. 7
( 13)


which were met by unsustainable coping strategies. The long-term mismatch
between income and expenditure was met by the sale of assets, but with each
consecutive sale, the ability to cope was further weakened. The outcome of
smooth decline is the result of cumulative causes where coping strategies with
each setback reinforce the impact of the next event. In considering the policy
implications of this pattern of decline we need to take a long-term perspective
when addressing the mismatch between downward pressures and the ability
to mitigate them. The prevention of denudation of coping power resources
would form an important part of a social protection strategy”in this case the

1 bigha = 1/3 of an acre.


Peter Davis
Independence Four children died 1974--1984 Nursery
war 1971 destroyed in
flood 1998

Daughter married,
dowry paid over 1
Father died
Father ill 1962--1977 year 1985
15 bighas of land sold

Brother has TB from

Working as a
rickshaw driver Wife ill for
in Kushtia one year
Working as a chowkidar 1979--
1950--1979 1974
present”bribes required to receive salary

Figure 7.1. Declining smooth: Sukur Ali
Poverty Dynamics in Bangladesh

sale of land”alongside the mitigation of common downward pressures such
as medical problems, medical expenses, court cases, and dowries.

Only three cases of ˜improving smooth™ appeared among the chosen ninety
life histories. In all cases the improvement in life circumstances was not so
much an emergence from poverty as the gradual consolidation of already
quite strong positions in communities. Smooth inclines generally re¬‚ected
either an ability to cope with the inevitable vicissitudes of life by having
assets, income, and other social and political power resources, or by being
fortunate enough to not be visited by major setbacks. However, what often
seems to be good fortune often has a structured component of being either in
an advantaged position with a strong power-resource base, or occupying a less
vulnerable a place in the life cycle.
One of these cases of a smooth incline is depicted in Figure 7.2. Jehangir
is a 40-year-old man who had secured a long-term position as a gardener
with a local Kushtia-based NGO earning approximately Tk. 3,900 per month.
He also acted as a shordar (local labour intermediary) organizing day labour
for work with the NGO in town when this was needed. This position of
economic security and a strong social network allowed consolidation of an
already strong position. His family owned 7 bighas of land of which 2 bighas
belonged to him personally. He had not yet faced dowry costs for daughters or
major medical costs for his parents and was one of ¬ve brothers, who all lived
locally. His parents were also still relatively healthy. A reliable income stream,
with some assets being consolidated, and a strong social network in the local
area contributed to a smoothly improving trajectory.
Smooth and level trajectories were not common and were usually associated
with interviews that had not gone well. A common response that poor people
gave when initially asked about their situations, particularly before a rela-
tionship of trust and rapport had been established, was that their condition
(obosta) was bad and had always been so with very little change.
One of the reasons I developed an approach based on life history interview-
ing was that I found that the same people produced stories which showed
that their life trajectories were by no means smooth and level once time had
been taken to build up trust and rapport with them and as they understood
that I really was interested in the details of their lives. However, in a small
number of cases, episodes of some lives re¬‚ected a fairly constant condition
even after detailed histories had emerged. The problem of distinguishing
between a truly smooth trajectory and a level saw-tooth highlights the dif-
¬culty of using a retrospective interviewing technique. With experience a
researcher can distinguish a problematic interview from a history which really
has few vicissitudes. However, problems of memory, differing perceptions of


Peter Davis
Sister married in 1995
1971 War Married in 1980 parents and all five
Started work with brothers contributed to
NGO in 1988 costs

Fairly secure job with

Figure 7.2. Improving smooth: Jehangir
Poverty Dynamics in Bangladesh

what constituted a crisis, and varied rapport with respondents make absolute
certainty dif¬cult to achieve.

7.7.2. Saw-tooth trajectories
Trajectories which resembled the teeth of a single-action saw were the most
common pattern in life trajectories and within episodes in trajectories. In
these, gradual improvements were interspersed with more abrupt declines.
Of what I judged to be the ninety most accurate life history interviews, forty-
eight (or 53 per cent) had signi¬cant episodes following this pattern.

The level saw-tooth pattern was usually the re¬‚ection of a number of short-
term improvements being reversed, usually suddenly, by intermittent crises,
producing a long-term trend of neither improvement nor decline. In most
cases the potential for long-term improvement was hampered by the common
problems identi¬ed, such as dowry or illness, often coupled with a low level
of resilience and low income making these full crises, rather than minor
The case outlined below (and Figure 7.3) illustrates a number of common
features which occurred with the level saw-tooth pattern. Improvements are
possible through household members being economically active; however
a number of lifecycle-related crises also occur, mostly involving illness and
dowry expenses.
Fuljan is a 33-year-old woman who lives with her husband, ¬ve children
(three boys and two girls), and her 66-year-old mother-in-law. Another daugh-
ter was married in 1998 and moved to her new in-laws. There were also six
other households of her husband™s brothers who lived in the same bari. Her
husband worked as a loom master and earned approximately Tk. 1,200 per
month and her eldest (17-year-old) son also earned about the same working
with a handloom. They owned about 1.5 kata 8 of land where the house was
built and their total assets were worth approximately Tk. 12,800. Their house
was in a bad state of repair and they are looked down on in the community
as ˜poor™ (goreb).
Fuljan was married at the age of 12 in 1981 after her father had died from
typhoid during the famine in 1975. Both Fuljan and her husband came from
poor families. In 1984 her mother-in-law (sasuri) became ill with a stomach
ulcer and they had to pay Tk. 13,000 for medical treatment. To fund this her
husband™s brothers sold a handloom. In 1987 Fuljan herself became ill after
a baby died before birth and Tk. 20,000 was spent on her medical treatment.
In 1990 the brothers divided the family property and in the following years
1 kata = 1/60 of an acre.

Peter Davis

their position improved. In 1998 her eldest daughter was married with a total
cost of Tk. 14,000 (10,000 dowry and 4,000 costs). To raise money for this an
advance was taken from an employer, Tk. 5,000 came from relatives, a goat
was sold, and an NGO loan was taken out. Now with her eldest son working
and contributing to the household income as well as her husband, she sees
her life condition improving over the coming years.
This level saw-tooth pattern tended to be the most common type of tra-
jectory for poor people who had maintained roughly the same level of life
condition over a long period of time. Downward crises were often experienced
as abrupt declines with gradual recovery between. For poor people with few
coping resources, shocks tended to be converted more directly into a decline
in well-being with more likelihood of a cumulative decline.

The declining saw-tooth trajectory has much in common with the multi-step
decline pattern discussed later and it is often dif¬cult to distinguish between
the two. However, it is useful to have two separate categories to distinguish
trajectories declining without much recovery between downward steps (multi-
step decline) from those where some recovery occurs but with further crises
nevertheless outweighing the improvements. As can be seen in the cases
considered below most real-life trajectories are a mix of the various ideal types.
The case of Amir Hossain below illustrates the saw-tooth pattern in an over-
all declining trajectory (Figure 7.4). His trajectory is beset with the common
crises such as illnesses, but times of improvement were also identi¬ed: land
was bought using a gift from in-laws and a new business was started. However,
the larger crises”a brother™s death in an accident and his mother™s illness
and death”outweighed the improvements, producing an overall downward
Since this pattern is the most common type for those who decline into
poverty, or from poverty into further destitution, a better understanding
of the long-term dynamics of patterns of decline within individual and
household life cycles is crucial for the better formulation of protective and
preventive social protection strategies. It is also particularly important in the
planning of formal social protection to seek a synergistic combination of
formal and informal strategies.

Improving trajectories of the saw-tooth pattern are the most common trajec-
tory of poor people when they are emerging from poverty. Table 7.4 shows
the main reasons for improvement in a sample of people with life trajec-
tories of this type. These highlight a number of important explanations for
an improved trajectory even while people were beset by intermittent crises.

Married in

law™s medical
Brothers divided
treatment 1984
Father died during Marriage of
1974 famine doughter
Tk. 14,000
Ill after baby
total costs
died in womb

Son also

Poverty Dynamics in Bangladesh
Husband working with working

Eye problems Tk. 4,000

Figure 7.3. Level saw-tooth: Fuljan”potential for improvement hampered by dowry and illness

Peter Davis
1971 War

10 Bigha of land, Wife ill over 2
Father ill from
trees and 4 cows years Brother died in
1979 and died Mother died from
stolen accident 2000
1982 hep. b
Tk. 15,000 cost
Wife ill

Bought 5 kata land with gift
Started a new business from
from sosur bari
mortgaged land

Figure 7.4. Declining saw-tooth: Amir Hossain
Poverty Dynamics in Bangladesh

Table 7.4. Reasons for improvement in saw-tooth trajectories

Age Sex Main reason(s) for an improved trajectory

20 F Suf¬cient regular work for husband with handloom, driving a van gari, and agricultural
day labour
26 M Started a new business bene¬ting from: receipt of dowry, NGO loan in wife™s name, and
illegal electricity connection
29 M Permanent job with an NGO
30 M Division of property between brothers, regular income from handloom
35 F Two sons begin working and a van gari was bought using an informal loan (howlat)
35 M Used an informal loan to open a tea shop in the bazar, and wife started working
39 M Loans taken out from relatives and an NGO and a new business was started
40 F Son starting to earn income
42 F Son starting to earn income, daughters all married
42 F Son starting to earn income
42 M Successful farming, hard work, and entrepreneurial use of other™s mortgaged land
49 M Regular income from agriculture, handloom, son™s dowry, and son beginning to work
55 M Regular income from agriculture and a number of sons working
56 M Continued small but regular income as a school teacher

Most improvements were associated with the achievement of a steady income
stream by a member, or members, of the household which provided a buffer
for intermittent crises. The lifecycle transition of sons beginning to work and
contributing to the household was particularly important, especially for the
women interviewed and especially for women who had lost a husband. The
ability to establish or consolidate small businesses using either informal or
NGO loans was also very important. In other cases, bene¬ts for some people
had costs attached for others. For example, the receipt of dowry money for
men was commonly used to start businesses, while the costs to the bride™s
family often produced a crisis. Also, in some cases the division of property
among brothers led to bene¬ts for some but with a decline for others who
had bene¬ted in the past from property being held in common.
Even though Table 7.4 shows that work-related activities of household
members are an important part of emergence from poverty, we should be
careful not to neglect other less visible aspects of this emergence. The absence
of crisis involves both upward opportunities and the lack of, or mitigation of,
downward pressures. What could have happened if something else was absent
will obviously not feature as strongly as what did happen.
The danger is that a super¬cial reading of stories of emergence from
poverty due to work-related activities can lead to policy priorities based solely
around promoting income generation”as is the case with the micro¬nance-
dominated poverty-reduction model which has existed for the last two
decades in Bangladesh. However, emergence from poverty has two sides:
slow upward episodes of opportunity interspersed with sudden, downward
crisis-causing events. The prevention or mitigation of downward crises can
also help convert declining saw-tooth trajectories into improving ones, but

Peter Davis

Table 7.5. Examples of causes of single-step decline patterns

Nature of crisis Gender Age

1. Husband poisoned on a train, became paralysed, and was ill for 10 years. F 45
Business, house, trees, sold to pay for husband™s medical care before his death.
2. Husband died when he fell from a tree F 38
3. Husband died from an undiagnosed chest infection F 37
4. Illness and death of husband from asthma leading to survival by begging F 50
5. Court case fought over disputed land for 5 years. Other land sold to pay for the F 60
case. When the case was lost her husband died from a heart attack leaving the
household landless and destitute
6. Father died leaving no money F 46
7. Husband attacked by a rival gang in a village dispute. Tk. 15,000 treatment costs F 55
followed by death from injuries sustained in the attack 8 months later.
8. Abandonment by husband. Kushtia textile mill closed leading to loss of husband™s F 45
job (alongside 1,000 other workers) which led to her husband abandoning her.
9. Tk 30,000 medical costs before death of father raised by the sale of 2 bighas of M 27
land. Now works as a landless day labourer.
10. Eye lost in an accident with a rice-husking machine. M 44
11. Land and property lost by theft and forcible occupation. M 60
12. Combination of illness of mother and costs associated with marriage of sister. M 33
13. Illness of child at the same time as division of property between brothers. M 29

because they prevent what could have happened, they can be overlooked when
tales of emergence are favoured in development literature. When stories of
success have disproportionate coverage, social protection measures can be
neglected. In the Bangladesh case the mitigation of the negative impacts due
to illness (particularly of the elderly), dowry costs, social con¬‚ict, marriage
breakdown, and household dissolution all need to be kept as social protection
priorities. Improvement is ˜caused™ by both upward drivers and the removal
of downward ones.
Figure 7.5 helps to contextualize improving saw-tooth patterns within
longer trajectories. It shows an initial declining trajectory for a woman due
mainly to a failed marriage. Improvement began with a second marriage and
a son™s contribution to household income, offsetting ongoing expenses for a
parent™s medical care.

7.7.3. Step trajectories
Single-step declines tended to occur when a serious single or composite crisis
event led to deterioration in life condition, which was usually irreversible
and catastrophic (Table 7.5). Distinguishing between composite single-step
declines and multi-step declines is a little arbitrary, particularly if inaccuracies
in memory recall con¬‚ate events which were in fact separated by months or
even years. This commonly occurs when the events recalled were several years
or decades in the past.

1971 Father fell
from a tree and Sister married,
died 6 months contributed to dowry
Married 1980 to a
wealthier man 1986 Married

Son started
Husband began to working
threaten her

Poverty Dynamics in Bangladesh
1971 War Left husband in
1983 with
3-month-old son

Mother ill for 10 years
1990--2000, contributed Tk. 200
per month for medical costs

Figure 7.5. Improving saw-tooth: Rena
Peter Davis

Allauddin works as a day labourer (Figure 7.6). He is 44 years old. He lives
with his wife Fatima who is 35 and their 15-year-old son Ruhul. In 1993
Allauddin injured his eye while working with a rice-threshing machine. He
received treatment in a private clinic and the doctor eventually, and probably
unnecessarily, removed his left eye. Now it is very dif¬cult for him to work
because he suffers from bad headaches if he is in the sun. His wife does road
day labour for the Union Parishad and earns approximately Tk. 1,200 per
month when there is work. Otherwise she does domestic work locally where
she can earn Tk. 18 per day and three meals (Tk. 35 per day without meals).
Per month she earns approximately Tk. 260. Allauddin manages to earn Tk.
400 per month driving a cycle van gari.
The eye injury had the following consequences: over eight months Tk.
13,500 was spent on treatment. Some of this money came from relatives,
some from the boss at the local tobacco godown, and some from loans. A loss of
income resulted due to Allauddin™s inability to do day labour any more. In this
work he would have earned approximately Tk. 1,700 per month; however now
he can only work enough to earn Tk. 400 per month. They have also taken a
loan from one of the largest NGOs in Bangladesh. In order to secure the loan
the money had to be taken out in Fatima™s name and they had to state that
the money was to be used to buy a van gari. In fact the money was used to
pay for Allauddin™s eye treatment. Now they are unable to pay the loan back
and the NGO worker has threatened to remove ten pieces of tin from the roof
of their house if the loan is not paid back. They are socially stigmatized for a
number of reasons: because they are poor, have defaulted on a loan, Allauddin
is disabled, and they are seen as being uncultivated”Allauddin is a rickshaw
driver, speaks with a rough accent, moved to the area from a neighbouring
thana, and Fatima does day labour. This stigma works its way out in a number
of guises which have signi¬cantly adversely affected their lives: they have few
signi¬cant connections in the area in which they now live and they have had
trouble gaining road access from their own home to the road in order to get
to town.

Multi-step declines are often experienced as an inexorable decline in resources
and well-being resulting from a combination of reinforcing adverse events.
The double and triple whammy episodes of multiple or cumulative causation
fall into this category. Dowry payments for daughters in households with
larger numbers of girls than boys often produce this pattern. With the mar-
riage of each successive daughter a further downward step is taken by the
remaining household. Downward steps are sometimes re¬‚ected in successive
land sales. With each step a block of land is sold to pay for the costs associated
with the event and with each decline in land ownership the resilience of

War 1971
Eye injury leading
to loss of left eye
in 1993

Unable to work due
to pain associated
costs and
with loss of eye and
lack of
general weakness

Poverty Dynamics in Bangladesh
Figure 7.6. Declining single step: Allauddin”accident and disability
Peter Davis

the household to further events is reduced, undermining the likelihood of
The ˜structural violence™ (to use Farmer™s (2003) expression) facing many
poor people in rural Bangladesh becomes particularly apparent in the exami-
nation of this type of trajectory.
Amit is a 46-year-old Hindu man (Figure 7.7) in the village of Gopalpur
(the only village site with a substantial Hindu population in the study). While
Amit belongs to the Brahmin caste, and derives some social capital from this,
he is still disadvantaged socially within the predominantly Muslim society.
The independence war was a particularly dif¬cult time for Hindus in
Bangladesh. Most of the Hindus I interviewed in Kushtia district travelled
to India during the war because of fear of the Pakistan army and so-called
rajakars. Amit was 22 in 1971 and his father died from dysentery while they
were sheltering in India during the war. When they returned they found
that their house had been burnt. This was also a common theme among the
Hindus I interviewed. After the war Amit built up his life with a rice-trading
business and managed to add 9 kata of land to the 8 kata he inherited from
his father. He was also able to buy twelve cows. When he was 26 in 1975 he
married Jayanti who was 14 at the time and received a Tk. 1,200 dowry. Over
the next eleven years four children were born, three girls and a boy.
In 1977 things started going wrong. Amit had been given 16 bighas of
land by his in-laws as they had chosen to leave the country and move to
India. This land was located in the neighbouring district of Rajbari. However,
local Muslims in the area had occupied the land and had produced false
papers claiming ownership. Amit started a court case attempting to secure
ownership of his in-laws™ land. However, while he was on the train on the
way to attending the court hearing he was kidnapped by members of the
group he was opposing. The case had cost him Tk. 17,750 in various fees and
payments to the advocate (ukil), the assistant advocate, the hajira for papers,
the police, and various witnesses. Another group of Muslims then assisted him
but also asked him for the land in return for their help. In the circumstances
he realized his life was in danger and he accepted the loss of the land. This
episode was the ¬rst of a series of about ¬ve successive downward steps.
The second was in 1983 when his house was burgled. His in-laws had also
left him with many of their possessions in terms of crockery and household
items after they left for India. Local criminals were aware of this and their
house was burgled and they lost Tk. 21,000 worth of goods.
In 1990 their ¬rst daughter was married at the age of 13. The total cost of
the wedding was Tk. 27,775. This included a dowry of Tk. 14,000 in cash and
gold. The money was raised from loans, mortgaging land, gifts, and sale of
assets (a cow and eight pieces of tin).
In 1994 Amit became ill and needed to have a liver stone operation. He sold
4 kata of land (Tk. 6,000) and his matobar helped raise the remaining money

War 1971

House burglary

Operation 1994

Poverty Dynamics in Bangladesh
Court case and
Marriage of
daughter #1 Marriage of
1990 daughter #2

Figure 7.7. Declining multi-step: Amit
Peter Davis

in the community with a collection. Without this help he said he would have
In 1997 his second daughter was married at a total cost of Tk. 28,000. To
raise this money a cow was sold, two NGO loans were taken out, and Tk. 5,000
was received in gifts.
He has one unmarried daughter remaining. His son is now working as a
helper for a goldsmith but he currently doesn™t get paid while he is learning
the trade. Amit and his wife earn about Tk. 1,700 per month from agriculture
(3 kata), conducting puja, and making small items. As a Brahmin from the
prestigious thakur gosti he performs puja and is respected among the Hindu
community. However, outside this small and relatively powerless community
he does not have wide in¬‚uence or power resources to draw from.

7.8. Concluding Remarks

There are a number of advantages of using life history methodology in studies
of poverty dynamics. A much longer and more detailed interview can take
place without boring the respondent, mainly because of the fascination we all
have with our life histories when they are discussed and depicted. The sense
of objecti¬cation which can be experienced is reduced because respondents
can take greater control of the interview process and enjoy a greater freedom
of expression than in more structured interviews. This leads to better commu-
nication of contextual detail and a more nuanced historical story which can
accurately re¬‚ect people™s own understanding of their experiences.
A life history approach also allows a different type of analysis to occur.
Lifecycle patterns can be identi¬ed, patterns of complex causation emerge,
and individual episodes of crisis, coping, and opportunity can be seen within
the interrelated and wider context of national, community, and family trajec-
tories. It allows patterns to emerge of diminishing or accumulating resources
and socially structured behaviour. These patterns point the researcher towards
the underlying social structures that support the informal social protection
Life history interviews also provide a rich source of contextually situated
(spatially and temporally) qualitative data providing the opportunity to bridge
the qual“quant divide through the use of medium-n case-based research and
categorical data analysis techniques (see Ragin, 2000). Such an approach
allows a re¬‚exive interchange between conceptual theory and substantive data
analysis. The use of trajectory categories can contribute to poverty dynamics
research and enhance its impact on social policy”particularly social protec-
tion policies. Such categorizations can also be used as heuristic tools to inform
A broader range of research approaches, including combinations of qual-
itative and quantitative methods, would help inform a range of protective,

Poverty Dynamics in Bangladesh

promotional, and redistributional social policies, aimed at enhancing and
complementing informal forms of social protection, and taking into account
the changing (structured and structuring) pro¬les of risk and coping that poor
people face over their life cycles. This leads to a rethinking of social protection
based on an understanding of dynamic risk pro¬les and their underlying
social structures. Interventions need to complement existing informal and
formal means of coping and be sensitive to lifecycle stages and transitions
when people are most vulnerable.
The present research uncovers the harsh and long-term impact of a number
of categories of crisis in a fresh way. These include: illness, dowry, underem-
ployment and low income, court cases, business failure, crop loss, divorce,
household breakdown, violence, con¬‚ict, and crime. Our understanding of
the causes and consequences of these can only be improved if they are
viewed from within a life course perspective. The different vantage point
taken addresses many of the blind spots of other methods and provides a
complement to the existing suite of research approaches already informing
social policy in development contexts.


Becker, S., and Mahmud, S. (1984), A Validation Study of Backward and Forward Pregnancy
Histories in Matlab, Bangladesh, Voorburg: International Statistical Institute.
Bourdieu, P. (1986), ˜The Forms of Capital™, in J. G. Richardson (ed.), Handbook of
Theory and Research for the Sociology of Education, New York: Greenwood Press, 241“
(1990), The Logic of Practice, Cambridge: Polity.
and Nice, R. (1977), Outline of a Theory of Practice, Cambridge: Cambridge
University Press.
Carney, D. (ed.) (1998), Sustainable Rural Livelihoods: What Contribution Can We Make?,
London: Department for International Development.
Chambers, R. (1983), Rural Development: Putting the Last First, London: Longman.
Conway, M. A. (1990), Autobiographical Memory: An Introduction, Philadelphia: Open
University Press.
Davis, P. (1997), ˜Social Exclusion in Bangladesh: New Problematic or Eurocentric Impo-
sition?™, M.Sc. dissertation, School of Social Sciences, University of Bath.
(2001), ˜Rethinking the Welfare Regime Approach: The Case of Bangladesh™, Global
Social Policy, 1: 79“107.
(2005), ˜Power-Resources and Social Policy in Bangladesh: A Life-History Per-
spective™, Ph.D. thesis, Department of Economics and International Development,
University of Bath.
(2006), ˜Poverty in Time: Exploring Poverty Dynamics from Life History Interviews
in Bangladesh™, CPRC Working Paper No. 69, Chronic Poverty Research Centre.
Dewilde, C. (2003), ˜A Life-Course Perspective on Social Exclusion and Poverty™, British
Journal of Sociology, 54(1): 109“28.

Peter Davis

Farmer, P. (2003), Pathologies of Power: Health, Human Rights, and the New War on the Poor,
Berkeley and Los Angeles: University of California Press.
Hulme, D., and Shepherd, A. (2003), ˜Conceptualizing Chronic Poverty™, World Develop-
ment, 31: 403“23.
Indra, D. M., and Buchignani, N. (1997), ˜Rural Landlessness, Extended Entitlements
and Inter-household Relations in South Asia: A Bangladesh Case™, Journal of Peasant
Studies, 24: 25“64.
Korpi, W. (1980), ˜Social Policy and Distributional Con¬‚ict in the Capitalist Democra-
cies: A Preliminary Comparative Framework™, West European Politics, 3: 296“316.
(1983), The Democratic Class Struggle, London: Routledge.
(1985), ˜Power Resources Approach vs. Action and Con¬‚ict: On Causal and Inten-
tional Explanation in the Study of Power™, Sociological Theory, 3: 31“45.
and Palme, J. (1998), ˜The Paradox of Redistribution and Strategies of Equality:
Welfare State Institutions, Inequality, and Poverty in the Western Countries™, American
Sociological Review, 63: 661“87.
Lawson, C. W., McGregor, J. A., and Saltmarshe, D. K. (2000), ˜Surviving and Thriving:
Differentiation in a Peri-urban Community in Northern Albania™, World Development,
28: 1499“514.
Lewis, D., and McGregor, J. A. (1992), Change and Impoverishment in Albania: A Report for
Oxfam, Bath: Centre for Development Studies, University of Bath.
McGregor, J. A. (2004), Researching Well-being: Communicating between the Needs of Policy
Makers and the Needs of People, ESRC Research Group on Well-being in Developing
Countries, University of Bath.
Miller, R. L. (1999), Researching Life Stories and Family Histories, Thousand Oaks, Calif.:
Ragin, C. (2000), Fuzzy-Set Social Science, Chicago: University of Chicago Press.
Shulman, D., Weinberg, A., and Gran, B. (2003), ˜Complexity, Generality and
Qualitative Comparative Analysis™, Field Methods, 15: 323“40.
Room, G. (2000), ˜Trajectories of Social Exclusion: The Wider Context™, in D. Gordon
and P. Townsend (eds.), Breadline Europe: The Measurement of Poverty, Bristol: Policy
Press, 407“39.
Rubin, D. C., and Baddeley, A. D. (1989), ˜Telescoping is not Time Compression: A Model
of the Dating of Autobiographical Events™, Memory and Cognition, 17: 653“61.
Scott, J. C. (1976), The Moral Economy of the Peasant: Rebellion and Subsistence in Southeast
Asia, New Haven: Yale University Press.
Sen, A. (1998), ˜Social Exclusion: A Critical Assessment of the Concept and its Rele-
vance™, Asian Development Bank, 49.
Sen, B., and Hulme, D. (2004), ˜Chronic Poverty in Bangladesh: Tales of Ascent, Descent,
Marginality and Persistence™, Bangladesh Institute of Development Studies, Dhaka,
and Chronic Poverty Research Centre, Institute for Development Policy and Manage-
ment, University of Manchester.
Wood, G. D. (1999), ˜Adverse Incorporation: Another Dark Side of Social Capital™, paper
presented to DfID/World Bank meeting on WDR 2000, Eynsham Hall, February.
(2003), ˜Staying Secure, Staying Poor: The “Faustian Bargain” ™, World Development,
31: 455“71.

Subjective Assessments, Participatory
Methods, and Poverty Dynamics
The Stages of Progress Method
Anirudh Krishna

8.1. Studying Poverty in Dynamic Contexts: The Need
for New Methods

Despite decades of studying poverty, it is still not possible to say how many
people were born poor in any country and how many others have become
poor within their lifetimes. Available poverty knowledge also does not tell us
how many formerly poor people have escaped from poverty in any country.
Because it is not possible to identify those who have escaped from poverty, it
becomes hard to compare them with those who have not. Thus, it becomes
hard to fathom why only some (but not other) poor individuals succeeded in
moving out of poverty.
Poverty has to be studied in dynamic context; otherwise the reasons for
escape are not properly known. Different reasons operate in diverse regional
contexts, and guessing these reasons is hardly enough. For policy to have
an impact upon poverty, reasons for escape must be known; they must
form the targets of policy. However, relatively little has been done so far
to study these reasons as they operate within speci¬c regional and local
New methods of studying poverty have been developed over the past few
years that are helping to build better micro-foundations for poverty knowl-
edge. I present below one of these new methods, developed with colleagues
in ¬ve countries where poverty is a signi¬cant problem.

Anirudh Krishna

8.2. The Stages of Progress Method

The ¬rst Stages of Progress study was undertaken in the summer of 2002 in
Rajasthan, India. I went into ¬eld research in April of that year, looking to
understand why some but not other poor households had been able to escape
from poverty. What had they experienced that others had not? Did education
make the difference in most cases, or was is it an increase in productivity, or
better market returns, or new opportunities, or fewer children?
How many people, not previously poor, had fallen into poverty within the
same time period? What reasons had operated to bring about their downfalls?
I started looking within a group of rural communities in Rajasthan, the part
of the world that I know best in terms of sensibilities and aspirations. But I
had little idea about how I would go about this study. I knew that having a
panel of data for two time periods would help, but it would take unbearably
long to assemble: a gap of seven or eight years, at least, is required for having
a panel capable of supporting effective comparisons over time (Walker and
Ryan, 1990). I went ahead hoping to uncover through innovation on the
ground a quicker and equally productive methodology yielding reliable and
useful results. It took six months of ¬eld research, including four months
experiencing nothing but failure, before a potentially workable methodology
started taking shape.
These initial formulations, implemented in the ¬rst Rajasthan study
(Krishna, 2003, 2004), were successively improved upon in additional studies,
undertaken with research partners and community groups in other parts of
India and later in Kenya, Uganda, Peru, and North Carolina, USA. These
research partners”notably Patti Kristjanson in the Kenya studies and the Peru
study; Mahesh Kapila, Mahendra Porwal, Sharad Pathak, and Virpal Singh
in the three India studies; Dan Lumonya in Uganda; Judith Kuan in Peru;
Milissa Markiewicz in Uganda and in North Carolina, together with Leslie
Boney, Christina Gibson, and students at Duke University”have contributed
in different ways to the present state of development of this methodology. I
am equally indebted to the thousands of individuals whom we interviewed
individually and in community groups, who gave freely of their time.
In each separate region we conducted investigations in the local language,
thus different teams of investigators were selected and trained separately in
each region. Up to three teams operated in tandem after training in each
region. Typically, each team was composed of two facilitators and between
four and six investigators, equally male and female. These facilitators are
mostly college graduates, while the investigators have usually eight to ten
years of school education.
Because so much depends upon the quality of interviewing”and upon
combining carefully results derived separately from individual interviews and
community groups”training is a very important aspect of this methodology.

The Stages of Progress Method

Training for a period of ten days was built in at the start of this exercise
in each study site. Following three days of classroom discussions and sim-
ulation, the study teams would go out to conduct practical exercises with
the methodology, ¬rst in one set of communities, then, following feedback
and discussions, in a second set of communities. 1 I remained with the study
teams for additional periods of up to two weeks, working with them and
watching them as they worked, and developing in discussion with them
further re¬nements to these methods. As practised today, the methodology
has seven successive steps, followed in order each time a study is conducted
within any community.
Step 1. Assembling a representative community group. A male and a female com-
munity group were convened separately in each community. We took partic-
ular care to ensure that all members of the village community, particularly
poorer and lower-status ones, were represented at these meetings. In some
cases, where women let men do all the talking in mixed groups, a separate
meeting was convened for women of the community.
Step 2. We presented our objectives, introducing ourselves as researchers. It
needed to be made clear that there would be no bene¬ts or losses from speak-
ing freely and frankly to us. We were not implementing any development
project or ˜selecting bene¬ciaries™. Making this clear would help remove, we
hoped, any incentive someone had to misrepresent himself or some other
person as being poor.
Step 3. Describing ˜poverty™ collectively. Community groups in each village were
asked to delineate the locally applicable stages of progress that poor house-
holds typically follow on their pathways out of poverty. ˜What does a house-
hold in your community typically do™, we asked the assembled community
members, ˜when it climbs out gradually from a state of acute poverty?™ ˜Which
assets or capacities are acquired ¬rst? Which expenditures are the very ¬rst
ones to be made?™ ˜Food,™ was the answer invariably in every single village.
Which expenditures follow immediately after? ˜Some clothes,™ we were told
almost invariably. As more money ¬‚ows in incrementally, what does this
household do in the third stage, in the fourth stage, and so on? Lively dis-
cussions ensued among villagers in these community groups, but the answers
that they provided, particularly about the ¬rst few stages of progress, were
relatively invariant across all communities of each region studied.
After crossing which stage is a household no longer considered poor, we
asked the assembled community members, after drawing up the progression
of stages? The placement of this poverty cut-off and the nature of the ini-
tial stages (i.e. those below the poverty cut-off) differed somewhat across

A detailed manual developed for these trainings can be downloaded free of charge from
the website: <www.pubpol.duke.edu/krishna>.

Anirudh Krishna

Table 8.1. Stages of progress and the poverty cut-off

Gujarat, Rajasthan,
Stage Peru Western Uganda Andhra
India India
(Cajamarca Kenya (West and Pradesh,
and Puno) Central) India

1. Food Food Food Food Food Food
2. Clothing Clothing Clothing House Clothing Primary
repairs education
3. House House Primary Debt Primary Clothing
repairs repairs education payments education
4. Purchase Primary House repairs Clothing Debt payments Debt payments
small education
animals ---------- ----------- --------------

5. Primary Small House
education animals repair/roof
6. Purchase Renting a small
small plot tract of land
of land to farm as
---------- sharecropper

Note: The dotted line corresponds to the poverty cut-off in each case. Households advancing past this threshold
are no longer considered poor, either by themselves or by their neighbours.

communities belonging to the different regions studied. However, remarkably
similar understandings exist across diverse communities within each particu-
lar region. Across regions as well, there were considerable similarities in terms
of these understandings of poverty, as Table 8.1 shows.
It was community members and not researchers who de¬ned these stages
of progress. The similarity in stages is more remarkable for this reason.
Notice the progression in stages as households gradually make their way
out of poverty. In villages of Rajasthan, India, for example, the ¬rst four
stages are food, primary education for children, clothing, and debt repayment.
The poverty cut-off is drawn immediately after the fourth stage. In Andhra
Pradesh villages, similarly, the poverty cut-off is drawn immediately after the
fourth stage. Three of these ¬rst four stages are similar between Rajasthan
and Andhra Pradesh villages, but instead of primary education, reported in
Rajasthan villages, another stage, corresponding to minor house repairs, was
reported among the ¬rst four stages in villages of Andhra Pradesh. Across
countries, as well, there is considerable similarity across stages, but there also
signi¬cant differences, re¬‚ecting diverse lifestyles and aspirations.
Later stages of progress beyond the ¬rst few are not reported in Table 8.1,
and these stages included, for example, digging an irrigation well on one™s
own land, purchasing larger animals, particularly cattle, starting a small
retail business, constructing a new house, purchasing jewellery, acquiring
radios, fans, and tape recorders, and so on. These are, however, discre-
tionary expenses, and depending upon the taste of a household™s members,

The Stages of Progress Method

purchasing a radio or a tape recorder can precede or come after acquiring
ornaments. There was, consequently, more variation in the ordering of these
later stages in different villages.
The ¬rst few stages of progress are not so discretionary: they are both
physically and socially obligatory. Physical needs”for food, for clothing, for
protection from the elements”combine with considerations of social recog-
nition to constitute the de¬nition of poverty that is prevalent within these
communities. 2 It is a commonly known and widely agreed-upon understand-
ing of poverty, and this everyday understanding of poverty is much more real
for these villagers than any de¬nition that is proposed from the outside.
These locally constructed understandings of poverty constitute the criteria
within these communities for identifying who is poor. They also constitute
a threshold or an objective that de¬nes the goals and the strategies of poor
people: what people do in order to deal with poverty depends on what they
understand to be the de¬ning features of this state.
Villagers participating in community groups developed these criteria among
themselves, and they used these well-understood and commonly known cri-
teria to classify which households are poor at the present time and which
households were poor twenty-¬ve years ago. We chose to work in most regions
with a period of twenty-¬ve years because it corresponds roughly to one
generation in time. Households™ strategies are made in terms of generational
time horizons. In addition to asking about twenty-¬ve years ago, however, we
also enquired about an interim period of eight to ten years ago.
Step 4. Treating households of today as the unit of analysis, enquiring about house-
holds™ poverty status today and twenty-¬ve years ago. In this step a complete list
of all households in each village was prepared. Referring to the shared under-
standing of poverty developed in the previous step, the assembled community
groups identi¬ed each household™s status at the present time, for twenty-¬ve
years ago, and also for an intervening period, eight to ten years ago. 3

Social recognition matters as much as economic conditions in de¬ning the shared
understandings of poverty within these (and other) communities. For instance, in Gujarat,
the ¬fth stage, ¬xing leaky roofs, usually entails an expenditure that does not in most cases
exceed an amount larger than Rs. 400“500 (about $10), and it is a one-time expense, not
often incurred year after year. Even as it is a relatively modest expense, however, its critical
signi¬cance is in terms of status and recognition: people who are not poor in this region
do not have leaky roofs. Similarly, the sixth stage reported in Gujarat villages”renting small
tracts of agricultural land on a sharecropping basis”also has a distinct social signi¬cance that
is related to the prevalence of debt bondage in this region. Possessing to rent even a small tract
of land helps elevate the social status of a household above those who currently are or might
later become bonded debtors in the village. It does not necessarily imply any considerable
increase in net income.
In order to denote the earlier periods clearly, we made reference to some signi¬cant event
that is commonly known. For instance, in India, we referred to the national emergency of
1975“7, which is clearly remembered particularly by older villagers. In Kenya, similarly, we
referred to the year of President Kenyatta™s demise.

Anirudh Krishna

Households of today formed the unit of analysis for this exercise. Household
composition has been relatively stable in all communities studied: relatively
few households, less than 2 per cent in all, have either migrated in or migrated
out permanently. Individual members of households, particularly younger
males, have left these communities in search of work, but very few members
have left permanently, and fewer still have left permanently along with their
Step 5. Assigning households to particular categories. After ascertaining their
poverty status for the present time and for twenty-¬ve years ago (or ten years
ago), each household was assigned to one of four separate categories:
Category A. Poor then and poor now (Remained poor);
Category B. Poor then but not poor now (Escaped poverty);
Category C. Not poor then but poor now (Became poor); and
(Remained not poor). 4
Category D. Not poor then and not poor now
Step 6. Enquiring about reasons for escape and reasons for descent in respect of a
random sample of households. We took a random sample of about 30 per cent
of all households within each category, and we enquired in detail from the
assembled community groups about causes and contributory factors associ-
ated with each selected household™s trajectory over the past twenty-¬ve years.
Step 7. Following up by interviewing household members. Reasons indicated by the
community groups for each selected household were cross-checked separately
through individual interviews with members of the household concerned. At
least two members of each household were interviewed separately in their
homes. Multiple sources of information were thus consulted for ascertaining
reasons associated with the trajectories of each selected household.
It took a team of six to eight individuals three to four days on average
to complete these enquiries in one rural community (which has on average
about 150 households). These were not standard eight-hour days, but it was
an enjoyable learning experience for me and for my colleagues.

8.3. Brief Synthesis of Results

Signi¬cant proportions of households have escaped poverty over the last
twenty-¬ve years. During the same period, large numbers of households have
also fallen into poverty (Table 8.2).
Such simultaneous up-and-down movements have occurred in every one
of more than 200 communities that we studied. Achieving poverty reduction
A residual category, E, was also de¬ned, and households that could not be classi¬ed
otherwise because of lack of information were assigned to this category. Very few households,
less than half of 1% in all, were placed within Category E.

The Stages of Progress Method

Table 8.2. Trends in poverty dynamics over twenty-¬ve years (%)

Escaped Became Change in
poverty poor poverty

Rajasthan (35 villages) 11 8 3
Gujarat (36 villages) 9 6 3
Andhra (36 villages) 14 12 2
W. Kenya (20 villages) 18 19
Uganda (36 villages) 24 15 9
Peru (20 communities) 17 8 9
North Carolina (13 communities, 10 years) 23 12 11

goals will require taking action aimed at helping poor people escape poverty”
but it will also call for actions that stem the ¬‚ow of people into poverty.
Depending upon the region studied, between 6 per cent and 19 per cent
of all households have fallen into poverty over the period examined. These
households were not poor at the start of the study period, but by the end of
this period they had joined the ranks of the poor.
The newly impoverished constitute a signi¬cant subgroup within each
region. In the thirty-¬ve Rajasthan villages, for example, almost one-third of
those who are currently poor were not born poor; they have become poor
within their lifetimes.
Introducing a separate focus on falling into poverty is an important contri-
bution of studies that used the Stages of Progress method. Very large numbers
of households are falling into poverty everywhere. Yet, very few policies
are directed speci¬cally toward reducing these frequent (and often needless)
descents. As discussed below, a separate set of policies will be required speci¬-
cally to curb descents into poverty. Understanding the reasons for descent will
help give shape to appropriate policies.

8.3.1. Reasons for descents
Descents into poverty occur generally (though not always) in a gradual and
cumulative fashion, and not from one moment to the next. No single reason
is usually associated with falling into poverty; multiple linked factors propel
most descents. Tackling these major factors should lead to large reductions
in the incidence and probability of descent. Important local-level factors
of descent included, in descending order of frequency, health and health-
related expenses, death of a major income earner, disability, marriage and new
household-related expenses, funeral-related expenses, high interest private
debt (in India), land division, and land exhaustion (Table 8.3).
Healthcare is overwhelmingly the single most important reason for house-
holds descending into poverty in every region studied. Ill health and health-
related expenses were mentioned as important reasons associated with nearly

Anirudh Krishna

Table 8.3. Principal reasons for falling into poverty (% of descending households)

Peru: Puno
Western Andhra, Uganda: Central
Reasons Rajasthan, Gujarat,
and Western and Cajamarca
India Kenya India
n = 335 n = 202 n = 252
n = 172
n = 364 n = 189

Poor health and 60 88 74 74 71 67
Marriage/dowry/new 31 68 69 18 29
Funeral-related 34 49 64 28 15 11
High-interest private 72 52 60
Drought/irrigation 18 44 19 11
failure/crop disease
Unproductive 38 8

60 per cent of all descents recorded in villages of Rajasthan, India, with
74 per cent of all descents examined in Andhra Pradesh, India, and with as
many as 88 per cent of all descents studied in villages of Gujarat, India.
In communities of Kenya, Uganda, and Peru that we studied, respectively
74 per cent, 71 per cent, and 67 per cent of all descents were associated with
ill health and health-related expenses.
Not only does ill health reduce the earning capacity of a household™s
members; in the absence of affordable and easy-to-access healthcare facilities,
it also adds considerably to the household™s burden of expenditure, thereby
striking a double blow, which quite often results in tragedy. The human body
is often poor people™s main productive asset, an indivisible and, in most cases,
an uninsured asset, which unlike most other assets can ¬‚ip or slide from
being an asset to being a liability. 5 The resulting dependence of survivors,
including orphans, upon other households contributed further to descent in
many cases.
Funeral expenses, especially expensive death feasts, were associated with a
high proportion of descending households in communities studied in Kenya
(64 per cent), Rajasthan (34 per cent), Gujarat (49 per cent), Andhra Pradesh
(28 per cent), and Peru (11 per cent). Marriage-related expenses were very
important in all three states studied in India. They were also cited as an impor-
tant factor in communities of Peru, affecting younger couples in particular.
Over a twenty-¬ve-year period ending in 2004, marriage and new household-
related expenses were associated with 29 per cent of all cases of households
falling into poverty in these forty Peruvian communities.
I thank Robert Chambers for suggesting this formulation.

The Stages of Progress Method

Land-related factors, including crop disease, land exhaustion, drought, and
irrigation failure, were also associated with a signi¬cant number of descents,
particularly in some regions. In communities of western and central Uganda
this set of factors was associated with 39 per cent of all observed descents
and in communities of western Kenya with 38 per cent of all descents. Other
reasons for descent included the loss of a job resulting from retrenchment,
sacking, or retirement.
Drunkenness and laziness, sometimes thought to be important causes of
poverty among the poor, were found to be relatively insigni¬cant reasons. In
all the communities investigated, these factors were associated with no more
than 5 per cent of all descents. 6
High-interest private debt is highly prevalent as a factor contributing
to descents in the three Indian states. Villagers deal with high healthcare
expenses and with expenses on marriages and death feasts by taking out high-
interest loans from private moneylenders, paying rates of interest as high
as 10 per cent per month. The high burden of debt that results helps push
households deeper into poverty.
Drought and irrigation failure constituted another important reason for
descent. However, the effect of this factor, as of many other factors reviewed
above, varies considerably across different parts of a region and country.
These reasons for descent are different everywhere from the reasons that
have helped take poor households out of poverty. This essential asymmetry
between escaping poverty and falling into poverty will require simultaneously
mounting two parallel sets of poverty policies, as discussed below.

8.3.2. Reasons for escapes
Income diversi¬cation has been the most important pathway out of poverty in
all areas studied (Table 8.4). Poor rural households diversi¬ed their livelihood
and income sources through two broad means: on-farm”through pursuing
new crop- and/or livestock-related strategies; and off-farm”through local
petty trade, small businesses, and, most importantly, through casual or tem-
porary employment within the informal sector in a city. Diversi¬cation of
income sources was related to 70 per cent of all escapes observed in commu-
nities of Rajasthan, India, 78 per cent of those observed in communities of
western Kenya, 69 per cent in Peru, and 54 per cent in Uganda.
In general, growth of private sector employment has not been the principal
or even a very prominent reason for escaping poverty. Even in Gujarat, India,
There might have been a few more households that hid this information successfully
from us, but I doubt that there are very many households of this type. In community
groups especially, villagers were hardly shy in talking about another person™s slothfulness or
penchant for drink, and gently probed, household members also came forth to speak frankly
about these aspects.

Anirudh Krishna

Table 8.4. Principal reasons for escaping poverty (% of escaping households)

Peru: Puno
Western Andhra, Uganda: Central
Reasons Rajasthan, Gujarat,
India and Western and Cajamarca
India India
n = 285 n = 172 n = 348 n = 398 n = 324
n = 499

Diversi¬cation of 70 35 78 51 54 69
Private sector 7 32 61 7 9 19
Public sector 11 39 13 11 6 10
Government 8 6 7 4
Irrigation 27 29 25

where economic growth rates have averaged 9 per cent over many years, only
about one-third of those who escaped from poverty could do so on account
of acquiring a regular job in the private sector (Krishna et al., 2005).
Growth in agriculture”related particularly to irrigation and land
improvement”has been more important as a reason for escape from poverty. 7
Improvements in productivity as well as diversi¬cation into cash crops
were quite important in both regions of Uganda, where ¬rst coffee, then
vanilla, were grown. Cash crop diversi¬cation was also important in west-
ern Kenya and in the Cajamarca region of Peru. Over one-quarter of all
escaping households in each of the three Indian states bene¬ted from large-
scale irrigation schemes or from small-scale irrigation activities on their
While most children are going to school in these communities, education
has hardly always amounted to an escape out of poverty. Information and
connections matter in addition to education, and the lucky few who have
found a job or business opportunities in the city have been assisted”usually
with information and sometimes also with a contact or two”by an uncle or
cousin established for many years in a city-based occupation.
It is disheartening that government as well as non-governmental assistance
and programmes are not contributing substantially to households™ move-
ments out of poverty. Perhaps these programmes are not well spread out over
all communities; perhaps lack of reach is made worse by lack of knowledge
about reasons to target.
Different trends and different causes operate in different regions and local-
ities, and pinpointed rather than blanket solutions need to be devised and

Ravallion and Datt (1996) show that 84.5% of the recent signi¬cant poverty reduction in
India was due to growth in the agricultural sector. See Timmer (1997) and Mellor (1999) for
cross-country comparisons yielding a similar conclusion.

The Stages of Progress Method

implemented. Disaggregated enquiries are important for this reason. Without
knowing what reasons are most prominent for escape and for descent in a
particular locality, appropriate interventions cannot be identi¬ed.
The Stages of Progress methodology helps critically with such locality-
speci¬c identi¬cation of reasons for escape and descent. Building on a rich
history of participatory approaches (including Salmen, 1987; Chambers, 1997;
Narayan et al., 2000), this methodology is rigorous but relatively simple to

8.4. Subjective Assessments in Relation to Panel Studies

Panel datasets have been used traditionally to examine households™ and
individuals™ movements into and out of poverty. Panel data on households™
consumption levels at two different points in time are very helpful for
comparing changes in monetary poverty among households. Because they
utilize a standardized de¬nition of poverty, it might be thought that they
deliver more precise numbers for escape and descent. But these numbers
are precise only in the terms of their de¬nition. Other de¬nitions are more
valid and precise for other observers; the poor themselves do not use dollar-
Depending upon the questions that some study is intended to address,
different methods are more appropriate and different de¬nitions more useful
to follow. 8 Stages of Progress is a preferable method to use in situations
where data for a prior period are simply not available, or as is often the case,
particularly in developing countries, where available data are hard to access,
unclear, or not rigorously obtained. In such situations, a panel dataset might
take unbearably long to assemble, and other methodologies, including Stages
of Progress, may be preferred.
It is also better to use Stages (or some other method like it) in situations
where it is important to identify household-level reasons. With a notable
few exceptions, including Sen (2003), panel data studies have not identi¬ed
reasons for escape and reasons for descent at the household level. In doing so,
they have missed out upon households™ strategies for dealing with poverty,
thereby de-linking the understanding of poverty dynamics from individuals™
own efforts.
In order to understand households™ strategies it helps to accept the de¬ni-
tions by which these strategies get de¬ned. Adopting a place-bound and local
understanding of poverty is better for this purpose.
Different understandings of poverty coexist, and whose reality we adopt in¬‚uences the
results that we obtain.People identi¬ed as being poor according to standardized monetary
measures do not always consider themselves poor in their own terms (Jodha, 1988; Chambers,
1997; Franco and Saith, 2003; Laderchi, Saith, and Stewart, 2003; McGee, 2004).

Anirudh Krishna

Third, because Stages is easy to apply, enjoyable in practice, 9 and its logic
is intuitively clear, it can help community groups assist with the analyses
that are undertaken. Applying this methodology and uncovering the reasons
for escape and descent helps to provide the rationale for selecting particular
investments over others.
Combining different methods will be important to understand different
facets of poverty. New and reliable methods must be developed that can
uncover more facets of poverty knowledge (O™Connor, 2001).

8.5. Comparability and Reliability

How reliable are the data from Stages of Progress? Recall can be quite imperfect
for an earlier period, and oral evidence may be faulty, incomplete, or deliber-
ately skewed. In order to deal with these possible sources of weakness, several
precautions have been built in, many as a result of experience.
To begin with, the methodology retraces large steps that are better remem-
bered compared to ¬ner distinctions. Each movement upward along the
Stages of Progress represents a signi¬cant improvement in material and social
status. People remember, for instance, whether their household possessed a
motorcycle or a radio set at the time when Kenyatta passed away; they can
recall clearly whether they lived in a mud or a brick house while growing up,
and whether they could afford to send their children to school. By seeking
recall data in terms of these clear, conspicuous, and sizeable referents, the
Stages of Progress method adds reliability to recall. Members of particular
households remember quite well where they were located along this clearly
understood hierarchy of Stages, and these recollections are veri¬ed by others
who have lived together with them for long periods of time.
One of the risks associated with subjective enquiries”which arises when
people think back to some mythical golden age: ˜everything was better in
the past™”gets limited because communities think in terms of distinct Stages
(and not in terms of better or worse). These stages are visible to all in the
community, so community members are able to say which households are at
each stage and where they were in previous time periods.

Wilson Nindo, who has implemented Stages of Progress in more than 40 Kenyan com-
munities, had the following to say when I interviewed him in Nairobi (21 May 2006): ˜I like
Stages of Progress because this study is never boring. Communities™ enthusiasm [for it] keeps
your own enthusiasm going . . . Communities have given you two“three things [names for
Stages]. You ask them “OK, which among them comes ¬rst?” Doing the stages is the most
challenging bit of this methodology. If you are not careful, you™ll just be making a list, a
wish list, not the Stages of Progress. You ask and probe a bit more. If you don™t make them
compare, they will just be adding items, not always in a [sequential] ¬‚ow. If you don™t take
care, you get confusion. Working with them is important. You work with them to come up
with a sequence acceptable to them [not to you].™

The Stages of Progress Method

Table 8.5. Stages of Progress and asset ownership
(thirty-six communities in Uganda)

Household™s stage at Average number of
the present time household assets (out of 10)

1 2.46
2 3.08
3 3.58
4 4.08
5 4.94
6 5.24
7 5.55
8 5.71
9 6.42
10 6.72
11 7.31
12 8.01

Triangulation of all data collected helps to further verify recall. Information
about each household is obtained separately at both the community and the
household level. Discrepancies, when found, bring forth repeat interviews;
community groups and the household verify each other™s account.
Corroboration with more ˜objective™ evidence was found by comparing
stages with asset holdings for households. Table 8.5 presents evidence in this
regard from the study conducted in thirty-six villages of Uganda (Krishna
et al., 2006a). Households were asked about ownership in respect of ten
different types of assets, including animals, radios, household furniture, and
so on. Table 8.5 shows that a monotonically increasing relationship exists
between a household™s ownership of assets and the stage at which they were
Communities™ gradations and rankings point in the same directions as the
grading schemes that ˜we™ (i.e. the experts and outsiders, in Robert Chambers™s
sense of the word) would prefer to employ, corresponding to ˜our™ preferred
de¬nitions of poverty. We found in other studies that some other visible
characteristics of material status”e.g. housing type and cattle ownership”
also align neatly (though hardly perfectly) with a household™s position on
the Stages of Progress. How well any household is doing in terms of material
achievement at the present time is thus re¬‚ected quite well by the stage
recorded for it in the current period.
But what about stage as recorded for a previous period? Does it also accord
quite well with what it actually was at that time?
In order to convert from this hypothetical question to one that could
actually be answered using the available evidence, I conducted a study in
2004 in the same group of sixty-one villages in Rajasthan, India, where I had
undertaken a previous study seven years before. I found households™ Stages

Anirudh Krishna

Table 8.6. Stages (as recalled) vs. assets actually possessed seven years ago (sixty-one
communities of Rajasthan, India)

Stage in 1997 (as recalled Assets actually possessed in 1997
in 2004)
Land (bighas) Large animals Small animals Kaccha (mud) house

Very poor (Stage 1“3) 3.6 1.8 2.8 86%
Poor (Stage 4“5) 5.5 2.5 3.7 77%
Middle (Stage 6“8) 8.1 3.1 5.1 51%
Better-off (Stage 9+) 10.6 4.3 3.1 22%

of Progress for 1997 (as recalled in the community meetings of 2004) to be
closely correlated with the number of assets possessed by them seven years
ago (as recorded in the survey conducted in 1997). Table 8.6 presents these
Objective data from a more distant past are not readily or abundantly
available (if they were, we would never have needed to develop any such
methodology). I knew of only instance”others may know more”of written
records from twenty-¬ve years ago, records which stated, in this instance,
the amount of land ownership of every household in some of the villages
that we had studied. By checking these land records for an earlier period it is
theoretically possible to map stages (as recalled) against landholdings actually
possessed twenty-¬ve years ago. In practice, this task is both complicated and
arduous. It can also be considerably expensive, especially if it is delegated to
someone else who is quali¬ed. Backtracking land ownership records requires
manually locating, collating, and compiling diverse handwritten registers,
which are most often not available at a single physical location. It also
requires matching present-day households with the individuals whose names
were recorded in the land registers of twenty-¬ve years ago (and in cases
where the household concerned has experienced subdivision, it also requires
calculating the share in the prior land holding of the household recorded
at the present time). Finding a match with land records for villages studied
in Rajasthan was simply not possible given the resources available. Instead,
I selected a random sample of feasible size, picking twenty-¬ve households
at random from among all those who have fallen into poverty in ¬ve vil-
lages, also randomly selected, in two districts of Rajasthan. With generous
assistance provided by the administration of Udaipur district, land owner-
ship for these twenty-¬ve households was tracked backward over twenty-¬ve
This examination of land records showed that of these twenty-¬ve house-
holds, all of which suffered descents into poverty, twenty-two households
(88 per cent) had simultaneously lost all or part of the land they had owned

The Stages of Progress Method

twenty-¬ve years ago. About half of these households had lost all of their
lands. The rest had to part with signi¬cant chunks of their holdings. 10
Observing this close match between land records and Stages data helped
justify the effort required to obtain this information. But I was not too
surprised upon learning of these facts. The villagers whom I had met and
interviewed had shared generously of their knowledge, and what I had learned
from these discussions, I felt instinctively, was true.
It helped to have local area residents working as the interviewers for these
studies. I have learned at my cost not to speak much at these community
meetings. Many questions cannot reasonably be asked by outsiders, but speak-
ing about misfortunes with ones who know and can empathize is easier and,
as I observed, also cathartic in some cases.
The interface between researcher and respondent is critical for this method
to work well. This is why intensive training is built in at the start of every such

8.6. Limitations and Planned Developments

Some limitations will need to be addressed as this methodology is extended
further. Some other limitations will not be easily overcome. I outline below
what I currently know about these limitations. I welcome comments and
suggestions about dealing with these limitations better and also about other
limitations that I may have failed to spot.
First, the methodology needs to deal better with intra-household differ-


. 7
( 13)