If modernization theory is correct, the probability that a country will become democratic should be higher in more developed countries, but this is a fallacy. Taiwan, for example, developed under an authoritarian regime, reached an annual per capita income of US$10,610 in 1996, and transitioned to democracy. But this is not yet evidence of the role of development. By 1996, the Taiwanese regime was 47 years old. If this regime had a 0.0226 chance of collapsing in any random year—the average probability for all dictatorships—it would have had only a 35 percent chance of surviving past 46 years even if it had remained as poor as it was in 1949. Most likely, the Taiwanese regime decided to hold elections because it needed to mobilize the support of democratic countries in its geopolitical conflict with China, a reason unrelated to development. In East Germany, the second wealthiest dictatorship fell in 1990, following the collapse of the Soviet Union. Spain, where the fourth wealthiest dictatorship collapsed in 1977, faced a crisis caused by the death of the founding dictator and the pressure to join Europe, for which democracy was a prerequisite. In Venezuela, the sixth wealthiest fell in 1958 because the United States turned against it. Dictatorships thus confront many hazards, independent of the development process, that can engender regime change.

These observations point to the underlying flaw in the modernization theory model that causally links development to democracy. Suppose that all dictatorships face the same chance of falling during a particular year for purely idiosyncratic reasons: death of the founding dictator, geopolitical crisis, foreign pressures, economic disaster, or a shift in policy by the United States or the Soviet Union, to take some examples. Some dictatorships succumb to these events, but since at any moment the probability of their occurrence is low, most escape them. Some of those that manage to survive grow economically. The observed result will be that dictatorships die in countries with high incomes. But they would have been unlikely to survive past old age just because of the accumulation of random hazards, independent of development. Thus, even if we observe a relation between incomes and the likelihood of transitions to democracy, the question remains whether dictatorships fall in wealthier countries because of their development or because the longer they persist, the more hazards they accumulate.

Even excluding the Persian Gulf states—where dictatorships survive at very high levels of income—while the probability of transitions to democracy increases as income grows until about US$5,000, dictatorships become more stable above this income level. Moreover, statistical analyses indicate that the observed relation between income and transitions to democracy is not due to income itself but to the fact that some of the factors that drive these transitions are related to income. First, countries with intermediate income levels are more likely to have experienced regime transitions in the past, and dictatorships last for shorter periods in countries that have previously experienced democracy. Secondly, dictatorships that emerge at intermediate income levels tend to be military dictatorships, which have shorter lives than civilian ones. Hence, the relation between income and transition to democracy is not monotonic, as modernization theory implies, and appears to be spurious.

Finally, if modernization theory is correct and if per capita income is a good indicator of modernization, it should be observed that stable democracies emerge around some distinct income threshold. Yet these thresholds vary enormously: India had a per capita income of about US$556 in 1947, the United States had an income of roughly US$1,100 in 1830—years when both countries established lasting democracies. Meanwhile East Germany, Taiwan, and Singapore remained under the grip of dictatorships even when their incomes surpassed US$10,000. Hence, lasting democracies can be established at highly disparate income levels.

In sum, even though some technical issues are entailed, there is no evidence that democracies are more likely to emerge when a country becomes modernized. Rather, the evidence is overwhelming that if democracy emerges in a country that is already modern, then it is much more likely to survive. No democracy ever fell in a country with a per capita income higher than that of Argentina in 1975—US$6055. This is a startling fact given that throughout history about 70 democracies have collapsed in poorer countries. In contrast, 35 democracies spent a total of 1,000 years under more affluent conditions, and not one collapsed. Affluent democracies survived wars, riots, scandals, and economic and governmental crises.

The probability that democracy survives increases monotonically with per capita income. Between 1951 and 1999, the probability that a democracy would fall during any particular year in countries with per capita income under US$1,000 was 0.089, implying that their expected life was about 11 years. With incomes in the range of US$1001 to US$3000, this probability was 0.037, for an expected duration of about 27 years. Between US$3001 and US$6055, the probability was 0.013, which translates into about 78 years of expected life. And above US$6055, democracies last forever.

Given these numbers, the observed relation between the incidence of democracy and per capita income is due almost entirely to the second mechanism of regime survival probabilities—not to the causal link between development and democracy. Dictatorships face multiple, heterogeneous risks. If one happens to fall in a country with a low per capita income, democracy is not likely to be long-lived. But if it falls in a more affluent country, democracy endures. Development does not generate democracy; rather democracies accumulate in the more developed countries.

Beware the Blueprints

In retrospect, both the pessimism about the effects of democracy on development and the optimism about the consequences of development for transitions to democracy were unjustified. Yet at the time, both of these views seemed to be supported by the available historical experience. If the idea that communism portended the future now appears ludicrous, if we now believe that dictatorships offer no developmental advantage over democratic alternatives, it is only because we now know what did transpire.

This is, then, a cautionary tale. Gripped by anti-communist panic, many US scholars and policymakers developed a view of the world that was scientifically unfounded and politically pernicious. Ideology dominated science. Remarkably, no statistical study published before 1982 found that democracies developed faster, and no study published after that date reported that they developed slower. This bias in the scholarly community provided a convenient rationale for US policymakers to overthrow elected governments and to support anti-communist dictatorships.