fact that the extensions of franchise in West European countries coincided with periods of the
historical peaks of inequality. In Feng and Zak (1999), a democratic transition is likely to
occur when a relatively wealthy middle class has a high enough demand for civil liberties
(which is a sort of luxury good) and can afford a destructive uprising. This approach implies
that the stability of authoritarianism is consistent with high inequality. Another implication is
that economic growth undermines a non-democratic regime. At the same time, McGuire and
Olson (1996) shows that a secure rational dictatorship promotes economic growth. Wintrobe
(1998) also suggests a possibility of a positive relationship between economic growth and the
political power of a dictator. The latter work, however, focuses on the loyalty of the
population “bought” by the dictator in exchange for public goods and on repression as the
major determinants of regime stability. A similar approach is adopted in Dmitriy Gershenson
and Herschel Grossman (2001), where the loyalty of the part of the population, coopted into
the ranks of the ruling elite, is a substitute for repression. Mark Harrison (2002) explains the
collapse of the Soviet regime by the increasing costs of coercion and decreasing reputation of
the rulers. The three latter works imply that a stable non-democratic regime is characterized
by high levels of public expenditure (which includes but is not limited to police expenditure).
Econometric studies of the determinants of the stability of non-democratic regimes
and democratic transitions are almost non-existent. One exception is Feng and Zak (1999),
which finds, in agreement with its theoretical predictions, positive effects of low inequality
and high levels of education and economic development on the probability of democratic
transitions in developing countries. Another cross-country study, Robert Barro (1999), yields
generally similar results with respect to the “propensity for democracy,” which is measured