Institutional Quality and Poverty Measures in A Cross-Section of Countries
Institutional Quality and Poverty Measures in A Cross-Section of Countries
Institutional Quality and Poverty Measures in A Cross-Section of Countries
(2000) 1: 123135
c Springer-Verlag 2000
(e-mail: [email protected])
1. Introduction
This paper makes two contributions. First, it provides empirical evidence on the
link between institutional quality and poverty, extending our understanding on
how these two variables may be associated. Such a question is not trivial as
there appears to be a growing conventional wisdom that they are independent,
reflected in the fact that a sentiment opposing institutional reform and favoring
populist policies has, once again, gained some support in many reforming coun-
Chong: The World Bank, and Adjunct Professor, Georgetown University Public Policy Institute;
Calderon: University of Rochester. We are grateful to Amihai Glazer, Luisa Zanforlin, and partic-
ipants at the Second Latin American and Caribbean Economic Association Meeting for comments.
A preliminary version of this paper was written while Chong was at IRIS, University of Maryland,
College Park. He wishes to acknowledge the late Mancur Olson for his hospitality. The views and
opinions in this paper are the authors and should not be attributed to the World Bank. The standard
disclaimer applies.
124 A. Chong C. Calderon
tries, particularly in Latin America1 . At the very least, this could be extremely
costly not only for political stability in the medium run, but also for economic
performance, as previous research by Mauro (1995), Clague (1997), and others
can attest.
Second, this paper builds on previous work on the link between institutional
efficiency and income inequality, an issue first pursued by Chong and Calderon,
(2000). These researchers find that in richer countries, the higher the quality of
a countrys institutions, the more equal the distribution of income. For poorer
countries, however, they find that the higher the quality of institutions, the worse
the distribution of income. They argue for the possibility that in the develop-
mental path of less developed countries income inequality may actually have to
worsen before improving when an institutional reform is first introduced. Another
explanation they mention is the existence of poverty traps in the development
path of less developing countries.
That inequality increases when institutional efficiency improves does not nec-
essarily mean that poverty will also increase. Poverty may well decrease in spite
of worsening distribution of income, provided that the income share of the richer
groups of the society increases by a larger proportion than that of the poorer
groups. Thus, stylized facts regarding income inequality may not necessarily be
extended to absolute poverty. The key question is what are the stylized facts
associated with poverty.
This paper is organized as follows. The next section provides an overview of
the recent relevant literature. The third section describes the empirical methodol-
ogy and the poverty measures employed. The fourth section presents our findings.
We find a significant and robust positive association between improvement in the
quality of institutions and reduction in poverty incidence. Thus, according to this
result, institutional reform may be worth pursuing. Finally, Section 5 concludes
and gives some policy recommendations.
Olson (1996) explains that the key elements in economic development are policies
and institutions. While industrialized economies seem to have achieved most of
their potential, many developing countries do not appear to have fulfilled theirs,
sometimes in spite of their vast resources. The problem, Olson argues, is that
poor countries do not have a well developed structure of incentives to bring forth
productive cooperation. Such a structure depends on economic policies as well
as on the quality of the institutional arrangements; for instance, legal systems
that enforce contracts impartially and make property rights secure over the long
run, political structures, constitutional provisions, and good enforcement systems
to monitor the extent of special-interest lobbies and cartels.
The argument provided by Olson to explain differences among countries may
also be applied to understand income differences within a country. Biased institu-
1 The Economist (1996)
Institutional quality and poverty measures in a cross-section of countries 125
tions where contracts are not enforced impartially but in favor of specific groups,
property rights that are not well secured but where special-interest lobbies have
the upper hand, corrupt bureaucracies that discriminate against the poor in such
a way that political and economic disparity is reinforced according to political
power and connections, and other institutional maladies (Klitgaard, 1988), may
produce great disparity of incomes, where a small elite reaps most of the benefits
of economic development while a large majority may have their incomes de-
cline. This may produce both a more unequal distribution of income and higher
absolute poverty to the lower shares of the population. Olsons argument gives
an indirect mechanism to understand the link between institutional quality and
poverty.
Pedersen (1997) uses a public choice analytical model to study a social struc-
ture where capitalists in the private sector and bureaucrats in the public sector
exploit agricultural producers, confiscate the benefits of foreign aid, and marginal-
ize the poor. His basic explanation relates to the distribution of political influence,
and the political processes that determine the distribution of such an influence.
The distribution of political influence between the private interest groups in his
model is determined by the extent of their rent-seeking and what this researcher
calls political endowment. The final distribution will depend on the extent to
which political decision makers manipulate the budget of the public sector to
their advantage. An implication of Pedersens model is that the distribution of
income may worsen along with an increase in poverty as long as the private
interest group is able to exert enough political influence to maximize its rents.
Another theoretical model that links political influence and distribution of
income is due to Bourguignon and Verdier (1997). These researchers develop
a model where the public decisions are taken by majority vote but where only
the educated part of the population is allowed to vote. Initially, an oligarchy of
educated and politically active individuals rule the country and the decision it has
to make is to foster growth, by subsidizing the schooling of the less educated and
consequently inactive majority. By doing so, the ruling elite gains an educational
externality, at the risk of losing some political control and income given the fact
that the newly educated individual will now be allowed to vote and thus, will tax
away the ruling elite. For a given set of conditions, their model may generate a
political Kuznets curve due to political rights redistribution (see Chong, 1999;
Acemoglu and Robinson, 1997). When the set of conditions is not met a Kuznets
curve may not occur and the uneducated may be trapped in poverty.
A somewhat counterintuitive explanation on the link between institutional
quality and poverty is provided by Chong and Calderon (2000). They argue
that an institutional reform may impose high transaction costs on the poorest
and consequently may first increase poverty before reform decreases it. The
informal mechanisms used to make the system work will no longer be useful
as new, formal, mechanisms may have to be learned to survive. Especially for the
poor, the transaction costs of this transition may be relatively high. Eventually,
improvements in the quality of institutions will reduce poverty. Better institutions
and lower uncertainty will ultimately improve the mechanisms, the effectiveness,
126 A. Chong C. Calderon
and the efficiency of public service delivery, especially in the rural areas and
marginal urban ones.
The development variables are from the World Development Indicators (World
Bank, 1996). The institutional quality data come from International Country Risk
Guide (ICRG), a private organization. The measures employed are: (i) Risk of
Expropriation, defined as the risk that the rules be abruptly changed; (ii) Repu-
diation of Contracts by the Government, defined as the risk of a modification
in a contract taking the form of postponement, scaling down or repudiation of
a contract due to change in government priorities; (iii) Law and Order Tradi-
tion, defined as the degree to which citizens are willing to accept the established
institutions to make and implement laws and adjudicate disputes; (iv) Corrup-
tion in Government, defined as the degree to which the officials will be more
likely to demand illegal payments; and, (v) Quality of Bureaucracy which rep-
resents the degree of autonomy from political pressure, strength, and expertise
to govern without drastic changes in policy or interruptions in government ser-
vices. These measures come from surveys of experts; the different variables are
assigned a score which results in subjective measures of a particular type of
institutional inefficiency. Among the many researchers who used these data are
Knack and Keefer (1995), Clague (1997), and Chong and Calderon (2000)2 . We
follow these researchers and rescale the institutional measures from zero to six3 .
Similarly, we take averages for each variable for the period 19821990, where
zero means lowest quality and six implies highest quality4 . Thus, if for a
particular country, the score is six this means that the country has very low
levels of corruption and vice-versa if the score were zero. Similarly, an overall
index is also created by calculating the simple averages of each of the measures.
The poverty indicators we use are members of the generic class of additive
indices proposed by Foster, Greer, and Thorbecke (1984). As it is well known,
this class of measures has the form
q
Px = 1/n [(z yi )/z ]x , (1)
i =1
are able to sell this kind of information year in and year out in exchange for high fees speaks in
favor of the reliability of the data.
3 When running regressions researchers typically take the average of the ICRG measures instead
of controlling for all single indices. This, because of the high correlation among them.
4 The ICRG variables are not available for the full period, but since 1982, only. Given the fact
that these variables tend to move slowly in time, the average for this period is believed to represent
the period 1960-1990 relatively accurately. Knack and Keefer (1995), Clague (1997), Mauro (1995),
and others, follow this same approach.
Institutional quality and poverty measures in a cross-section of countries 127
The poverty measures we use are three, when x = 0, 1, and 2. In other words, we
use the headcount index, the poverty gap index, and the Foster-Greer-Thorbecke
P2 index, respectively. The first is a measure of prevalence of poverty and is
defined as the proportion of the population for whom a certain measure of living
standard is less than an specific poverty line (Ravallion, 1992). The main limita-
tion of this measure is that it cannot measure how poor are the poor. For instance,
someone in the neighborhood of the poverty line threshold level is not in the
same situation as someone with no income at all (Kakwani, 1980). The second
poverty measure, or poverty-gap index, deals with this limitation. This measure
is based on the aggregate poverty deficit of the poor relative to the poverty line.
It depends on the distance of the poor below the poverty line and thus gives a
better indication of depth of poverty (Ravallion, 1992). A shortcoming of this in-
dex is that it does not take into account any possible inequality among the poor.
This measure will be unaffected by transfers from someone poor to someone
less poor. To address this last problem, we also use the Foster-Greer-Thorbecke
P2 measure which is an indicator of the severity of poverty and helps compare
policies that aim to reach the poorest. (Ravallion, 1992)5 .
In (1) it is clear that the critical variable required in order to obtain the dif-
ferent indices mentioned above is the poverty threshold, z. Poverty lines vary
according to the criteria employed, the geographic location considered, and the
degree of development of a society. The group considered poor in the United
States may not be considered poor in, say, Peru. When doing intercountry com-
parisons, there is agreement that the issues related with comparability are more
important than issues related with the precise objective criteria to be employed
in order to construct poverty lines (Ahluwalia et.al., 1979; Ravallion, 1992). This
is the approach we also take.
To construct an absolute poverty line that is comparable among countries
we follow a methodology first employed by Ahluwalia, et.al., (1979) and more
recently used by Psacharopoulos and Arriagada (1986) and Londono and Talvi
(1997). We exclude all the OECD countries from our original country sample
and keep 49 countries. This avoids problems of different poverty concepts in
developed countries as compared to underdeveloped ones. The focus on the latter
also allows us to use absolute poverty lines, which do not vary with changes in
living standards, and which are considered more appropriate for less developed
countries (Ravallion, 1992).
We follow Ahluwalia (1976) and use India as our reference case. These
researchers argue that a typical poverty line is defined by the total consumption
expenditure needed to ensure a daily supply of about 2250 calories per person,
which ranges around forty and fifty percent of the total population (Ahluwalia,
1976). They set the poverty level to be applied across countries as the income per
head accruing to the forty-fifth percentile which, they explain, yields conservative
and comparable estimates of poverty6 . We follow a similar approach, but use
5 For additional details please see Ravallion (1992) or Kakwani (1981)
6 As has been frequently pointed out, the definition of a poverty line is arbitrary and should be taken
with caution. An obvious weakness of our approach is that there might be wide variation in calorie
128 A. Chong C. Calderon
Arriagada and Psacharopoulos (1986) criteria and set the poverty level as the
income per head accruing to the fifty-first percentile of the Indian population7 .
Using a recent data set on income distribution by Deininger and Squire
(1996), and data on the equivalent purchasing power conversion rates and aver-
age per-capita income from Summers and Heston (1991) we calculate poverty
lines for 1990 for each country in our sample through estimations of parametrized
Lorenz curves8 . Finally, the headcount, poverty gap and Foster-Greer-Thorbecke
P2 measures are easily computed from the estimated poverty lines.
4. Results
Table 1 presents basic summary statistics of our estimated poverty measures us-
ing parametrized Lorenz Curves. The poverty line estimates, as explained above,
were calculated based on an income per head accruing to the fifty-first percentile
of the population. Table 2, on the other hand, show simple correlations for the
institutional and poverty measures. Regardless of the particular type of insti-
tutional measure or poverty index, the corresponding sign is always negative.
Given the high correlation among institutional measures (see footnote 3) it is not
surprising to find that the simple correlation between the poverty measures and
each institutional measure is somewhat similar. Overall, the institutional quality
average measure, has a correlation of -0.44 with the headcount index, -0.32 with
the poverty gap measure, and -0.29 with the Foster-Greer-Thorbecke P2 measure.
intake depending on the level of expenditure. Moreover, caloric requirement for the same individual
may change over time. On the other hand, purchasing power ratios may vary for different income
groups within a country. Finally, this income based approach makes no allowance for achievement of
minimum levels for essential public services (Ahluwalia, et al., 1979). In our favor, we should mention
that for a subsample of available countries (28) we computed the correlation between our estimated
poverty measures and those available from household surveys from the World Bank, adjusting for
calorie intake when necessary, and found that it was 0.89 for the headcount index.
7 All the results are robust to local change of the absolute poverty line. This is expected given the
fact that the country ranking does not change. Also, this result is consistent with that of Ravallion
(1992).
8 Following Datt (1992), we order Deininger and Squires data according to per-capita income
quintiles and calculate parametrized Lorenz curves using the generalized quadratic model of Vil-
lasenor and Arnold, (1989) and Kakwanis Beta Model (Kakwani, 1980). We use the estimation that
best adjusts to the data.
Institutional quality and poverty measures in a cross-section of countries 129
Headcount
Maximum 0.065 0.036 0.071 0.025 0.022 0.521
(3.059) (2.156) (4.956) (1.120) (1.562) (3.517)
Minimum 0.031 0.029 0.049 0.012 0.009 0.029
(3.126) (2.065) (3.564) (0.845) (1.286) (3.216)
Robust Robust Robust Fragile Fragile Robust
Pov. Gap
Maximum 0.355 0.212 0.273 0.120 0.012 0.184
(3.002) (3.198) (3.200) (1.698) (1.074) (2.902)
Minimum 0.253 0.137 0.228 0.102 0.071 0.176
(3.698) (2.022) (3.981) (1.110) (1.361) (2.564)
Robust Robust Robust Fragile Fragile Robust
FGT P2
Maximum 0.252 0.107 0.180 0.073 0.022 0.206
(4.479) (2.185) (2.492) (1.005) (1.001) (2.971)
Minimum 0.158 0.069 0.153 0.036 0.019 0.182
(2.002) (1.516) (2.477) (1.269) (1.984) (3.598)
Robust Fragile Robust Fragile Fragile Robust
Ordinary least squares regressions. Numbers in absolute value. T-stats in parenthesis.
Table 4 uses the same specification (2) but applies a two-stage least squares
method. Instruments are government spending on defense as a percentage of
GDP, and the legislative tradition of the country. Similar to Chong and Calderon
(2000), increases in defense spending are taken as diversion of resources (Knight,
et al., 1996). On the other hand, it has been shown that the legislative tradition
of a country is a significant regressor of institutional quality, but not of income
inequality (La Porta, et al., 1998; Chong and Calderon, 2000). The results using
2SLS are similar to those using OLS. Corruption and law and order tradition are
not significant explanatory variables of poverty, but quality of the bureaucracy,
risk of expropriation, and repudiation of contracts are important explanatory
variables of poverty indices.
Institutional quality and poverty measures in a cross-section of countries 133
5. Sensitivity Analysis
We use the specification (2) and apply Levine and Renelts (1991) sensitivity
analysis. To do this, we include an additional group of controls. We consider
about a dozen variables, all from the World Bank (1996). In particular, we use
rate of inflation, the standard deviation of the rate of inflation, fiscal deficit,
exchange rates, trade volume, international openness, black market premium, and
development indicators, such as the number of doctors per inhabitants, number
of teachers, the share of industry on GDP, the share of agriculture on GDP, and
a few others. Results are shown in Table 5. We find that (i) there is a negative,
significant, and robust relationship between the different poverty measures and
our overall index of institutional quality; (ii) the specific robust measures are risk
of expropriation and quality of the bureaucracy. The other individual measures,
such as corruption in government, law and order tradition, and repudiation of
contracts are not robust at the five percent significance level.
6. Conclusions
Appendix 1
Countries in sample
References
Acemoglu, D., Robinson, J. (1997) Why did the west extended its franchise. Working Paper, Depart-
ment of Economics, MIT
Ahluwalia, M. (1976) Inequality, poverty and development. Journal of Development Economics 3(3):
307342
Ahluwalia, M., Carter, N., Chenery, N. (1979) Growth and poverty in developing countries. Journal
of Development Economics 6(3): 299342
Bourguignon, F., Verdier, T. (1997) Oligarchy, democracy, inequality and growth. Working Paper,
97-10. DELTA Forthcoming, Journal of Development Economics
Chong, A., Calderon, C. (2000) Institutional quality and distribution of income. Economic Develop-
ment and Cultural Change. July, vol. 42, forthcoming
Chong, A. (1999) Inequality, democracy, and persistence: Is there a political Kuznets curve?.
Manuscript. The World Bank
Clague, C., ed., (1997) Institutions and economic development: Growth and governance in less
developed and post-socialist countries. Johns Hopkins University Press, Baltimore, MD
Datt, G., (1992) Computational tools for poverty measurement and analysis. Mimeo, World Bank
Deininger, K., Squire, L. (1996) Measuring income inequality: A new database. The World Bank
Economic Review 10(3): 565591
Easterly, W. (1997) Life during growth. Journal of Economic Growth 4(3): 239276 (September)
Foster, J. Greer, J., Thorbecke, E. (1984) A class of decomposable poverty measures. Econometrica
52(3): 761765
Kakwani, N. (1980) On a class of poverty measures. Econometrica 48(2): 437446
Kakwani, N. (1981) Note on a new measure of poverty. Econometrica 49(2): 525526
Klitgaard, R. (1988) Controlling corruption. University of California Press, Berkeley, CA
Knack, S., Keefer, P. (1995) Institutions and economic performance: Cross-country tests using alter-
native institutional measures. Economics and Politics 7(3): 207228
Institutional quality and poverty measures in a cross-section of countries 135
Knight, M., Loayza, M., Villanueva, D. (1996) The peace dividend: Military spending cuts and
economic growth. International Monetary Fund Staff Papers 43(1): 137
La Porta, R., Lopez de Silanes, F., Shleifer, A., Vishny, R. (1998) Law and finance. Journal of
Political Economy 106(6): 11131155 (December)
Levine, R., Renelt, D. (1992) A sensitivity analysis of cross-country growth regressions. American
Economic Review 82(2): 942963
Londono, J., Talvi, E. (1996) Inequality and structural reform in Latin America. Manuscript, Inter-
American Development Bank
Mauro, P. (1995) Corruption and growth. Quarterly Journal of Economics 82(2): 681712
Olson, M. (1996) Big bills left on the sidewalk: Why some nations are rich, and others poor. Journal
of Economic Perspectives 10(2): 324
Otero, M., Rhyne, E. (eds.) (1996) The new enterprise microfinance. Kumarian Press, Boulder, CO
Pedersen, K. (1997) The political economy of distribution in developing countries: A rent-seeking
approach, Public Choice 91(4): 351373
Persson, T., Tabellini G. (1992) Is inequality harmful for growth. American Economic Review 82(3):
537555
Psacharopoulos, G., Arriagada, M. (1986) The educational composition of the labor force: An inter-
national comparison. International Labour Review 125(5): 561574
Ravallion, M. (1992) Poverty comparisons: A guide to concepts and methods. Living standards
measurement study. The World Bank
Summers, R., Heston, A. (1991) The penn world table (mark 5): An expanded set of international
comparisons, 19501988, Quarterly Journal of Economics 106: 327368
The Economist (1996) The backlash in Latin America 31(7994): 1921
Villasenor, J, Arnold, B. (1989) Elliptical lorenz curves. Journal of Econometrics 40(2): 327338
World Bank, (1996) World development indicators. CD ROM