ŮÚū. 8
(‚ŮŚ„Ó 13)



ences, particularly those based on gender. 11 Second, it will need to be adapted
for dealing better with newly formed communities, particularly those located
in large urban centres. To some extent, the study undertaken in North Car-
olina helped develop amendments in the methods appropriate for studying
urban areas. Because ‚Ępoverty‚Ä™ is less easily discussed publicly here than in
the other countries we had studied, and because communities are less stable
here, the Stages of Progress methodology needed to be modiÔ¬Āed for North
Carolina. Relatively more reliance was placed upon household interviews.
While the stages themselves were ascertained in community meetings, house-
holds‚Ä™ rankings and reasons were elicited mostly through household inter-
views (Krishna et al., 2006c). This procedure helped us to go forward with
the process of enquiry, but it compromised to some extent the triangulation
The detailed results, not provided here for lack of space, are available on request from
the author.
In our 36-village study in Gujarat, we interviewed members of a random sample of 133
female-headed households, Ô¬Ānding that 74% have remained poor over twenty-Ô¬Āve years, and
another 15% have become poor during this time, making a total of almost 90% poor in 2003.

Anirudh Krishna

and veriÔ¬Ācation that was possible. 12 Further reÔ¬Ānements have been made for
an ongoing study in Kenya, where communities in Nairobi and Mombasa are
being studied along with several others.
Applying Stages in a community setting helps abate to a considerable
extent the danger of stigmatization. By categorizing people as occupying
a particular Stage (1‚Ä“13) or belonging within some particular category (A‚Ä“
D), we have no need to refer to some individual as ‚Ępoor‚Ä™ or ‚Ęrich‚Ä™. Tracy
Rhoney, an enthusiastic and well-regarded community organizer in Burke
County, North Carolina, explained to me in her unforgettable accent: ‚ĘHoney,
it‚Ä™s almost like asking a woman about her dress size: Are you a Ô¬Āve or a
four? Are you Stage 5, or are you Stage 4? It‪s that simple.‪ She was right in
this regard; people who attended these North Carolina community meetings
spoke freely about their own positions along the Stages of Progress. They
were wary and close-mouthed, however, when someone else‪s situation was
Another danger for community-based studies is that of elite capture. In the
Stages process, we make clear at the very start of community discussions that
no tangible beneÔ¬Āts will be given out by us to anybody. This reduces the
incentives that people might have to fabricate or distort the facts, but the
danger of elite domination is not fully averted nevertheless. We have main-
tained some balance in the composition of the community group. In India,
for instance, we did not commence formal discussion until lower and upper
castes were both present. We also learned techniques for rotating community
respondents and isolating domineering speakers by taking them aside for
separate interviewing. One other part of the Stages process helped to reduce
the impact of elite domination: all facts ascertained in the community meet-
ing were separately veriÔ¬Āed in privately held household interviews. To the
extent the fear of elites does not also extend into private spaces, imbalances
arising in the community group were ironed out at this point in the study
Another potential weakness, common to all longitudinal studies, arises
on account of the changing compositions of communities and households.
Households twenty or even ten years hence will not be the same as the house-
holds of today. Some new households will be set up by young adults, and some
others will not be in the same place when a later study is conducted. Because
households do not remain the same over time, some simplifying assumptions
have to be made in longitudinal studies. Panel data studies consider house-
holds in the starting year of the study. They compare these households over
time, neglecting all households newly arisen. This neglect does not, however,
detract from the purpose of these studies, which is to understand and trace

It was comforting to observe that asset ownership continued to be closely related to

The Stages of Progress Method

households‚Ä™ trajectories over time. The Stages of Progress method involves
an equal though opposite neglect. By considering households at the end of
this period, this method neglects all households that have faded away during
the period. We have found in a few locations where we enquired about this
disappearance that it was undergone by roughly equal numbers of very rich
and very poor households, with members of both groups leaving to try their
luck in some city. By studying households that exist at the present time, we
could elicit, particularly in the case of younger households, the difference
between some individual‪s inherited and acquired status: did a person who
was born to poverty remain poor at the end of the period, or did she or he
manage to escape from poverty? Is another person who was part of a non-
poor household ten years ago still non-poor, or has she, regrettably, fallen
into poverty during this time? Compiling these trajectories‚Ä”of stability and
of change‚Ä”helped us to assess the overall situation of poverty over time.
More important, learning about the reasons for change in each individual
case helped to identify chains of events associated with escaping or falling
into poverty.
It needs to be mentioned that the reasons for escape and descent identiÔ¬Āed
in these studies are all micro-level and proximate, as experienced by house-
holds and individuals. More distant and macro-level reasons operating on
account of national policies and international economic conditions are not
directly identiÔ¬Āed using the Stages of Progress methodology; thus combining
these micro-level analyses together with macro-level examination of policies
and structures will help Ô¬Āll out a more complete picture. It would be useful to
undertake such a synthetic micro‚Ä“macro study.
It would also be interesting to undertake a study that combines monetary
measures of poverty together with community-based ones. In future work I
intend to undertake such comparisons. It would be useful to consider the
extent to which these different measures‚Ä”and perhaps also a third one, based
on an asset-ownership index‚Ä”offer the same or a different identiÔ¬Ācation of
No single method is ever adequate, I feel, for studying poverty in all its
complexities and dimensions. Different combinations of methods work better
for different ends. Which method or methods one elects to adopt must be
guided by the nature of questions that are addressed. For instance, Stages is
not useful for making cross-country (and in some cases, even cross-regional)
comparisons. Because somewhat different poverty lines are identiÔ¬Āed in differ-
ent countries, cross-country comparisons are not precise using this method. 13
Stages of Progress is also not very useful for looking at dimensions of poverty
other than material ones. Communities‚Ä™ rankings of households are enquired

Although, according to Reddy and Pogge (2002) and Wade (2004), comparability prob-
lems are also severe when some standardized metric is used.

Anirudh Krishna

after in terms related to material poverty. Other dimensions of poverty, includ-
ing social exclusion and political disempowerment, are not reÔ¬‚ected within
these assessments.
Combinations of methods for studying poverty will be required to Ô¬Āll
important gaps in poverty knowledge. Different methods are variously suited
for studying different facets. No one true method or deÔ¬Ānition of poverty
exists or can exist. Adopting a problem-solving approach is better than striv-
ing for purity of technique.


Barrett, C., and Clay, D. (2003), ‚ĘHow Accurate is Food-for-Work Self-Targeting in the
Presence of Imperfect Factor Markets? Evidence from Ethiopia‚Ä™, Journal of Development
Studies, 39(5): 152‚Ä“80.
and Maxwell, D. (2005), Food Aid after Fifty Years: Recasting its Role, New York:
Reardon, T., and Webb, P. (2001), ‚ĘNon-Farm DiversiÔ¬Ācation and Household Liveli-
hood Strategies in Rural Africa: Concepts, Dynamics, and Policy Implications‚Ä™, Food
Policy, 26: 315‚Ä“31.
Carter, M., and Barrett, C. (2006), ‚ĘThe Economics of Poverty Traps and Persistent
Poverty: An Asset-Based Approach‚Ä™, Journal of Development Studies, 42(2): 178‚Ä“99.
Chambers, R. (1988), ‚ĘPoverty in India: Concepts, Research and Reality‚Ä™, Discussion
Paper 241, Brighton: Institute of Development Studies.
(1997), Whose Reality Counts? Putting the First Last, London: Intermediary Technol-
ogy Publications.
Franco, S., and Saith, R. (2003), ‚ĘDifferent Conceptions of Poverty: An Empirical
Investigation and Policy Implications‚Ä™, available at <www.wider.unu.edu/conference/
Hulme, D., and Shepherd, A. (2003), ‚ĘConceptualizing Chronic Poverty‚Ä™, World Develop-
ment, 31(3): 403‚Ä“24.
Jodha, N. (1988), ‚ĘPoverty Debate in India: A Minority View‚Ä™, Economic and Political
Weekly, November: 2421‚Ä“8.
Krishna, A. (2003), ‚ĘFalling into Poverty: The Other Side of Poverty Reduction‚Ä™, Economic
and Political Weekly (Bombay), 8 February.
(2004), ‚ĘEscaping Poverty and Becoming Poor: Who Gains, Who Loses, and Why?‚Ä™,
World Development, 32(1): 121‚Ä“36.
(2006). ‚ĘPathways out of and into Poverty in 36 Villages of Andhra Pradesh, India‚Ä™,
World Development, 34(2): 271‚Ä“88.
Kristjanson, P., Radeny, M., and Nindo, W. (2004), ‚ĘEscaping Poverty and Becoming
Poor in 20 Kenyan Villages‚Ä™, Journal of Human Development, 5(2): 211‚Ä“26.
Kapila, M., Porwal, M., and Singh, V. (2005), ‚ĘWhy Growth is not Enough: House-
hold Poverty Dynamics in Northeast Gujarat, India‚Ä™, Journal of Development Studies,
41(7): 1163‚Ä“92.

The Stages of Progress Method

Lumonya, D., Markiewicz, M., Mugumya, F., Kafuko, A., and Wegoye, J. (2006a),
‚ĘEscaping Poverty and Becoming Poor in 36 Villages of Central and Western Uganda‚Ä™,
Journal of Development Studies, 42(2): 346‚Ä“70.
Kristjanson, P., Kuan, J., Quilca, G., Radeny, M., and Sanchez-Urrelo, A. (2006b),
‚ĘFixing the Hole in the Bucket: Household Poverty Dynamics in Forty Communities
of the Peruvian Andes‚Ä™, Development and Change, 37(5): 997‚Ä“1021.
Gibson-Davis, C., Clasen, L., Markiewicz, M., and Perez, N. (2006c), ‚ĘEscaping
Poverty and Becoming Poor in Thirteen Communities of Rural North Carolina‚Ä™,
working paper, Sanford Institute of Public Policy, Duke University. Available at
Laderchi, C., Saith, R., and Stewart, F. (2003), ‚ĘDoes it Matter that We Don‚Ä™t Agree on
the DeÔ¬Ānition of Poverty? A Comparison of Four Approaches‚Ä™, Queen Elizabeth House
(QEH) Working Paper Series 107, Oxford: QEH.
McGee, R. (2004), ‚ĘConstructing Poverty Trends in Uganda: A Multidisciplinary Perspec-
tive‚Ä™, Development and Change, 35(3): 499‚Ä“523.
Mellor, J. W. (1999), ‚ĘPro-Poor Growth: The Relation between Growth in Agriculture
and Poverty Reduction‚Ä™, report prepared for USAID, Washington.
Narayan, D., Patel, R., Schafft, K., Rademacher, A., and Koch-Schulte, S. (2000), Voices of
the Poor: Can Anyone Hear Us?, New York: Oxford University Press.
O‪Connor, A. (2001), Poverty Knowledge: Social Science, Social Policy, and the Poor in 20th
Century U.S. History, Princeton: Princeton University Press.
Ravallion, M., and Datt, G. (1996), ‚ĘHow Important to India‚Ä™s Poor is the Sectoral
Composition of Economic Growth?‚Ä™, World Bank Economic Review, 10(1): 1‚Ä“25.
Reddy, S., and Pogge, T. W. (2002), ‚ĘHow Not To Count the Poor‚Ä™, <www.
Salmen, L. (1987), Listen to the People: Participant‚Ä“Observer Evaluation of Development
Projects, New York: Oxford University Press.
Sen, B. (2003), ‚ĘDrivers of Escape and Descent: Changing Household Fortunes in Rural
Bangladesh‚Ä™, World Development, 31(3): 513‚Ä“34.
Timmer, P. C. (1997), ‚ĘHow Well do the Poor Connect to the Growth Process?‚Ä™, CAER
Discussion Paper No. 178, Cambridge, Mass.: Harvard Institute for International
Van Schendel, W. (1981), Peasant Mobility: The Odds of Life in Rural Bangladesh, Assen:
Van Gorcum.
Wade, R. H. (2004), ‚ĘIs Globalization Reducing Poverty and Inequality?‚Ä™, World Develop-
ment, 32(4): 567‚Ä“89.
Walker, T. S., and Ryan, J. G. (1990), Village and Household Economies in India‪s Semi-Arid
Tropics, Baltimore: Johns Hopkins University Press.

This page intentionally left blank
Part III
Explanatory Frameworks for
Understanding Poverty Dynamics
This page intentionally left blank
Bringing Politics Back into
Poverty Analysis
Why Understanding of Social Relations Matters More for
Policy on Chronic Poverty than Measurement
John Harriss

Poverty becomes what has been measured and is available for analysis.
(Robert Chambers)
[It is] a matter of a knowledge base that, however unintentionally, has
opened itself to conservative interpretation by locating the crux of the
poverty problem in the characteristics of the poor.
(Alice O‪Connor)

9.1. Reconceptualizing Poverty: The Story so Far

My Ô¬Ārst epigraph comes from a paper written by Robert Chambers for the
World Bank in Delhi twenty years ago, with the title ‚ĘPoverty in India:
Concepts, Research and Reality‚Ä™ (Chambers, 1988), which contributed to
rethinking on the concept and the nature of poverty in the 1990s (shown
up in the differences between WDR 1990 and WDR 2000). Chambers argued
that there are two possible starting points for understanding poverty and
ways of reducing it: with the perceptions of professionals‚Ä”social scientists
and development practitioners‚Ä”or with the perceptions of poor people them-
selves. His paper compares these two sets of perceptions. The professionals
deÔ¬Āne poverty in terms of deprivation and ‚Ęthe poor‚Ä™ are those who are
in various ways deprived. But in practice the professionals have concerned
themselves with those aspects of deprivation that are most readily measured‚Ä”
Ô¬‚ows of income or consumption‚Ä”and a huge amount of intellectual energy

John Harriss

and resources has gone, and continues to go, into poverty research which is
concerned with reÔ¬Āning these measures (involving a chain of assumptions).
But as Chambers says, the poverty line‚Ä”which is what so much research
has been about deÔ¬Āning‚Ä”is not concerned with wealth or material posses-
sions, nor with aspects of deprivation relating to access to water, shelter,
health services, education, or transport, nor with debt, dependence, isolation,
migration, vulnerability, powerlessness, physical weakness or disability, high
mortality, or short life expectancy; nor with social disadvantage, status, or
self-respect (1988, p. 3). Many possible aspects of deprivation are left out of
conventional poverty measurement, therefore‚Ä”and thus it was that Cham-
bers argued that poverty has come to be equated with what can most readily
be measured. When it comes to action too, he thought, professionals also tend
to focus on poverty deÔ¬Āned in terms of lack of income, and perhaps physical
weakness and isolation, rather than on those aspects of poverty that have to
do with vulnerability and powerlessness, perhaps because ‚ĘMembers of elite
groups . . . Ô¬Ānd [these] less threatening aspects of deprivation to measure and
tackle‚Ä™ (1992, p. 10).
Chambers contrasts this ‚Ęprofessional‚Ä™ way of thinking about poverty with
the concepts of the poor themselves, as these have been interpreted ethno-
graphically. He shows that if ‚Ęwe‚Ä™ (professional outsiders) take account of poor
people‪s own concepts and concerns then we should give much greater weight
to qualitative social and psychological aspects of well-being. He summed
up by arguing that we should think about poverty in terms of different
dimensions that are all relevant to poor people themselves. Incomes and
consumption do matter (he labels this dimension ‚Ęsurvival‚Ä™), but so do net
assets and security 1 (labelled ‚Ęsecurity‚Ä™), and beyond even security there is the
dimension of independence and self-respect (‚Ęself-respect‚Ä™).
These arguments contributed to recognition of the multidimensional char-
acter of poverty, and of not allowing it to be understood simply in terms of
the Ô¬‚ows that are most easily measured. They also helped to bring much more
sharply into focus the importance of taking account of the perceptions and
understandings of poor people themselves‚Ä”recognition of which led in the
next decade to that major programme of participatory research undertaken
by the World Bank that gave rise in the end to celebrated publications on
the ‚ĘVoices of the Poor‚Ä™. These adumbrated parts of Chambers‚Ä™s original
argument‚Ä”reÔ¬‚ected in the list of chapters that discuss the ten dimensions
of powerlessness and ill-being that emerged from the study, and in the sum-
ming up ‚Ęcall to action‚Ä™ in the volume Crying out for Change (Narayan et al.,

At about the same time that Chambers was writing, analyses of the ways in which people
respond to the stress of drought and famine showed, of course, that in these circumstances
they may choose to forgo consumption in order to maintain assets, striving to balance out
immediate survival and longer-run security: see e.g. de Waal (1989).

Bringing Politics Back into Poverty Analysis

2000). The chapter headings in the book recall Chambers‪s earlier listing of
different dimensions of deprivation or poverty (quoted above) rather closely.
The results of the ‚ĘVoices‚Ä™ research, however, display the same features that
characterize the literature on the measurement of poverty: causes and effects
are muddled up, and the characteristics of individuals, or of households, that
are associated/correlated with poverty are represented as causal. There is no
analysis of the structures and relationships that give rise to the effects that are
taken to deÔ¬Āne poverty.
This is also the principal limitation of Chambers‪s analysis of the concep-
tualization of poverty, and of poverty research: in his account of it poverty
remains a characteristic of individuals or of households (it is individuals or
households that lack incomes, security, and self-respect) and the effects of
poverty are sometimes represented as causes. Still, his paper does show that
‚Ępoverty‚Ä™ is a construct, and that it is construed in different ways by different
actors; he does begin to recognize that these constructions are profoundly
political‚Ä”in the passing remark about those ideas of poverty that maybe
suit the interests of elites; and there is more than a suggestion there that
conventional poverty analysis rests on a mistaken view of ‚Ęscience‚Ä™ that
elevates measurement and disregards contextualization. The last is a point
to which I return later in this chapter. The other points that I have raised
here have been taken up by several other writers in the more recent literature
of poverty, perhaps notably by Maia Green and David Hulme (Green, 2005;
Green and Hulme, 2005). The core of their arguments is that through the way
in which it is conceptualized in mainstream research, poverty is not seen as
the consequence of social relations or of the categories through which people
classify and act upon the social world. Notably the way in which poverty
is conceptualized separates it from the social processes of the accumulation
and distribution of wealth, which depoliticizes it‚Ä”and depoliticization is of
course a profoundly political intellectual act. Poverty is then treated as a kind
of social aberration rather than as an aspect of the ways in which the modern
state and a market society function.
What has been going on in mainstream poverty research in the time since
Chambers wrote his seminal critique? How much has research been changed
by the changing ideas that Chambers‚Ä™s paper exempliÔ¬Āes? In practice, though
there has been more exploration of alternative approaches, a great deal of
intellectual effort has continued to be expended on poverty measurement,
and on the related analysis of ‚Ępoverty dynamics‚Ä™ by comparison of the
characteristics of individuals or households that have remained poor (in the
sense of being below the conventionally deÔ¬Āned poverty line) over time, or
that have moved into or out of poverty. The focus of poverty research in
the World Bank, for notable instance, remains‚Ä”according to the information
given on the Bank‚Ä™s website‚Ä”on measurement, relying still on the headcount

John Harriss

measure, and based on nationally representative income and/or expenditure
surveys. There is also work going on dealing with risk and vulnerability, and
aspects of social exclusion, but it appears to be somewhat peripheral to the
main thrusts of World Bank poverty research.
Another vein in recent research that departs signiÔ¬Ācantly from the main-
stream work of World Bank researchers is that of the asset-based approach
developed by Christopher Barrett, Michael Carter, and their associates (see
the special issue of the Journal of Development Studies, February 2006). This is
based on the persuasive view that

Ô¬‚ow measures tend to be more subject to considerable measurement error than stock
variables, even in well-run surveys, because they can only rarely be directly observed
and veriÔ¬Āed. Moreover, productive assets are the durable inputs used to generate
income . . . Understanding the dynamics of assets is thus fundamental to understanding
persistent poverty and longer-term socio-economic dynamics.
(Barrett, Carter, and Little, 2006, p. 169)

The asset-based approach‚Ä”which in fact recalls in some respects work done
in the 1970s on differentiation and class formation in agrarian economies
(a point discussed later)‚Ä”has come up with impressive results, drawing on
longitudinal data of both qualitative and quantitative kinds, 2 that show up
the factors inÔ¬‚uencing movements into and out of poverty and highlight
the existence of poverty traps. The possession of assets, whether of land,
labour, livestock, human, or social capital, greatly inÔ¬‚uences the capacities
of individuals and households to withstand shocks, such as drought or‚Ä”as
is shown very often to be of particular signiÔ¬Ācance‚Ä”episodes of ill health
(reÔ¬‚ecting the fact of the particular dependence of the very poor on their own
bodies). Greater attention is paid in this work to structural determinants of
poverty but it is a moot point whether it has much to say about ‚Ęthe dynamics
of those underlying structural positions‚Ä™‚Ä”as Barrett, Carter, and Little claim in
their introduction to the special issue (2006, p. 169)‚Ä”as opposed to treating
precipitating causes of movements into or out of persistent poverty.
I will come back to the asset-based approach later in this chapter but turn
now to examine two country cases in which a lot of effort has gone into
poverty analysis on the lines suggested by the World Bank poverty research
programme: Vietnam and India. I aim to point up difÔ¬Āculties that derive from
the model of knowledge that underlies the poverty research industry.

The way in which this work has sought very deliberately to build links between quanti-
tative and qualitative research in the way suggested in the Conversations between Economists
and Anthropologists, orchestrated by Pranab Bardhan (1989), is very welcome, though in some
cases the use of qualitative cases studies is only to provide descriptive support to arguments
drawn from quantitative analysis.

Bringing Politics Back into Poverty Analysis

9.2. Questioning the Mainstream Model of Poverty Knowledge

Vietnam is widely regarded as a success story of economic liberalization (see,
for instance, The Economist, 5‚Ä“11 August 2006). Economic reform and integra-
tion into the global economy are held to have brought about growth that has
been remarkably pro-poor (Klump and Bonschab, 2004). Indeed, according to
data from the Vietnam Household Living Standards Surveys (VHLSS) poverty
fell by one-third between 2002 and 2004, which scarcely seems credible. The
Vietnam story depends on analysis of the Vietnam Living Standards Surveys
of 1992‚Ä“3 and 1997‚Ä“8, and then of the two rounds of the VHLSS. Though
sample designs and sample sizes in these surveys have changed they are held
to provide comparable results and they are widely used and widely respected.
Yet Pincus and Sender (2006) have recently shown that there are serious
problems with the design of these surveys that are likely to have resulted
in underestimation of the total numbers of very poor people in Vietnam.
These authors do not deny that rapid growth in Vietnam has improved
living standards for many, but they show that there are strong grounds for
believing that there are many more very poor people in the country than are
represented in the surveys on which poverty measurements are based. It is
particularly those who migrate for wage work who are likely to have been
missed, and Pincus and Sender argue that ‚Ęthe failure to capture migrants in
surveys that aim to measure living standards in a rapidly urbanising country
in which the structure of the labour force is experiencing profound change
leads to questions concerning the intent, representativeness and accuracy of
the surveys‚Ä™ (2006, p. 7, emphasis added). Migrants are excluded because
the sampling frame consists only of the ofÔ¬Ācial lists of registered households
in communes and urban wards of Vietnam, who must have lived in the
enumeration area for at least six months. The problem is compounded by the
fact that these lists of registered households are anyway often outdated. The
resulting exclusion of mobile people reÔ¬‚ects the precarious legal position of
migrants in Vietnam where the ho khau system of registration of households,
designed to control migration to cities, makes it difÔ¬Ācult for people to migrate
legally. The two authors show by means of comparison of VHLSS data with
the evidence of surveys conducted by the Statistics OfÔ¬Āce of Ho Chi Minh
City that the former excludes quite large numbers of young migrants; and that
VHLSS population estimates and census Ô¬Āgures don‚Ä™t match up, especially for
those aged 20‚Ä“9. Yet in an experimental survey conducted in rural areas of
Hanoi and four neighbouring provinces Pincus and Sender found that they
were able easily to identify relatively large numbers of ‚Ęillegal‚Ä™ migrants, in
spite of the blocking tactics in some cases of local administrators, and they
show that such migrants‚Ä”not all of whom, by any means, are poor‚Ä”have
very diverse characteristics. The survey still makes it clear that ‚Ęlarge numbers
of desperately poor people are living in geographical areas that conventional

John Harriss

analysis has classiÔ¬Āed as ‚Äúnon-poor‚ÄĚ ‚Ä™, and that ‚ĘIt seems likely that VHLSS has
mis-estimated poverty by excluding a large number of very poor and vulnera-
ble households‚Ä™ (2006, p. 40). Pincus and Sender further make the point that
it is difÔ¬Ācult to square claims from the living standards surveys regarding the
rapid decrease in the incidence of income poverty with anthropometric data
or with data on child malnutrition.
Analysis of the ‚Ędeterminants‚Ä™ of poverty in Vietnam, based on the liv-
ing standards surveys‚Ä”it is rather analysis of the characteristics of those
who are shown as still being poor‚Ä”highlights geographical factors (Klump
and Bonschab, 2004, for example, refer to emerging regional imbalances in
Vietnam), and those of household size, ethnicity, and educational attainment
(the ‚Ęusual suspects‚Ä™); and this analysis has led through to policy recom-
mendations in the Comprehensive Poverty Reduction and Growth Strategy that
is Vietnam‪s PRSP. Poverty reduction is expected to be driven in future by
private sector development, especially of household enterprises; and it will
be assisted by better targeting to ensure that poor people get access to basic
services, by the provision of infrastructure for poor and remote communes,
and by giving ethnic minorities greater voice in the design of anti-poverty pro-
grammes. Pincus and Sender argue that what is striking about these ‚Ęstandard
policy recommendations‚Ä™ is what is omitted. The emphasis is on household
enterprise, when

Studies from a range of developing countries show that the most secure route out
of poverty for the majority of the poor is access to regular waged employment [the
argument is spelled out in Sender, 2003; but the point is also made in a recent study of
poverty reduction in Bangladesh, in which it is argued that it has been waged employ-
ment in the rural non-farm sector rather than self-employment that has been associated
with poverty reduction: Sen et al., 2004]. Although the standard recommendations cite
job creation as a major objective, no attempt is made to account for labour market
dynamics, the determinants of the growth of unskilled wage employment and real
wages. (2006, p. 22)

It is very odd indeed, as Pincus and Sender say, that in a country like Vietnam
where so much emphasis is placed on urbanization and the development of
labour-intensive industries, poverty rates should be calculated based on data
‚Ęthat systematically exclude migrants to cities and industrial areas‚Ä™ (2006, p.
41). It is not that the signiÔ¬Ācance of migration has not previously been recog-
nized (it is discussed for instance by Klump and Bonschab, 2004), and the
two authors argue that ‚ĘIt is inexcusable that the poverty analyses for Vietnam
should make no reference to the fact that the VHLSS sample is limited to long-
term, legally registered households‚Ä™ (2006, p. 41). It is for this reason that they
see deliberate intent on the part of poverty analysts in the World Bank and the
government to paint a particular picture of poverty reduction in the country.

Bringing Politics Back into Poverty Analysis

This may be going too far, but it is easy to understand how one narrative
of change, attractive to those persuaded by theoretical arguments in favour
of particular policies, comes to drive the construction and interpretation of
Just how politically charged the apparently scientiÔ¬Āc task of counting the
poor can become is shown up very starkly in what Deaton and Kozel (2004)
refer to as ‚Ęthe Great Indian Poverty Debate‚Ä™. This is the debate over the impact
of India‪s liberalizing economic reforms, initiated in 1991, on the incidence
of poverty. Different perceptions have become highly politicized in circum-
stances in which the gap in terms of the measure of average consumption
derived from the National Accounts, on the one hand, and from the results,
on the other, of the regular household income and expenditure surveys run by
the National Sample Survey Organization has grown wider, and the reporting
periods used in the sample surveys for different categories of consumption
have been changed. One set of changes in reporting periods in an experiment
conducted by the NSSO increased estimates of per capita incomes by 15‚Ä“18
per cent, thus halving the numbers of the poor. Those who are supportive
of the economic reforms prefer one interpretation of inconsistent datasets,
while the critics of reform prefer another. Deaton and others have attempted
a considered reconciliation of the data, but the debate as a whole shows just
how sensitive measures of poverty are to statistical problems and the different
ways in which these problems are addressed. It also exempliÔ¬Āes Chambers‚Ä™s
point that ‚Ępoverty becomes what has been measured‚Ä™. Even if there were not
the particular technical problems that have arisen because of changes in the
design of the sample surveys in the 1990s, so that successive rounds of the
NSS are not easily compared with each other, it would still be the case that
the measurement of trends in the incidence of poverty is highly sensitive to
judgements made in a whole string of assumptions.
What puzzles many observers of Indian development is what the economic
processes are that can have brought about the kind of reduction in income
poverty that is claimed by some. How can it be that poverty has declined
as much as some maintain when, as is widely recognized, India has been
experiencing high rates of growth but without the creation of many regular
jobs‚Ä”‚Ęjobless growth‚Ä™, as it is described‚Ä”and when the agricultural economy
over much of the country is reasonably understood as being in a state of crisis?
The problem is brought out in studies of employment and poverty trends
in the city of Ahmedabad, once known as ‚Ęthe Manchester of India‚Ä™. In the
last twenty years of the last century as many as 100,000 ‚Ęgood jobs‚Ä™ were
lost in the cotton textile industry, and there has been extensive casualization
of employment‚Ä”as has happened very widely. Ethnographic research shows
that in these circumstances households have very often become more depen-
dent upon women‪s work for their survival. This has posed a serious threat to

John Harriss

the dignity and the self-esteem of men, who have reacted in ways that may
have harmful social and political consequences (an important theme that I
cannot pursue here). There are scholars, however, who argue that the evidence
from Ahmedabad shows that the policy of Ô¬‚exibilizing labour markets‚Ä”which
leads to casualization of labour‚Ä”is working, because there has been (in the
1990s) substantial growth in employment, a rise in the level of real wages,
and greater participation of both men and women in the labour process. Such
positive conclusions from the analysis of National Sample Survey data (by
the Deshpandes, by Dutta and Batley, and by Kundu) conÔ¬‚ict with those
from the ethnographic research of Jan Breman (2001). This shows that the
increased vulnerability of households has led to the greater involvement of
dependent members of families, both women and children, in work, and
that while workers may have ‚Ęregular‚Ä™ jobs in the dynamic sectors of the
urban economy such as power loom units, diamond ateliers, and garment
workshops, they can be dismissed at any time and do not enjoy the social
provisions that have historically accompanied ‚Ęformal sector‚Ä™ employment.
Breman further points out, on the basis of survey evidence, that underem-
ployment and low pay are extensive and that the percentage of the population
living in slum areas almost doubled between 1981 and 1996‚Ä“7. These differing
views of what is going on in Ahmedabad reÔ¬‚ect precisely the point made by
Kanbur in his comparable observations of the radical differences that exist
in perceptions of poverty trends in Ghana (2002): both the ‚Ęoptimists‚Ä™ and
the ‚Ępessimists‚Ä™ can in a sense be ‚Ęright‚Ä™ because they are looking at different
things. The optimists may be right. Employment in Ahmedabad may have
increased. But what about the quality of that employment, asks Breman? If
more people‪s livelihoods are more vulnerable, doesn‪t this connote deterio-
ration in levels of well-being, even if real wages have risen? Doesn‪t it mean
that they are more likely to enter into relations of dependence on particular
patrons, with negative implications for their self-respect and psychological

9.3. Going against ‚ĘNormal Science‚Ä™ and Making
Social Science Matter

ReÔ¬‚ection upon these examples of the kinds of difÔ¬Āculties that arise in poverty
measurement‚Ä”difÔ¬Āculties that are ontological as well as practical problems
of methodology‚Ä”points to the underlying problem with the whole model
of knowledge on which conventional poverty analysis rests. It is the model
of what we may call ‚Ęnormal science‚Ä™ which aims at developing explanatory
and predictive theory of universal application, based on generalization from
empirical observation. This is a model that has worked well in the natural

Bringing Politics Back into Poverty Analysis

sciences which have been characterized by the cumulation of knowledge as
well as by shifting paradigms. It is quite clear, I believe, that the social sciences
have not done nearly so well as the natural sciences in developing explanatory
and predictive theory of universal application (there are few lawlike general-
izations that can be made about human behaviour). Neither have they done
very well in cumulating knowledge, while they are characterized‚Ä”as Bent
Flyvberg has put it‚Ä”not so much by paradigm shifts as by style changes:
‚Ęit is not a case of evolution [in the social sciences] but more of fashion‚Ä™
(Flyvberg, 2001, p. 30). And there are powerful reasons for this difference
which have to do essentially with the nature of the phenomena with which
social scientists deal‚Ä”the actions/behaviour of self-reÔ¬‚ecting human beings‚Ä”
while the background conditions of the natural sciences are physical facts.
In social science the object of analysis is a subject, whereas the objects of
research in the natural sciences don‪t talk back. Of course studies of science
have shown that there is no radical distinction between the natural and the
social sciences, 3 and hermeneutics is now recognized as applying to natural
science too‚Ä”but it can still be demonstrated that the natural sciences are
relatively cumulative and predictive, and the social sciences not. This has
been a source of considerable anxiety for many social scientists‚Ä”reÔ¬‚ected
in my own experience, in regard to poverty research, in the ‚ĘConversations
between anthropologists and economists‚Ä™ set up by Pranab Bardhan (Bardhan,
1989). In these conversations some argued tenaciously that it must be possible
to establish ‚Ęthe facts‚Ä™ about poverty and to develop predictive theory of
universal application, whereas others (not all of them anthropologists) argued
that knowledge about poverty must always be context dependent. Now there
is even greater fear amongst many social scientists of a descent to relativism,
which the currents of postmodernism over the last two decades have served
to intensify.
Flyvberg‪s argument in Making Social Science Matter: Why Social Inquiry Fails
and How it can Succeed again (2001) is that social scientists set themselves an
impossible task in seeking to emulate the natural sciences. The crux of the
difÔ¬Āculty for the social sciences is that human beings are ‚Ęskilful‚Ä™‚Ä”referring
essentially to the ability of human beings to make judgements, and change
their ways of thinking and of behaving. Human skills go well beyond follow-
ing rules; they are context dependent. The kind of theory that is developed
in ‚Ęnormal science‚Ä™, on the other hand, depends on freedom from context
and the existence of rules (see Flyvberg‪s summary of his arguments, 2001,
p. 47). The social sciences, however, have distinctive strengths in areas where

This is brought out eloquently in Stephen Jay Gould‪s history of geology in Time‪s Arrow,
Time‪s Cycle (1987), which shows how archetypal differences between scholars in terms of
their conceptions of time and history led to very different theories about the ‚Ęfacts‚Ä™ of geology.

John Harriss

the natural sciences are weak‚Ä”precisely in dealing with reÔ¬‚exive analysis and
discussion of values and interests. Such analysis is necessarily context depen-
dent; but recognizing the centrality of context does not mean descending into
Flyvberg‚Ä™s aim is ‚Ęto help to restore social science to its classical position
[based on Aristotelian concepts] as a practical intellectual activity aimed at
clarifying the problems, risks and possibilities we face as humans and societies,
and at contributing to social and political praxis‚Ä™ (2001, p. 4). On the face of
it this is probably not a radically different ambition from that of those like
the poverty measurers who seek to pursue social-science-as-normal-science.
But the latter work with a model of knowledge which implies that scientiÔ¬Āc
analysis can establish, for example, whether or not policy changes in India
in the 1990s have led to a reduction in poverty, and that policy making
can be an exercise in rational problem solving. Flyvberg‪s view, however, of
what sort of knowledge is possible about people and societies is that it is
interpretative, and dialogical. In social-science-as-normal-science the key task
is taken to be the making of deductions and discovering of general principles
across large samples, and detailed case study research is often regarded as
unproductive (as it was by some of the economists at the ‚ĘConversations‚Ä™
conference to which I referred earlier). If we recognize the context dependence
of human action, however, then the kind of concrete, context-dependent
knowledge that may be derived from careful case study research is seen
as being ‚Ęmore valuable than the vain search for predictive theories and
universals‚Ä™ (Flyvberg, 2001, p. 72). Interestingly, a very similar conclusion
is reached by two economists in a recent review of theory and of empirical
research on economic growth. Kenny and Williams argue that ‚Ęthe social
world is more causally complex than the natural world‚Ä™ and that ‚Ęevents
rarely, if ever, have a single cause, but are rather the result of a conjuncture
of several factors or conditions [so particular historical analysis is essential]‚Ä™
(2001, p. 13); and they conclude that ‚Ęmore energy should be directed toward
understanding the complex and varied inner workings of actual economies
rather than trying to assimilate them into abstract universal models‚Ä™
(2001, p. 16).
The approach to research that Flyvberg advocates, therefore, is to address
real-world problems of particular societies, probably using a case study
methodology, in an interactive and engaged way (not to be equated, however,
with ‚Ęaction research‚Ä™) and to be ready to use a good deal of bricolage in draw-
ing on the work of other professional social scientists‚Ä”doing ‚Ęwhat works‚Ä™ to
address the key underlying questions: (i) where are we going?; (ii) who gains,
and who loses, by what mechanisms of power?; (iii) is this desirable?; (iv) what
should be done? Let me come back to this approach later in this chapter when
I outline a different approach to poverty research from the currently prevalent

Bringing Politics Back into Poverty Analysis

9.4. Poverty Research and the ‚ĘAnti-politics Machine‚Ä™

There are very strong similarities between the history of poverty research
and (less clearly so, perhaps) policy practice in the context of international
development, and that of ‚Ępoverty knowledge‚Ä™ in the United States, as this
has been analysed by Alice O‚Ä™Connor (2001)‚Ä”and I Ô¬Ānd it quite striking
that O‚Ä™Connor‚Ä™s suggestions about ‚Ęwhat is to be done‚Ä™ in poverty research
are closely comparable with Flyvberg‚Ä™s general propositions for ‚Ęmaking social
science matter‚Ä™. O‚Ä™Connor‚Ä™s argument starts with the observations that ‚ĘThe
idea that scientiÔ¬Āc knowledge holds the key to solving social problems has
long been an article of faith in American liberalism [and that] Nowhere is
this more apparent than when it comes to solving the ‚Äúpoverty problem‚ÄĚ ‚Ä™
(2001, p. 3). As I have suggested earlier, the international poverty research
industry, too, rests on the same article of faith‚Ä”that scientiÔ¬Āc knowledge
holds the key to solving the poverty problem. O‪Connor shows that although
early work on poverty in the United States linked it with unemployment,
low wages, labour exploitation, and political disenfranchisement‚Ä”‚Ęand more
generally [with] the social disruptions associated with large-scale urbanization
and industrial capitalism‚Ä™ (2001, p. 18) (note the similarity to Pincus and
Sender‚Ä™s arguments about contemporary Vietnam)‚Ä”it was quite soon turned
away from these matters of political economy. Latterly this has been associ-
ated with the inÔ¬‚uence of research foundations and government agencies,
which have provided large amounts of funding for poverty research, and
have been able to set the agenda. They have required that research should
be ‚Ępolicy relevant‚Ä™, ‚ĘscientiÔ¬Āc‚Ä™, and free from ideology‚Ä”but in all the work
that they have Ô¬Ānanced poverty has never been deÔ¬Āned as anything other
than an individual condition. Poverty knowledge rests on an ethos of scientiÔ¬Āc
neutrality, but it is very clearly distinguished by what it is not:

[C]ontemporary poverty knowledge does not deÔ¬Āne itself as an enquiry into the politi-
cal economy and culture of late Twentieth Century capitalism; it is knowledge about the
characteristics and behaviour, and, especially in recent years, about the welfare status of
the poor. Nor does it much countenance knowledge honed in direct action or everyday
experience . . . [which] kind of knowledge does not translate into measurable variables
that are the common currency of ‚Ęobjective‚Ä™, ‚ĘscientiÔ¬Āc‚Ä™ and hence authoritative poverty
research. (2001, p. 4)

The technically very sophisticated survey research on poverty that has been
carried on has by now built up a very accurate statistical portrait of poverty
in America, but the results of the interactions between politicians and pol-
icy makers, research foundations, and researchers have been to ensure that
poverty is seen as the failure of individuals or of the welfare system ‚Ęrather
than of an economy in which middle- and working-class as well as ofÔ¬Ācially
poor Americans faced diminishing opportunities‚Ä™ (2001, p. 241).

John Harriss

Very similar features characterize poverty knowledge in the international
context as well. Here too early studies of poverty‚Ä”such as Dadabhai Naoroji‚Ä™s
Poverty and Un-British Rule in India (1901) which sought explanation for
endemic poverty in India in the political economy of colonialism‚Ä”were
concerned with the structural conditions that caused the effects of poverty,
but the poverty research industry that became established in the 1970s has
turned to analysis primarily of the characteristics of the poor and of the
correlates of poverty. Studies of the causes of poverty, or latterly of ‚Ępoverty
dynamics‚Ä™, establish correlations between the characteristics of individuals
and households and poverty‚Ä”generally understood in terms of Ô¬‚ows of
consumption. Such studies have tended to highlight much the same broad
set of factors: features of households (high dependency ratios; female head-
ship; ill health of members); assets (holding few productive assets); educa-
tion (illiteracy); nature of occupations (lack of regular waged employment
amongst household members whether resident or working elsewhere); some-
times factors having to do with ethnicity and/or geography (e.g. being a
‚Ętribal‚Ä™/indigenous person in a remote area)‚Ä”and the signiÔ¬Ācance of crises
or of other idiosyncratic factors which in turn highlights the general problem
of the lack of insurance. What international poverty research has not done
very much has been to explain how and why these factors have the effects
they do, in the context of an analysis of the political economy of the locality
and of the state. Poverty research does not usually address the processes of
accumulation in contemporary capitalism and evades the problems of the
distribution of economic resources and of political power, apparently offering
technical solutions to the problem in a way that is not threatening to the elites
who beneÔ¬Āt from existing structures and relationships. The current mantra
about the role of ‚Ęprivate business‚Ä™ in growth and‚Ä”it is hoped‚Ä”in ‚Ępro-poor
growth‚Ä™ and the achievement of the Millennium Development Goals is only
one, particularly egregious instance of how language matters. International
poverty research, too, aims to be ‚Ęobjective‚Ä™ and ‚ĘscientiÔ¬Āc‚Ä™ and‚Ä”Chambers‚Ä™s
paper and the later ‚ĘVoices of the Poor‚Ä™ work notwithstanding‚Ä”has not much
countenanced knowledge deriving from direct action or everyday experience.
Just as it is striking that in the United States the problems of the poor
have not been connected with the economics of rising inequality but rather
have been ‚Ęcentred squarely on issues framed as ‚Äúfamily values‚ÄĚ ‚Ä™ (O‚Ä™Connor,
2001, p. 10), so it is striking that in Vietnam, for instance, contemporary
poverty knowledge should ignore the everyday experience which teaches
that the industrializing, urbanizing economy draws in large numbers of
migrant workers who are likely to be missed out in a sampling frame drawn
from lists of only registered households. As O‪Connor says in the second
of the epigraphs to this chapter, poverty knowledge in the United States
has opened itself to conservative interpretation. Poverty knowledge in the
international context, too, opens itself to conservative interpretation, at least

Bringing Politics Back into Poverty Analysis

in the sense that by reducing the problem of poverty to the characteristics
of individuals, abstracted from class and other power relationships‚Ä”note
the language of ‚Ęprivate business‚Ä™ rather than of ‚Ęcapitalism‚Ä™‚Ä”it has the
effect of depoliticizing it. The poverty research industry constitutes a part
of what James Ferguson (1990) memorably described as ‚Ęthe anti-politics
Alice O‪Connor concludes her history of poverty knowledge in the United
States by arguing that this knowledge needs to be reconstructed, and she
suggests Ô¬Āve important steps towards this reconstruction:

1. Shifting from explanation of individual deprivation to explanation of
inequalities in the distribution of power, wealth, and opportunity;
2. Recognizing that studying poverty is not to be equated with ‚Ęstudying
the poor‚Ä™;
3. Getting away from the research industry model;
4. Challenging the privilege attached to hypothesis-testing models of
5. Recognizing that the ideas of value-free social science and of Ô¬Ānding
scientiÔ¬Āc ‚Ęcures‚Ä™ for social problems are chimeras.

The last three of these points correspond very closely with Flyvberg‪s gen-
eral critique of the attempt to establish social-science-as-normal-science, and
all are entirely apposite in the case of international poverty research. Very
signiÔ¬Ācant amounts of money and of intellectual resources continue to be
poured into surveys like the Vietnam Household Living Standards Survey, for
the production of poverty headcounts based on detailed expenditure surveys
that are prone to enormous errors‚Ä”think, for instance of the impact on
poverty estimates for India of changing reporting periods. To what end? They
can never provide deÔ¬Ānitive answers to a question like ‚Ęwhat has been the
impact on well-being/ill-being of liberalizing economic reforms?‚Ä™, and they
actually provide very little information on the causes of poverty. Insofar as it
is important to monitor trends in income and its distribution then there may
be simpler and cheaper methods, such as collecting visually conÔ¬Ārmed data
on the consumer durables owned by households, or collecting information on
the education of all household members. There is a growing body of research
showing that the ranking of households by these means is not signiÔ¬Ācantly
different from that obtained by collecting information on household income
per capita (work by Filmer and Pritchett, 1998; Sahn and Stifel, 2000; Stifel and
Christiaensen, 2006; all cited by Pincus and Sender, 2006). And how many
more studies are needed to test hypotheses on poverty dynamics using data
obtained from living standards surveys? Such studies have often tended to
conÔ¬Ārm what Pincus and Sender reasonably describe as more or less ‚Ęstan-
dard‚Ä™ policy recommendations deriving from demographic and geographic

John Harriss

explanations that downplay the role of class formation and factors such as
gender discrimination in the labour market.

9.5. Refocusing Poverty Research

Rather than devoting international poverty research to the reÔ¬Ānement of
measurement (in the way that happened years before in poverty research in
the United States) and to hypothesis testing aimed at establishing predictive
theory, it will be more productive to redirect research so as give greater
attention to the analysis of the social processes, structures, and relationships
that give rise to poverty‚Ä”recognizing that the creation and re-creation of
poverty is inherent within the dynamics of capitalism (Harriss-White, 2006).
Such research will often be based on strategically selected case studies, in
which researchers build up familiarity with social practice in particular con-
texts (as Hulme and Shepherd suggest is likely to be necessary in analysing
chronic/persistent poverty: 2003)‚Ä”and desirably will help people themselves
to question the relations of knowledge and power that give rise to poverty
(though it is clearly essential in this case that this is done responsibly so that
poor people are not left to be victims of reprisals at the hands of the power
Within current poverty research some of the most interesting work is that
around the assets-based approach. It is quite striking, however, that this recalls
in signiÔ¬Ācant respects a much earlier vein of research on differentiation and
class formation in rural societies. Assets researchers construct indices of assets
and then identify thresholds‚Ä”such as the ‚Ęasset poverty line‚Ä™ used by Adato,
Carter, and May, ‚ĘdeÔ¬Āned as the level of assets needed to generate an expected
living standard equal to the poverty line‚Ä™ (Adato, Carter, and May, 2006, p.
230). This is similar to the procedure adopted by scholars who sought to
study peasant differentiation (see for example Harriss, 1982a). Recent work,
encouraged by development agencies, on ‚Ęlivelihood diversiÔ¬Ācation‚Ä™ is also
anticipated in the differentiation literature‚Ä”which was concerned with the
portfolios of livelihood activities of peasants in different presumptive classes,
and drew attention to the importance of rural non-farm activity at an early
stage (see e.g. Harriss, 1985; Bhaduri et al., 1986). Indeed, the analysis of
processes of differentiation in rural societies in some respects went beyond the
livelihoods approach that has found such favour with development agencies.
The latter ‚Ęis less (well) able to grasp the external inÔ¬‚uences on [the] disparate
components [“of income that rural people have to pull together in order to
make a living of sorts‚ÄĚ and] the extent to which rural dwellers are embedded
in regional and transnational economies‚Ä™ (Green and Hulme, 2005, p. 868).
Precisely these ‚Ęexternal inÔ¬‚uences‚Ä™ are brought into the analysis of the

Bringing Politics Back into Poverty Analysis

reproduction of households in the context of the development of capitalism‚Ä”
for example by Deere and de Janvry (1979).
The point of drawing attention to the ways in which some aspects of con-
temporary poverty analysis are anticipated in this older literature is because
this literature has certain strengths that are less apparent in contemporary
assets-based approaches. These do help to identify structural determinants of
poverty and they are ‚Ędynamic‚Ä™ insofar as they show how households move
in and out of poverty. But their dynamic analysis remains quite descriptive,
and though they are sometimes concerned with social relationships (as in
the work on social capital and social exclusion by Adato, Carter, and May,
2006), they do not address questions of political economy, nor do very much
to link up local patterns with wider processes of capitalist accumulation (see
also Green and Hulme, 2005, p. 9 of web copy). This is attempted in the
older literature on the political economy of agrarian change. For instance,
the work on African rural economies of Henry Bernstein and others shows
how places for petty commodity production are continually destroyed and
re-created with the development of capitalism, and his analysis of the ‚Ęsimple
reproduction squeeze‚Ä™ to which such producers may be subject places them
into relation with other classes: it is a relational analysis, showing how poverty
is reproduced under capitalism to the beneÔ¬Āt of owners mainly of money cap-
ital (see e.g. Bernstein, 1977, 1990). The analysis has the qualities that Green
and Hulme look for in the concept of chronic poverty, identifying ‚Ęthose in
society who have minimal or no prospects for economic and social mobility
and are structurally constrained by the social relations which produce poverty
effects‚Ä™ (2005, p. 9 of web copy). Work of this kind has some of the features,
at least, of the sort of social science advocated by Flyvberg.
To give a further, more detailed example: analysis of what was labelled
‚Ęsemi-feudalism‚Ä™ in West Bengal has been concerned with the relationships
that give rise to poverty rather than with measurement (though it also came
up with convenient measures of assets). This is a context in which the large
majority of rural people who own very small holdings of land, or whose liveli-
hoods are based upon agricultural and other forms of casual labour, depend
upon their relationships with the small class of larger land holders who are
themselves subordinate to the overarching power of the numerically tiny but
economically overwhelmingly preponderant group of rice millers. Household
reproduction in this context is described in village studies from the 1950s
(Bhattacharjee et al., 1958), and was analysed and modelled by Amit Bhaduri
some years later (Bhaduri, 1973). He shows how relationships of dependence
(and the ‚Ęcompulsive involvement‚Ä™ in markets, or ‚Ęforced commerce‚Ä™, that it
entails 4 ) ensure that the class of larger land holders comes to control most of

I refer here to the comparable arguments of Krishna Bharadwaj (1985) on ‚Ęcompul-
sive involvement‚Ä™ in markets, and of Amit Bhaduri (1986) on ‚Ęforced commerce‚Ä™. Both

John Harriss

the product of the region through rents from sharecropping and interest on
loans for subsistence and for production, so that they are then able to earn
speculative proÔ¬Āts from trading in rice. He then sought to show, more con-
troversially, how in these circumstances the larger land holders would have
no incentive to invest in productivity-raising technology, because this could
relax the dependence upon them of the small producers‚Ä”but in the present
context what is signiÔ¬Ācant about this work is the way in which it shows
how processes of accumulation bring about the reproduction of poverty. The
wealth of some is causally linked to the crushing poverty of others. 5 Some
years later I showed how, in spite of changes in the rural economy that
followed from the modest land reforms that had been brought about by the
then recently elected Left Front government of West Bengal, the reproduction
of households depended upon the same mechanisms (Harriss, 1982b). This
analysis also showed how a variety of non-crop agriculture-based activities,
and some non-farm activities, were involved in the survival of ‚Ępoor peasant‚Ä™
and agricultural labour households (‚Ęlivelihoods analysis‚Ä™, according to the
more recent terminology). Work by Barbara Harriss(-White) on the paddy and
rice trade (1983), conducted at the same time in the early 1980s, showed how
legislation enacted to ensure rice supplies to Calcutta underpinned the over-
arching power of the rice millers, on whose capital the entire rural economy
ultimately rested. Connections were made, therefore, with wider processes of
capitalist accumulation, and the whole body of literature and the analysis
it develops shows how poverty is reproduced through these processes. More
recent work has shown how agrarian reform in West Bengal, the institution
of panchayats and (in some instances) political mobilization of agricultural
labour, have been instrumental in relaxing the conditions of ‚Ęsemi-feudalism‚Ä™
and in bringing about higher levels of agricultural productivity and the
reduction of income poverty‚Ä”with the development of rural capitalism (see
Harriss, 2006, for a short review of literature). The widely attested relative
success of the state of West Bengal in reducing poverty (see Besley, Burgess,
and Esteve-Volart, 2004) has come about signiÔ¬Ācantly as a result of struc-
tural reforms and innovations rather than through programmes focused on
‚Ęthe poor‚Ä™.
Considerations of space preclude the development of further examples of
research that shows how relationships that arise in the context of the devel-
opment of capitalism inÔ¬‚uence the reproduction of poverty. The commentary
above on the work of Pincus and Sender on Vietnam refers to their emphasis

are concerned with the implications of the ways in which the commercialization of rural
economies takes place, in circumstances in which there are big disparities in entitlements.
As Maureen Mackintosh has put it in a critical commentary on the nature of markets,
‚ĘproÔ¬Āts of a few thrive in conditions of uncertainty, inequality and the vulnerability of those
who sell their labour power, and of most consumers‚Ä™ (Mackintosh, 1990, p. 50).

Bringing Politics Back into Poverty Analysis

on labour markets and how they work. Another example of research that
traces the links between the operations of labour markets and poverty is
in work by Gillian Hart in Indonesia (Hart, 1986), and later by Jonathan
Pincus (1996). In all of this it is extremely important to bring gender relations
into the analysis‚Ä”as, for instance, Ann Whitehead shows in relation to West
African societies (e.g. 1981), and Bina Agarwal in regard to South Asia (e.g.

9.6. Conclusion

I have argued here that mainstream research on poverty in international
development suffers from the same Ô¬‚aws as those that Alice O‚Ä™Connor brings
out in her analysis of ‚Ępoverty knowledge‚Ä™ with regard to the United States‚Ä”
and for similar reasons. O‪Connor refers to the role of research-funding agen-
cies in bringing about a preoccupation with measurement that has abstracted
poverty from its context in the way in which a particular capitalist economy is
functioning, and to the mistaken appeal to ‚ĘscientiÔ¬Āc neutrality‚Ä™ as the means
of justifying this. She then shows how this kind of poverty knowledge has
suited conservative interests. The same conclusions can reasonably be drawn
in regard to international development‚Ä”and they substantially explain the
persisting dominance of ‚Ęmeasurement approaches‚Ä™ in spite of the strength of
critiques, like that of Robert Chambers, that were developed more than twenty
years ago. Poverty knowledge exempliÔ¬Āes the kind of social science that is
critiqued powerfully by Bent Flyvberg, and there is reason for taking seriously
his arguments about building ‚Ęsocial science that matters‚Ä™‚Ä”arguments that
converge with O‪Connor‪s on the reconstruction of knowledge about poverty.
These are worth recalling here: shifting from explanation of individual depri-
vation to explanation of inequalities in the distribution of power, wealth,
and opportunity; recognizing that studying poverty is not to be equated
with ‚Ęstudying the poor‚Ä™; getting away from the research industry model;
challenging the privilege attached to hypothesis-testing models of enquiry;
recognizing that the ideas of value-free social science and of Ô¬Ānding scientiÔ¬Āc
‚Ęcures‚Ä™ for social problems are chimeras. Though the ‚Ęassets-approach‚Ä™ in the
recent literature has brought some advances it too fails to examine the social
and political-economic relationships that bring about the effect of poverty.
I have argued that the earlier and now largely disregarded literature on the
development of capitalism in rural economies (discouraged, of course, by
what O‚Ä™Connor refers to as the ‚Ęresearch industry‚Ä™) does develop the analysis
of these relationships‚Ä”which is so strikingly lacking in mainstream research
on poverty in international development. It is a literature, it is true, that is
concerned very much with ‚Ęprocess‚Ä™ rather than with ‚Ęoutput‚Ä™ (following one

John Harriss

of the distinctions between anthropological and economics-based approaches
recently made by Bardhan and Ray, 2006)‚Ä”but this seems more likely to
be conducive to practical action (including ‚Ępolicy‚Ä™) to address the causes of


Adato, M., Carter, M., and May, J. (2006), ‚ĘExploring Poverty Traps and Social Exclusion
in South Africa Using Qualitative and Quantitative Data‚Ä™, Journal of Development
Studies, 42(2): 226‚Ä“47.
Agarwal, B. (1994), A Field of One‪s Own: Gender and Land Rights in South Asia, Cambridge:
Cambridge University Press.
Bardhan, P. (1989), Conversations between Economists and Anthropologists: Methodological
Issues in Measuring Economic Change in Rural India, Delhi: Oxford University Press.
and Ray, I. (2006), ‚ĘMethodological Approaches to the Question of the Commons‚Ä™,
Economic Development and Cultural Change, 54: 3.
Barrett, C., Carter, M., and Little, P. (eds.) (2006), Understanding and Reducing Persistent
Poverty in Africa, Special Issue of the Journal of Development Studies, 42(2); and these
authors‚Ä™ ‚ĘIntroduction‚Ä™, pp. 167‚Ä“77.
Bernstein, H. (1977), ‚ĘNotes on Capital and Peasantry‚Ä™, Review of African Political Economy
10: 60‚Ä“73.
(1990), ‚ĘTaking the Part of Peasants?‚Ä™, in H. Bernstein et al. (eds.), The Food Question:
ProÔ¬Āts versus People?, London: Earthscan.
Besley, T., Burgess, R., and Esteve-Volart, B. (2004), ‚ĘOperationalising Pro-Poor Growth:
India Case Study‚Ä™, Department of Economics, London School of Economics.
Bhaduri, A. (1973), ‚ĘA Study in Agricultural Backwardness under Semi-Feudalism‚Ä™, Eco-
nomic Journal, 83(329): 120‚Ä“37.
(1986), ‚ĘForced Commerce and Agrarian Growth‚Ä™, World Development, 14(2): 267‚Ä“
et al. (1986), ‚ĘPersistence and Polarization: A Study in the Dynamics of Agrarian
Contradiction‚Ä™, Journal of Peasant Studies, 13(3): 82‚Ä“9.
Bharadwaj, K. (1985), ‚ĘA View on Commercialization in Indian Agriculture and the
Development of Capitalism‚Ä™, Journal of Peasant Studies, 12(4): 7‚Ä“25.
Bhattacharjee, J. P., and associates (1958), Sahajapur, West Bengal: Socio-economic Study of
a Village, Santiniketan: Agro-Economic Research Centre for East India, Visva Bharati
Breman, J. (2001), ‚ĘAn Informalised Labour System: End of Labour Market Dualism‚Ä™,
Economic and Political Weekly, Review of Labour, 29 December.
Chambers, R. (1988), ‚ĘPoverty in India: Concepts Measurement and Reality‚Ä™, IDS Work-
ing Paper (repr. in B. Harriss et al. (eds.), Poverty in India: Research and Policy, Delhi:
Oxford University Press, 1992).
Deaton, A., and Kozel, V. (2004), ‚ĘData and Dogma: The Great Indian Poverty Debate‚Ä™,
World Bank.
Deere, C., and Janvry, A. de (1979), ‚ĘA Conceptual Framework for the Analysis of
Peasants‚Ä™, American Journal of Agricultural Economics, 61(4): 601‚Ä“11.

Bringing Politics Back into Poverty Analysis

Ferguson, J. (1990), The Anti-Politics Machine: ‚ĘDevelopment‚Ä™, Depoliticisation and Bureau-
cratic Power in Lesotho, Cambridge: Cambridge University Press.
Flyvberg, B. (2001), Making Social Science Matter: Why Social Inquiry Fails and How it can
Succeed again, Cambridge: Cambridge University Press.
Gould, S. J. (1987), Time‪s Arrow, Time‪s Cycle: Myth and Metaphor in the Discovery of
Geological Time, Cambridge, Mass.: Harvard University Press.
Green, M. (2005), ‚ĘRepresenting Poverty and Attacking Representations: Some Anthro-
pological Perspectives on Poverty and Development‚Ä™, Global Poverty Research Group,
Working Paper No. 9.
and Hulme, D. (2005), ‚ĘFrom Correlates and Characteristics to Causes: Thinking
about Poverty from a Chronic Poverty Perspective‚Ä™, World Development, 33(6): 867‚Ä“
Harriss, B. (1983), ‚ĘPaddy and Rice Marketing in a Bengal District‚Ä™, Cressida Transac-
tions, 2.
Harriss, J. (1982a), Capitalism and Peasant Farming: Agrarian Structure and Ideology in
Northern Tamil Nadu, Bombay: Oxford University Press.
(1982b), ‚ĘMaking out on Limited Resources: or, What Happened to Semi-Feudalism
in a West Bengal District‚Ä™, Ecoscience: Cressida Transactions, 2(1‚Ä“2): 16‚Ä“76 (repr. in part
in J. Harriss, Power Matters: Essays on Institutions, Politics and Society in India, Delhi:
Oxford University Press, 2006).
(1985), ‚ĘWhat Happened to the Green Revolution in South India? Economic
Trends, Household Mobility, and the Politics of an Awkward Class‚Ä™, Discussion Paper
No. 175, School of Development Studies, University of East Anglia.
(2006), ‚ĘPostscript on Agrarian Reform and Agricultural Development in West
Bengal‚Ä™, in Power Matters: Essays on Institutions, Politics and Society in India, Delhi:
Oxford University Press.
Harriss-White, B. (2006), ‚ĘPoverty and Capitalism‚Ä™, Economic and Political Weekly, 41(13):
Hart, G. (1986), Power, Labor and Livelihoods, Berkeley and Los Angeles: University of
California Press.
Hulme, D., and Shepherd, A. (2003), ‚ĘConceptualising Chronic Poverty‚Ä™, World Develop-
ment, 31(3): 403‚Ä“24.
Kanbur, R. (2002), Introduction to the special section on Cross-Disciplinary Approaches
in International Development, World Development, 30(3): 477‚Ä“86.
Kenny, C., and Williams, D. (2001), ‚ĘWhat Do We Know about Economic Growth? Or,
Why Don‚Ä™t We Know Very Much‚Ä™, World Development, 29(1): 1‚Ä“22.
Klump, R., and Bonschab, T. (2004), Operationalizing Pro-Poor Growth: A Country Case
Study on Vietnam, AFD, BMZ, DFID, and the World Bank.
Mackintosh, M. (1990), ‚ĘAbstract Markets and Real Needs‚Ä™, in H. Bernstein, B. Crow, M.
Mackintosh, and C. Martin (eds.), The Food Question: ProÔ¬Āts versus People?, London:
Narayan, D., et al. (2000), Voices of the Poor: Crying Out for Change, New York: Oxford
University Press for the World Bank.
O‪Connor, A. (2001), Poverty Knowledge: Social Science, Social Policy and the Poor in
Twentieth Century US History, Princeton: Princeton University Press.
Pincus, J. (1996), Class, Power and Agrarian Change: Land and Labour in Rural West Java,
Houndmills: Macmillan.

John Harriss

Pincus, J., and Sender, J. (2006), ‚ĘQuantifying Poverty in Vietnam: Who Counts?‚Ä™, paper
presented at the Annual Meeting of the Association of Asian Studies, San Francisco.
Sen, B., et al. (2004), Operationalising Pro-Poor Growth: Country Case Study on Bangladesh,
AFD, BMZ, DFID, and the World Bank.
Sender, J. (2003), ‚ĘRural Poverty and Gender: Analytical Frameworks and Policy Propos-
als‚Ä™, in H.-J. Chang (ed.), Rethinking Development Economics, London: Anthem Press.
de Waal, A. (1989), Famine That Kills: Darfur, Sudan, 1984‚Ä“5, Oxford: Oxford University
Whitehead, A. (1981), ‚Ę ‚ÄúI‚Ä™m hungry, mum‚ÄĚ: The Politics of Domestic Budgeting in
Northeast Ghana‚Ä™, in K. Young, C. Wolkovitz, and R. McCullagh (eds.), Of Marriage
and the Market, London: CSE Books.

Poverty Measurement Blues
Beyond ‚ĘQ-squared‚Ä™ Approaches to Understanding
Chronic Poverty in South Africa‚ą—

Andries du Toit

We don‪t want complicated stories. What we need is a number. One num-
ber, if possible. One indicator that tells us where the poor and vulnerable
are. That‪s what we need.
(Member of the Regional Vulnerability Assessment Committee
(RVAC) for Botswana at an October 2004 planning meeting
of the Southern African Vulnerability Initiative)

10.1. Introduction

Discussions about the limits of econometric approaches to understanding
poverty are often framed as if the central differences are those between quan-
titative and qualitative method and as if the key issue up for discussion is the
best way of ‚Ęintegrating‚Ä™ them (see e.g. Kanbur, 2002; Hulme and Shepherd,
2003; Kanbur and Shaffer, 2007; Shepherd, 2007). This chapter argues that it
is necessary to go further. It considers the difÔ¬Āculties that arise out of the
domination of development studies and poverty research by what is here
called the ‚Ęeconometric imaginary‚Ä™: an approach that frames questions of
social understanding as essentially questions of measurement. But, although
the limitations of the econometric imaginary clearly illustrate the need for
qualitative modes of research and understanding, I argue here that more is
This chapter is based on research funded by the Chronic Poverty Research Centre (see
<www.chronicpoverty.org>). An earlier version was presented to the First International Con-
ference on Qualitative Inquiry (see <www.qi2005.org>). Many thanks to those who saw and
commented on these early drafts, including Tony Addison, Philippa Bevan, Colleen Crawford
Cousins, David Hulme, Uma Kothari, and Jeremy Seekings.

Andries du Toit

needed than various methods of combining or ‚Ęintegrating‚Ä™ qualitative and
quantitative approaches, as if these are traditions that can be connected to
one another without themselves being transformed or affected; or as if they
proceed from a set of underlying assumptions that can seamlessly merge.
Some of the differences that are often named in references to the split between
qualitative and quantitative go deeper than method or even epistemology.
What matters are also the larger explanatory metanarratives: the paradigms,
theoretical frameworks, and underlying ontological assumptions about the
nature of poverty, society, social knowledge, and judgement that guide the
process of ‚Ęintegration‚Ä™. Meeting this challenge may also require us to consider
the ways in which applied social science research in the twenty-Ô¬Ārst century
is shaped by the architectures of power and knowledge in modern states
and donor institutions. In South Africa these limitations, I argue, are part
of a fertile yet hazardous terrain for engagement and contestation by critical
scholars and researchers.
These threads of argument are hung from the rather humble ediÔ¬Āce of a
consideration of some years of ‚Ęchronic poverty‚Ä™ research conducted in South
Africa (see Aliber, 2001; Arnall et al., 2004 De Swardt, 2004a,b; du Toit, 2004,
2005; De Swardt et al., 2005; du Toit, Skuse, and Cousins, 2007). In the Ô¬Ārst
place, I argue that dominant approaches to the conceptualization of chronic
poverty are undermined by their reliance on a mystiÔ¬Ācatory theoretical
metanarrative that tries to imbue poverty judgements with a spurious aura
of objectivity, and by the fact that they direct attention away from structural
aspects of persistent poverty. Second, I argue that if the analysis of structural
poverty is to avoid either reductionism or a vitiating abstraction we need
to come to grips with the extent to which the structural conÔ¬Āgurations of
poverty are relational and socially meaningful: shaped through and through
by the complexities of culture, identity, discursive practice, and agency. Third,
I propose that this implies that more is needed than the simple addition
of qualitative data to existing measurement-based accounts: instead, critical
theory allows a reimagining and reframing of the way in which inequality
and poverty are conceptualized in the Ô¬Ārst place. The chapter closes with
a consideration of some of the obstacles and limitations in the way of
an attempt to bring these alternative ways of imagining poverty into the
mainstream of applied poverty work in South Africa.

10.2. Imagining and Understanding Chronic Poverty

10.2.1. Conceptualizing and measuring chronic poverty
Our research on persistent poverty in South Africa is essentially framed by
the organizing concept of chronic poverty. This is often given a fairly broad

Poverty Measurement Blues

meaning‚Ä”in the work of the Chronic Poverty Research Centre, for instance,
it refers inter alia to poverty of long duration, the poverty of those who
are poor for most of their lives and ‚Ętransmit their poverty‚Ä™ (sic) to subse-
quent generations, to the situation of those caught in poverty traps, and
to those who number among the ‚Ęhard-to-reach poor‚Ä™, etc. (see e.g. Hulme
and Shepherd, 2003; CPRC, 2004). Ultimately, however, chronic poverty is
usually understood in its canonical econometric sense, where it is deÔ¬Āned in
contradistinction to transitory poverty. Though the econometric analysis of
chronic poverty is possible on the basis of ‚Ęstatic‚Ä™ indicators that are robust
to change over time (e.g. Chauduri and Ravallion, 1994; see also McKay
and Lawson, 2003), a preferred strategy‚Ä”indeed the litmus test by which
other strategies are evaluated‚Ä”is to aggregate static snapshots in a way that
might allow a composite ‚Ęmoving‚Ä™ picture to emerge. A typical approach is
to run a panel dataset and to use a poverty line (most commonly monetary
in nature) to develop a dichotomous indicator which is then used to divide
the individuals in the population in each wave of the panel study into two
groups‚Ä”usually ‚Ęthe poor‚Ä™ and ‚Ęthe non-poor‚Ä™. Those who move above (or
dip below) the poverty line are held to have ‚Ęescaped poverty‚Ä™ (or to have
‚Ęentered‚Ä™ it); those who are counted as poor in every wave of the survey, or
who on average remain below the poverty line, are counted as the ‚Ęchronically
poor‚Ä™ (see Bane and Ellwood, 1986; Baulch, 1996; Baulch and Masset, 2003).
This approach dominates the ways in which ‚Ęthe chronic poor‚Ä™ are identiÔ¬Āed:
although other ways of approaching persistent poverty exist they are often
treated simply as complementary.
In this chapter I argue that important as the distinction between chronic
and transitory poverty can be, it is also very limited, focusing attention away
from other matters critical to the understanding of persistent poverty. It is also
tied up with some deeply problematic‚Ä”indeed, thoroughly mystiÔ¬Ācatory‚Ä”
underlying metanarratives about poverty itself, what it is and how it can be
scientiÔ¬Ācally known. To go beyond the limitations of the econometric concept
of chronic poverty, then, it is necessary to engage with the ways in which the
econometric imaginary dominant in applied social science frames the concept
of poverty itself.

10.2.2. Some chronic problems with poverty measurement
Let us begin this engagement by considering the practices of ‚Ępoverty mea-
surement‚Ä™ upon which the deÔ¬Ānition of chronic poverty‚Ä”and the identiÔ¬Āca-
tion of ‚Ęthe chronic poor‚Ä™‚Ä”depend. These involve two key operations. First,
they require the identiÔ¬Ācation of an indicator which stands as a proxy for the
state of poverty; and, second, they require the division of a ‚Ępopulation‚Ä™ into
two groups on the basis of this indicator.

Andries du Toit

These operations raise three key difÔ¬Āculties. First, poverty judgements‚Ä”
judgements as to whether someone is poor, and about what it is that consti-
tutes their poverty‚Ä”are ordinarily moral and political judgements: they derive
their import and are invested with signiÔ¬Ācance and consequence by virtue
of being embedded in and drawing on a rich and diverse current of over-
lapping and divergent underlying moral, philosophical, social, and religious
discourses about (inter alia) the nature of society, the identity of its members,
the nature of the obligations and claims that membership enables, and the
relationships between material lack, human suffering, and the claims of sol-
idarity. Any judgement about whether or not a particular person is poor‚Ä”or
about what the ‚Ęessentials of life‚Ä™ are, the lack of which constitutes poverty‚Ä”is
always a political judgement, and is as such almost always contested (Noble,
Ratcliffe, and Wright, 2004). Furthermore, poverty judgements are always
made by particular social actors acting in a particular strategic context and
are therefore always part of some larger social and political agenda.
This has important implications. It means, inter alia, that ‚Ępoverty‚Ä™ as a
concept in political and social discourse is an inherently messy notion‚Ä”one
that cannot simply and without loss be reduced to any one of its sometimes
contradictory and competing underlying threads. This is not a bad thing:
indeed, some of the power and importance of poverty as a concept in debates
about social justice, policy, and legitimacy in the present global order probably
lies precisely in its protean many-facetedness and its breadth of potential
meaning, which render it available for mobilization in a diversity of contexts
and allow it to be used to problematize and focus on a wide range of social
issues and phenomena.
But this also means that there is no objective, uncontroversial, value-free,
and unitary concept of poverty directly available for transparent operational-
ization by ‚Ęsocial science‚Ä™. Scholarly and applied research about poverty can-
not disregard this. The claims to truth, resources, time, and attention made
by ‚Ępoverty experts‚Ä™ are dependent‚Ä”even parasitic‚Ä”upon these broader and
essentially contested pre-existing political and moral metanarratives. Trying
to impart a spurious cut-and-dried ‚Ęobjective‚Ä™ scientiÔ¬Ācity to poverty measure-
ment is not to make it rigorous, but to mystify it.
This is not simply an abstract point. Consider the role played by poverty
lines in the attempt to make poverty judgements rigorous and objective. Evi-
dently this immediately raises the issue of just where the poverty line should
be set (for a South African discussion see e.g. Leibbrandt and Woolard, 2001).
Some have developed interesting approaches that attempt to ground this
decision in local consensus(es) about ‚Ęsocially accepted necessities‚Ä™ (Noble,
Ratcliffe and Wright, 2004), but quite often (see e.g. Baulch and Masset, 2003)
this decision seems to be informed by the assumption that value judgements
can be avoided altogether and that it is possible to develop a ‚ĘscientiÔ¬Āc‚Ä™
standard based on some ‚Ęobjective‚Ä™ reality (e.g. dietary needs, caloric intake

Poverty Measurement Blues

requirements, and the like). Almost inevitably this leads not to an uncontro-
versial but to a punishingly conservative poverty line‚Ä”one in which only those
who are at risk of starvation or malnutrition will ever really formally count as
poor‚Ä”and a situation where, paradoxically, there is widespread poverty above
the poverty line.
Second, one important consequence of the inherently political and moral
character of poverty judgements is that they ordinarily, ‚Ęin the wild‚Ä™, involve
a wide space for nuance and indeterminacy. It is part of the logic of the
concept of poverty that we can speak of someone as being, for example,
‚Ęnot very poor‚Ä™, ‚Ęalmost poor‚Ä™, or (in South Africa, for example) ‚Ępoor‚Ä”for
a white person‚Ä™. The econometric habit of dividing ‚Ępopulations‚Ä™ into ‚Ępoor‚Ä™
and ‚Ęnon‚Ä™-poor‚Ä”a distinction absolutely central to the way in which chronic
poverty is distinguished from transitory‚Ä”involves a misrecognition of this
essential feature. Though some have attempted to recognize the space for
indeterminacy in poverty judgements, e.g. by using fuzzy set theory (Qizil-
bash, 2002), these involve a doomed attempt to shoehorn them into a binary,
two-tailed form. 1
Third, poverty judgements are complex, theory-rich, and layered interpreta-
tions‚Ä”not simply of one aspect of a person or group‚Ä™s existence (how much
they earn, for instance) but of complex and dynamic states of well-being or
suffering. Though those states of being typically involve aspects of depriva-
tion, some of which may be quantiÔ¬Āable, these are moments in a complex
non-linear interactive process‚Ä”‚Ętransient elements in the moving now‚Ä™, as
Bevan (2004, p. 28) puts it‚Ä”a process in which they Ô¬Āgure both as momentary
outcomes of complex interactions and as determinants of further interactions.
What is central in understanding people‪s prospects and situation is not any
particular aspect of deprivation but how all the facets of their existence and
experience come together in a complex and always historically situated way
to produce a state of lack, powerlessness, suffering, or need which can then
(always in a particular context, always within the framework of meanings of
a particular political or moral discourse, and always by particular people with
their agendas and interests) be called poverty. This is why poverty is not just
a contested concept, but an essentially contested one.
Econometric deÔ¬Ānitions of poverty on the other hand are, as Bevan (2004)
has pointed out, measurement based, relying on the interpretation of ‚Ęindica-
tors‚Ä™ which in turn are created through abstracting and isolating particular
elements of people‪s overall situation from the broader context in which
they exist and assigning meanings to these in their own right. This kind of
abstraction is a tricky enterprise at the best of times, in which a lot depends on

This is because, contra Qizilbash and the fuzzy set theorists, saying ‚Ęsomeone is to some
extent part of the group of the poor‚Ä™ is not the same as saying ‚Ęsomeone is part of the group
of the to-some-extent poor‚Ä™. The later Wittgenstein had a lot to say about this.

Andries du Toit

the ability to use those indicators in an informed way‚Ä”and it is particularly
difÔ¬Ācult when used to Ô¬‚ag a condition such as ‚Ępoverty‚Ä™, which is highly
complex, comprising a number of different determinants, mechanisms, and
long-term trajectories. In practice what this approach comes down to is that
the deÔ¬Ānition of poverty is essentially collapsed into its indicator‚Ä”and the
indicator then taken for the condition it tries to measure: a circular operation
that directs attention away from the complex underlying causal dynamics
that link particular aspects of deprivation with the social experience of lack
disempowerment, need, and suffering.

10.2.3. Capabilities and multidimensionality
The problems pointed out here apply most trenchantly‚Ä”and most
obviously‚Ä”to that most familiar of ‚Ępoverty indicators‚Ä™: income or expen-
diture measured at household level. One approach that attempts to transcend
some of the limitations of this approach involves a focus, deriving from
the work of Amartya Sen, on ‚Ęmultidimensional‚Ä™ poverty and on people‚Ä™s
‚Ęcapabilities‚Ä™. Sen famously argued that the study of poverty should focus, not
on attempting to measure income and expenditure, but on the underlying
capabilities without which it is not possible to live a fully human life (e.g.
Nussbaum, 1999; Sen, 1999). This offers the potential for an account of
poverty that is alive to the complex and time-bound dynamics of deprivation,
suffering, and need. But though the capabilities approach has fundamentally
challenged some of the underlying assumptions of welfare economics its
implications have only been followed through in limited ways. Sen‪s frame-
work is notoriously hard to operationalize (see e.g. Martinetti, 2000), and
many attempts at operationalization have fallen foul of similar problems to
those described in the previous section. Typically, attempts to put it into
practice have involved identifying various capabilities (e.g. health, nutrition,
education, political participation), matching these to quantiÔ¬Āable indicators
(longevity, anthropometric measurements, school enrolments, democratic
institutions), and then trying to assess whether people are deprived or not
according to these criteria (see e.g. Klasen, 2000; UNDP, 2002; Barrientos,
2003; Qizilbash, 2003; McGillivray, 2003). This can shed valuable additional
light on the extent and nature of poverty, making visible aspects of depri-
vation not discernible from a monetary perspective alone‚Ä”but ultimately
the underlying problem has not been transcended, and sometimes leads to
approaches that just seem to miss the point. McGillivray (2003), for instance,
has endeavoured to use correlations between ‚Ęnon-economic dimensions of
well-being‚Ä™ (life expectancy, adult literacy, gross school enrolment) to empir-
ically identify ‚Ęthe variation not accounted for by income per capita‚Ä™, and
then take this variation as an ‚Ęaggregate measure of non-economic well-
being‚Ä™‚Ä”assuming, in other words, that there is some abstract thing called

Poverty Measurement Blues

‚Ęnon-economic well-being‚Ä™ which all these indicators partly measure. Another,
less extreme example is again Baulch and Masset (2003), who understand the
idea that ‚Ęmonetary and non-monetary indicators of poverty tell different
stories about chronic poverty‚Ä™ to mean that there are ‚Ędifferent subgroups‚Ä™ of
the chronic poor, or even different kinds of chronic poverty (e.g. ‚Ęnutritional
poverty‚Ä™, ‚Ęchronic education poverty‚Ä™‚Ä”Baulch and Masset, 2003, pp. 449,
Aside from the conceptual difÔ¬Āculties involved in describing capability
deprivation in this way (how can hunger, for example, be described as ‚Ęnon-
economic‚Ä™?) this approach produces intractable problems when used to try
to identify ‚Ęthe chronic poor‚Ä™ on the basis of panel studies. Are ‚Ęthe chronic
poor‚Ä™ only those who show up as deprived every time along every dimension
measured? If we do not wish to adopt such a rigorous criterion, should
we disaggregate ‚Ęthe chronic poor‚Ä™ into ‚Ęthe chronic monetary poor‚Ä™, ‚Ęthe
chronically malnourished‚Ä™, and so on? And how are we to understand the
difference between those who are deprived in ‚Ęonly one‚Ä™ dimension and those
who suffer multiple forms of deprivation? (Is someone who is educationally
deprived, chronically sick, and food insecure three times as poor as someone


ŮÚū. 8
(‚ŮŚ„Ó 13)