Jump to ContentJump to Main Navigation
One Illness AwayWhy People Become Poor and How They Escape Poverty$

Anirudh Krishna

Print publication date: 2010

Print ISBN-13: 9780199584512

Published to Oxford Scholarship Online: April 2015

DOI: 10.1093/acprof:osobl/9780199584512.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2017. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 26 February 2017

(p.164) Appendix Measuring Poverty

(p.164) Appendix Measuring Poverty

Testing Stages-of-Progress

Source:
One Illness Away
Publisher:
Oxford University Press

Poverty has a complex, multi-faceted reality, and it is not feasible to examine all of these facets within any single study. Depending upon its specific purpose, each study focuses upon some particular facet or facets and neglects to attend to other important facets. The choice of methodology and measurement tool depends upon this prior selection of study objectives: What part of poverty one selects to study helps determine what should be measured and how.

No single measure of poverty can be universally useful nor indeed is any such measure universally utilized. Diverse measures of poverty co-exist, serving different objectives. The best known of these measures, the World Bank’s dollar-a-day index, helps make comparisons of consumption poverty among countries. However, this possibility is gained at the cost of neglecting other important facets of poverty. The United Nation’s Human Development Index (HDI) takes account of a broader concern by looking, in addition to income, at aspects related to education and health. By taking account as well of these additional facets, HDI scores can better reflect the overall living conditions of poorer people. Since these calculations do not address equity and distributional concerns, a third set of measures has been developed to examine these particular facets. A fourth set of measures has been developed to investigate gender-based differences in well-being. Social and political correlates of poverty are examined using yet another set of measures.

This ever-expanding set of poverty measures should come as no surprise. Poverty is not like blood pressure or height, for which some single definition or measure can be established by convention.

(p.165) Methodological disputes are common. The apparent precision and simplicity of dollar-a-day sit atop a number of procedural disagreements. Which items of consumption should be included within these calculations, and which ones are better left out? Should one include only what each individual purchases directly for herself? Or should public goods—such as higher-quality public education, better-maintained public gardens, cleaner air, and fairer governance—also be counted within these calculations of individual well-being? Should expenditures on health care be included, or should they be excluded, being seen as indicators of pathology rather than components of well-being? Over how many days or weeks should people’s consumption be averaged? Which seasons of the year should be considered? How should differences in the cost of living be harmonized across countries? Which particular basket of goods and services—those typically consumed in richer countries or in poorer ones—should be considered while making cross-country comparisons?1

No universally accepted or clear-cut answers are available for these questions. ‘Poverty lines unavoidably retain an element of arbitrariness and inevitably embody some implicit or explicit normative judgments.’2 The kinds of thumb rules that are adopted critically affect the results that are obtained. In India, for example, changing the period of recall—considering people’s consumption expenditures over one week instead of one month—resulted in sharply lowering the calculation of the national poverty rate.3 Because they are ‘very sensitive to both survey design and post-survey analysis,’4 seemingly precise quantitative poverty rates can be imperfect and capricious. Analysts are aware that estimates of ‘the magnitude of the increase or decrease in the extent of world poverty since 1990 are crucially dependent on the assumptions made.’5

International comparisons are particularly error-prone. ‘Finding a poverty line that is representative and comparable across countries and regions is an impossible task,’ according to T. N. Srinivasan, a leading expert.6 Understandings of poverty are inherently related to specific societal contexts. ‘Poverty is not a certain small amount of goods, nor is it just a relation between means and ends; above all, it is a relation between people. Poverty is a social status. As such, it is the invention of civilizations.’7 Thus, people’s understandings of poverty are related to the context in which they live. A person who considers herself relatively poor in one context can suddenly discover that in another context she is relatively rich.

No particular measure or method helps overcome all of these different pitfalls. Depending upon the specific context that is studied and the particular facets of poverty that are explored, particular methods and measures will be useful. Instead of pushing (p.166) ahead in every case with the same methodology, one should prefer to use the method and measure that help make a particular set of questions tractable.

The Stages-of-Progress methodology is a useful tool to employ in situations when one is interested in examining changes in households’ circumstances over time and to ascertain context-specific reasons for escape and descent. It is especially good for examining these facets of poverty. It is less helpful for looking at some other facets.

Two Judgment Calls

Two methodological choices that were made while developing Stages-of-Progress are important to discuss in some detail. First, Stages utilizes a place-bound or local understanding of poverty that is different from the usual consumption- or income-based measures. Second, it employs a retrospective design and considers present-day households as its units of analysis. The rationale for each of these choices is explained in more detail below, starting with the selection of the scale of measurement.

A common procedure for measuring household poverty, including the well-known dollar-a-day calculations, relies upon calculating households’ consumption expenses. A list of items commonly consumed in the area is drawn up, and household members are asked to recall how much of each item they have consumed over the previous 15 days or one month. Multiplying the quantities reported by a particular household with the set of prices prevailing in the particular region results in generating an estimate for this household’s consumption expenditure. Households who fall below a certain threshold of consumption expenditure (e.g., US$1/day) are classified as poor. Others, who have higher consumption expenditures, are classified as non-poor.

This procedure seems relatively straightforward, and I could have adopted such a technique for assessing households’ poverty status at the time of inquiry. But I was also interested in assessing their poverty status at some point in the past, and it would have been foolish to ask people to recall their consumption patterns—the precise quantities of cabbage and wheat flour and cooking oil that they had purchased—at some previous point of time, ten years (or, especially, 25 years) ago. Assets, being lumpy and limited in number, are easier to recollect with more precision. Instead of working with a consumption-based measure of poverty, I selected to work with an assets- and capacities-based measure.

My preference for working with such a measure gained strength after I examined the results of participatory poverty assessments and ethnographic studies. These initiatives have worked more closely with people’s own understandings and assessments of poverty which, far from being standardized or uniform, vary considerably across (p.167) different cultural contexts.8 Researchers conducting such studies have found that local understandings of poverty are most often expressed in terms of discrete assets or capabilities, which are ‘intrinsically important for people while low income is only instrumentally significant.’9 The scales that I developed initially in Rajasthan were also phrased by the people in these villages in terms of assets and capabilities. Thus, an assets- and capabilities-based scale is advantageous in that it corresponds better with people’s own understandings of poverty.

Further, working with assets and capabilities provides a more stable and reliable index for assessing longer term changes in material well-being. Incomes and consumption expenditures tend to fluctuate substantially from month to month and year to year, especially for poorer people, hardly any among whom get fixed monthly salaries.10 Seasonality plays a large part in the lives of the rural poor,11 resulting in ‘damaging fluctuations’ that limit the extent to which consumption (averaged over any particular fortnight or month) can serve as a stable or reliable measure of well-being.12

Observing these aspects, scholars have drawn an important distinction between structural and stochastic poverty.13 The structural poor are those who lack assets. On average, their income and consumption levels fall below the poverty line. The stochastic poor, on the other hand, have sufficient asset endowments, so on average their incomes are above the poverty line. In some particular month or year, their incomes can dip, occasionally falling below the poverty line. However, measuring such temporarily lowered incomes does not give an accurate indication of this household’s usual (or structural) conditions. Considering assets and capabilities is more reliable.14

For all these reasons, including both practical and theoretical ones, I selected to forego the usual consumption-based measure. Instead, I utilized measures, developed with the participation of the people concerned, which draw upon context-specific assets and capabilities.

A second methodological choice resulted in adopting a retrospective design. Households existing at the time of inquiry constituted the units of analysis for Stages-of-Progress. Pathways traveled by present-day households were traced backward in time. Events, processes, and household characteristics associated with different households’ experiences were identified through extensive interviews. Comparing the trajectories of households with different experiences helped to recognize commonly occurring reasons for escape and descent. Policies and programs of assistance appropriate to particular contexts can be designed on the basis of this knowledge.

The robustness of these results can be compromised in situations where households migrate in or migrate out on a large scale. In the regions studied, relatively few (p.168) households had, in fact, migrated into or out of their home communities. Individual members of households, particularly younger males, had left rural communities in significant numbers, but relatively few individuals had left permanently, leaving no trace behind, and fewer still had taken along their entire household.

When considered in terms of individuals, the proportions involved in migration can be quite large, but when calculated in relation to entire households these proportions are much smaller. The Dutch scholar, Jan Breman, who has followed developments over a long period of time in the Indian state of Gujarat, observed that he had ‘seldom come across cases of households who left in their entirety to seek a new life elsewhere.’15

Such trends are not peculiar to the regions selected for this study. Similar results—showing a clear difference between individuals’ and households’ migration patterns—have been reported as well for other communities and other countries. Household surveys undertaken in 13 developing countries found that most people do not leave their home villages for very long, and those who migrate do not usually take along their dependents.16 In 1979, scholars studied a group of 240 households in one village of the Indian state of Tamil Nadu. Returning 25 years later to the same village, they found ‘233 households out of the originally selected 240. Of the 233 households traced, some still remain under the same head. Others remain in the village but have a new head. Yet others have migrated but left enough traces in the village to enable us to find out to where they went.’17

Clearly, one can never expect to run a fully controlled experiment in which no one moves out, or dies, or is born during the period of study. Some level of flux is inevitable, and any longitudinal study, particularly one that considers longer periods of time, will have to find suitable ways of dealing with this risk. Considering households, rather than individuals, as the units of analysis helps render migration a less important source of risk. I will discuss some other aspects of this issue later in this chapter. Meanwhile, let me detail how some other potential biases were addressed.

Triangulation and Verification

Initially, while developing a new methodology, it is natural to experience some anxiety. Several adjustments and revisions were incorporated before I became easier in my mind that Stages-of-Progress worked reasonably well.

It helped to raise my faith that common stages of progress—at least, common initial stages—were consistently reported by different community groups. The small differences that arose were related to higher level stages of progress, which are reached long (p.169) after poverty is overcome. People’s needs are more varied and expenses are discretionary at these higher levels, because culture and human biology are less prescriptive.

I continued to be concerned about the reliability of recall. We were working with fairly long periods of time. Oral evidence can be faulty, incomplete, or deliberately skewed. Further, fear of stigma or elite domination of community groups could skew these results. A number of precautions were incorporated that helped deal reasonably effectively with each of these risks. In addition, I looked for sources of verification, seeking out objective evidence recorded in the past. These precautions and verification procedures are recounted briefly below.18

Intentionally, the Stages-of-Progress methodology has been designed to retrace large steps that are better remembered rather than finer distinctions that are more easily forgotten. Each movement upward along the Stages-of-Progress represents a significant improvement in material and social status. People in Kenya remembered quite easily, for instance, whether they lived in a mud or a brick house while growing up, whether their parents were in a position to send children to school, and so on.

Considering one’s own case will help make clear what I mean about larger steps resulting in easier recall. Casting my mind back to 25 years ago, I can clearly locate my household’s position on the Rajasthan communities’ stages of progress. I remember, as I suppose each of us can, where my household was situated in terms of these clear referents. Did we: have the capacity to regularly eat enough food (yes); clothing (yes); could send children to school (yes); own a television set (yes); own the house in which we lived (no).

By seeking recall data in terms of clear, conspicuous, and sizeable referents, the Stages-of-Progress method helps add reliability to recall. The downside is that one misses out on smaller changes. Such smaller changes are better measured when consumption- or income-based measures are used. Since Stages-of-Progress works with a discrete number of levels of well-being, finer improvements in households’ well-being cannot be captured using this methodology, although larger increments are reliably recorded.

Members of particular households remembered quite well where they were located along this hierarchy of stages, and these recollections were verified by others of the same community. It was rarely hard or controversial to locate any household’s position, either at the present time or in the past.

Still, people’s judgments, even though they were spontaneously given and corroborated by others, may be colored by the presence of some collective myth, for example, people might feel that ‘everything was better in the past.’ The risk of glorifying the past was limited by the design of these exercises, because communities were not asked to think in terms of better or worse. Distinct stages of progress were put up publicly on (p.170) large charts, and community members were asked to locate where along these well-specified stages each particular household was located in the previous time period. Study team members could verify hard facts: ‘Did they have a brick house at that time? Did they own cattle or only small animals?’ Considerations of ‘better or worse’ were entirely avoided. Indeed, the terms ‘poverty’ and ‘poor and rich’ were hardly ever used.19

Triangulating all data by consulting multiple independent sources helped to guard against the risk of partial or biased data. Information about each household was obtained separately at both the community and the household levels.

Corroborating what we found with other available forms of evidence helped further verify the results that we obtained. As part of these verification exercises, I compared the stages recorded for particular households with the physical assets that these households possessed. Table A1 presents evidence in this regard from the study conducted in communities of Central and Western Uganda. Interviewed households were asked to provide information about ten different types of assets, including animals, radios, household furniture, and so on.

Not surprisingly—since Stages is an assets- and capabilities-based measure—a consistent relationship exists between a household’s asset holdings and its stage reported by the community group. Other visible characteristics of a household’s economic status—for example, its livestock ownership and the kind of house in which it lived—also align closely with its reported stage of progress. How well any household is doing

Table A1 Stages-of-Progress and Asset Ownership (36 Communities in Uganda)

Households Stage at the Present Time

Average Number of 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

(p.171) in terms of material achievement at the present time is thus quite well reflected by the stage recorded for the current period.

But what about the stage recorded for a previous period? Does it also align equally well with asset holdings in existence at that time?

In order to convert from this hypothetical question to one that could actually be addressed I conducted a study in 2004 within a group of 61 villages in Rajasthan, India where I had conducted a previous study seven years ago (that is, in 1997). This previous study was intended to examine social capital, economic development, and some other outcomes.20 A random sample of individuals in each of these villages was interviewed, and information related to a number of different items, including asset ownership, was collected in 1997. The second study, undertaken in 2004, implemented Stages-of-Progress. A seven-year recall period was selected for this study. Each household’s stage for the present time (that is, for 2004) and for seven years ago (that is, for 1997) was recorded. These recalled stages were compared with the results of the 1997 survey. Table A2 presents what we found.

There is a close match between the recall data and the data recorded seven years previously. Households’ stages of progress for 1997 (as recalled in the community meetings of 2004) are closely correlated with asset holdings recorded by the 1997 survey. Whether considered in terms of agricultural land, large or small animals, or home construction quality, households who were recalled to have a lower stage of progress actually possessed fewer assets at that time.

Objective data from a more distant past are not readily available—and if they were, there would be no need for a recall-based methodology! I could think of only one

Table A2 Stages (as Recalled) v. Assets Possessed Seven Years Ago (61 Communities of Rajasthan, India)

Stage for 1997 (as Recalled in 2004)

Assets Possessed in 1997

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%

Very Poor (Stage 9+)

10.6

4.3

3.1

22%

(p.172) possibility of locating reliable written records for 25 years in the past. These are the official land registers that have existed unbroken for several decades in Rajasthan and in many other parts of India. Each register relates to a particular village, and it records the amount of land owned by each household. Land registers are handwritten in black or blue ink, with changes (‘mutations’) being inserted in red ink. Because paper tends to wear down with constant use, a new register is written up every four years, with the old registers being preserved, as carefully as possible, in government records rooms that have their own specialized staffs and procedures.21

By checking the official land registers for an earlier period it is theoretically possible to map stages (as recalled) against landholdings actually possessed 25 years ago. In practice, however, this task is both complicated and arduous. It can also be extremely expensive. Backtracking land ownership requires manually collating diverse handwritten registers, not all of which are available at a single location. It also requires matching present-day households with the households or individuals whose names were recorded in land registers of the past. Because households and their landholdings tend to get sub-divided over time, it becomes necessary as well to calculate the present-day household’s notional share in the original household’s landholding.

Finding a match with the historical land record for all 61 villages studied in Rajasthan was simply not possible given the resources available. Instead I selected a random sample of feasible size, picking 25 households at random from among all those who have fallen into poverty in five villages, which were also randomly selected from two districts of Rajasthan. I also selected 25 other households at random belonging to the other three categories (escaped poverty, remained poor, and remained not poor). With generous assistance provided by concerned government departments, landownership for all of these households was tracked backward over 25 years. Table A3 provides these results.

Notice that of these 25 households, all of which suffered descents into poverty, 22 households (88 percent) lost all or part of the land that they had owned 25 years ago. About half of these households lost all the land they had owned. The rest had to part with significant chunks of their landholdings. Hardly any other household in these villages has lost anywhere close to the same proportion of its prior-period land.

Observing this close match between land records and Stages data helped justify the enormous effort that went into obtaining this information. But I was not surprised to learn of these facts. The community members whom I met had given generously of their knowledge, and I had not detected any widely shared impulse to dissemble, distort, or fabricate.

(p.173)

Table A3 Impoverishment and Reduced Land Holdings

Village

Household Head

Stage 25 Years

Stage in 2004

Land Owned in 1980 (Hectares)

Land Owned in 2004 (Hectares)

Change in Land Holding (Hectares)

Aamliya

Detali Beeram Das

4

1

0.65

0.00

-0.65

Aamliya

Hakri Vala

4

1

2.33

0.00

-2.33

Aamliya

Harda Pratha

4

1

0.66

0.00

-0.66

Aamliya

Kakudi Bai Lalu Ji

4

1

0.29

0.00

-0.29

Aamliya

Lalu Limba

4

1

0.38

0.00

-0.38

Aamliya

Laluji Nanka

4

1

0.75

0.75

0.00

Aamliya

Nukki Jala

4

1

0.32

0.00

-0.32

Aamliya

Phoola Bhima Ji

4

1

3.25

2.33

-0.92

Cheerwa

Ram Lal Bheru Lal

6

4

5.12

4.68

-0.44

Cheerwa

Hamira Geva

7

5

1.20

1.20

0.00

Cheerwa

Keshar Hemer Singh

7

5

0.75

0.00

-0.75

Cheerwa

Devoo Kalyan

4

1

2.20

0.00

-2.20

Khempur

Deva Lakhma

5

3

1.37

0.75

-0.62

Khempur

Ramji Kannaji

4

2

1.10

0.57

-0.53

Khempur

Logerlal Pemaji

8

4

2.20

1.00

-1.20

Khempur

Laluram Pema

8

4

2.10

0.90

-1.20

Namri

Mangni Ukarlal

9

3

3.69

2.56

-1.13

Namri

Heera Bai Roopa Ji

7

4

0.66

0.00

-0.66

Namri

Logari Bai Bhaga Ji

7

4

1.25

0.25

-1.00

Namri

Balu Kalu Ji

7

4

0.75

0.75

0.00

Shyampura

Mool Chand Kalu Ram

11

4

2.55

0.81

-1.74

Shyampura

Mava Ji Vaja

7

2

1.01

0.00

-1.01

Shyampura

Ratni Bai Kush

7

4

0.00

0.00

0.00

Shyampura

Balki Bai Dharmi Lal

6

2

0.35

0.00

-0.35

Shyampura

Mangla Chamna

4

1

2.92

0.70

-2.22

(p.174) It helped to have local area residents working as the interviewers for these studies. Many questions cannot reasonably be asked by outsiders. People are more comfortable discussing important events in their lives with others who have a similar cultural and socioeconomic background; there is more empathy and sensitivity, and less embarrassment, on both sides. The interface between researcher and respondent is critical for this method to work well, which is why intensive training is built in at the start of every Stages-of-Progress exercise.

On occasion, it was heart-rending for both the interviewee and the interviewer to discuss particular events, particularly those involving the sickness and ultimate death of loved ones. Often, the interviewers had to stop and comfort their interviewees. In a small number of cases, these interviews had to be adjourned to the next day.

Risks and Remedies

Some of the risks involved were addressed reasonably successfully. Some other risks and limitations will not be as easily overcome. First, the methodology needs to deal better with intra-household, particularly gender-based, differences. Research has shown that females within households as well as female-headed households are likely to be poorer on average than their male counterparts.22 It is important, therefore, to probe these differences further.

Using Stages-of-Progress, I was able to uncover the differences between male- and female-headed households, finding the latter group of households to be considerably worse off than the former group. However, considering entire households as the units of analysis made it difficult to drill further down. Differences within households, between male and female members, are not easily detected using this process. Further improvements are necessary that will make possible such intra-household comparisons. I continue to seek a viable means for incorporating these additional steps within Stage-of-Progress.

Another set of refinements were made necessary when newly formed and urban communities were selected for investigation. The North Carolina study was the first one of this type. Because poverty is less easily and less self-consciously discussed publicly in communities of the United States than in communities of the other countries studied, and because community composition also tends to be less stable in this part of the world, the Stages-of-Progress methodology was modified in a few respects when we conducted the study in North Carolina. First, we worked with a shorter time period, considering no more than ten years in all. Second, we relied more than we had in other (p.175) contexts upon household interviews. Reasons for escape and descent were ascertained entirely through interviews with multiple household members; community groups did not provide this information. Thus, one element of triangulation was not available to us in North Carolina. Another set of improvements is required on this account.

Applying Stages within community settings helped abate to a considerable extent the danger of stigmatization. By categorizing people as occupying a particular stage (1–13) or a particular category (A–D), we had no need to refer to particular individuals as ‘poor’ or ‘rich.’ Tracy Rhoney, an enthusiastic and well-regarded community organizer in Burke County, North Carolina, explained to me in her unforgettable accent: ‘It’s almost like asking a woman about her dress size: Are you a five or a four? Are you Stage 5, or are you Stage 4? It’s that simple. People don’t mind talking about these things.’ 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 more wary and close-mouthed, however, when someone else’s situation was being discussed. It was quite different in communities of the other regions studied, where people spoke more freely and openly about each others’ situations, perhaps because in these contexts poverty is less often considered as an indicator of personal failure.

I referred earlier to the possibility of elite domination. In the Stages process, it is made clear at the start of every community meeting that no tangible benefits will be given out to anyone. This reduces the incentives that people might have to distort the facts, but the danger of elite domination is still present. We have maintained some balance in the composition of the community groups. In India, for instance, we did not commence discussions until lower and upper caste people were both present at the community meeting. We have also learned techniques for rotating community respondents and isolating domineering speakers by taking them aside for separate interviews. One other part of the Stages process helped to reduce the possibility of elite domination. All facts ascertained in the community meeting were separately verified in privately held household interviews. To the extent that the fear of elites does not also extend into private spaces, imbalances arising in the community group were ironed out at this point. I welcome suggestions about other safeguards that can be employed to deal better with this source of risk.

Another likely pitfall, common to all longitudinal studies, arises on account of the changing compositions of communities and households. Migration is one reason why the composition of households and communities can change over time, but there are also some other reasons, having to do with natural processes of birth, death, and aging. (p.176) Such effects will be more pronounced the longer is the period that is studied. Over a period of 25 years, new households will be inevitably set up and some old households will cease to exist.

Because households do not retain their original composition, some simplifying assumptions are made by all longitudinal studies. The objective of the study guides the assumptions that are made. Prospective studies consider households in the starting year of the study. They compare this original set of households over time, usually neglecting to study other households that get established in subsequent years. This assumption about the appropriate units of analysis (and the consequent neglect of newly established households) does not, however, detract from the basic objective of these studies, which is to understand what happened to the original households at later times.

Retrospective studies, such as those that use Stages-of-Progress, risk incurring the opposite neglect. Because they regard present-day households as their units of analysis, such studies can fail to take account of households that existed in the past but faded away. We found in a few locations where we inquired about this disappearance that it was undergone by roughly equal numbers of very rich and very poor households. On average, around 2 percent of households had vanished from both tails of the asset distribution. Once again, the assumptions that we made did not compromise the basic objective of these studies. They were primarily intended to answer the two why questions, that is, to identify the natures of reasons involved in escapes and descents, and this objective has been adequately served by these studies.

To conclude this relatively brief discussion of methodological issues, I will mention one last cause for concern, which is also the hardest one to address. Understandings of poverty can change over time. As the world around them advances, people’s conceptions of what it means to be poor are commensurately revised. Can comparisons be validly made between two periods of time that are characterized by different poverty definitions?

For the sake of comparability, analysts generally evade this question, preferring to work with one fixed measure of poverty. Usually, they adopt the measure that is prevalent at the time when the study is commenced. Valid concerns can be raised about how much relevance any such measure has for some period in the past (or some period in the future). Practically, some such (admittedly arbitrary) selection has to be made. For if we worked with separate definitions (and different measures) for the previous period and the present time, then what exactly would we end up comparing?

The methodological complications associated with studying change over time are many. Nevertheless, such analyses have to be undertaken. In order to design more (p.177) effective policies, it is essential to identify the factors that propel change over time. While remaining aware of all that can go wrong in longitudinal investigations, one must try to uncover as much as possible that is reliable and useful. Different methodologies can help fill this knowledge gap in some part. Stages-of-Progress is important among these methods.

Notes:

(1.) For a discussion of these and other issues that tend to complicate the tasks of poverty measurement, see, for example, Chaudhuri and Ravallion (1994); Lanjouw (1998); Lok-Desallien (1999); Schelzig (2001); Johnson (2002); Reddy and Pogge (p.194) (2002); Karshenas (2003); Krishna (2003); Kakwani and Son (2006); Reddy and Minoiu (2007); Wade (2004); and Reddy (2008).

(18.) A more detailed methodology manual, which also provides guidance on training procedures, can be downloaded from the author’s web site.

(19.) I thank Patti Kristjanson for drawing my attention to this important point.

(21.) Sporadic experiments to computerize land records have been ongoing since the late 1980s.