The Individual Poverty Incidence of Growth
| Author | Maria C. Lo Bue,Flaviana Palmisano |
| Date | 01 December 2021 |
| Published date | 01 December 2021 |
| DOI | http://doi.org/10.1111/obes.12362 |
1295
©2020 UNU-WIDER. OxfordBulletin of Economics and Statistics published by Oxford University and John Wiley & Sons Ltd.
This is an open access article under the terms of the Creative CommonsAttribution-NonCommercial-ShareAlike License, which permits use and distribution in any
medium, providedthe original work is properly cited, the use is non-commercial and the content is offered under identical terms
The copyright line for this article was changed on 9 July2020 after original online publication.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 82, 6 (2020) 0305–9049
doi: 10.1111/obes.12362
The Individual Poverty Incidence of Growth
Maria C. Lo Bue† and Flaviana Palmisano‡
†UNU-WIDER, Helsinki, 00530, Finland. (e-mail: Lobue@wider.unu.edu)
‡Department of Economics and Law, Sapienza University of Rome, Via del Castro
Laurenziano 9, Rome, 00161, Italy. (e-mail: flaviana.palmisano@uniroma1.it)
Abstract
The canonical approach to analyse the povertyimpact of g rowthis based on the comparison
of poverty before and after growth. Measurement tools endorsing this approach fail to
capture the different experiences of povertydynamic in the population: there can be g roups
of the population made poorer or non-poor made poor by growth. We propose an approach
that allows measuring this individualpoverty incidence of growth and show how it is related
with existing models. We apply our framework to evaluate the poverty impact of growth in
Indonesia, by comparing the 1993–2000 with the 2000–07 and 2007–14 growth spells.
I. Introduction
A highly debated issue in the economic literature concerns the evaluation of the distribu-
tional implications of growth and dates back to the Kuznets curve (the Kuznets hypothesis
elaborated in 1955). This curve was used to understand the nature of the link between
aggregate indicators of the distribution, such as the growth rate in mean income and the
level of inequality. The increasing availability of survey data has more recently made the
scientific community better awarethat individuals, rather than a representative aggregate of
the whole population, should be the focus of analysis for evaluating the impact of growth
on the distribution (see, among others, Ravallion, 1998, 2012; Benjamin, et al., 2001;
Ravallion and Chen, 2007). This individual-based perspective is now also central in the
political agenda: for instance, one of the targets of the Sustainable Development Goals
is to promote ‘inclusive economic growth’– that is, growth that benefits all segments of
society.
On the basis of this observation, a more recent literature has tried to go beyond the anal-
ysis of the nexus of aggregate indicators of the distribution and to understand whether and
how each part of this distribution benefits from the overall growth process.This approach
finds its roots in the so-called growth incidence curve (GIC) introduced by Ravallion and
Chen (2003). The GIC plots the percentile-specific rate of income growth in a givenperiod
of time against each percentile. Different ways of aggregating each coordinate of the GIC
JEL Classification numbers: D31, D63, I32.
1296 Bulletin
produce different measures of growth pro-poorness: the extent to which poverty changes
over time because of growth (see Kraay, 2006).
These criteria are based on the comparison of each income percentile at two different
points in time. Therefore, although based on individual data, this procedure—built on the
anonymity axiom (see inter alia Ravallion and Chen, 2003; Son, 2004; Essama-Nssah,
2005; Duclos, 2009; Essama-Nssah and Lambert, 2009)—ignores the individuals’ identi-
ties and does not allow tracing their income dynamics. In terms of pro-poorness evaluation,
this implies that, for a given initial distribution of income, if after growth the final distri-
bution of income is exactly the same as the initial distribution of income except for the
fact that there is complete re-ranking, this growth process would be judged to be neutral
from a pro-poorness perspective. However, we might want to judge this process differently
as, because of re-ranking, the poverty of each individual changes over time. The same
counter-intuitive result is obtained when comparing this growth process to a process in
which every individual maintains the same income and rank: the two growth processes
would be judged to be identical from a pro-poorness perspective.
More recently, the literature has started to move from the study of the distribution of
quintile-specific growth rates to the analysis of individual-specific growth rates and of
individual transitions between twodistributions. This has been done by relaxing the axiom
of anonymity (see inter alia Grimm, 2007;Van Kerm, 2009; Bourguignon,2011; Palmisano
and Peragine, 2015; Jenkins and Van Kerm, 2016; Palmisano andVan de Gaer, 2016) with
the conceptualization of non-anonymous GICs (na-GIC, or mobility profiles) introduced
independently by Grimm (2007), Van Kerm (2009), and Bourguignon (2011).The na-GIC
measures the individual-specific rate of economic growth between two points in time,
thus plotting the income trajectories of all the individuals which were in the same initial
position, independently of the position they acquire in the final distribution of income. To
be more specific, the na-GIC is obtained by keeping the ranking of statistical units constant.
Instead, comparing the initial and terminal quantile functions, as the standard GIC does,
is equivalent to re-ranking individuals, with the result that it is not the income of the same
individual that is compared, but the income of the same quantile.
Some recent contributions propose normative characterizations for the non-anonymous
approach and hence provide a normativejustification for the use of the na-GIC and a variety
of aggregate measures based on these curves.1However, with few exceptions that will be
discussed below, no contribution has explicitly considered the extent to which economic
growth hinders the upward income trajectories of the poorest segments of society. Hence,
for instance, to measure the pro-poorness of growth, the poverty levels are computed in the
two periods of time and then compared. If this procedure is valid when one is interested in
measuring the pure distributional change that takes place, it can make a big difference if
the poor people in the first period are still the same poor people in the following period, or
if there has been a substantial reshuffling of the individual positions in the population. To
capture this aspect, one needs to relax anonymity and account for each individual’s poverty
dynamics along the distribution. This information allows us to find out whoare the winners
and the losers from growth, useful data, for example, in the evaluation of the efficacy of
policy reforms. For instance, in the case of a continuous monitoring of poverty dynamics,
1See also Barcena and Canto (2018) and Creedy and Gemmell (2018).
©2020 UNU-WIDER. Oxford Bulletin of Economics and Statistics published byJohn Wiley & Sons Ltd.
To continue reading
Request your trial