Now here is a potential dilemma. Would you frown your eyebrows if you came across a text where societies are said to consist of two groups — the elite and the citizens — and democracy is conceptualized as rule by the latter, and the history of Great-Britain between 1688 and 1932 was treated in two pages? But would you be impressed if you read a book that applied the implicit function theorem and subgame perfect equilibria with the greatest of ease? If the answer to both questions is affirmative, then the dilemma has become real, as two eminent economists, Acemoglu and Robinson, combine exactly these two features in their book. What is more and urges the dilemma to be addressed is that they aim to contribute to the long-standing political debate on the origin of democratization, by offering an encompassing, economic and formal framework.
The history of Britain is one of four paths to democracy, discussed in the first chapter. Britain exemplifies a steady albeit long development towards democracy, the main driving force for which, the authors suggest, is (the threat of) social disorder. This path contrasts with the path of Argentina, where democracy and non-democracy frequently switched place, due to — again — the threat of revolution and coups. The third and fourth routes are illustrated by Singapore and South Africa. Both were non-democracies, but the later moved towards democracy, whereas Singapore has not.
Many countries follow essentially four paths, which, the authors hold, may very well be explainable by one theory. The theory is discussed verbally in the second chapter, and its shortest version is that democracy solves the commitment problem the elite (alternatively dubbed the rich) faces when the citizens (or the poor) threaten to topple a non-democratic regime. In the dichotomous distinction between democracies and non-democracies, the latter are understood as situations of (relative) political inequality where the elite hold all de jure political power. In contrast to that, democracy means government by the people and for the people, resulting in political equality. Although political institutions determine de jure power, that does not hold for de facto power. Windows of opportunities, such as social and economic crises and (aftermaths of) war, enable to solve the coordination problem revolutionaries face, and so may temporally give the citizens the de facto power to overthrow the elite. The elite may want to avoid that by promising redistribution policies. However, that promise is not credible, for once the threat resides it is not in their advantage to follow the policies through. The only way for the elite to avoid revolution by credibly committing to policies they will not pursue when in charge is to surrender their de jure power, that is, to democratize. Crucial for this reasoning is that the threat of revolution is transitory (otherwise the citizens can force redistributional policies every period), that political institutions are not transitory (otherwise democratization would be equally incredible) and that elites are better off in democracies than after revolutions.
In the end, both the elite and the citizens make a cost–benefit calculation, and moves to democracy, revolutions and coups are the outcome of the relative costs attached to the different alternatives. What affects these costs, thereby affects the outcomes. This leads to the main empirical prediction: a non-monotonic relationship between inequality and democratization, where the latter takes place when the first is neither too high nor too low. In a relative equal but non-democratic society, citizens do not have much to gain by pressing for democracy and in a very unequal non-democratic society, elites have everything to lose and will find repression an option to think about. So, Singapore does not democratize because of its (relatively) high levels of equality, whereas South Africa's slightly decreasing inequality resulted in a move towards democracy.
Chapter 3 documents one part of the relationship by assessing that democracy is associated with lower inequality (as captured by the income share of labour in GNP and the Gini coefficient), and points out that democracy is more likely to arise after social-economic crises. This part forms the empirical motivation for the game-theoretic models proposed in Chapters 4–7. The models formalize the points already made. In the dynamic models, agents differ in their income, and therefore have different preferences over the policy variable in the model: the level of taxes. In a democracy, taxes are expected to be close to the level preferred by the median voter, while in a non-democracy the tax rate is set by the elite. Apart from the policy variable, elites can engage in (costly) repression or coups and citizens can choose costly revolutions. Outcomes differ, depending on the parameter values of inequality and costs. An extension of the workhorse model allows political identities to arise from other factors than income. This is relevant in for example South Africa. Then the racial dividing line and poor–rich dichotomy divides society in four groups rather than two. Still, the tax rate remains the sole policy variable throughout the analysis and identities are exogenous as it is left unexplained where they come from. These chapters are a joyful read for those who like game theory and do not mind some calculus, but may be safely skipped by those who want to go directly from the first three chapters to the three final ones, where different extensions are discussed.
Inequality is the main but far from only determinant of democratization. On the list of contributing factors are, among others, the presence of a middle class, globalization and economic structure. The middle class actually plays different roles. It may effectively press for partial democracy, and it can mitigate the redistributive effect of full democracy, thereby decreasing the cost of democracy for elites and, consequently, their inclination for coups. On the other hand, if the middle-class is relatively poor and presses for redistributive taxation, the poorest do not have much to gain by full democracy. A large middle class facilitates democratization, without being the necessary condition Barrington Moore, to whom the title of the book is a tribute, proposed it to be ('No bourgeoisie, no democracy'). Another factor entering the equation is economic structure. Repression and coups are more disruptive in capital-intensive than land-intensive economies. In addition to these cost-considerations, elites have more to loose in a democratizing agrarian society, as the production factor land can be taxed relatively easy — landowners are essentially lame ducks — whereas labour and especially capital are more elastic, thereby placing restrictions on the rate of taxation. Globalization makes capital more mobile, thereby decreasing the possibility to tax the elite. Again, this lowers the costs of democracy for elites. International trade — in theory equalizing local factor prices to world market prices — may also close the gap between poor and rich in non-democratic, poor societies, which are typically labour-intensive and scarce in capital.
This pretty much ends the story on the origins of democracy. So, let us return to the starting dilemma. The answer to the question how convincing the account is, first depends crucially on whether you deem deductive modelling a useful approach in political and historical research to begin with. Whoever holds that history is indeed one damn fact after another, probably will not like the one damn model after another approach. However, one not to be underestimated advantage of formal modelling is that models discipline our thinking by forcing internal consistency. Of course, internal consistency should not be confused with external validity. Or as the authors state, they 'do not disagree that democratizations (...), are tremendously complex social phenomena. (...) To develop any systematic understanding of the world, one must proceed by simplifying (...) and abstracting (...). Perhaps (...) we make the wrong decisions about which factors to emphasize and which to ignore, but whether we do can only be answered by the scientific and empirical usefulness of the theory and not by a priori assessments of how complicated the phenomena of democratization is'.
The main mechanism the authors propose and rigorously work out — that democracy is a solution to a commitment problem — is appealing. It is comprehensible without being trivial, and the models are intuitive and elegant while the examples are relevant and empirically meaningful. The mechanism is indeed very stylized, but then: reality is complex, that's why models shouldn't. Insofar, it definitely has earned a place among the social scientific mechanisms describing democratization that historians and political scientist will only ignore to their own disadvantage.
However, the aim to go beyond description and provide 'unified explanations' and 'general propositions' appears overstretched, for at least two reasons. First, the two-player model of society is meant to provide micro-foundations to the theory, but the opposite seems the result: assuming rationality of two theoretical entities at the macro-level without giving micro-foundations for this assumption. This not only ignores heterogeneity of agents, but also assumes that coordination problems and social hierarchies (divide-and-rule) do not exist. To be fully satisfying on a theoretical level, coordination within groups or classes should be derived, not assumed.
The second point concerns the empirical evidence. Although correlations are adequate for motivating a theory, in a historical account correlations, as the authors point out themselves, mean little and are in any case no substitute for careful econometric analysis. To make the case that the mechanism is general, it would even be more convincing if the model would give testable predictions about (future) events that could not possibly be known while constructing the theory. In the last chapter three reasons are given why the future of democracy is bright (increasing importance of human capital, globalization and the end of the cold War), but no predictions are made.
In the end, formal modelling aimed at reproducing stylized facts is not that different from the inductive historical method of for example B. Moore. The irony then is that whether or not the general mechanism applies in a particular situation has to be assessed on a case-by-case basis, which is another way of saying that it has to follow from a historical account. And that pretty much solves the dilemma: if you like the application of game theory to political history, you should read the book. If you prefer reading historical accounts, you probably shouldn't.
