(3) The impact of economic growth on the survival of democracy is hard to determine.
Empirical patterns show that democracy is more fragile in countries where per capita
income stagnates or declines. But the direction of causality is not clear: do democracies die
because they perform poorly or do they perform poorly because they are about to die? I
could not determine which way the causality runs.
(4) Any conclusions about the impact of income distribution on the survival of democracy
are hampered by the paucity and the poor quality of the data. Nevertheless, among the few
cases for which comparable data are available, democracies are more likely to survive when
the Gini coefficient or the ratio of incomes of top-to-bottom-quintile are lower. Data
concerning functional distribution are more extensive and they show the same: democracy
is four times more likely to survive in countries in which the labor share of value added in
manufacturing is greater than 25 percent.
(5) Democracies are more likely to succumb when one party controls a large share (more
than two-thirds) of seats in the lower house of the legislature. Moreover, democracies are
most stable when the heads of governments change every so often, more often than once in
five years (but not as often as every two years). These observations indicate that democracy
is more likely to survive when no political force dominates completely and permanently.
(6) Since Montesquieu (1995 [1748]), much has been said about the importance of culture
for the emergence and durability of democracies. Recent discussions of this topic revolved
mainly around cultures identified by dominant religions. Even if Weber (1958 [1904-5])
himself said almost nothing about political institutions (Przeworski, Cheibub, and Limongi
1997), the idea that he saw in Protestantism the wellspring of modern democracy is
widespread among contemporary political scientists. Lipset (1959: 165) claimed that “It has
been argued by Max Weber among others that the factors making for democracy in this area
[north-west Europe and their English-speaking offsprings in America and Australasia] are a
historically unique concatenation of elements, part of the complex which also produced
capitalism in this area,” since “The emphasis within Protestantism on individual
responsibility furthered the emergence of democratic values.” In turn, Catholicism, in
Lipset's (1960: 72-73) view, was antithetical to democracy in pre-Second World War
Europe and Latin America. Yet even Catholicism is not the worst enemy of democracy:
Islam and Confucianism hold the palm (Eisenstadt 1968: 25-27). Thus, Huntington (1993:
15) reported that “No scholarly disagreement exists regarding the proposition that
traditional Confucianism was either undemocratic or antidemocratic.” Similar views about
Islam abound (Gellner 1991: 506, Lewis 1993: 96-98).
None of these assertions defends itself against facts. Indeed, the only effect of religions that
emerges from the statistical examination is that democracies are more likely to emerge in
countries in which there are more Catholics. Neither Protestantism nor Islam has an effect
on the emergence or the durability of democracy.
(7) Colonial legacy has little effect on regime stability once all the other factors are
considered. The rates of democratic failure are much higher among democracies that were
established after 1950. Yet this effect vanishes when controlled for income, indicating that