. 9
( 13)


who is just food insecure? Is someone who is deprived in two ˜dimensions™
as poor as someone who is deprived in two others?) These may seem like
silly questions, but they are precisely the ones that arise in any attempt
to develop an aggregate multidimensional poverty score, or to rank poor
people”activities that are routine in econometric approaches to poverty (see
e.g. Atkinson, 2003; Bourguignon and Chakravarty, 2003).
Surely all this misses Sen™s point. The relationship between human capa-
bilities and the ˜full human life™ that they enable is complex and dynamic.
To treat the absence of a particular capability, or the lack of access to the
resources required for it, as an ˜indicator™ of ˜poverty™ is to reify it and to miss
its signi¬cance. Those who lack education are not suffering from ˜education
poverty™; and those who have poor health are not ˜the health poor™. This is
to do violence to words. They are caught in a process of lack, deprivation,
or suffering which may (or may not!) lead to a severe impairment of their
social and economic agency and functioning in the world”and the different
dimensions of their deprivation re¬‚ect the diverse material roots and deter-
minants of that state. It may well be that those who are deprived in more
dimensions than one are less likely to escape poverty”but this depends on
the local structural context and the actual, empirical ways in which different
aspects of deprivation play into and feed into one another. The signi¬cance of
a variation, for example, in literacy or access to water lies not in the fact that
they are ˜indicators™ or transparent re¬‚ections of ˜non-economic well-being™
(whatever that is!), but in their implications and consequences for what people
can do”which are in the ¬rst place always shaped by a dynamic and complex
interplay; and which are irreducibly different and therefore non-substitutable.

Andries du Toit

More to the point, all these dif¬culties undermine the ability of poverty
researchers to achieve what is, after all, their avowed aim, which is to develop
a more disaggregated picture of ˜the poor™, and in particular to highlight the
existence of ˜the chronic poor™, imagined as a distinct subgroup, systematically
different from ˜the transitory poor™ in signi¬cant and policy relevant ways
(Hulme and Shepherd, 2003). Given the shaky ground underneath the notion
that a poverty line marks any really signi¬cant discontinuity in people™s ability
to make a living, such a claim seems dubious indeed. Surely it is true that
those who show up under a poverty line in repeated measurements are worse
off, in a general sense, than those who do not. But does it mean anything
more? It does not in itself constitute an argument that they are a socially
distinct group; indeed, depending on where the poverty line is drawn, those
with stable incomes (above and below the ˜poverty™ line) may have more
in common with one another than with those whose incomes are highly

10.3. From Chronic to Structural Poverty

10.3.1. Vulnerability, agency, and structural poverty
The above arguments seem to indicate that if we want to identify ˜the chronic
poor™ and understand what keeps them poor, measurement-based approaches,
then, offer only a slippery grasp. This, of course, is not to say that mea-
surement is irrelevant. The problem arises when we think that what we are
measuring is not income (or any other variable that we can construct as an
˜indicator™) but poverty itself ; when we believe that a distinction based on an
(inevitably arbitrary, conventional, formal) threshold genuinely can be said
to divide ˜the poor™ from ˜the non-poor™; and when this one marker is used
as the critical and paradigmatic criterion by which ˜the chronic poor™ are
distinguished from all the rest and imagined to constitute a distinct popu-
lation with its own special policy needs. A different approach is to recognize
that many of those who show up as ˜transitorily poor™ in a panel study may
still be held to be chronically poor if their underlying situation”the way they
are structurally inserted in society”means that they are unlikely to get out
of poverty in the long run. Such an approach requires an engagement with
the causal dynamics and processes that drive and shape livelihood careers.
Understanding who is likely to sink into poverty, who is likely to stay out
of it for long periods of time, and who is able to make the investments
required to ensure that a subsequent generation gets out (and stays out) of
it requires not only the post hoc tracking of actual welfare over time, but
also an assessment of the underlying factors that shape their likely welfare.
This means that the study of chronic poverty”and the identi¬cation of the

Poverty Measurement Blues

chronically poor”is inseparable from the study of structural poverty and
Development economics and econometrics are not disciplines well geared
towards understanding the structural con¬gurations of vulnerability. Sen™s
approach and the currently popular ˜livelihoods framework™ at least orient
enquiry towards an exploration of the material systems that underlie poverty
and well-being”but even these offer scant guidance, partly because they offer
very abstract and decontextualized ways of thinking about the particular ways
in which individuals and groups are situated in society.

10.3.2. From ˜distributions™ to relationships
Here, it may be instructive to look at one of the more innovative attempts in
South African poverty scholarship to use econometric analysis to develop an
assessment, not simply of whether or not people are poor, but of their under-
lying ˜structural poverty™: Carter and May™s (2001) analysis of the KwaZulu-
Natal Income Dynamics Study panel dataset (see also Carter and Barrett,
2005). Their analysis goes well beyond the limitations explored above, partly
because it uses a component analysis to explore the underlying aspects of
people™s livelihood situation. Rather than simply look at income, expendi-
ture, or capability deprivation, Carter and Barrett look at the assets (land,
human capital, ¬nancial wealth, social claims, and grain stocks) upon which
households rely to generate their income and argue that households whose
assets fall below the level required to generate an income equal to the poverty
line are ˜structurally poor™ even though a windfall may cause them to show
up above the income poverty line during a particular measurement. They
further postulate that though some people may suffer transitory structural
poverty (in other words, ˜structural poverty™ from which it is possible to escape
by accumulating suf¬cient assets) there may be a ˜Micawber Threshold™”a
level of asset deprivation so severe as to render escape through accumulation
This is an important corrective to the ˜structure blindness™ of de¬nitions of
chronic poverty that rely entirely on poverty spells”but it begs an important
question. Although the notion of ˜the steady level of well-being™ a household
˜can expect™ based on a particular level of assets or asset combination is a
useful ¬ction”one that can be adopted to great effect in the kinds of thought
experiments that econometric approaches are so good at”a ¬ction it to some
extent remains. Any attempt to ˜derive™ an expected income level from an
assessment of a given household™s asset base will be dogged by uncertainty”
particularly if we want to start including notions like ˜social capital™ in that
asset base. Although there is a link between the assets over which someone
disposes and the income one may expect them to generate from it, that link

Andries du Toit

is not linear and is mediated in complex ways by a host of other often non-
quanti¬able factors.
This is something not well recognized in the econometric approach to
poverty and social understanding. For all the innovativeness of their approach
May, Carter, and Barrett still see inequality statistically: as a matter of distribu-
tion. But as important as access to assets and resources are the social power
relations that govern this access within ˜households™ and in society more
broadly. This is where the seductive language of ˜household assets™, ˜social
capital™, and ˜human capital™ becomes dangerously misleading. For one thing,
households are not natural units but small, open systems (Bevan, pers. comm.)
that are internally contested, that change and re-form over time”and whose
access to resources are powerfully mediated by networks and connections that
extend outside the supposed household boundaries, so that there is often not
a very clear line between household members and non-members (Spiegel,
Watson, and Wilkinson, 1996; Russel, 2004; Ross, 2005; du Toit, Skuse, and
Cousins, 2007). For another, ˜social capital™ is not a quanti¬able resource, like
a seed bank or a herd of cattle, which exists in greater or lesser amounts and
which can be cashed or converted into other forms of capital in predictable
ways. Rather, it is a (theoretically contested) term for a wide range of variously
structured human relationships”kinship networks, friendships, af¬liation to
formal and informal bodies, patron“client relationships, political alliances”
that can be used to make claims and counter-claims (Meagher, 2005; du Toit
and Neves, 2006; du Toit, Skuse, and Cousins, 2007). And these, crucially,
are ˜meaning-ful™ relationships: in other words, relationships that are what
they are partly because of the way they are shaped by discursive practices,
underlying ideologies, moral metanarratives, and cultural paradigms that
come together to form a more or less consensual or contested ˜moral economy™
(Thompson, 1964; Scott, 1985) that de¬nes them and that speci¬es which
expectations can legitimately be based upon them.
A consideration of the different social landscapes explored as part of PLAAS™s
ongoing poverty research in South Africa (see the map in Figure 10.1) high-
lights how these complex webs of relationship and power work in very dif-
ferent ways in different contexts. On the commercial fruit farms of Ceres,
for instance, one very important form of ˜social capital™ is constituted by
the patron“client relationships between the coloured workers who work on
deciduous fruit farms and the white people who manage and own them (du
Toit, 2004). These relationships are shaped by discourses and practices of
paternalism that took shape in the course of a century and a half of slavery
and that adapted and mutated into new forms in the course of a century
and a half more of capitalist modernization (du Toit, 1993, 1998; Ewert and
Hamman, 1999; du Toit and Ewert, 2005). Paternalist discourse sets in place an
underlying ˜moral community™ between black workers and white owners that
is highly racialized and hierarchical; which also allows for the formulation of

Poverty Measurement Blues

Figure 10.1. PLAAS and CPRC™s research sites in South Africa (map by John Hall)

claims for resources and protection dependent on personal histories of loyalty
and service, and which requires a complex politics of moral suasion, hidden
resistance, and subtle negotiation beneath the fa§ade of racial deference. This
racialized ideology shapes relationships among white and black, between
African and coloured, and among the powerful and the powerless even off the
farms (du Toit, 2004). People with highly similar levels of ˜asset endowment™,
as the livelihood framework (or Carter and May) would describe them, will
have wildly different fortunes depending on their ability to negotiate these
relationships and to secure their interests.
In Mount Frere in the remote Eastern Cape, ˜social capital™ is also central”
but here what matter are complex traditional networks of kinship, clan
membership, village history, and patronage shaped by local history; and the
ways in which these have adapted and mutated in response to moderniza-
tion and change (du Toit, Skuse, and Cousins, 2007; Skuse and Cousins,
2007a). ˜Social capital™ exists nowhere but in a vast, complex, relational econ-
omy involving deeply embedded and socially coded practices of reciprocity
within and between ˜stretched™ households (Spiegel, Watson, and Wilkinson,
1996): extensive trade in goods, services, favours, labour, and sometimes even

Andries du Toit

money, shaped by more than a century of con¬‚ict and migrant labour. The
local cultures that shape this relational economy and that de¬ne people™s
expectations about themselves and one another are thoroughly different
from those one would ¬nd in Ceres: though Xhosa culture has not persisted
unchanged into modernity, local traditions about identity gender and isidima
(loosely translated as ˜status™ or dignity), for instance, play a powerful role in
shaping aspirations and behaviour. Again, households that look very similar
in a livelihood survey can have very different fortunes depending on their
links to local elites, their ability to make claims and to exploit sometimes
tangential kinship networks and so on (du Toit and Neves, 2006).
In Cape Town™s African suburbs (de Swardt, 2004b; du Toit and Neves,
2006; Skuse and Cousins, 2007b) survival also depends on an informal rela-
tional economy, but here things work very differently again. Whereas in
the countryside, community clan and village connections play a huge role
in the practices of reciprocal exchange whereby vulnerability is mediated,
neighbourliness is much less depended upon in town. And while kinship
is important, it is only one of a wide range of social relations, patron“
client arrangements, af¬liations, alliances, and enmities that structure and
are structured by informal exchange. Xhosa cultural forms and practices are
still important, but the ethos is much less shaped by traditionalism and is
infused with a valorization of youth and an assertive, street-smart urbanity
(Sagner and Mtati, 1999; Skuse and Cousins, 2007b). What matters here is
the ability to ˜work™ the urban system to get access to social services; the
ability to juggle debts and obligations and the ˜politics of intimacy™ in the
dance of the relational economy; the ability (crucially) to manage risk and
violence; and the ability to interface effectively with the still white-dominated
formal economy. The ability to insert oneself in complex local development
processes; the ability to claim membership of particular sub-communities and
interest groups; one™s history of belonging in Khayelitsha and the alliances
and allegiances thus formed: all have a major impact on the resources one
can mobilize.
In all three of these contexts, the local logic of ˜social capital™ leads to
the identi¬cation of very different groups of vulnerable people. In Mount
Frere, for instance, women and girl children bear the brunt of the impact
of gender roles that assign to them most of the responsibility for care work
and household reproduction (du Toit and Neves, 2006; du Toit, Skuse, and
Cousins, 2007). At the same time, those gender roles have given them, after
more than a century of migrant labour, a very real centrality in the networks
of civil society, while young men are no longer as able to use migrancy as
a path to full adult manhood, and are in some ways much more peripheral
to village politics and society (Banks, 2005). A different vulnerable group is
comprised by older people who end up being the heads of HIV/AIDS-affected
households. In urban Cape Town, it is young women and female household

Poverty Measurement Blues

heads who are particularly at risk, partly because gender roles dictate that
they should be dependent on men; and they often lack either the social or
the material resources that allow them to accumulate social capital. And in
Ceres, African men and women are disadvantaged by a racialized local culture
that constructs them as outsiders (du Toit, 2005).
Clearly an attempt to deduce ˜expected incomes™ from ˜asset combinations™
by running regressions on household survey data stands a poor chance of
uncovering any of this complexity. The point is not merely that there is
plenty that does not show up in the radar of any particular dataset. It is
also that the incorporation of these additional factors involves, not merely
their addition to an existing analysis of correlations, but the development of
a critical and theoretical account of power, ideology, culture, and inequality
in these contexts. The thumbnail sketches provided above derive from an
analysis informed by a family of (Geertzian and Foucauldian; agent centred
and structuralist) theoretical frameworks very different from the econometric
one”a theoretical imaginary that emphasizes the role of structure, agency,
antagonism, and social change, and according to which the perspectives and
stated experiences of various social actors are not taken simply at face value
but are seen as complex social creations, shaped by social power relations and
in turn impacting upon them. This has crucial implications for the prospects
of building more robust accounts of the nature of structural and chronic

10.4. Poverty Measurement and the Government of Poverty

10.4.1. Beyond Q-squared
In one sense, of course, none of the above arguments is very new. Argu-
ments about the limitations of purely quantitative research are probably as
old as ˜quantitative social science™ itself and have recently become common-
place again even within the development mainstream (see e.g. Kanbur, 2002;
Kanbur and Shaffer, 2007). This is undoubtedly a good thing: forays into
˜q-squared™ attempts to integrate qualitative and quantitative work can clearly
add to the rigour, depth, reach, and accuracy of poverty research (see e.g.
Adato, Lund, and Mhlongo, 2007). At the same time, this recognition is often
quite circumspect and the integration between ˜qualitative™ and ˜quantitative™
often takes place in restricted ways. The recognition of the limitations of
econometric analysis usually takes quite a limited form”being con¬ned, for
instance, to the idea that it is enough for econometric analysis to be sup-
plemented, corrected, or added to in some way by ˜qualitative™ research. For
writers like Thorbecke, for instance, qualitative data seem to be understood as
being equivalent to doing some PRAs”and the role of qualitative data seems

Andries du Toit

to be limited to generating hypotheses that can be quantitatively tested (Thor-
becke, 2004; see also Parker and Kozel, 2007; Hargreaves, et al., 2007). Others
admit of a wider range of methods and highlight a number of different ways in
which qualitative and quantitative work can illustrate, con¬rm, refute, enrich,
and illuminate one another (Carvalho and White, 1997; see also Adato, Lund,
and Mhlongo 2007; Howe and McKay, 2007; Jha, Rao, and Woolcock, 2007;
London, Schwartz, and Scott, 2007; Place, Adato, and Hebinck, 2007; Rew,
Khan, and Rew, 2007). On the whole, however, ˜qualitative data™ has been
seen to have an essentially supplementary and illustrative role in accounts of
poverty still essentially shaped by the econometric imaginary. The tensions
produced by the underlying differences in epistemological approach and
normative theory that exist between ˜qualitative approaches™ and econometric
research are only rarely noted within this literature (Kanbur and Shaffer, 2007;
see also Rew, Khan, and Rew, 2007; for a more searching and critical account,
see Bevan, 2004). Even more problematically, ˜qualitative data™ themselves are
all too often still understood, very simplistically, as if they are transparently
meaningful in themselves, and as if what emerges from PRAs, life histories,
focus groups, and the like can be taken at face value”and even quantita-
tively analysed (Hargreaves et al., 2007; Parker and Kozel, 2007), without an
engagement with the need to interpret these as textual artefacts, themselves
the products of con¬‚icts, antagonisms, and other encounters that are shaped
by social power relations and concrete social interests.
There is a danger, therefore, that attempts to assert the value of qualitative
research can simply take us back to a new positivism, in which slightly more
methodologically diverse research strategies (household surveys plus focus
group interviews; panel datasets plus life histories; econometric regressions
plus PRAs) ¬gure within accounts of society and social change essentially
still caught within the ahistorical, power-blind, technicist, and rational-choice
imaginary of econometric analysis and mainstream development economics
(see e.g. Kothari, 2001). What the calls for ˜integration™ ignore is that the
real issue is not whether we need to connect qualitative and quantitative
research”obviously we do”but that any attempt at integration is always
theory rich, dependent on underlying narratives about the nature of society,
agency, power, poverty, and social change.
Two observations issues arise out of this point. The ¬rst is that this theoreti-
cal poverty is not necessary. If the purpose is indeed to understand chronic
(and therefore structural) poverty, and to understand how social relations
shape people™s chances of getting into or out of poverty, then the ¬eld of social
science and critical social theory offers rich resources. A veritable academic
industry exists in which the links between power, agency, culture, identity,
and history are explored and which offers fertile space for re¬‚ective and
incisive accounts of the ways in which these are linked to the distribution
of resources in society.

Poverty Measurement Blues

The second is that, in spite of this promise, these critical traditions are
to a large extent marginalized in the ¬eld of development research and
poverty studies, relegated to a fairly well-de¬ned circuit of institutions and
journals which development economists and poverty scholars seem to feel
they can safely ignore. In South Africa, for instance, there is a rich legacy
of critical debate and research dating from the 1980s and 1990s on the rela-
tionships between capital accumulation, identity, ideology, social change, and
inequality”a legacy that has been radicalized and extended more recently,
in the work of institutions such as WISER, into searching re¬‚ections on
postcoloniality, racism, and identity. Yet at the same time it is possible for
scholars who to all intents and purposes are clearly deeply committed to social
justice and the eradication of inequality and poverty to produce an account
of labour market vulnerability and poverty in South Africa that reads as if
the ˜revisionist™ debates of the 1980s never took place and as if the liberal
orthodoxy of the 1960s and 1970s was never subjected to critique (Bhorat, et
al., 2001; for a discussion, see du Toit, 2005).

10.4.2. Power, knowledge, and the government of poverty
What scope is there for this state of affairs to be addressed and for applied
social science in general (and policy-oriented poverty research in particular)
to become more sensitive to the need for”and the power of”critical and
agent-centred accounts of structural poverty and the prospects for getting out
of it? Recently, the author (see du Toit, 2005) has to some extent attempted to
name and problematize the marginalization described here, which is all too
often seen as the natural order of things.
One of the most prominent stated reasons for the failure of critical social
theory to seriously challenge the hegemony of the econometric imaginary is
that there is no clearly hegemonic ˜critical theory™ approach even within the
margins. In contrast to the disciplines of economics and econometrics, where
debates and discussions are underpinned by a widely shared and hegemonic
framework setting the boundaries of a generally accepted ˜normal science™
(and also in contrast to the ¬eld of development studies, which lacks its
own rigour but is thoroughly governed by the changing orthodoxies and
frameworks adopted by leading donor institutions), critical social theory and
anthropology have since the mid-1980s been characterized by a ¬‚owering
of increasingly divergent and sometimes competing explanatory paradigms
and ontologies, subdisciplines and specialities: post-colonial, gender, and cul-
tural studies, social constructionism, critical realism, post-structuralist theory
and discourse analysis, actor network theory, global value chain analysis,
convention theory, to name but a few”with no speci¬c approach decisively
succeeding in establishing itself as central or dominant.

Andries du Toit

Norman Long has argued that, rather than being seen as fragmentation
and crisis, this diversity should be recognized as a fundamental condition
of social enquiry and welcomed as an opportunity for innovation (Long,
1992). Nevertheless, this diversity means that there is no single generally
accepted ˜qualitative™ or ˜non-positivist™ or ˜post-foundational™ approach. Calls
by economists for examples of generally accepted (or ˜best practice™) ways in
which social theory can help them understand chronic poverty unfortunately
have to be met by the answer that there is no master paradigm. Any attempt
to ˜operationalize™ the insights of qualitative sociology and critical social
theory has to be partial and local, and will require the case-by-case theoretical
concepts and approaches that can help illuminate particular problems.
But this is of course only part of the story, for the demand for ˜nor-
mal science™ in social research”for powerful, uncontroversial, and replicable
methodologies and schemata that can be used to produce reliable, policy-
relevant knowledge about poverty”has its own political economy. Para-
doxically many of the most problematic features of poverty measurement
described in previous pages are precisely those that make it attractive to
governments and donor institutions. Some of the features of econometric
approaches to de¬ning chronic poverty that I have criticized above arise to
some extent out of the underlying logic of the social technologies of knowl-
edge and power which make poverty measurement necessary and possible as
an enterprise in the ¬rst place. Poverty measurement has a complex history,
but a very important role in this history has been played by what we might
call the historical project of the ˜government of poverty™. The need for uni-
versal measurements and easily replicable indicators is linked to the project
of constituting poverty as an object of management and government”as
something whose presence in society needs to be recognized in ways that
render it subject to regulation and which can contain and limit its potential
as a radically disruptive political problematic.
As such, the discipline of poverty measurement is caught on the horns of a
dilemma, a double bind that arises out of the ˜optics™ of modern government
described by James C. Scott (Scott, 1998; but see also Foucault, 1987). This
way of looking at poverty is partly driven by the need to make society ˜legible™
in a regular, homogeneous, and universalizing way. In order to be useful for
the process of government and planning at all, technologies of measurement
and assessment have to be developed that can be formalized and routinized,
and which can be de-linked from the complexity and intransparency of local
context. Economies of scale in government, in decision making, in judgement
and assessment require the development of embodied techniques of know-
ing and decision making that can be ˜ported™ from one context to another,
that make it possible to compare one individual (or household or region)
with another against a homogeneous, calibrated scale, that allow them to be
ranked, that can inform decisions about the allocation of resources”and that

Poverty Measurement Blues

allow all of these operations to be done in the shadow of the authority of
˜science™: apparently free of bias, objective, and incontrovertible.
The problem, as Scott points out, arises when this process of abstraction
and decontextualization leads not to legibility but to misreading: when, for
example, imposing the template of monoculture on forestry management
destroys the underlying ecological base of biodiversity on which the forest
depends, or when dirigiste city planners misunderstand the local dynamics
that make neighbourhoods liveable and attractive (Scott, 1998). In such cases
the preference for certain kinds of information”information that is read-
ily quanti¬able and standardized, that abstracts from local complexity and
appears to sidestep intransparency”leads not to an accurate grasp of the
dynamics of a situation, but to distorted and misleading accounts that miss
crucial dynamics.
The question is, what follows from the recognition of these distortions
and misunderstandings? What scope is there for what Scott called metís”for
more ¬‚exible forms of knowledge that allow for an understanding of some of
these complex dynamics and which are by their very nature more provisional,
more embodied and localized, more connected with speci¬c histories and
relationships, more value laden and political? What scope is there for ˜the
state™ and donor agencies to graduate from the ˜short-cut™ methodologies of
quantitative surveys and PRAs (Rew, Khan, and Rew, 2007) and to learn other
ways of seeing and imagining poverty and vulnerability?
The struggle is an uphill one, if recent attempts to build governmental
capacity to understand food insecurity and vulnerability in South Africa are
anything to go by. A case study of the development of a Food Insecurity and
Vulnerability Information and Mapping System (FIVIMS) for the ˜social clus-
ter™ of departments in South Africa shows that, in spite of the recognition of
the role of local history, power relations, and in spite of the acknowledgement
of the importance of practical local knowledge embedded in institutions on
the ground, very little could be done to shift the perception on the part
of the of¬cials involved that ultimately, what was practical was a GIS-based
system that would provide information about ˜indicators™ of ˜structural vul-
nerability™ in unambiguous, mappable, quanti¬able terms. This institutional
inertia seemed to be produced partly by what one could call the mystique of
quantitative data”a wholly misplaced faith in what one could learn from the
quantitative data that are available for use in a GIS-based system”but partly
also by underlying totalizing narratives about the place of ˜integrated plan-
ning and implementation™ and centralized knowledge in the exercise of state
power (du Toit, Vogel, and Drimie, 2005). Asking governments and donor
institutions to make space for critical accounts of social change”accounts
that are more sensitive to the nature and dynamics of power relations”seems
inevitably to come up against the limitations that arise out of the present-day
logic of forms of power-knowledge and modes of governmentality that seek

Andries du Toit

to de-link claims to authority from knowledge from locality, and that depend
on technologies for decontextualizing and homogenizing social and political
space (for a broader discussion see Kothari, 2005).

10.5. Conclusion

What, then, is the scope for ˜decolonizing™ methodologies that are so clearly
linked to formations of power and knowledge deeply shaped by their links to
postcolonial (and still imperial!) forms of governance and governmentality?
In the long run, there is only one way of ¬nding out: by actually trying to
contest homogenizing positivist narratives by developing powerful and con-
vincing counter-hegemonic accounts. In South Africa, at least, it is possible to
imagine that the terms of this engagement do not run only one way. Rather
than being the stage for a seamless ˜ordering of dissent™ (Kothari, 2005) in
which the institutions of globalized corporate power are always and inevitably
able to contain criticism by incorporating it, the ¬eld of applied social science
research in South Africa seems to embody a fruitful, if hazardous, terrain for
engagement. Given the urgency of addressing persistent poverty in South
Africa and the recognition by the ruling party that modernizing narratives
about trickle-down are not working (see Mbeki, 2003), there is a wide scope for
critical scholars to interrupt and to problematize the apparent self-evidence
of normalizing metanarratives about growth, modernity, security, and the
like. It is part of both the fertility and the hazard of this terrain that all
such interventions needs must be themselves situated and informed by an
awareness of their own dependency on and inevitable complicity with a
history steeped in con¬‚ict and suffering.


Adato, M., Lund, F., and Mhlongo, P. (2007), ˜Methodological Innovations in Research
on the Dynamics of Poverty: A Longitudinal Study in Kwazulu-Natal, South Africa™,
World Development, 35(2): 247“63.
Aliber, M. (2001), ˜Study of the Incidence and Nature of Chronic Poverty and Devel-
opment Policy in South Africa: An Overview™, Chronic Poverty Research Centre
Background Paper 3, Manchester: CPRC.
Arnall, A., Furtado, J., Ghazoul, J., and de Swardt, C. (2004), ˜Perceptions of Informal
Safety Nets: A Case Study from a South African Informal Settlement™, Development
Southern Africa, 21(30): 443“60.
Atkinson, A. B. (2003), ˜Multidimensional Deprivation: Contrasting Social Welfare and
Counting Approaches™, Journal of Economic Inequality, 1: 51“65.
Bane, M. J., and Ellwood, D. T. (1986), ˜Slipping into and out of Poverty: The Dynamics
of Spells™, Journal of Human Resources, 21(1): 1“23.

Poverty Measurement Blues

Banks, L. (2005), ˜On Family Farms and Commodity Groups: Rural Livelihoods, House-
holds and Development Policy in the Eastern Cape™, Social Dynamics, 31(1): 157“
Barrientos, A. (2003), Non-Contributory Pensions and the Well-Being of Older People: Evi-
dence on Multi-dimensional Deprivation from Brazil and South Africa, mimeo, Institute
for Development Policy and Management,Manchester, UK: University of Manchester.
Baulch, B. (1996), ˜Neglected Trade-Offs in Poverty Measurement™, IDS Bulletin, 27(1):
and McCulloch, N. (1998), ˜Being Poor and Becoming Poor: Poverty Status and
Poverty Transitions in Rural Pakistan™, IDS Working Paper No. 79, Brighton: Institute
of Development Studies.
and Masset, E. (2003), ˜Do Monetary and Non-Monetary Indicators Tell the Same
Story about Chronic Poverty? A Study of Vietnam in the 1990s™, World Development,
31(3): 441“54.
Bevan, P. (2004), ˜Exploring the Structured Dynamics of Chronic Poverty: A Sociological
Approach™, WeD Working Paper 06, Bath: Well-Being in Developing Countries ESRC
research group, University of Bath.
Bhorat, H., Leibbrandt, M., Maziya, M., van der Bergh, S., and Woolard, I. (eds.) (2001),
Fighting Poverty: Labour Markets and Inequality in South Africa, Cape Town: UCT Press.
Bourguignon, F., and Chakravarty, S. R. (2003), ˜The Measurement of Multidimensional
Poverty™, Journal of Economic Inequality, 1: 25“49.
Carter, M., and Barrett, C. (2005), ˜The Economics of Poverty Traps and Persis-
tent Poverty: An Asset-Based Approach™, unpublished typescript, Basis CRSP. See
and May, J. (2001), ˜One Kind of Freedom: Poverty Dynamics in Post-Apartheid
South Africa™, World Development, 29(12): 1987“2006.
Carvalho, S., and White, H. (1997), ˜Combining the Quantitative and Qualitative
Approaches to Poverty Measurement and Analysis: The Practice and the Potential™,
World Bank Technical Paper No. 366, Washington: World Bank.
Chaudhuri, S., and Ravallion, M. (1994), ˜How Well Do Static Indicators Identify the
Chronically Poor?™, Journal of Public Economics, 53: 367“94.
Chronic Poverty Research Centre (CPRC) (2004), The Chronic Poverty Report 2004“05,
Manchester: Chronic Poverty Research Centre.
Cichello, P. L., Fields, G. S., and Leibbrandt, M. (2003), Earnings and Employment
Dynamics for Africans in Post-apartheid South Africa: A Panel Study of KwaZulu-Natal,
Development Policy Research Unit, May, Working Paper No. 03/77, Cape Town:
De Swardt, C. (2004a), ˜Chronic Poverty and the Basic Income Grant™, Chronic Poverty
and Development Policy Working Paper No. 2, Bellville: Programme for Land and
Agrarian Studies.
(2004b), ˜Cape Town™s African Poor: Chronic Poverty and Development Policy™,
Working Paper No. 3, Bellville: Programme for Land and African Studies.
Puoane, T., Chopra, M., and du Toit, A. (2005), ˜Urban Poverty in Cape Town™,
Environment and Urbanization, 17(2): 101“11.
Du Toit, A. (1993), ˜The Micropolitics of Paternalism: Discourses of Management and
Resistance on Western Cape Fruit and Wine Farms™, Journal of Southern African Studies,

Andries du Toit

19(3): 314“36.
(1998), ˜The Fruits of Modernity: Law, Power and Paternalism in Western Cape Fruit
and Wine Farms™, in A. Norval and D. Howarth (eds.), South Africa in Transition: New
Theoretical Perspectives, London: Macmillan, 149“64.
(2004), ˜Forgotten by the Highway: Globalization and Chronic Poverty in Ceres™,
CPRC Working Paper No. 49, Bellville: PLAAS, CPRC, and CSSR.
(2005), ˜The Sociology of Chronic and Structural Poverty in South Africa: Chal-
lenges for Action and Research™, CPRC Working Paper No. 56, Bellville: PLAAS, CPRC,
and CSSR.
and Ewert, J. (2005), ˜A Deepening Divide in the Countryside: Restructuring and
Rural Livelihoods in the South African Wine Industry™, Journal of Southern African
Studies, 31(2): 315“32.
and Neves, D. (2006), ˜Vulnerability and Social Protection at the Margins of the
Formal Economy: Case Studies from Khayelitsha and the Eastern Cape™, report for
National Department of Treasury, unpublished typescript, Bellville: PLAAS, CPRC, and
Skuse, A., and Cousins, T. (2007), ˜The Political Economy of Social Capital: Chronic
Poverty, Remoteness and Gender in the Rural Eastern Cape™, Social Identities, 13(4):
Vogel, C., and Drimie, S. (2005), ˜Institutional Arrangements and Utilization of
FIVIMS-ZA™, unpublished typescript, Pretoria; FIVIMS-ZA.
Ewert, J., and Hamman, J. (1999), ˜Why Paternalism Survives: Globalisation, Democrati-
sation and Labour on South African Wine Farms™, Sociologia Ruralis, 39(2): 202“21.
Foucault, M. (1987), Discipline and Punish: The Birth of the Prison, Harmondsworth:
Hargreaves, J. R., Morison, L. A., Gear, J. S. S., Makhubele, M. B., Porter, J. D. H., Busa,
J., Watts, C., Kim, J. C., and Pronyk, P. M. (2007), ˜ “Hearing the Voices of the Poor”:
Assigning Poverty Lines on the Basis of Local Perceptions of Poverty. A Quantitative
Analysis of Qualitative Data from Participatory Wealth Ranking in Rural South Africa™,
World Development, 35(2): 212“29.
Howe, G., and McKay, A. (2007), ˜Combining Quantitative and Qualitative Methods in
Assessing Chronic Poverty: The Case of Rwanda™, World Development, 35(2): 197“211.
Hulme, D., and Shepherd, A. (2003), ˜Conceptualizing Chronic Poverty™, World Develop-
ment, 31(3): 403“24.
Jha, S., Rao, V., and Woolcock, M. (2007), ˜Governance in the Gullies: Democratic
Responsiveness and Leadership in Delhi™s Slums™, World Development, 35(2): 230“46.
Kanbur, R. (2002), ˜Economics, Social Science and Development™, World Development,
30(3): 477“86.
and Shaffer, P. (2007), ˜Epistemology, Normative Theory and Poverty Analysis:
Implications for Q-Squared in Practice™, World Development, 35(2): 183“96.
Klasen, S. (2000), ˜Measuring Poverty and Deprivation in South Africa™, Review of Income
and Wealth, 46(1): pp. 33“58.
Kothari, U. (2001), ˜Power, Knowledge and Social Control in Participatory Develop-
ment™, in B. Cooke and U. Kothari (eds.), Participation the New Tyranny, London: Zed,
(2005), ˜Authority and Expertise: The Professionalisation of International Develop-
ment and the Ordering of Dissent™, Antipode, 37(3): 425“46.

Poverty Measurement Blues

Kritzinger, A., and Barrientos, S. W. (2004), ˜Squaring the Circle: Global Production and
the Informalization of Work in South African Fruit Exports™, Journal of International
Development, 16: 81“92.
Leibbrandt, M., and Woolard, I. (2001), ˜Measuring Poverty in South Africa™, in
H. Bhorat, M. Leibbrandt, M. Maziya, S. van der Bergh, and Ingrid Woolard (eds.),
Fighting Poverty: Labour Markets and Inequality in South Africa, Cape Town: UCT Press.
London, A. S., Schwartz, S., and Scott, E. K. (2007), ˜Combining Quantitative and Qual-
itative Data in Welfare Policy Evaluations in the United States™, World Development,
35(2): 342“53.
Long, N. (1992), ˜From Paradigm Lost to Paradigm Regained: The Case for an Actor
Oriented Sociology of Development™, in N. Long and A. Long (eds.), Battle¬elds of
Knowledge: The Interlocking of Theory and Research in Social Theory and Development,
Wageningen: Wageningen Agricultural University, 17“42.
McCulloch, N., and Baulch, B. (1999), ˜Distinguishing the Chronically from the Tran-
sitorily Poor: Evidence from Pakistan™, Working Paper No. 97, Brighton: Institute of
Development Studies.
McGillivray, M. (2003), ˜Capturing Non-Economic Dimensions of Human Well-Being™,
UNI-WIDER Conference on Poverty, Inequality and Well-Being, United Nations World
Institute for Development Economics Research, Helsinki, Finland, 30“31 May 2003.
McKay, A., and Lawson, D. (2003), ˜Assessing the Extent and Nature of Chronic
Poverty in Low Income Countries: Issues and Evidence™, World Development, 31(3):
Martinetti, E. C. (2000), ˜A Multidimensional Assessment of Well-Being Based on Sen™s
Functioning Approach™, Revista internazionale di scienze sociali, 2.
Mbeki T. (2003), ˜Letter from the President™, <http://www.anc.org.za/ancdocs/
Meagher, K. (2005), ˜Social Capital or Analytical Liability? Social Networks and African
Informal Economies™, Global Networks, 5(3): 217“38.
Noble, M., Ratcliffe, A., and Wright, G. (2004), ˜Conceptualising, De¬ning and Mea-
suring Poverty in South Africa: An Argument for a Consensual Approach™, Oxford:
Nussbaum, M. (1999), ˜Women and Equality: The Capabilities Approach™, International
Labour Review, 138(3): 227“45.
Parker, B., and Kozel, V. (2007), ˜Understanding Poverty and Vulnerability in India™s
Uttar Pradesh and Bihar: A Q-Squared Approach™, World Development, 35(2): 296“311.
Place, F., Adato, M., and Hebinck, P. (2007), ˜Understanding Rural Poverty and Invest-
ment in Agriculture: An Assessment of Integrated Quantitative and Qualitative
Research in Western Kenya™, World Development, 35(2): 312“25.
Qizilbash, M. (2002), ˜A Note on the Measurement of Poverty and Vulnerability in the
South African Context™, Journal of International Development, 14: 757“72.
(2004), ˜On the Arbitrariness and Robustness of Multi-dimensional Poverty Rank-
ings™, WIDER Research Paper No. 2004/37, Helsinki: World Institute for Development
Economics Research, United Nations University.
Ravallion, M. (1996), ˜Issues in Measuring and Modelling Poverty™, Economic Journal,
106: 1328“43.
Rew, A., Khan, S., and Rew, M. ( 2007), ˜ “P3 > Q2 ” in Northern Orissa: An Example
of Integrating “Combined Methods” (Q2 ) through a “Platform for Probing Poverties”

Andries du Toit

(P3)™, World Development, 35(2): 281“95.
Ross, F. C. (2005), ˜Urban Development and Social Contingency: A Case Study of Urban
Relocation in the Western Cape, South Africa™, Africa Today, 51(4): 19“31.
Russel, M. (2004), ˜Understanding Black Households in Southern Africa: The African
Kinship and Western Nuclear Family Systems™, CSSR Working Paper No. 67, Cape
Town: CSSR.
Sagner, A., and Mtati, R. (1999), ˜Politics of Pension Sharing in Urban South Africa™,
Aging and Society, 19: 393“416.
Scott, J. C. (1985), Weapons of the Weak: Everyday Forms of Peasant Resistance, New Haven:
Yale University Press.
(1998), Seeing Like a State: How Certain Schemes to Improve the Human Condition have
Failed, New Haven: Yale University Press.
Sen, A. (1999), Development as Freedom, Oxford: Oxford University Press.
Shepherd, A. (2007), ˜Understanding and Explaining Chronic Poverty: An Evolv-
ing Framework for Phase III of CPRC™s Research™, Working Paper 80, Manchester:
IDPM/Chronic Poverty Research Centre.
Skuse, A., and Cousins, T. ( 2007a), ˜Managing Distance: The Social Dynamics of Rural
Telecommunications Access and Use in the Eastern Cape, South Africa™, unpublished
typescript submitted to Journal of African and Asian Studies, (42): 185“207.
(2007b), ˜Getting Connected: The Social Dynamics of Urban Telecommunica-
tions Use in Khayelitsha, Cape Town™, New Media and Society, Sage Publications.
Spiegel, A., Watson, V., and Wilkinson, P. (1996), ˜Domestic Diversity and Fluidity
among Some African Households in Greater Cape Town™, Social Dynamics, 22(1): 7“30.
(1999), ˜Speaking Truth to Power: Some Problems Using Ethnographic
Methods to In¬‚uence the Formulation of Housing Policy in South Africa™, in A.
Cheater (ed.), The Anthropology of Power: Empowerment and Disempowerment in Chang-
ing Structures, London: Routledge, 175“90.
Thompson, E. P. (1964), The Making of the English Working Class, Harmondsworth:
Thorbecke, E. (2004), ˜Conceptual and Measurement Issues in Poverty Analysis™, WIDER
Discussion Paper No. 2004/04, Helsinki: World Institute for Development Economics
Research and United Nations University.
Qizilbas, M. (2004), On the Arbitariness and Robustness of Multi-dimensional Poverty
Rankings™. WIDER Research Paper No. 2004/37. (Helsinki: World Institute for Devolp-
ment Economics Research, United Nations University).
United Nations Development Program (UNDP) (2002), Human Development Report, New
York: Oxford University Press.

When Endowments and Opportunities
Don™t Match
Understanding Chronic Poverty
S. R. Osmani

11.1. Introduction

Recent attempts to understand the causes of chronic poverty have centred
largely on the concept of poverty trap. In this perspective, the main focus is
on the paucity of initial wealth or endowments, which under certain plausible
conditions can create a trap from which a poor person will ¬nd it hard to
escape without help from outside. While acknowledging the value of insights
gained from this perspective, this chapter proposes to draw attention to a
different perspective that can also prove useful in both understanding the
causes of chronic poverty and looking for its solution. Instead of focusing
on the level or the magnitude of endowments possessed by the poor, this
new perspective focuses on the structure or composition of endowments.
The crucial insight offered by this perspective is that chronic poverty can
arise not just from a low level of endowments but also from a mismatch
between the structure of endowments possessed by the poor and the structure
of opportunities open to them. This chapter is an attempt to elaborate on this
insight and to draw out some of its implications.
Section 11.2 sets the scene by offering a formal characterization of chronic
poverty, and in the process it also tries to clarify the distinction between
chronic and transitory poverty more precisely compared to the usual way
this distinction is drawn in the literature. Section 11.3 constitutes the core of
the chapter in which the new perspective is introduced and its implications
are analysed. First, the point is made that chronic poverty can exist even
without a poverty trap. Next, an attempt is made to examine the genesis
of chronic poverty, in which the idea of a mismatch between endowments

S. R. Osmani

and opportunities plays an important role. Finally, the relationship between
economic growth and chronic poverty is discussed in the light of the notion
of structural mismatch discussed earlier. Section 11.4 offers some brief con-
cluding observations.

11.2. Characterization of Chronic Poverty

Chronic poverty is essentially a dynamic concept”the idea of a time dimen-
sion is inherent in it. And yet the standard literature on chronic poverty
seldom brings out the time dimension explicitly. For how long does a person
have to be poor in order to be counted as chronically poor? This would seem
to be a natural question to ask but is rarely asked. A particular strand of the
literature, which equates the concept of chronic poverty with that of poverty
trap, implicitly assumes an inde¬nitely long time span. The very concept of
trap implies that a person will never escape poverty unless some exogenous
event helps her to break out of the trap. Suppose, however, that there is no
such trap for some poor person, so that under the prevailing circumstances
she should be able to break out of poverty, given suf¬cient time. But she only
has a ¬nite lifespan and an even shorter working life. So even if she could
in theory escape poverty given suf¬cient time, in reality she may never do
so because the time required could be longer than she can reasonably expect
to live or be ¬t enough to work. For all practical purposes, she too should be
counted as chronically poor even though she may not be caught in a trap from
which there is no endogenous escape route. The concept of chronic poverty
must, therefore, explicitly incorporate the idea of a lifespan”strictly speaking,
the span of working life. 1
Yet another essential feature of the concept of chronic poverty is the
recognition that income is subject to stochastic variation. This feature is of
course duly recognized in the standard literature; indeed it is with reference to
such variation that chronic poverty is generally distinguished from transitory
poverty. I would argue, however, that the manner in which this distinction
is generally made is not rigorous enough. Typically, a person is said to suffer
from transitory poverty if her income sometimes falls below the poverty line
and sometimes rises above the poverty line, whereas a chronically poor person
is said to be one whose income is always below the poverty line. The basic idea
behind this way of looking at the distinction is sound enough, but the way
it is formulated lacks rigour. Consider, for example, a person whose income
occasionally rises above the poverty line but remains below it most of the
Throughout this chapter, we shall focus exclusively on the so-called ˜working poor™,
leaving out poor people outside the labour force”such as the very young or very old or
disabled people”for whom the problem of poverty is qualitatively different in nature, and
calls for a rather different kind of analysis.

Understanding Chronic Poverty

time. It seems reasonable to argue that such a person should be counted as
chronically poor for all practical purposes, but the standard de¬nition doesn™t
allow it.
In defence of the standard procedure, one could argue that if a person™s
income goes above the poverty line from time to time, she is probably living
pretty close to the poverty line even if staying below it most of the time.
That™s possible, and if true this may disqualify her from being classi¬ed as
extremely poor or ultra-poor in some sense but not from being classi¬ed as
chronically poor. It™s important to keep the concepts of extreme poverty and
chronic poverty separate. Extreme poverty relates to the notion of depth or
intensity of deprivation, whereas chronic poverty relates to the duration of
deprivation. In the real world, there may be considerable overlap between
the two concepts”many of the chronically poor people may be found to be
extremely poor as well. But for analytical purposes the two concepts must
be kept distinct, allowing for the possibility that some people suffering from
chronic poverty (in terms of duration) may be only moderately poor (in terms
of intensity).
By combining the two features”namely time dimension of deprivation and
stochastic variation of income”it is now possible to formalize the notion of
chronic poverty. Let the income of a person at any point in time t be denoted
by Yt , which depends on two sets of factors:
r A set of observable individual-speci¬c variables such as age, gender, assets,
education, skills, etc. and observable household-level variables such as the
number of dependants, the size of household labour force, etc. We call it
the endowment set of the individual and denote it by Dt .
r A set of observable variables exogenous to the individual and the house-
hold. These could be village-level or community-level or national-level or
even global-level variables that affect the income of the person one way
or the other. Denote this set by Xt .

Some of the endowments (D) may be time invariant (e.g. gender), others may
have a deterministic time trend (e.g. age), and yet others may be amenable
to choice on the part of the individual (e.g. physical assets, skills, etc.).
The endowments that are amenable to choice will in general depend upon
three sets of factors: the value of all the D variables in the preceding time
period (Dt’1 ), income of the preceding period (Yt’1 ), and the set of exogenous
factors (Xt ).
Then the income generation process over time can be described by the
following recursive system of dynamic equations:

Yt = F (Dt , Xt ) + et (11.1)
Dt = G(Yt’1 , Dt’1 , Xt ) + µt (11.2)

S. R. Osmani

where G is a vector of functions”with one function for each element of Dt ”
and et and µt are random error terms, with all the classical properties of white
It is through these error terms that stochastic variation is introduced in the
income generation process.
Because of stochastic variation people may move into or out of poverty from
time to time, and even a chronically poor person may occasionally rise above
the poverty line, but what distinguishes her from the transitorily poor is that
more often than not she would be expected to remain below the poverty line.
The expression ˜more often than not™ is obviously rather vague, but there are
several possible ways of making it precise. One simple way of doing so is to
think in terms of expected income E (Yt ). If the time path of expected income
is such that it always stays below the poverty line (Z), then we would expect
a person™s actual income to be below the poverty line more often than not.
That™s the approach we take below in order to de¬ne chronic poverty.
Let ™ denote the point in time at which the existence of chronic poverty is
being assessed and T denote the end-point of a person™s working life. If the
exogenous variables obtaining at time ™ were to remain unchanged over the
remainder of a person™s working life”denoted by the interval [™, T]”then
the income generation process can be expressed as:

Yt = F (Dt , X™ ) + et ; t ∈ [™, T]
for all (11.3)

Dt = G(Yt’1 , Dt’1 , X™ ) + µt t ∈ [™, T]
for all (11.4)

Note that taking expectation of income Yt , conditional on Dt and X™ , would
eliminate the stochastic variation arising from different realizations of et , but
it would still retain an element of stochastic variation arising from µt , which
would operate through Dt . In order to eliminate that element of variation,
so as to arrive at the essence of chronic poverty, we will have to work with
expectation of Dt rather than particular realizations of Dt . In other words, we
shall have to think in terms of expectation of income conditional on X™ and
the expectation of Dt .
Let Dt— = E (Dt ). Then the criterion by which we can identify the chronically
poor is the time path of E (Yt |Dt— , X™ ). The formal de¬nition of chronic poverty
can now be given as follows.
De¬nition: A person is chronically poor if E (Yt |Dt— , X™ ) < Z, for all t ∈ [™, T],
where Z is the poverty line income, ™ denotes the point in time at which the
existence of chronic poverty is being assessed, and T is the end-point of a
person™s working life.
The time path of chronic poverty according to this de¬nition is shown
in Figure 11.1. 2 To see exactly what this de¬nition entails, recall that Dt—
The time path has been drawn as upward rising on the assumption that a rational
person will try to accumulate some capital”either physical or human or both”to the extent

Understanding Chronic Poverty



™ T t

Figure 11.1. Chronic poverty

contains two types of variables. There are some over which a person has no
control, such as gender and age; these can be seen as part of a person™s initial
conditions. There are other variables over which she does have a degree of
control such as accumulation of different types of assets and skills, size of the
household, labour force participation, and so on. The time paths of the latter
group of variables are determined by the person herself, presumably operating
within some kind of optimizing decision framework under the constraints
imposed by the initial conditions of Y and D and the status of exogenous
variables X prevailing at the time of observation (™). Thus considering both
types of variables, the evolution of Dt— as a whole can be seen to depend on
the initial conditions (including preferences which guide the optimization
process) and the state of the exogenous variables prevailing at the time of
observation. This in turn implies that the evolution of the time path of the
conditional expectation of income E (Yt |Dt— , X™ ) also depends solely on the

permitted by her resources and in line with her subjective rate of time preference, which will
enable her to earn higher income over time.

S. R. Osmani

initial conditions and the state of the exogenous variables prevailing at the
time of observation.
The formal de¬nition of chronic poverty can now be translated informally
as follows. A person is chronically poor if her initial conditions ensure that her
income will fall below the poverty line more often than not for the rest of her
working life, if the exogenous variables prevailing at the time of observation
remain unchanged. This de¬nition clearly involves a prediction about the
time path of income.
However, the prediction involved is not about how income will actually
change over time. It™s a conditional prediction which assumes that the state
of exogenous variables prevailing at the time of observation will remain
unchanged for the remainder of a person™s working life. In reality, the exoge-
nous variables may change, which will alter the time path of expected income
and may even ensure that a person identi¬ed as chronically poor at the time
of observation will escape poverty at some point in future. Chronic poverty
is thus not a statement about a person™s actual future, but a statement about
the income dynamics that the current state of exogenous variables can be
expected to generate given the initial conditions of a person.
A salient feature of the de¬nition given above is that it is forward look-
ing instead of being backward looking. Much of the empirical literature on
chronic poverty adopts a backward-looking view, trying to discern from panel
data what kind of people have remained stuck below the poverty level over a
given period of time. For policy purposes, however, it is the forward-looking
view that must be of primary interest.
For the sake of completeness, we may now de¬ne transitory poverty in a
manner analogous to chronic poverty. Transitory poverty refers to a situation
where the time path of (conditional) expected income always stays above
the poverty line but suf¬ciently close to it that the actual income would fall
below the poverty line fairly frequently (but not most of the time), if the
exogenous factors remained unchanged. In Figure 11.2, this case is demon-
strated by the line TP, in contrast to chronic poverty, which is shown by the
line CP.
It should be apparent that this way of de¬ning the transitory poor involves
a certain degree of ambiguity while trying to distinguish them from the non-
poor. After all, the non-poor too would have a time path of (conditional)
expected income that always stays above the poverty line and if the (condi-
tional) density function of income has a suf¬ciently long tail at the lower end
then they too might slip into poverty from time to time. The main difference
with the transitory poor would be that they would have a higher time path of
expected income and would slip below the poverty line less frequently. That
makes the difference one of degree rather than of different kinds of dynamics.
The resulting ambiguity would, however, seem to be inherent in the concepts

Understanding Chronic Poverty

CP ’ Chronic poverty
TP ’ Transitory poverty
LP ’ Lifecycle poverty



™ t

Figure 11.2. Varieties of poverty

themselves”perhaps the difference between the transitory poor and the non-
poor is essentially one of degree rather than of substance.
Finally, it may be noted that the approach adopted here brings into light
a rather different kind of poverty that is often lost sight of in the discussion
on chronic versus transitory poverty. Consider the time path of (conditional)
expected income denoted by the line LP in Figure 11.2. This line is neither
always below the poverty line nor always above; instead it is below the poverty
line for some part of working life and above it for the rest. Obviously, it
does not ¬t in the category of either chronic or transitory poverty, and it
is certainly not the time-path of a non-poor person either. This is the case
of lifecycle poverty. A person may experience poverty in the early part of
life when the resources at her disposal, such as assets, skills, experience, and
labour force within the household, are not adequate to yield a suf¬ciently
high income to live above the poverty line. However, over time as some or all
of these resources increase through the process of accumulation and learning-
by-doing, the same person may be able to escape poverty in the mature years
of her working life.

S. R. Osmani

11.3. Endowments, Opportunities, and the Causation
of Chronic Poverty

11.3.1. Chronic poverty with or without trap
The dominant form of theorizing, at least by economists, on the nature and
causes of chronic poverty has centred on the concept of poverty trap. A person
is said to be caught in a poverty trap when the endogenous dynamics of
the economic system within which she operates does not offer any escape
route out of poverty. In trying to discern the reasons for such trap-creating
dynamics, the theories of poverty trap have drawn heavily upon the wider
literature on persistent poverty of nations that is observed in many parts of
the developing world.
The causes of persistent stagnation are analysed predominantly within the
framework of poverty traps at the national level, by drawing upon the insights
of endogenous growth theory. 3 In turn, the insights gained from studying
poverty traps at the national level are applied by some analysts to poverty
traps at lower levels”e.g. the levels of a community, a household, or an
individual. 4
There is no doubt that the idea of poverty trap has much to contribute
to our understanding of chronic poverty, for after all there is something
unmistakably chronic about being caught in a trap. However, there is a danger
of being carried away by the novelty and elegance of poverty trap theories,
derived in part from the novelty and elegance of endogenous growth theories,
and to be lulled into thinking that poverty trap is all there is to chronic
poverty. As shown below, chronic poverty can exist even without a poverty
In the framework developed in the preceding section, the idea of a poverty
trap is best captured by a time path of expected income that either hits
a ceiling that lies below the poverty line or approaches it asymptotically.
Formally, a poverty trap exists, if

E (Yt |Dt— , X™ ) ¤ W ¤ Z, t≥™
for all (11.5)

where W is an asymptote that is no higher than the poverty line income Z
and ™ is the point in time at which the observation is being made.
Time paths A and B in Figure 11.3 represent this case. In A the conditional
expected income path actually hits the ceiling below the poverty line and in B
it approaches the ceiling asymptotically. Obviously, if the dynamics generated
by the combination of a person™s initial endowments and exogenous factors

For a recent and comprehensive account of this literature, see Azariadis and Stachurski
See, for example, Barrett (2003); Barrett and Swallow (2003); Carter and Barrett (2005);
Bowles, Durlauf, and Hoff (2006).

Understanding Chronic Poverty

lead to time paths of expected income like these, then the person is caught in
a poverty trap from which there is no escape unless there is some favourable
change in the exogenous factors. Chronic poverty follows inevitably from
such a trap.
The special feature of such a trap is that the persistence of chronic poverty
becomes independent of the length of the working life or even the lifespan of
a person. No matter how long a person lives and keeps on working, there is no
way out of poverty, even with all the assets and skills she might accumulate
along the way. In the asymptotic case, the accumulation of assets and skills
will allow her to raise her income inde¬nitely, but it will do so at such a
sharply diminishing rate that despite an ever-increasing income she will not
be able to escape poverty”even if she were to live forever. This is chronic
poverty with a vengeance!
There is, however, a milder version of chronic poverty. In this case, either
there is no ceiling to the time path of expected income or if there is such a
ceiling then it lies above the poverty line. In other words, there is no poverty
trap, and yet a person may live the life of a chronically poor and die as one,
only because the time path does not rise above the poverty line during her
lifetime. The time paths C and D in Figure 11.3 represent this case.
There is a simple reason why this type of chronic poverty can exist. It is
that people have a ¬nite life. Even if the dynamics of the income generation
process were such that it would some day take the expected income above
the poverty line, a person may simply not live long enough to see that
day. Chronic poverty would be the outcome. In their preoccupation with
poverty traps, most analysts have failed to take adequate notice of this type
of non-trap chronic poverty that has to do with the ¬niteness of human
life. Perhaps the reason for this neglect lies in the fact that the literature on
poverty traps at individual and household levels has drawn inspiration from
the wider literature on persistent poverty at the level of the nation state or
large communities, where the idea of an in¬nite lifespan has more salience.
In any given context, the set of chronic poor may be made up of both
types of people”those who are caught in a trap and those who are not. The
relative importance of the two types of chronic poverty is an empirical matter.
On purely theoretical grounds, there is no reason for thinking that one type
or the other will dominate. More importantly, as we argue below, for policy
purposes this distinction does not really matter”what matters is the nature
of the constraints she faces, which could be similar for those who are in a trap
and for those who are not.

11.3.2. Causes of chronic poverty: macro versus structural constraint
Chronic poverty is a characteristic of the time path of expected income gen-
erated by the constrained choices made by individuals at each point in time

S. R. Osmani






™ T t

Figure 11.3. Varieties of chronic poverty

as to how to allocate their income between consumption and accumulation
of assets of various types (including human capital), given the constraints of
initial conditions and exogenous factors. The nature of the emerging time
path depends on the evolution of two sets of factors, namely (a) the portfolio
of assets”i.e. the quality and quantity of various types of physical, natural,
and human capital”a person can employ and (b) rates of return on different
types of assets. The evolution of the portfolio of assets will itself depend,
however, on the rates of return earned (or expected to be earned) on the assets
because these rates will determine both the ability and incentives of a person
to invest in different types of assets.
Therefore, the future income stream would depend essentially on the evo-
lution of the rates of return, as determined by individual choices made under
the constraints imposed by the initial portfolio of assets and the exogenous
factors. In other words, the time path of (conditional) expected income, which
determines whether or not a person will be chronically poor, is primarily a
function of the current and future rates of return on different types of assets.
If a person is chronically poor it must be because these rates of return are very

Understanding Chronic Poverty

low. This is true regardless of whether the person is caught in a poverty trap
or not.
The varieties of constraints that may depress the rates of return can be use-
fully classi¬ed into two groups”namely macro-level constraint and structural
constraint”leading to two different types of chronic poverty. One way of
seeing this distinction is to draw an analogy with the typology of unemploy-
ment. The distinction I wish to make here is analogous to the one between
Keynesian or demand-de¬cient unemployment and structural unemploy-
ment. Keynesian unemployment arises from de¬ciency of aggregate demand
at the macroeconomic level, which imposes an overall limit to how much
employment can be sustained by the economy. Structural unemployment, by
contrast, emerges not from the existence of an overall macroeconomic limit
to employment but from mismatch between the supply side and demand side
of the labour market. The mismatch consists in the fact that some workers
may not possess the kinds of skills demanded by the changing structure of an
economy: as a result, they fail to get employed.
The distinction I have in mind”between the types of chronic poverty
caused by macro-level versus structural constraints”has something in com-
mon with these notions of overall limit and structural mismatch. There
is one major difference, however. The overall limit in the present context
originates not so much from de¬cient aggregate demand, which is essen-
tially a short-run phenomenon, as from the limitations of market size,
which can be a longer-term constraint. What is relevant here is the famous
insight of Adam Smith that division of labour is limited by the size of the
The macro constraint, in the form of the size of the market, can be an
important limiting factor on the rates of return. For instance, if the size of
the overall market is small, then there will be limited opportunities for wage
employment, resulting in low wages. In the case of self-employed poor, if the
market in which they sell their products is small in size, any attempt to earn
higher income by expanding the scale of operation may be subject to rapidly
diminishing returns, because in a small market more can be sold only by
lowering the price. This will result in perennially low rates of return. If these
rates happen to be lower than either the subjective rate of time preference or
the rate of interest at which a person is able to borrow, new investments will
not be made and the scale of production will not rise. Poverty will become
Striking evidence of how market limitations can prevent the poor peo-
ple from escaping poverty has been found by studies on micro¬nance, for
example. An early evaluation of the Grameen Bank found that as the scale
of operation expanded the rate of return fell rapidly (Hossain, 1984). Many
subsequent studies have found that quite often poor borrowers would not opt
for borrowing at a higher scale even if the opportunity was offered to them,

S. R. Osmani

presumably because they were concerned about the falling rate of return. This
implies that so long as the limitation of market size remains a serious problem
the time path of expected income will not rise enough to enable the poor to
escape poverty altogether.
The limitation of market size is like an envelope that sets a limit to how far
individuals can go in their effort to improve their lot even if they have the
means to advance further, and it is all-embracing in nature in the sense that
it affects everyone although perhaps not equally. As the market size becomes
bigger, the envelope is pushed forward, thereby expanding the opportunity
set of everyone.
Structural constraint differs from the macro constraint of market limitation
in two ways. First, it affects speci¬c groups of people rather than economic
actors in general. Second, the constraint arises not from the overall lack of
opportunities but from a mismatch between the structure of opportunities
that become available and the structure of endowments possessed by a speci¬c
An example can be given from the ¬eld of micro¬nance again. Studies on
micro¬nance have found that even successful credit programmes fail consis-
tently to serve one group of the poor”namely landless wage labourers. In
principle, the opportunities for micro¬nance are open to the landless wage
labourers as well; in fact the Grameen Bank model was speci¬cally designed
to cater to the needs of the landless and near-landless people. The prob-
lem, however, is that wage labourers are unable to take advantage of the
opportunities opened up by micro¬nance. The opportunities that are created
are in the nature of self-employment”mainly in trading, but to a smaller
extent in handicrafts and farming as well. This may help those who already
have past experience in self-employment and have, through that experience,
accumulated some human capital speci¬c to their ¬elds. The landless wage
labourers, who have no such experience, do not possess this particular type
of human capital. Since the structure of their endowments does not match
the structure of expanding opportunities, they remain outside the reach of
micro¬nance programmes. 5
The problem of mismatch between endowments and opportunities is actu-
ally quite a pervasive one and manifests itself in many different ways. In
a different context, this has been described as the ˜integrability™ problem
(Osmani, 2006). The idea is that as the economy grows and the overall
economic opportunities expand, some people may ¬nd it dif¬cult to integrate
with the growth process since the structure of their endowments does not
match the structure of opportunities that are being opened up. The mismatch
can occur for a variety of reasons”some of them have to do with the nature of

Both the market limitation aspect and the structural mismatch aspect of the impediments
to poverty reduction in the context of micro¬nance have been discussed in Osmani (1989).

Understanding Chronic Poverty

technology, some with economic organization, and some with deeply rooted
social and cultural practices.
An example of limited integrability arising from the mismatch between the
skills demanded by an expanding economy and the skills possessed by the
poor is offered by the recent attempts at economic liberalization by some
Latin American countries. As these countries have tried to open up their
economies, they have found that, unlike the countries in East and South-
East Asia which had opened up their economies earlier, their comparative
advantage does not lie in the activities that are intensive in relatively unskilled
labour. The emergence of the poor and populous countries such as China and
India on the global scene has prevented Latin America from being compet-
itive in these types of products. Instead, they have found their comparative
advantage in activities involving technologies that use relatively more skilled
labour (Wood, 1997). The opening up of these economies has, therefore, led
predictably to an expansion of these skill-intensive activities, but the poorest
segments of the society have bene¬ted rather little from this expansion so far,
as they do not possess the skills that are needed by the expanding sectors. 6
For them, the structure of endowments does not match the structure of
Many of the disadvantages suffered by women in the economic sphere can
also be seen as a mismatch between endowments and opportunities when
gender is seen as an invariant endowment of a person. Culturally determined
phenomena such as gendered stereotyping of occupations, time constraint
imposed on women by the burden of combining productive and reproductive
activities, and discrimination in various spheres of life render the integrability
problem especially severe for poor women. 7 For instance, where poverty is
concentrated mostly among women, and yet the types of jobs for which
demand rises are culturally de¬ned as men™s jobs, poor women would bene¬t
little from overall expansion of opportunities. This problem is especially acute
in much of Africa, where crop production has acquired a gendered pattern”
with many cash crops being identi¬ed as men™s crops and subsistence food
crops being identi¬ed as women™s crops. As some of these countries move
toward greater liberalization of trade and commercialization of agriculture,
employment potential in the cash crop sector may receive a boost, but to
the extent that gendered pattern of crop production remains a constraint,
poor women will ¬nd it hard to take advantage of the new opportunities.
The overall economy may expand but many rural women will still remain
chronically poor because the structure of their endowments does not match
the structure of opportunities.
The resulting phenomenon of widening wage differentials between skilled and unskilled
workers has been analysed, among others, by Behrman, Birdsall, and Sz©kely (2000).
The speci¬c disadvantages faced by poor women are discussed, among others, by DAW/
UN (1999, 2001) and World Bank (2001).

S. R. Osmani

It may be useful at this point to explore the relevance of the macro
and structural constraints discussed above for the insights gained from the
poverty trap literature. The essential feature of poverty traps is the existence of
thresholds in the rates of return to assets. The thresholds have the important
property that diminishing rates of return become increasing rates once the
threshold is crossed. As more assets are accumulated, diminishing returns
might set in again, but this may not be a problem because there may exist yet
another threshold at a higher level after which increasing rates of return can
be enjoyed again. Those among the poor who can cross the initial thresholds
move out of poverty riding on the back of increasing returns; those who can™t,
remain caught in a poverty trap.
The important question is why some people fail to cross the thresholds. The
answer lies in a combination of several factors. First, the livelihood strategies
that would enable someone to enjoy the increasing rates of return beyond a
threshold usually involve some entry barrier, typically requiring lumpy invest-
ments. Second, those who are below the threshold to begin with do not earn
a suf¬ciently high rate of return to be able to accumulate the capital necessary
to overcome the entry barrier. Third, the poor people are suf¬ciently credit
constrained not to be able to borrow the required capital. The combination
of these three factors together ensures that those who start off below the
initial thresholds are condemned to remain below the threshold”caught in a
poverty trap.
Our discussion on macro and structural constraints is relevant to this story
of poverty trap at several levels. First, once the notion of structural constraint
is grasped, it becomes clear that the entry barrier to livelihood strategies offer-
ing increasing returns does not have to consist only in lumpy investments.
It may arise simply from a mismatch between the structure of endowments
and the structure of opportunities. If lumpy investment happens to be the
barrier, then the root of the problem appears to lie in ˜inadequate™ assets
in a quantitative sense”a person doesn™t have enough assets to make the
necessary investment. But if the mismatch creates the barrier, then the focus
must fall not on the size or magnitude of endowments but on their structure
or composition. For example, when a poor woman in Africa fails to make
the transition from low-paying food crop production to high-paying cash
crops, the barrier lies in the endowment of gender in the context of a society
that has evolved a gendered pattern of crop production. Gender rather than
lumpy investment acts as the entry barrier here. If lumpy investment were
the problem, it could in principle be solved either by transfer of assets or
by innovations in the credit market. But when gender is the problem, the
remedy involves an altogether different kind of social action. The implication
of all this is that the analytical underpinning of the concept of poverty trap
needs to be extended far beyond the constraint of lumpy investment so as to

Understanding Chronic Poverty

consider structural constraints of various kinds if the concept is to make sense
of chronic poverty more generally.
Second, the poverty trap literature recognizes that there is nothing inex-
orable about the low and diminishing rates of return below the threshold
and that appropriate changes in exogenous factors can lead to an improve-
ment, even to an extent that escape from the poverty trap might become
possible, but exactly how this might happen is not clearly explained. A little
re¬‚ection will show that what keeps the rates of return low and diminishing
below the threshold is nothing other than either the macro constraint or
the structural constraint discussed earlier. Softening of whichever happens to
be the binding constraint in a particular context will help raise the thresh-
old and thereby create the possibility of escape from poverty. Investigation
of these constraints and identi¬cation of the binding constraint in speci¬c
contexts should, therefore, form an integral part of any enquiry into poverty
Third, once it is understood that it is either the macro constraint or the
structural constraint that lies behind the poverty trap, it also becomes clear
that there is nothing special about poverty traps as an explanation of chronic
poverty. These same constraints are also at the root of the non-trap type of
income trajectories that fail to take a poor person above the poverty line
during her working life even though at some distant future they might have
done so if the person had lived long enough. In other words, there is no
fundamental difference in the causal stories behind the trajectories A and B
in Figure 11.3. Trajectory A leads to a trap while trajectory B doesn™t, but
both of them make a person chronically poor for much the same reasons.
Therefore, especially from this point of view it doesn™t really matter whether a
chronically poor person is caught in a trap or not. The much more important
distinction relates to the causality of chronic poverty, namely whether it is
the macro constraint or the structural constraint that is binding in a speci¬c
context, because as discussed below this will have a crucial bearing on the
choice of policy response.

11.3.3. Economic growth and chronic poverty
The relationship between growth and poverty has been much discussed in
recent times. There is an increasing recognition that while growth by itself
may not be suf¬cient to bring about a rapid reduction of poverty, high and
sustained growth is at least necessary for sustained poverty reduction. How-
ever, when it comes to chronic poverty, as distinct from poverty in general,
there is sometimes a presumption that economic growth may not be of much
help. Indeed, the whole research agenda on chronic poverty seems to be based
on the premiss that there is something very special about this type of poverty

S. R. Osmani

that at least weakens, if not nulli¬es, the potency of growth as a remedy. The
increasingly common tendency to identify chronic poverty with poverty trap
makes that presumption even stronger.
The analysis presented in this chapter suggests, however, that growth is
not necessarily impotent in reducing chronic poverty. Much depends on
the causation behind the genesis of chronic poverty. The distinction made
above between the macro constraint and structural constraint is especially
relevant here. The impact of growth on chronic poverty would depend cru-
cially on which of the two happens to be the binding constraint in a speci¬c
First consider the case where the macro-level limitation of market size is the
binding constraint”the envelope of opportunities available in the economic
system is simply too crushing to allow most people to earn a high enough rate
of return to whatever endowments they possess. For wage earners, it would


. 9
( 13)