Guillermo O'Donnell was one of the most creative social scientists of the past generation. His relatively early work concerned Argentine politics and the reasons for persistent democratic breakdown in that country. He went on to propose a structural theory of the spread of military dictatorship throughout South America during the 1960s and 1970s. These military regimes came to be known as ‘bureaucratic authoritarian’ regimes. Although ultimately flawed, it is fair to say that this idea raised the study of authoritarian government in the region to a new level. However, just when the flaws of this theory were beginning to show, the majority of Latin American countries moved to democratise. O'Donnell, along with Philippe Schmitter and Laurence Whitehead, then produced a major, path-breaking work on pacted democratic transitions. As if this were not achievement enough for one career, O'Donnell then went on to pioneer the study of issues of democratic accountability, coining the phrases ‘delegative democracy’ and ‘vertical and horizontal’ forms of accountability. He has also written perceptively on weaknesses in the rule of law in Lain America. Just about everything O'Donnell has written on Latin America in the past generation, therefore, is worth taking seriously, although some of it is controversial.

Now O'Donnell has attempted a work of political theory dealing with democracy as a concept. It must be said at the outset that this is a difficult book, erudite in places but also surprisingly tentative in others. Its main claim to originality lies in his discussion of agency. This term is not quite defined in the work but it seems to mean something like democratic effectiveness. How can a voter in a democracy make sure that his/her vote is respected by those who win elective office? It is far from uncommon, not least in Latin America, for successful electoral candidates to ‘bait and switch’, that is, to say one thing to get elected and do quite another when in office.

Discussion in this work is sometimes cloudy but always interesting. On p. 203 O'Donnell admits that his book has ‘traversed a winding road’. If anything, this is understatement. Nevertheless, the book is worth reading for its digressions and its wonderfully erudite footnotes, even though they can detract from the main argument, which is sometimes foggy.

There is also only one chapter on Latin America. This makes the important but by now well-established point that public confidence in democratic institutions (including state institutions) in the region tends to be low, despite which democracy in the region seems mostly secure albeit with some crumbling at the Andean edges. Latin American respondents to surveys show a culture that is assertive of the virtues of citizenship and rights, side by side with a lack of confidence in institutions. This might prove unsustainable. Without at least some degree of institutional effectiveness, the ability of the popular vote to shape or discipline what governments actually do is likely to be a matter of contingency at best. O'Donnell finds it paradoxical that more than a generation of democratic governance has not brought these two conflicting features (widespread citizen distrust of government but equally widespread support for democracy as a concept) closer together. What is also noteworthy is that there is less variance in the pattern of public opinion, as revealed by surveys, than there are different patterns of organisation in the different Latin American countries. The ‘Latin American’ pattern of high levels of distrust of parties, legislatures, the legal process and even electoral tribunals, combined with support for ‘strong’ leaders, can be seen not only in countries that are ‘delegative democracies’ but also in countries such as Brazil where checks and balances are real. One can even see a similar pattern in countries such as Mexico that have moved from hyper-presidential to weak presidential systems and in countries such as Chile and Uruguay which would be regarded by most academics as consolidated democracies. This might indicate a problem of democratic vulnerability, but democracy in the region seems to be becoming more rather than less secure.

As he freely admits, O'Donnell does not know quite what to make of these apparent contradictions. He intimates that something has got to give. Either democratic institutions will become more respected by the population or else the democratic progress of the region in the last generation will be thrown into reverse. The more optimistic of the two potential outcomes is possibly the more likely. O'Donnell does mention one significant point, which is a broadening of participation that was not widely anticipated at the time of democratisation. This is true both of the enhanced role of social movements in countries such as Bolivia and the emergence of unlikely figures as individual political leaders in much of the region. A generation ago it would have seemed fanciful to suppose that in 2012 South American presidents would include two women, two left-of-centre military officers (both ex-golpistas), one coca grower, one ex-guerrilla and one uncelibate priest. There are admittedly a few besuited conservatives too, but they seem almost like an endangered species. It is surely one consequence of the general distrust of political institutions in the region that the presidential palace has become accessible to a much wider range of individuals, but it may at the same time underpin democracy in a symbolic way. More could have been made of this point.

When considering the variants of democracy in the wider world, something may depend on economic performance. It is clear that electorates the world over hold their governments responsible for economic setbacks. It may be, moreover, that the politics of economic setback will be less of an issue in Latin America during the next decade and more of one in Europe.