Geoscience Reference
In-Depth Information
the highest levels geographical disparities in the incidence of income inequality
bear no impact on levels of decentralization of redistribution.
In conclusion, the hypothesis that the effects of inequality on the decen-
tralization of interpersonal redistribution are contingent on the way different
political institutions shape the balance of power between the center and the
regions receives a good deal of empirical support. These findings, however,
are based on a small subset of advanced industrial societies. If the relation-
ships identified hold only in a small subgroup of rich democracies, and not
across political unions around the world, the argument would not be neces-
sarily wrong but the realm of its validity would be much narrower. In other
words, this sampling restriction could potentially cast doubts on the scope of
the argument and findings in this topic.
BEYOND THE OECD: ECONOMIC GEOGRAPHY AND INTERPERSONAL
REDISTRIBUTION IN POLITICAL UNIONS AROUND THE WORLD
To alleviate these concerns, I replicate the analysis on a different sample, one
including nineteen political unions from both the developed and the developing
world during the period 1978-2001.
10
To the extent that the data allow, I fol-
low the approach presented in equations
(8.1)
and
(8.2)
. There are important
differences concerning measurement issues that condition the analysis through-
out: while I operationalize the dependent variable (IPR) in the same way as
before, I rely now on IMF data as opposed to the better quality OECD ones.
In addition, in the absence of household income surveys that are representa-
tive at the subnational level, comparable, and accessible, I cannot measure the
incidence of inequality across regions, much less calculate the ration between
most and the least unequal region. As an alternative, I rely on a more limited
measure. For every union included in the analysis, I proxy the geography of
income inequality with the p90/p10 ratio of its members' GDP per capita.
Determinants of Interregional Inequalities
Because of the nature of the data and the sample of countries under considera-
tion, I introduce important modifications in the first equation. The geography
of income inequality is modeled as a function of the levels of interindustry
labor mobility, the country's level of oil dependence, and the distribution of
the ethnic population across regions. The rationale behind the selection of these
variables follows from established findings in economic geography.
10
The countries included in the analysis are: Argentina (11), Australia (22), Austria (20), Belgium
(3), Brazil (2), Canada (20), Czechoslovakia (3), Germany (17), India (20), Malaysia (5), Mexico
(20), Pakistan (2), Russia (2), South Africa (5), Spain (19), Switzerland (9), United States (19),
Venezuela (2), and Yugoslavia (12). In parentheses I report the number of years for which
I actually have valid observations for all the variables included in the analysis. The panel is
obviously unbalanced.
