Empirical issues in lifetime poverty measurement

TL;DR: In this paper, the authors analyze some specific proposals that take the so-called spells approach and consider how they differ in the manner in which they address issues of lifetime poverty, most notably the measurement of chronic poverty.

Abstract: Recently there has been increased interest in developing methods for measuring poverty in a way that takes into account the lifetime experience of individuals. We analyze some specific proposals that take the so-called spells approach and consider how they differ in the manner in which they address issues of lifetime poverty, most notably the measurement of chronic poverty. Comparing these specific measures by applying them to a US panel data set, we provide important insights for further research in conceptualizing and measuring lifetime poverty.

Summary

I. Introduction

• It has long been recognized that the impact of poverty experienced by an individual for a long and sustained period of time is very di¤erent than poverty experienced for one or more relatively short, intermittent, periods within a person's lifetime.
• Much of the focus of this research has been on the identi…cation of chronic and transient poverty and the measurement of their relative importance.
• Besides addressing the property of chronic poverty over a person's lifetime, another consideration that has been deemed important in comparing the temporal pattern of poverty is that of the importance of poverty experienced early in life.
• In conjunction with other axioms that re ‡ect standard concerns or properties of poverty from the literature on measurement of poverty in a static setting, one obtains a simple structure for a lifetime poverty measurement approach.
• Empirical applications are explored and reported on in Section IV, followed by a section with remarks and conclusions.

II. The Measurement of Individual Lifetime Poverty

• In this section the authors explain the properties of the lifetime poverty measure used in their empirical analysis.
• The measurement for individual lifetime poverty consists of three steps: (1) the measurement of each individual's "snapshot poverty"at each time period in life, (2) the aggregation of these snapshot poverty spells across all periods, and (3) the inclusion of a retrospective view of poverty over the lifetime as a whole.
• When viewed from a lifetime perspective, the su¤ering and deprivation of an individual in each period transmits into the lifetime evaluation of poverty.
• In each period, the individual's poverty status is determined by comparing his consumption level with the poverty line 0 < z < 1 which is exogenously given and remains constant throughout the T periods.
• The authors choice also re ‡ects the fact that poverty orderings at third and above orders may collapse to second-order if the poverty line is uncertain and expands over a large interval (Zheng, 1999).

III Elaboration of Early and Chronic Poverty Concerns

• (t; T ), the way in which one can incorporate a concern with chronic and early poverty in a lifetime poverty measure.the authors.
• Et al. (2010) in de…ning a money metric cost of poverty as the "equally-distributed equivalent"(EDE) poverty gap; that is, the level of poverty gap if distributed equally to all persons in all periods of life that would produce the same measure of poverty as for the actual lifetime pro…les of poverty experienced across the population.the authors.
• It is not and should not generally be possible to separate in a simplistic manner concerns with early poverty, chronic poverty, and di¤erent intensities of poverty across di¤erent time periods.
• Comparing the above two weighting functions, based on the family (t; T ) = (1 t T +1 ) with = 0:2 and 0:5, the authors can see, roughly speaking, that the weighting function that has a higher degree of curvature also has less sensitivity to early poverty.
• As shown earlier, one can easily generate weighting functions for which this is not the case.

IV. Empirical Results

• The authors results above suggest that moving to a lifetime poverty measurement approach in which one needs to compare many time periods of possible poverty experiences between individuals rather than simply a single period, cross-sectional application, may reduce one's ability to say when one person has experienced more lifetime poverty than another.
• Moreover, among those experiencing any poverty, the percentage incurring …ve or more spells of poverty is 28.6% for whites and 63% for non-whites.
• One interesting …nding of the analysis performed by Bishop, Formby, and Thistle (1992,1994) is that the South's income distribution either converged or moved signi…cantly closer to the income distribution of the rest of the country between 1969 and 1979.
• There is a sense in which breaking up the period into two subintervals indirectly allows for an approximate comparison of the two cohorts of families and their children of the two subintervals.

V. Some Remarks and Conclusion

• 494 individuals over 26 consecutive years in order to illustrate the implications of measuring poverty from a lifetime perspective.the authors.
• By using a weighted sum of the individual snapshot poverty experiences, the authors are able to re ‡ect the sensitivity of how any poverty spells are distributed over a person's lifetime on lifetime poverty through the pattern of weights.
• The early poverty axiom re ‡ects the well-established argument that poverty early in life is more critical than poverty later in life while the chronic poverty axiom re ‡ects the idea that, for example, two spells of poverty of a given intensity are more harmful to an individual's well-being the closer in time that these spells occur.
• One can also, of course, depart from their axioms.
• In fact, the appropriate set of weights may depend on country-speci…c matters such as availability of age-related public goods and services.

Figures

Table 1: EDE poverty gapbg( ; "); " = 1

Table 2: EDE poverty gapbg( ; "); " = 2

Table 18: West EDE poverty gapbg(c1; c2; "); " = 1

Table 16: South EDE poverty gapbg(c1; c2; "); " = 1

Table 17: West EDE poverty gap (1967-1992)bg(c1; c2; "); " = 1 c1

Table 15: South EDE poverty gap (1967-1992)bg(c1; c2; "); " = 1 c1

Table 14: North Central EDE poverty gapbg(c1; c2; "); " = 1

Table 7: Percentage experiencing at least one spell of poverty

Figure 1: Weights for (t; T ) = (1 t T+1 ) with = 0:2 (i.e., 1(t)) and = 0:5 (i.e., 2(t))

Table 13: North Central EDE poverty gap (1967-1992)bg(c1; c2; "); " = 1 c1

Table 10: EDE poverty gap for regions (1980-1992)bg( ; "); " = 1

Table 11: Northeast EDE poverty gap (1967-1992)bg(c1; c2; "); " = 1 c1

Table 12: Northeast EDE poverty gapbg(c1; c2; "); " = 1

Table 6: EDE poverty gap for non-whitesbg(c1; c2; "); " = 2 c1

Table 4: EDE poverty gap for non-whitesbg(c1; c2; "); " = 1 c1

Table 3: EDE poverty gap for whitesbg(c1; c2; "); " = 1 c1

Table 5: EDE poverty gap for whitesbg(c1; c2; "); " = 2 c1

Table 8: EDE poverty gap for regions (1967-1992)bg( ; "); " = 1

Table 9: EDE poverty gap for regions (1967-1979)bg( ; "); " = 1

Full text

Copyright
©
UNU-WIDER 2010
1
University of Guelph, E-mail: mhoy@uoguelph.ca;
2
Ryerson University, E-mail: brennan@ryerson.ca;
3
University of Colorado, Denver, E-mail: buhong.zheng@cudenver.edu
This study has been prepared within the UNU-WIDER project on the Frontiers of Poverty Analysis,
directed by Tony Shorrocks.
UNU-WIDER gratefully acknowledges the financial contributions to the research programme by the
governments of Denmark (Royal Ministry of Foreign Affairs), Finland (Ministry for Foreign Affairs),
Sweden (Swedish International Development Cooperation Agency—Sida) and the United Kingdom
(Department for International Development—DFID).
ISSN 1798-7237 ISBN 978-92-9230-311-2
Working Paper No. 2010/73
Empirical Issues in Lifetime Poverty
Measurement
Michael Hoy,
1
Brennan Scott
Thompson,
2
and Buhong Zheng
3
June 2010
Abstract
This paper demonstrates the implications of adopting an approach to measuring poverty that
takes into account the lifetime experience of individuals rather than simply taking a static or
cross-sectional perspective. Our approach follows the theoretical innovations in Hoy and
Zheng (2008) which address various aspects of the specific pattern of any poverty spells
experienced by an individual as well as a possible retrospective consideration that an
individual might have concerning his life experience as a whole. For an individual, our
perspective of lifetime poverty is influenced by both the snapshot poverty of each period
and the poverty level of the permanent lifetime consumption; it is also influenced by how
poverty spells are distributed over the lifetime. Using PSID data for the US, we demonstrate
empirically the power of alternative axioms concerning how lifetime poverty should be
measured when making pairwise comparisons of individual lifetime profiles of consumption
(income) experiences. We also demonstrate the importance of taking a lifetime view of
poverty in comparing poverty between groups by use of the classic FGT ‘snapshot’ poverty
index in conjunction with period weighting functions that explicitly reflect concerns about
the pattern of poverty spells over individuals’ lifetimes.
Keywords: Lifetime poverty, snapshot poverty, chronic poverty, early poverty, poverty
measurement
JEL classification: I32

The World Institute for Development Economics Research (WIDER) was
established by the United Nations University (UNU) as its first research and
training centre and started work in Helsinki, Finland in 1985. The Institute
undertakes applied research and policy analysis on structural changes
affecting the developing and transitional economies, provides a forum for the
advocacy of policies leading to robust, equitable and environmentally
sustainable growth, and promotes capacity strengthening and training in the
field of economic and social policy making. Work is carried out by staff
researchers and visiting scholars in Helsinki and through networks of
collaborating scholars and institutions around the world.
www.wider.unu.edu publications@wider.unu.edu
UNU World Institute for Development Economics Research (UNU-WIDER)
Katajanokanlaituri 6 B, 00160 Helsinki, Finland
Typescript prepared by the Authors.
The views expressed in this publication are those of the author(s). Publication does not imply
endorsement by the Institute or the United Nations University, nor by the programme/project sponsors, of
any of the views expressed.
Acknowledgements
An earlier version of this paper was presented at the UNU-WIDER conference on
Frontiers of Poverty Analysis, September 2008. We thank participants for many helpful
comments and UNU-WIDER for financial support to attend the conference. Also, a
version of this paper was presented at the 8th International Meeting of the Society for
Social Choice and Welfare (July 2006, Istanbul). For very helpful comments we
especially thank Walter Bossert, Conchita D’Ambrosio, and Serge Kolm.

E
m
pirical Issues in Lifetime Poverty Measurement
I. Introduction
It has long been recognized that the impact of poverty experienced by an in-
dividual for a long and sustained period of time is very di¤erent than poverty ex-
perienced for one or more relatively short, intermittent, periods within a person’s
lifetime. These two temporal characterizations of poverty have come to be known as
chronic and transient p overty, respectively. Clark and Hume (2010) provide a very
useful background to the historical development of concern about, and progress with,
the conceptualization and measurement of individual poverty over time. They note
(Clark and Hume, 2010, p. 351) that “breakthroughs in terms of conceptualizing the
depth and breadth of poverty were not generally matched by equivalent, systematic
advances with regard to time prior to the late 1980s.”
1
However, even the terminology
of chronic poverty is long-standing. For example, Godley (1847, p. 2) describes his
concern with the persistent class di¤erences between the “English Protestant aris-
tocracy and middle class, and the Irish Roman Catholic peasantry, ... (being) two
nations within Ireland.”He describes the result of this social relationship (p. 10) as
creating “political and religious anomalies .... (that have) produced chronic anarchy,
with its necessary concomitant, chronic poverty.”
As a result of increased availability of panel data, there has recently also been sig-
ni…cant empirical research performed regarding the transitions of individuals into and
out of poverty and comparing duration of poverty for di¤erent individuals within and
between countries or various population subgroups. Much of the focus of this research
has been on the identi…cation of chronic and transient poverty and the measurement
of their relative importance. These approaches typically start by aggregating all
poverty spells of all individuals in order to obtain a measure of “total poverty.”One
can then establish an estimate of each individual’s permanent income and use this
as the basis for determining the level of chronic poverty for each person in the pop-
ulation. The di¤erence (residual) between total poverty and chronic poverty is then
frequently referred to as transient poverty. Methods for creating these types of dis-
tinctive measures can include a concern with intensity of poverty in both dimensions
of chronic and transient poverty by adopting an appropriate poverty index just as is
generally done in static poverty measurement exercises.
2
Since policy prescriptions
that deal with these two types of poverty are likely to di¤er, these exercises are both
useful and important.
Although important and insightful research on poverty dynamics of both a theo-
retical and empirical nature has been gaining in quantity and sophistication, devel-
opment of complete axiomatic characterizations for measuring poverty through time
1
Se
e
also Addison, Hulme, and Kanbur (2008, p. 8) who note that “The introduction of time
into the economic theory of poverty measurement is relatively recent.”
2
Good examples of this approach are Jalan and Ravallion (1998, 2000), Ravalliaon (1998), Baulch
and Ho d dinott (2000), and Duclos, Araar, and Giles (2010). See also the seminal paper of Rodgers
and Rodgers (1993).
3
1

is
an area of research in its infancy. The axiomatic foundations of static poverty
analysis has b een well established for some time.
3
Developing a complementary ap-
proach that incorporates sensitivity to the pattern of poverty spells through time can
borrow from that literature. However, there are many new challenges in developing
an axiomatic characterization that incorporates a temporal perspective. In order to
re‡ect chronic poverty concerns, one must address the manner and extent to which
the clumping of poverty spells together or in “nearby periods” should matter. The
permanent income approach, although useful empirically, does not distinguish, for
example, the di¤erence between someone who experiences poverty in every second
period of her lifetime and someone who spends the …rst half of her life constantly in
poverty and the second half out of poverty. Both individuals could have the same
permanent income and hence be assigned the same measure of chronic poverty even
though one could well argue that the second person su¤ered more chronic poverty
over her lifetime. Other questions that arise when considering how to measure chronic
poverty include the extent to which chronic poverty relief results from an interruption
of several consecutive periods of poverty with one or two periods of non-poverty.
Besides addressing the property of chronic poverty over a person’s lifetime, an-
other consideration that has been deemed important in comparing the temporal pat-
tern of poverty is that of the importance of poverty experienced early in life. There
is substantial empirical evidence that poverty in earlier stages of life not only a¤ects
consumption in later periods but also leaves an inherently deeper mark on lifetime
deprivation. Recent research in neuroscience (e.g., see Farah, et al., 2006) suggests
that children growing up in poor families with low social status not only su¤er from
inadequate nutrition and exposure to environmental toxins but also su¤er from ele-
vated stress hormones that generally impair neural development, including e¤ects on
language and memory.
4
This suggests at least two channels for adverse e¤ects on the
individual later in life. First is a direct lasting impact from poverty early in life in
that, due to the vulnerability of children, the physiological and psychological e¤ects
noted above reduce their future enjoyment from life for any given future (continu-
ing) stream of consumption levels. Secondly, early poverty worsens an individual’s
capability to generate higher consumption later in life due to a compromised ability
to accumulate human capital and obtain favorable employment opportunities. This
second e¤ect is captured implicitly if one includes all future levels of consumption in
the measurement of lifetime poverty. However, to account for the …rst (direct) e¤ect,
one may require that any lifetime measure of poverty place greater weight on poverty
experienced earlier in life.
We believe that developing an axiomatic approach to measuring poverty in a way
that captures the subtle but important nature of the temporal pattern in which an
3
See
for example the surveys of Zheng (1997, 2000a).
4
See also the reports on this line of research from the 2008 meetings of the American Asso c iation
for the Advancement of Science (Boston) described in an article by Clive Cookson (The Financial
Times, February 16, 2008) and also by Paul Krugman (New York Times, February 18, 2008).
4
2

in
dividual experiences poverty during her lifetime can advance the way in which we
conceptualize and measure lifetime poverty. Examples of papers that take this ap-
proach and investigate the implications of precisely stated axioms relating to chronic
poverty include Calvo and Dercon (2007), Foster (2007), Bossert, Chakravarty, and
D’Ambrosio (2007), and Hoy and Zheng (2008).
5
In the early stages of this research
program we believe it is important to (i) demonstrate the normative appeal of ax-
ioms adopted through use of simple hypothetical examples, (ii) illustrate the structure
imposed by axioms on the implied measurement approach, (iii) explain how di¤er-
ent poverty measures imply di¤erent e¤ects from changes in the pattern of lifetime
poverty experiences, and (iv) demonstrate through examples the empirical value of
the measurement approach. In this paper we develop these themes for the approach
of Hoy and Zheng (2008). We review and expand on the normative implications of
the axioms adopted in their approach and provide empirical applications using the
PSID data set for the US.
In this paper we focus on the axioms from Hoy and Zheng (2008) that bring out
a concern with measuring lifetime poverty in a way that emphasizes the presence of
chronic poverty and poverty experienced early in life for an individual. In conjunction
with other axioms that re‡ect standard concerns or properties of poverty from the lit-
erature on measurement of poverty in a static setting, one obtains a simple structure
for a lifetime poverty measurement approach. Speci…cally, the axioms together imply
that lifetime poverty is measured as a weighted sum of all spells or “snapshot” ex-
periences of poverty over the individual’s lifetime in conjunction with a retrospective
property of lifetime poverty based on average lifetime or permanent consumption.
We also show how pairs of individuals’lifetime poverty experiences can be compared
on the basis of these axioms by use of relevant dominance conditions that are related
to the familiar “concentration dominances” but based on vectors constructed from
the time pro…les of consumptions of pairs of individuals.
We demonstrate our approach using PSID data for the US from which we are
able to create income pro…les over twenty-six years for 1,494 individuals. Without
any structure on the weights, a pairwise lifetime poverty ordering requires vector
dominance of lifetime poverty pro…les. We …nd that few pairwise comparisons are
unambiguous. The structure imposed by both the chronic and early p overty axioms
signi…cantly extends the fraction of unambiguous pairwise comparisons that can be
made. Further, by adopting speci…c poverty measures that are consistent with our
axioms, we show how the comparisons of poverty between subpopulations can be
in‡uenced by a concern with the temporal pattern of poverty experiences. The implied
cost of poverty di¤ers from that implied by an approach that simply averages each
person’s poverty experiences over time.
In Section II we lay out the structure of our lifetime poverty measurement ap-
5
See
also Cruces (2005), Grab and Grimm (2007) and Carter and Ikegami (2007). Bossert,
D’Ambrosio, and Peragine (2008) and Hoy and Zheng (2008) are, to our knowledge, the only papers
that provide a complete axiomatic characterization of lifetime poverty measurement.
5
3

Frequently asked questions

Q1. What are the contributions in this paper?

This paper demonstrates the implications of adopting an approach to measuring poverty that takes into account the lifetime experience of individuals rather than simply taking a static or cross-sectional perspective. Their approach follows the theoretical innovations in Hoy and Zheng ( 2008 ) which address various aspects of the specific pattern of any poverty spells experienced by an individual as well as a possible retrospective consideration that an individual might have concerning his life experience as a whole. Using PSID data for the US, the authors demonstrate empirically the power of alternative axioms concerning how lifetime poverty should be measured when making pairwise comparisons of individual lifetime profiles of consumption ( income ) experiences. The authors also demonstrate the importance of taking a lifetime view of poverty in comparing poverty between groups by use of the classic FGT ‘ snapshot ’ poverty index in conjunction with period weighting functions that explicitly reflect concerns about the pattern of poverty spells over individuals ’ lifetimes. 

Q2. What are the future works in this paper?

The increased emphasis on the earlier and later poverty experiences form each set trumps the possibility that relief from chronic poverty should imply less lifetime poverty. Their approach explicitly recognizes such con icts and, the authors believe, emphasizes the need to consider such matters about the temporal pattern of poverty spells in the conceptualization of lifetime poverty. The authors have used the PSID data set and explored the power of orderings for pairwise comparisons as implied by alternative combinations of properties that can be made for any set of well-behaved snapshot poverty indices in conjunction with their axioms on how the temporal pattern of poverty spells may in uence lifetime poverty. 

Q3. What is the property of early poverty sensitivity?

The property of early poverty sensitivity, if it holds globally over a person s lifetime, means the weights must be nonincreasing over the time that an individual lives, while sensitivity to chronic poverty means the weights must be concave in time. 

Q4. What is the sensitivity of the implied lifetime poverty measure to early poverty?

As one reduces the value of from 1 to 0, the degree of curvature of the weights increases but there is also a change in the sensitivity of the implied lifetime poverty measure to early poverty. 

Q5. What is the memory parameter used in the paper?

To compute the individual s lifetime poverty index using (1) which will be used in the rest of the paper, the memory parameter must be speci ed. 

Q6. What is the weighting function for the lifetime poverty measure?

In the case of = 1, the weighting function will be linear and decreasing and so the associated lifetime poverty measure would be sensitive to early poverty but not chronic poverty. 

Q7. What are the usual possibilities for the properties of p(xt); z?

The authors consider two usual possibilities for the properties of p(xt; z) and p(x; z); namely, p0 < 0 for xt < z - the notion that poverty deprivation decreases as consumption increases - and p00 > 

Q8. What is the experience axiom used in the measurement of snapshot poverty?

This experience axiom is akin to the monotonicity axiom or the subgroup consistency axiom typically used in the measurement of snapshot poverty for a single period of time. 

Q9. What is the effect of the poverty gap index on the extent of comparisons?

Their results demonstrate that the extent to which this has occurred, as well as comparisons overall between regions, depends on the assumed sensitivity to the temporal pattern of poverty spells used in constructing lifetime poverty measures. 

Q10. What is the effect of recognizing temporal aspects of poverty in a lifetime poverty measurement exercise?

Since it is well established from cross-sectional studies that there is more poverty among non-whites than whites, it is interesting here to determine whether recognizing temporal aspects of poverty in a lifetime poverty measurement exercise changes the extent to which these sub-groups experience poverty. 

Q11. How many whites experience some poverty in their lifetime?

In terms of numbers of individuals who experience some poverty in their lifetime, the percentage for whites is 27.4% while that for non-whites is 70.7%. 

Q12. What is the weight function for the equally-distributed equivalent?

The authors now explain how the equally-distributed equivalent (EDE) poverty gap can be interpreted as a cost of lifetime poverty and how to associate various aspects of18It is clear that the weight function with the smaller value of can be written as a strictly concave function of the other. 

Q13. What is the effect of a weighting function on the measure of lifetime poverty?

when moving to a weighting function that is more sensitive to early poverty, the impact on lifetime poverty measurement depends on whether one does this using a function with a greater or lesser degree of curvature. 

Q14. How does the cost of poverty for whites change as the age of the population increases?

In the case of " = 1 (Table 1), the authors see that as rises, the cost of aggregate lifetime poverty for whites falls by over 10% while that for non-whites rises by about 3%. 

Q15. What is the effect of comparing lifetime poverty?

Their results above suggest that moving to a lifetime poverty measurement approach in which one needs to compare many time periods of possible poverty experiences between individuals rather than simply a single period, cross-sectional application, may reduce one s ability to say when one person has (unambiguously) experienced more lifetime poverty than another.