Political Instability and Economic Growth
Not everyone will be surprised to learn that political regimes have no effect on average rates
of growth of total income. One generally held view, made influential by Huntington (1968),
is that what matters for economic development is political stability, rather than the particular political institutions. Any system of political institutions promotes development
as long as it maintains political order. The danger is “political instability.”
Political instability, as measured by past or expected changes of chief executives, the
frequency of strikes, demonstrations, or riots, is much greater in democracies. Yet political
instability affects economic performance only under dictatorships. Changes of chief
executives, while much less frequent in dictatorships, are economically costly only in these
regimes. Whether because of institutional constraints or of motivations of those who govern
democracies, neither past nor expected changes of heads of governments affect growth
under democracies. But under dictatorships economic growth slows down significantly
when the tenure of rulers is threatened. The same is true of various forms of “socio-political
unrest”: strikes, anti-government demonstrations, and riots occur more often in democracies
but they retard growth only in dictatorships. Finally, the effect of threats to the stability of
regimes on investment shows that investors fear democracy and hope to find a safe haven in
many dictatorships: prospective demise of a dictatorship causes investors to flee while its
prospective advent makes them flock.
Thus, political instability retards growth exclusively under dictatorships. Huntington and
other “realists” got it all wrong. Studies of political instability are guided by the hypothesis
that, to state it in the words of Alesina and Perotti (1997: 21), “what influences growth is
not so much the type of regime (dictatorship or democracy) but regime instability, that is,
the propensity to coups and major changes of government.” This hypothesis is logically
incoherent, since “political instability” cannot be even defined independently of political
institutions. Alternation in office or other manifestations of popular opposition, whether
strikes or anti-government demonstrations, do not constitute “instability” under democracy.
Such phenomena are frequent in democracies because democracy is a system in which
people are free to express their dissatisfaction with governments. And they are rare in
dictatorships because it is precisely to prevent such forms of expression that dictatorships
are established and maintained. Democracies are inherently “unstable”: what constitutes
anomalies, breakdowns of rule, under dictatorship is just an essential, definitional, feature of
democracy. A change of chief executives in democracies occurs as a consequence of
elections or other regularized procedures, while under most dictatorships the only way
rulers can change is by a coup. To discover, as Alesina at al. (1996) do, that coups reduce
growth but regular alternations in office do not, is not to find that some forms of “political
instability” affect development, but only that some political events that constitute instability
under dictatorship do not under democracy.
Once it is understood that the same political phenomena have different meanings under
different regimes, it is not surprising that economic actors react to them differently. Under
dictatorships, whenever the regime is threatened, whenever rulers in fact change or are
expected to change, whenever workers muster the courage to strike or masses of people to
demonstrate their opposition to the government, the economy suffers. Under democracy,
everyone knows that the government will change from time to time, that workers may
strike, and people may express their dissatisfaction with the government in a variety of
ways. Hence, when such phenomena do transpire, they evoke at most an economic yawn.
Mortality, Fertility, and Population Growth
The most surprising impact of regimes is on the growth of population, which is much faster
under dictatorships. Moreover, this difference does not disappear when controlled for per
capita income and a whole host of other factors, including the colonial heritage, the
frequencies of Catholics, Protestants, and Muslims, labor force participation of women,
human capital, and female enrollment in secondary education. Moreover, diachronic
patterns show the same: the rate of population growth increases slightly when countries
experience transitions from democracy to dictatorship, while it falls when they transit from
dictatorship to democracy.
Now, note that the rate of growth of population (net of migration) can be decomposed as the
difference between the birth rates and the death rates. Hence, one reason population grows
faster under dictatorships could be that they have lower mortality rates. Yet, as shown in
Table 4, this is not the case. Indeed, the observed gross mortality rates (“deaths” in Table 4)
are higher under dictatorship. Infant mortality is also higher under dictatorships and life
expectancy is much shorter (not shown). Moreover, these differences do not vanish when
controlled for per capita income and the passage of time. If dictatorships and democracies
had the same per capita income and if they existed during the same year, mortality rates
would be 10.4 for dictatorships and 9.2 for democracies, while life expectancy (“lives”)
would be 61.2 under dictatorship and 67.1 under democracy. These are enormous
differences.
Hence, dictatorships have higher mortality rates. If their population grows faster than under
democracies, it is because dictatorships have even higher crude birth rates (“births”). Again,
this difference does not disappear when controlled for per capita income and year.
Moreover, this difference is not due to the age structure of the population but to differential
fertility: at any age, women have more children under dictatorships.
The average observed fertility per woman is 5.2 under dictatorships and 2.7 under
democracies. This difference does become reduced when controlled for per capita income
and a whole host of other variables. But whatever one controls for, women under
dictatorships have on the average 0.5 children more than under democracies. This difference
seems to be due to regime per se, rather than to any conditions under which the regimes find
themselves or to anything else they do. Under both regimes, fertility is strongly reduced
when women attend secondary schools but it is greater where public educational
expenditures are larger. A simple explanation is that secondary education increases the
market incomes of women and hence the opportunity costs of having children, while public
expenditures on education lower these costs. What is most striking is that public
expenditures on social security and on health strongly reduce fertility under democracy but
have no impact under dictatorship. Moreover, once these expenditures are considered, labor
force participation of women has no impact on fertility. The only explanation of this finding
I can think of is that under democracy public expenditures are seen as a form of insurance,
while they do not reduce uncertainty under dictatorships. This explanation is consistent with
the fact that both the variability of economic performance and of economic policies is greater under dictatorships. Hence, households respond to the uncertainty inherent in
dictatorships by hoarding the least risky asset they can, namely, children.