Geoscience Reference
In-Depth Information
a causal chain independent of the scope of cross-regional economic external-
ities. This alternative causal chain is purely political: it concerns the relative
position within the political system of those actors whose optimal choice is
a fiscal structure that privileges large levels of interregional transfers at the
expense of large scale systems of interpersonal redistribution. The theoretical
analysis in
Chapter 2
identified those actors as the social and political elites
in backward regions. They want to maximize the resources they extract from
the rest of the union without giving up control over the tools for interpersonal
redistribution and service provision. Backward regions tend to be rural and
underpopulated, and therefore these elites would be particularly privileged in
systems of representation in which malapportionment empowers them at the
national level, both within formal representative institutions and, as impor-
tantly, within the organization of countrywide political parties. To the extent
that this process is at work, one should expect that in those unions with weaker
national party systems in which local elites are directly elected, representatives
of relatively poorer and underpopulated regions will be more capable of car-
rying their preferences through into the design of fiscal structures. In sum,
controlling for all other determinants of interpersonal redistribution, more
centrifugal systems should facilitate a larger extraction capacity by local lead-
ers. In contrast, in a more centripetal system, local leaders have less leverage
vis-`a-vis national elites to carry their will through the system. Recent case study
analyses of a number of developing federations offer results that are consistent
with this expectation (Beramendi and Diaz-Cayeros
2008
; Gibson, Calvo and
Falleti
2004
).
I now offer what, to the best of my knowledge, is the first statistical analysis
of the influence of these two factors on the levels of interregional redistribution
in political unions around the world. The dependent variable measures the
overall amount of transfers between the center and the regions or between
the regions themselves as a percentage of the union's GDP. This variable is
the result of multiplying two of the variables facilitated in the IMF dataset on
decentralization: 1) transfers from other levels of governments as a percentage
of total regional revenues, and 2) the value of total regional revenues expressed
as a percentage of GDP. Admittedly, these data are less than ideal (Rodden
2004
), but for the quantities of interest in this section they constitute the only
available source for a large number of countries.
16
In turn, for the key independent variables of interest (mobility and repre-
sentation) I use the same two indicators as previously; that is, representation
defined as the level of party centralization weighted by the number of directly
elected officials (see expression (8.3)), and mobility defined as the elasticity of
labor to interregional wage differentials. Again, the use of the latter as a proxy
is justified on the idea that production activities tend to be regionally clustered
16
Contrasting these data with national sources in those cases for which these exist, for instance
Germany, Canada, the United States or Spain, does substantiate the point that the quality of
these data is very uneven. In some cases, the information provided by the IMF data seems
accurate whereas in others, like Germany, the gap is significantly larger.
