Published by EH.NET (February 2005)

Carles Boix, Democracy and Redistribution. Cambridge: Cambridge University Press, 2003. xiii + 264 pp. $70 (hardback), ISBN: 0-521-82560-1.

Reviewed for EH.NET by Jari Eloranta, Department of History, Appalachian State University.

Carles Boix’s book is a path-breaking study of how and why democracies have come into existence, even into the dominant regime type of the twentieth century, combining insights from various social science fields. As such, the book attempts to answer four key questions: 1) How is stable democracy achieved? 2) Why do authoritarian regimes survive? 3) Why do revolutions take place? 4) What are the distributional consequences of various types of regimes? As he points out in the introductory chapter, despite the large body of theoretical study on these issues over centuries and the growing body of quantitative and qualitative analysis especially during the last fifty years, we still do not have “a convincing theory of political development and transitions” (p.1). His book first sets up a testable theoretical framework and then proceeds to engage in empirical analysis and discussion of extensions of the basic model. The introduction and Chapter 1 provide the key arguments of the study, whereas Chapters 2 and 3 delve into empirical evaluation of the model on the basis of (mainly) nineteenth and twentieth century data and cases. Moreover, Chapter 4 extends the argument to include the impact of economic growth, trade, and political institutions; Chapter 5 analyzes the interaction between democracy and the (size of) the public sector; and Chapter 6 examines the threat of rent expropriation vis-?-vis the state and development. Boix then concludes the book with a brief summary of the findings.

Why is Boix’s “unified theory” of political transition and regime survival relevant for economists and historians? To begin with, the model (based on game theory) developed in the book is remarkably clear and persuasive. In essence, he argues that political regimes (and transitions) are the outcome of the preferences and resources of actors in a given society. Thus he isolates income inequality, mobility of assets, and distribution of political resources as the key variables driving the emergence of democracies. The logic is as follows: High income equality and capital mobility are characteristics of a democracy, whereas high income inequality and asset specificity (for example, the presence of oil) provide support for an authoritarian regime. Moreover, a strong repressive elite implies a durable authoritarian regime, yet an abundance of political competition leads to revolutions and civil wars (p. 3). In addition, the transition from an authoritarian regime to a democratic one can be peaceful if there is only a moderate level of economic inequality or the assets are highly mobile (p. 13). Boix argues that the challenge to the ruling elites comes from the lower classes, who wish to change the redistributional equilibrium in the society (p. 171). He perceives three possible redistributional outcomes: 1) Right-wing authoritarian regimes, in which practically no transfers take place (also, limited taxes); 2) Democracies, which have more redistribution of wealth, based on the aspirations of the median voter; 3) Revolutionary regimes, which are prone to the follies of nationalization and central planning, eventually leading to corruption and economic stagnation (p. 3).

The historical dimensions and possibilities of these problems are duly recognized by Boix, including the use of both the theoretical and empirical contributions of notable economic historians, such as Jeffrey Williamson and Kevin O’Rourke. Nonetheless, although at one point he maintains that certain aggregate correlations between the key variables (for example, between economic development and democracy) may not hold for the pre-Second World War era (p. 2), he later claims to have developed a model that explains the existence and evolution of regimes from Classical Athens through the Middle Ages all the way to the twenty-first century (for example, p. 20). This is certainly an ambitious endeavor and cannot be verified with such a cursory and selective survey of historical cases (for example, Chapter 3). In general, Boix criticizes earlier research for being too general in scope (i.e., Huntington, Dahl), and for not being able to go beyond the analysis of simple correlations when we should be looking into the role of other, theoretically more relevant variables. He illustrates this by focusing on the three key variables (inequality, asset specificity, repression costs), citing the U.K. as an example, where in fact we see a dramatic decline in inequality during the nineteenth century (p. 37-38) and universal male suffrage finally being introduced in 1918 (similar to many other countries after the First World War). Yet, is this a credible model for understanding political transitions? Can we test it?

I do not have much to criticize in terms of the model itself (for example, the game theoretical modeling or the underlying assumptions), except when it comes to the institutional aspects of the theoretical framework. However, the empirical testing of the model has both strengths and weaknesses. How does Boix go about doing that? In general, he utilizes panel data regressions extensively, with numerous combinations of dependent (usually a binary variable, whether country was or was not a democracy in a given year) and independent variables (usually various proxies pertaining to income inequality, asset specificity, and political mobilization), and in fact calculates yearly probabilities of democratic transition for the period 1950-1990. The second data set covers the period 1850-1980, with which he, by and large, replicates the analysis for a longer time period in order to investigate the causes of the move towards democracy by most Western nations in the beginning of the twentieth century. He subsequently utilizes more panel data regressions to analyze the redistributional hypotheses arising from his theory in the later chapters.

The first problem with Boix’s analysis pertains to the choice of data and definitions. For example, he employs a fairly generous definition of democracy, based on three characteristics: 1) Legislature elected in free multiparty elections; 2) Executive directly or indirectly elected in popular elections, and responsible to voters or legislature; 3) Majority (at least 50 percent of adult men) allowed to vote (p. 66). Curiously enough, Boix does not discuss other alternatives, such as the Polity, Freedom House, Dahl, Vanhanen etc. data.[1] In addition, the underlying assumptions as to what constitutes a democracy vary greatly among such data sets and scholars. For example, the Polity data provides much more variability in the dependent variable (score ranging from -10, complete autocracy, to 10, full democracy).

Furthermore, the use of various types of proxies, especially in the analysis of the longer period, for the independent variables is problematic. Even for the more recent period, Boix has to rely on proxies for the measurement of asset specificity, thus utilizing agricultural share of GDP (land immobility), ratio of fuel exports over total exports (fixed energy production), years of schooling (human capital), and economic concentration as variables. This raises several issues that cannot all, however, be discussed here due to space constraints. For example, how mobile is human capital? Why? Or, as Boix several times attempts to explain away, can GDP per capita really be thought of as a “first approximation of asset specificity” (p. 82)? He seems determined to maintain throughout the volume that income is not a relevant measure to explain any aspect of the phenomena under evaluation (see e.g. pp. 82, 91-92, 128) in comparison with income inequality, for instance. He also does not give any justification for his choice of control variables or dedicate too much attention to analyzing them, even though the role of religion as an explanatory force (which even a cursory perusal of the regression statistics would suggest) could have been investigated further (p. 78). Moreover, why measure pre-Second World War income inequality via the distribution of agrarian property and the quality of human capital? Or, how were the Angus Maddison income data combined with the Penn World Tables figures (p. 90)? Despite these weaknesses, the results seem to support the underlying model quite well and are certainly persuasive at least.

Chapter 3, however, does not leave the reader as confident. First of all, this chapter, which claims to compliment the statistical analyses with a historical/institutional overview of the progress toward democracy of the Swiss cantons (from the Middle Ages to the mid-nineteenth century) and the United States (from the colonial times to the twentieth century), is quite short in comparison with some of the other sections of the book. Second, there seems to be no clear justification for focusing on just these two cases per se, or on these two particular time periods. The discussion of racial voting barriers in the U.S. case is woefully brief and should have been linked to the discussion of the problems inherent in many of the democracy indices. For example, Boix’s definition does consider universal suffrage as an essential component in a democracy. In fact, Switzerland, one of the cases scoring a full 10 on the Polity scale (like the United States every since year 1800) throughout the nineteenth century, did not allow women the right to vote until 1973! Thus, following Juan Linz[2], I think it would strengthen the empirical dimension of the book to allow for more flexibility in analyzing and interpreting the regime types, such as using the sliding Polity scale. Furthermore, it would be interesting to test for certain threshold levels for the various variables, since it is very difficult to conceptualize low versus moderate levels of income inequality or variability between limited versus full democracies.

The theoretical insights of the book on institutions also leave something to be desired. For example, Boix claims that institutions (formal or informal) do not matter in the progress towards democracy if income equality and asset mobility are both high (p. 15). I would argue that institutions are a key reason why this would be the case in the first place. Furthermore, he claims that “whether institutions play any role in ensuring democratic outcomes depends mainly on whether they modify the underlying balance of power among any of the contending parties” (p. 144). Again, I find this to be a curiously minimalist statement given the extensive work by Avner Greif, Douglass C. North et al., who view institutions (especially informal) as a key variable explaining why economic interaction is possible at all. Boix tries to incorporate an element of institutional analysis, for example, in the section attempting explain the size of general and central government (revenue, total spending, nonmilitary spending etc.). However, the use of the democracy variable or constitutional dummies is simply not enough (p. 179), especially to substantiate his claims on the role of institutions. Moreover, although the regressions on government spending seem fairly robust overall, their R-squared values are often low. The introduction of other variables, arising from numerous competing explanations on the size of government, might improve the fit. In addition, the author does not spend much time discussing, for example, panel data techniques (he uses OLS with panel corrected standard errors), possible specification problems, or the risk of spurious correlation due to, for example, the possible presence of unit roots. Finally, Chapter 6 on rent expropriation leaves this reader somewhat confused — why is rent seeking necessarily synonymous with corruption? Or, concerning the concluding chapter, what are the concrete policy recommendations arising from this, at times brilliant, study?

In the aggregate, this study is definitely a worthwhile read, especially its theoretical section. The notion that income inequality, the mobility of assets, and the distribution of political resources are the key variables driving the transitions to democracy is persuasive. Boix’s strength lies in his clarity of expression and style, as well as his ability to bring forth a logical, testable model. However, some of the choices made in the testing of these hypotheses can be challenged and decrease the overall value of the empirical findings. Nonetheless, Boix’s model can certainly be applied to various historical cases and periods. Yet, as usual, only the lack of data stands in our way.


1. See especially Robert A. Dahl, On Democracy. New Haven: Yale University Press, 1998; Polity IV website:; Freedom House data:; Tatu Vanhanen, “A New Dataset for Measuring Democracy, 1810-1998” (2000) 37:2, Journal of Peace Research.

2. See, for example, Juan J. Linz, Totalitarian and Authoritarian Regimes, Lynne Rienner Publishers: Boulder, Colorado, 2000.

Jari Eloranta is assistant professor of economic and business history in the Department of History, Appalachian State University, Boone, North Carolina. His research interests include corporate political action in the long run, defense economics and the financing of wars, as well as the analysis of government spending in the nineteenth and twentieth centuries. Recent publications include: (with Juha-Antti Lamberg, Mika Skippari and Saku M?kinen) “The Evolution of Corporate Political Action: A Framework for Processual Analysis” (2004) Business and Society; “Filling the Void? Implications of Hegemonic Competition and the Lack of American Military Leadership on the Military Spending of European Democracies, 1920-1938,” History and Change, edited by Kirsi Vainio-Korhonen and Anu Laitinen, SKS: Turku, 2004; and “Warfare and Welfare? Understanding Nineteenth and Twentieth Century Central Government Spending,” Studying Economic Growth: New Tools and Perspectives, edited by Peter Vikstr?m, Occasional Papers in Economic History, Ume? University, 2004.