
precise calculations given their small sample size, and potential for errors in recall and
measurement. However they do provide a means to generate additional facts about poverty and
its causes in diverse local contexts.

Non-income measures of poverty dynamics and mixed methods

An alternative to community-based enquiries is the use of both qualitative and quantitative

methods. An early example of this “mixed method” approach includes Sen (2003) who collected
household-level panel data from a survey of 21 rural villages in Bangladesh between 1987 and
2000, and finds that moving out of poverty meant combining different exit routes, including fast
accumulation of human and physical assets, diversification of the asset base favoring higher
income-yielding nonagricultural assets, a general reorientation from agricultural to
nonagricultural activities, and the pooling of household incomes from different sources. In
addition to observing the sources of income for the poor versus the non-poor, respondent‟s
perceptions as to why their economic well-being had improved among the economically
 - 14 -

ascending households were reported. Simple tabulations show that 64 percent of households
exiting poverty attributed their upward mobility to issues related to the availability of better
work, on account of increases in human or physical assets, favorable market conditions, or a
greater number of labor force participants.

More recent examples of this approach include Davis and Baulch (2011), who show how
different methods lead to very different assessments of socio-economic mobility. They embark
on what is often referred to as mixed-methods research into poverty dynamics, or „qsquared‟
methods. In particular, they apply a methodologically integrated approach which involves
undertaking qualitative and quantitative fieldwork simultaneously or sequentially, with
integrated analysis and write-up.21 Their analysis highlights the fact that a small shift in the peak
of the expenditure distribution or of poverty lines, can lead to the impression of a large number
of people moving out of, or into, poverty. Their qualitative findings, on the other hand tend
towards a much more pessimistic view of the tangible poverty reduction in rural Bangladesh. By
looking at the mismatches between quantitative and qualitative assessments they find different
types of ill-being which are not well detected in standard expenditure-based measures, including
ill health, dowry pressure, disability, domestic violence, social isolation or stigma. Similarly,
Quisumbing (2011) and Lohano (2011) conduct qualitative interviews as part of their studies in
Bangladesh and Pakistan. These interviews point to cases where there was a sequence of
negative shocks that were important for downward mobility, and would have typically not have
been picked up in standard panel data. Given these caveats, some authors are now strongly
advocating for the use of mixed methods (Addison et al, 2009; Davis and Baulch, 2011;
Kristjanson et al, 2011).

Finally, several authors have argued that poverty dynamics can look very different when non-
income measures are used, and these are critical as both a cross-check on trends in income
measures, as well as giving us a broader picture of how well-being in all its dimensions is
moving over time (Addison et al, 2009). In particular, Barrett et al. (2006) provide empirical
evidence that household welfare dynamics differ significantly depending on whether an income-
based measure is used versus an asset-based welfare measure. Carter and Barrett, (2006 ) argue
that asset-based approaches have several advantages over income-based measures, and show the
linkages between the depth of poverty, in terms of material and social assets, and duration with a
focus on household-level poverty traps. Moreover, given that assets like land and livestock
represent both accumulated past wealth and security in the future, Davis and Baulch (2011),
argue that assets play a vital role in most household‟s strategies for accumulation and
consumption smoothing. They show that using an asset-based measure, such as land ownership,
improves their ability to detect actual poverty transitions and therefore the reliability of the
poverty assessments substantially in Bangladesh.

21
Some examples of methodologically integrated q-squared studies include Devereux et al. (2003) in Ethiopia, and
Parker and Kozel (2005) in India.
 - 15 -

VI. Conclusions

Although the methodologies reviewed in this paper are diverse, they all point to a similar pattern:
more and better paid work is critical in lifting people out of poverty. More specifically,
improving the returns to labor is critical in lifting people out of poverty. Whether additional labor
income comes from greater diversification, higher earnings per hour, or a greater number of
hours worked, jobs are at the core of what counts for poverty reduction (Figure 3). The
substantial literature on economic mobility and poverty dynamics that has emerged in the past
decade provides a better understanding of the circumstances that allow the poor to pull
themselves up.

Figure 3. Events related to Poverty Exits

100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
India (Rajhastan) (1975-2000)

Uganda (1980-2005)

USA (1970-1982)

USA (1989-1993)

Spain (1985-1995)
Chile (1996-2001)

Brazil (2003-2006)
UK (1991-1999)

Peru (2002-2006)
Kenya (1978-2003)

Sweden (1991-1996)

Argentina (2003-2006)

Costa Rica (2006-2008)

Bangladesh (1987-2000)

South Africa (Kwazulu-Natal) (1993-1998)

Canada (1990-95)

Ecuador (2002-2006)
Germany (1991-96)

United Kingdom (1991-96)

Argentina (B.A.) (1991-2003)

Vietnam (1992 - 1998)
Peru (1980-2005) 3/

Stages of Progress Perceptions MNL Bane and Ellwood (1986) Jenkins and Schluter (2003) method
method

Labor-related events 1/ Non labor-related events 2/

1/ For stages of progress methods labor-related events include diversification of income sources, obtaining a job, improving productivity,
employment in private sector. For perceptions methodology, labor-related events include increases in human or physical assets, an
increase in the labor force, or favorable market conditions. For MNL methods, labor-related events include improvements in human capital,
improvements in the labor market and the changes in the employment share in the export sector. For the Bane and Ellwood and Jenkins
and Schluter methods, labor-related events include changes in household members' labor market attachment and earnings.
2/ Includes both changes in non-labor earnings (e.g., rents, pensions, land assets) and demographic changes.
3/ Only includes diversification out of livestock.
 - 16 -

The methodologies used to answer these questions have been varied, ranging from models
estimating the determinants of transitions across poverty states using household panel surveys, to
qualitative surveys based on community-based inquiries. However, the consensus seems to be
that improved returns to endowments, particularly the returns to human and physical capital, are
critical for exiting poverty. Although non-labor incomes and demographic changes can help to
lift the poor out of poverty, there is no substitute for providing an enabling environment for
better work and pay for the poor.
 - 17 -

References

Aassve, A., Betti, G., Mazzuco, S., and Mencarini, L. (2007). „Marital Disruption and Economic
Well-Being: A Comparative Analysis‟, Journal of the Royal Statistical Society, Series A,
170: 781–99.

Addison, T., Hulme, D., & Kanbur, R. (Eds.) (2009). Poverty dynamics: Interdisciplinary
perspectives. Oxford, UK: Oxford University Press.

Ananat, E. O. and Michaels, M. (2007). „The Effect of Marital Breakup on the Income
Distribution of Women with Children‟, Journal of Human Resources, 43: 611–29.

Antman, F. and J. McKenzie (2007),

Bane, M. and D.Ellwood (1986) “Slipping Into and Out of Poverty: the Dynamics of Spells”,
Journal of Human Resources, 21(1): 1-23.

Barrett, C., M. Carter and P. Little (2006), “Understanding and reducing persistent poverty in
Africa: introduction to a special issue”, Journal of Development Studies, 42(2): 167-77.

Barros, R., Carvalho, M. Franco, S. and R. Mendonca (2006). “Uma Análise das Principais
Causas da Queda Recente na Desigualdade de Renda Brasileira.” In: Revista Econômica.
Volume 8, número 1, p.117-147. Universidade Federal Fluminense. Rio de Janeiro.
Available at: http://www.uff.br/revistaeconomica/V8N1/RICARDO.PDF.

Baulch, F. and J. Hoddinott (2000). “Economic Mobility and Poverty Dynamics in Developing
Countries.” Journal of Development Studies, 36(6): 1-24.

Baulch, B. and H.D. Vu (2011) “Poverty dynamics in Vietnam, 2002 to 2006” in B. Baulch
(Ed.), Why Poverty Persists: Poverty Dynamics in Asia and Africa, Northhampton:
Edward Elgar Publishing.

Beccaria, L., and R. Maurizio (2009) “ Factors associated to poverty mobility in Greater Buenos
Aires”, Journal of Income Distribution. An International Quarterly. 18(2)

Beccaria, L., Maurizio, R., Fernandez, A., Monsalvo, P. and M. Álvarez (2011). “Dynamics of
Poverty, Labor Market and Public Policies in Latin America.” PMMA Working Paper
2011-05.

Bigsten, A. and Shimles, A. (2004). “Dynamics of Poverty in Ethiopia” WIDER Research Paper
No. 2004/39. June 2004.

Bourguignon François, Ferreira Francisco H. G. and Lustig Nora 2005. Introduction. In François
Bourguignon, Francisco H.G. Ferreira and Nora Lustig (eds), The Microeconomics of
Income Distribution Dynamics in East Asia and Latin America, pp. 17-46 Washington,
D.C.: The World Bank
 - 18 -

Calvo,C. and S. Dercon (2009). “Chronic poverty and all that: the measurement of poverty over
time”, in Addison, T., D. Hulme and R. Kanbur (eds), Poverty Dynamics:
Interdisciplinary Perspectives, Oxford: Oxford University Press.

Capellari, L. and S. Jenkins (2004) “Modelling Low Income Transitions” Journal of Applied
Econometrics, 19(5): 593-610.

Cantó, O. (2003). “Finding Out the Routes to Escape Poverty: The Relevance of Demographic
vs. Labor Market Events in Spain.” Review of Income and Wealth, 49(4): 569-588.

Cantó, O., C. del Río and C. Gradín (2007) “What helps households with children in leaving
poverty?: Evidence from Spain” Research on Economic Inequality, 14: 1-29.

Carter M, and Barrett C. (2006). The economics of poverty traps and persistent poverty: an asset-
based approach. Journal of Development Studies 42(2): 178–199.

Cherdchuchai, S. and K. Otsuka (2006) “Rural income dynamics and poverty reduction in Thai
villages from 1987 to 2004” Agricultural Economics, 35: 409-423.

Cunningham, Wendy, and William F. Maloney. 2000. “Measuring Vulnerability: Who Suffered
in the 1995 Mexican Crisis?” Washington: World Bank.

Davis, B. et al (2010). “A Cross-Country Comparison of Rural Income Generating Activities”

World Development, 38(1): 48-63.

Davis, P. and B. Baulch. (2011). “Parallel Realities: Exploring Poverty Dynamics Using Mixed
Methods in Rural Bangladesh”. Journal of Development Studies, 47(1): 118-142.

Deininger, K. and J. Okidi. (2003). “Growth and Poverty reduction in Uganda, 1999-2000: Panel
Data Evidence”. Development Policy Review, 21(4): 481-509.

Denisova, I. (2007). “Entry to and Exit from Poverty in Russia: Evidence from Longitudinal
Data” Centre for Economic and Financial Research at New Economic School,
CEIFR/NES Working Paper No 98.

Dercon, S., (2004 ). “Growth and Shocks: Evidence from Rural Ethiopia,” Journal of
Development Economics, 74: 309–29.

Dercon and Krishnan., (1996). “Income Portfolios in Rural Ethiopia and Tanzania: Choices and
Constraints” Journal of Development Studies. 32(6): 850-875.

Dercon, S. and C. Porter, (2011) “A poor life? Chronic Poverty and downward mobility in rural
Ethiopia, 1994 to 2004” in B. Baulch (Ed.), Why Poverty Persists: Poverty Dynamics in
Asia and Africa, Northhampton: Edward Elgar Publishing.
 - 19 -

Devereux, S., Sharp, K. and Amare, Y. (2003) Destitution in Wollo, Ethiopia. Research Report,
No. 55 Institute of Development Studies, University of Sussex.

Duclos, J.Y., A. Araar and J. Giles (2010), “Chronic and transient poverty: measurement and
estimation with evidence from China”, Journal of Development Economics, 91(2): 266-
77.

Escobal, J. (2001). “The Determinants of Nonfarm Income Diversification in Rural Peru” World
Development, 29(3): 497-502.

Essama-Nssah, B. (2012). “Identification of Sources of Variation in Poverty Outcomes”, World

Bank Policy Research Working Papers, No. 5954.

Estudillo, J., Sawada, Y., and K. Otsuka. (2008) “Poverty and Income Dynamics in Philippine
villages, 1985-2004” Review of Development Economics, 12(4): 877-890.

Fields, G., Cichelo, P., Freije, S., Menéndez, M. and D. Newhouse (2003). “Household Income
Dynamics: A Four-Country Story.” Journal of Development Studies, 40(2): 30–54.

Fields, G., Hernández, R., Freije, S., Sánchez Puerta, M. (2007). “Intergenerational Income
Mobility in Latin America.” Economía, :101-154.

Fields, Gary S., and Efe A. Ok. 1999. “Measuring Movement of Incomes.” Economica 66(264):
455–72.

Fortin Nicole, Lemieux Thomas and Firpo Sergio. 2011. Decomposition Methods in Economics.
In: Ashenfelter Orley and Card David (eds) Handbook of Labor Economics, Vol. 4A , pp.
1-102. Northolland, Amsterdam.

Foster J. (2009) “A class of chronic poverty measures.” in Addison, T., D. Hulme and R. Kanbur
(eds), Poverty Dynamics: Interdisciplinary Perspectives, Oxford: Oxford university
Press.

Glewwe, P., M. Gragnolati, and H. Zaman (2002). “Who Gained from Vietnam‟s Boom in the
1990s?” Economic Development and Cultural Change, 50(4): 773–92.

Gottschalk, Peter, and Enrico Spolaore. (2002). “On the Evaluation of Economic Mobility.”
Review of Economic Studies, 69(1): 191–208.

Gunning, J.W., J. Hoddinott, B. Kinsey, and T. Owens. (2000). “Revisiting forever gained:
Income dynamics in the resettlement areas of Zimbabwe, 1983-1997”. Journal of
Development Studies, 36: 131-154.

Günther, I. and S. Klasen, (2009). “Measuring Chronic Non‐Income Poverty” in Addison, T., D.
Hulme and R. Kanbur (eds), Poverty Dynamics: Interdisciplinary Perspectives, Oxford:
Oxford University Press.
 - 20 -

Haggblade, S., Hazell, P. and T. Reardon. (2007). Transforming the rural non-farm economy:
Opportunities and threats in the developing world. Baltimore: Johns Hopkins University
Press.
_______________________________. (2010).” The Rural Non-farm Economy: Prospects for
Growth and Poverty Reduction”, World Development, 38(10): 1429-1441.
Jalan, J. and M. Ravallion. (2000) “Is Transient Poverty Different? Evidence from China”
Journal of Development Studies, 36(6): 82-99.

___________________________ (2004). Household Income Dynamics in Rural China.” In

Insurance against Poverty, edited by Stefan Dercon, pp.108–24. Oxford University Press.

Jenkins, S. (2000). „Modelling Household Income Dynamics‟. Journal of Population Economics,

13: 529–67.

_________ (2011). Changing Fortunes: Income Mobility and Poverty Dynamics in Britain. New
York: Oxford University Press Inc.

Jenkins, S. and J.A. Rigg (2001). The Dynamics of Poverty in Britain. Department for Work and
Pensions Research Report No. 157. Leeds: Corporate Document Services.
〈http://research.dwp.gov.uk/asd/asd5/rrep157.pdf〉.

Jenkins, S. and C. Schluter (2003). “Why are Child Poverty Rates Higher in Britain than in
Germany? A Longitudinal Perspective.” Journal of Human Resources, 38(2): 441-465.

Justino, P. and J. Litchfield (2003). “Poverty Dynamics In Rural Vietnam: Winners and Losers
During Reform”, PRUS Working Paper No. 10, Department of Economics, University of
Sussex, Brighton, UK.

Justino, P., J. Litchfield and H. Thai Pham (2008). "Poverty Dynamics During Trade Reform:
Evidence from Rural Vietnam" Review of Income and Wealth, 54 (2): 166-192.

Juhn Chinhui, Murphy Kevin M. and Brooks Pierce. 1993. Wage Inequality and the Rise of
Returns to Skill. Journal of Political Economy, Vol. 101, No.3 : 410-442.

Kijima, Y., Matsumoto, T., and T. Yamomano (2006) “Nonfarm employment, agricultural
shocks, and poverty dynamics: evidence from rural Uganda”. Agricultural Economics,
35: 459-467.

Krishna, A. (2004) “Escaping poverty and becoming poor: who gains, who loses, and why?
Accounting for stability and change in 35 north Indian villages”, World Development,
32(1): 121–136.

Krishna, A., P. Kristjanson, M. Radeny and W. Nindo, (2004). “Escaping Poverty and Becoming
Poor in 20 Kenyan Villages” Journal of Human Development, 5(2): 211-226.
 - 21 -

Krishna A, Kristjanson P, Radeny M, Kuan J, Quilca G, Sanchez-Urrelo A. (2006a). “Escaping

poverty and becoming poor in forty communities of the Peruvian Andes”. Development
and Change 37(5): 997–1021.

Krishna A, Lumonya D, Markiewcz M, Mugumya F, Kafuko A, Wegoye J. (2006b). “Escaping

poverty and becoming poor in 36 Villages of central and western Uganda”. Journal of
Development Studies 42(2): 346–370.

Kristhjanson, P., Mango, N., Krishna, A., Radeny, M., and N. Johnson. (2010). “Understanding
Poverty Dynamics in Kenya”. Journal of International Development, 22: 978-996.

Lanjouw, J. and P. Lanjouw, (2001). “The rural nonfarm sector: Issues and evidence from
developing countries.” Agricultural Economics, 26(1): 1-23.

Lanjouw, P. (2007). “Does the rural non-farm economy contribute to poverty reduction?”. In S.
Haggblade, P. Hazell, & T. Reardon (Eds.), Transforming the rural nonfarm economy:
Opportunities and threats in the developing world (pp. 55-79). Baltimore: Johns Hopkins
University Press.

Lanjouw, P., J.Luoto, and D. McKenzie (2011). “Using Repeated Cross-Sections to Explore
Movements into and out of Poverty” World Bank Policy Research Working Paper No.
5550.

Lawson, D., A. McKay and J. Okidi (2006), “Poverty persistence and transitions in Uganda: a
combined qualitative and quantitative analysis”, Journal of Development Studies, 42(7):
1225-51.

Lillard, L. and R. Willis (1978), “Dynamic Aspects of Earnings Mobility.” Econometrica, 46:
985-1012.

Lohano, H.R. (2011) “Poverty dynamics in rural Sindh, Pakistan, 1987-88 to 2004-05” in B.
Baulch (Ed.), Why Poverty Persists: Poverty Dynamics in Asia and Africa,
Northhampton: Edward Elgar Publishing.

Lokshin, M. and Ravallion, M. (2004) Household income dynamics in two transition economies.
Studies in Nonlinear Dynamics and Econometrics, 8(3).

Machado, A. and R. Perez Ribas (2010). “Do Changes in the Labor Market Take Families Out of
Poverty? Determinants of Exiting Poverty in Brazilian Metropolitan Regions” Journal of
Development Studies, 46(9): 1503-1522.

Maloney, William F., Wendy V. Cunningham, and Mariano Bosch. (2004). “The Distribution of
Income Shocks during Crises: An Application of Quantile Analysis to Mexico, 1992–
95.” World Bank Economic Review 18(2): 155–74.
 - 22 -

Mankiw, N.G., Romer, D., Weil, D.N. (1992) “ A contribution to the empirics of economic
growth”, Quarterly Journal of Economics, 107: 409-437.

McKay, A. and D. Lawson (2003). “ Assessing the extent and nature of chronic poverty in low
income countries: issues and evidence”, World Development, 31(3): 425-39.

Nargis, N. and M. Hossain. (2006). “Income Dynamics and Pathways Out of Rural Poverty in
Bangladesh, 1988-2004” Agricultural Economics, 35: 425-435.

Neilson, C., D. Contreras, R. Cooper, and J. Hermann, (2008). “The Dynamics of Poverty in
Chile” Journal of Latin American Studies, 40(2): 251-273.

Oxley, H., Thanh Dang, T. and P. Antolin (2000). “ Poverty Dynamics in Six OECD Countries”
OECD Economic Studies, No. 30, 2000/1.

Otsuka, K., Estudillo, J. and Y. Sawada, (2009). Rural Poverty and Income Dynamics in Asia
and Africa. Routledge Studies in Development Economics. London: Routledge.

Parker, B. and Kozel, V. (2005) Understanding poverty and vulnerability in India‟s Uttar
Pradesh and Bihar: a Q-squared approach. Q-Squared Working Paper Series, No. 9,
Centre for International Studies, University of Toronto.

Porter C. and N. Quinn (2008). “Intertemporal poverty measurement: tradeoffs and policy
options”, Centre for the Study of African Economies working paper series 2008-21,
Oxford.

Quisumbing,A. and B. Baulch (2009).”Assets and poverty traps in rural Bangladesh”, Chronic
Poverty Research Centre research working paper no. 143, Manchester.

Quisumbing,A. (2011). “Poverty transitions, shocks and consumption in rural Bangladesh, 1996-
97 to 2006-07”, in B. Baulch (Ed.), Why Poverty Persists: Poverty Dynamics in Asia and
Africa, Northhampton: Edward Elgar Publishing.

Reardon, T, Taylor, E., Stamoulis, K. Lanjouw, P. and A. Baliscan (2000). “Effects of Non-Farm
Income Inequality in Developing Countries: An Investment Perspective”. Journal of
Agricultural Economics. 51(2): 266-288.

Sen, B. (2003). “Drivers of Escape and Descent: Changing Household Fortunes in Rural
Bangladesh.” World Development, 31(3): 513–534.

Stevens, A. H. (1994).” The Dynamics of Poverty Spells: Updating Bane and Ellwood.”
American Economic Review, 84(2): 34-37.

Stewart, M. and J. Swaffield (1999) “Low pay dynamics and transition probabilities”
Economica, 66: 23-42.
 - 23 -

Thorbecke, E. (2009) “Review of Keijiro Otsuka, Jonna P. Estudillo, and Yasuyuki Sawada, eds.
Rural Poverty and Income Dynamics in Asia and Africa. Routledge Studies in
Development Economics.” Economic Development & Cultural Change.

Ulrick, S. W. (2008). „Using Semi-Parametric Methods in an Analysis of Earnings Mobility‟,

Econometric Journal, 11: 478–98.

Woolard I., and S. Klasen (2005). “Determinants of Income Mobility and Household Poverty
Dynamics in South Africa.” The Journal of Development Studies, 41(5): 865 – 897.

World Bank. 2004. Poverty in Mexico: An Assessment of Conditions, Trends, and Government
Strategy. Washington: World Bank.