I’m going to revisit with you the study of democratization as I myself have experienced it, since one learns by looking back at what one did not see at the time. Basically I’m going to give you two themes: one is the study of transitions to democracy and the other is the current thinking about economic reforms which, I think, is very rapidly being transformed by the understanding that economic reforms must meet political reforms.

Personal/historical background

Let me start by telling you how I got involved. I’m Polish and I got involved with democratization for the first time by being beaten by police in 1957 at a student demonstration when the government closed a student newspaper. I left Poland; I came here; I went to Chile; spent some time in Chile; saw democracy being destroyed there; and came back to the United States. In 1978 I received a phone call from a friend, Philippe Schmitter, in which he invited me to join a project which he and Guillermo O’Donnell were organizing under the auspices of the Wilson Center to study the possibility of liberalization and perhaps even democratization of authoritarian regimes in southern Europe and Latin America. This was just after Portugal and Greece became democracies. Portugal had abolished the Salazar authoritarian system, Greece became a full-fledged democracy, and Spain was in the process of transition, while in Latin America ferocious military dictatorships still reigned. About 40 people were there from many countries, including the United States. The first thing that struck me was that nobody even mentioned the names of Barrington Moore or Seymour Martin Lipset, the two major thinkers at that time on democracy and democratization who had previously dominated the intellectual perspective. That’s what we all read, that’s what we all studied, that’s what we all wrote about, and yet here you had a group of 40 scholars who discussed the topic and the work without even mentioning them. Why was that? Because both of these perspectives were just too deterministic for our taste. If you remember, Moore says that whether a country ends up being a democracy or a dictatorship depends on the agrarian class structure a couple of centuries ago. And Lipset says that a country has to develop, it has to modernize; only then can democracy come about.

“We wanted to know what movements in different countries could do to bring dictatorships down rather than simply wait.”

Now this group of people were militants for democracy in one form or another, at least vicariously. And the notion that we would have to wait for the conditions to produce it and that there was nothing for us to do was untenable. I think that is why these names did not appear and why the whole project was cast very much in the vein of what somebody dubbed as “thoughtful wishing.” We wanted to know what movements in different countries could do to bring dictatorships down rather than simply wait.

But even though the mood was one of open inquiry, the dominant intellectual perspective at this time was hostile to any kind of analytic or formalized strategic thinking. Probably the dominant theoretical and methodological perspective, in that group at least, was something called in Spanish the historical structural method, el método histórico structural. That method proceeded from the analysis of the economic structure of societies, to an analysis of interests, to an analysis of groups, to an analysis of alliances. So it was strategic and interest-based and asked questions about alliances. Yet, there was a kind of emphasis and insistence that rejected any notion of studying these things with some kind of abstract approach. A year before the Wilson Center conference I was in a meeting with Fernando Henrique Cardoso, now president of Brazil, and he was outlining and analyzing Latin American countries: given these groups and those interests, these were the coalitions we could observe. I asked him: “How do you know that with these interests you get these coalitions.” He said, “Oh, you’re just looking for empty formulas.” So this is the sort of spirit predominating at that time, and what happened in the project was that the classificatory approach prevailed, which was something my friend Jon Elster eventually called the “botanical approach.” That is, the question was: what were the boxes and how would you put the countries into the right boxes?

The first hypothesis that was formulated at the meeting was that the modes of transition depend on types of authoritarian regime. And the second was that the eventual result depends on modes of transition. So we had types of authoritarian regimes, modes of transition, and then eventual outcomes; the question was how to best to classify them. And a competition developed in which big prizes were given for suggesting categories that other people would accept and small prizes were given for putting countries into the particular boxes. The great winner in this competition was what was not accidentally called in Spanish ruptura pactada, negotiated breakthrough, since this was the only feasible pattern to follow. Much of the discussion concerned conditions for making this negotiated breakthrough possible, as well as actual strategies for negotiation.

“We were concerned first and foremost about stopping the killing; that’s really what democracy meant for us.”

This neglect of long-term processes, as well as our willingness to compromise, eventually evoked widespread and sometimes virulent political and intellectual criticism. Both the Wilson Center volumes—which I learned recently are cited six times as often as any other work on democratization—and my own book, published in 1991, have been subject to that criticism.1 I have to say that I am not sympathetic to the political criticism. Most of the people engaged in that project were socialists of one kind or another. The understanding that the only way to bring dictatorships down and to establish some kind of system of rule of law entailed possible compromises in the social and economic realm was painful to many of us. We knew what was at stake and we were prudent, justifiably so I think. It is very easy in retrospect to charge “betrayal,” now that the dictatorships are overthrown and some of the perpetrator of the crimes are or were in jail. You have to remember that this was a time when Franco was still killing miners in Asturias; the Argentine dictatorship was killing randomly; the Chilean dictatorship was killing systematically. We were concerned first and foremost about stopping the killing; that’s really what democracy meant for us. And we were ready to compromise.

Dynamics of dictatorship

The intellectual issue, which is much more complicated, is: how does one think about the combination of long-term factors and short-term ones? The way to think about it is that you have a dictatorship and the dictatorship evolves and it has its own dynamic. The strategic situation is such that some forces rule at the time, and some forces make up the opposition. But the strategic structure of the situation is such that whoever is to be the winner would want to rule by force—or at least dictators do.

What must be happening in these long-term processes is that somehow the structure of the situation is altered in such a way that a negotiated solution, some kind of compromise, suddenly becomes possible. To apply a game-theoretic framework, the payoffs to the dictator from maintaining the monopoly of power change in such a way that if a dictatorship can negotiate its way out with guarantees, then that solution is preferable to an overthrow.

So the question becomes: what are these processes? Modernization theory points to long-term economic development. Many of us believe short-term economic crises play a role. It could be the death of the founding dictator. Some regimes are just as stable as their founder (Franco). In some cases it’s exhaustion in a civil war, as in El Salvador. In some cases it’s probably international pressure, which certainly played a role in South Africa. In other cases it is purely geopolitical, as in Taiwan. Evidently, dictatorships run many competing risks; they can die for all kinds of reasons. The result is that it’s very hard, if not impossible, to predict transitions to democracy. I’ve gathered lots of statistics on the topic, trying to include anything I can observe—economic factors, political factors, and so on—to statistically predict whether dictatorships will fall. I can tell you that the level of economic development, at least as measured by per capita income, predicts zero transitions. And if you bring in the whole kitchen sink, including cultural variables, and all kinds of economic variables, you predict two out of 49 collapses that occurred between 1950 and 1990.

If we don’t have a good long-term view of transitions, do we have a better short-term view? Let me give you a little game theory because there is something really puzzling about this problem. Say there is a dictatorship and the dictator has the following choice: he can try to open the system, or keep the status quo. This is the opening, “liberalization” as it was called at that time. The dictator offers a limited form of participation to some groups of civilians who are not part of the power group, in exchange for their support. Or, there is no opening and the result is the status quo. If the dictator opens, then the opposition moves. And the opposition can either accept this opening and “enter” or it can mobilize. If the opposition enters, the result is a broader dictatorship. If the opposition mobilizes, then the dictatorship has to decide what to do. It can repress, which I treat as keeping the status quo, or it can yield and the result is then transition.

There are some forces, at least in the dictatorship, that prefer the broader dictatorship, but dictators are not democrats so they prefer the status quo to transitions. Now, the opposition prefers transition to the broader dictatorship to the status quo. If the opposition mobilizes, the dictator will go for repression. But if the opposition knows that the result is going to be repression, the opposition will go for a broader dictatorship which is preferable to the status quo. If the dictator knows that the opposition will go for broader dictatorship, and since the dictator prefers broader dictatorship to the status quo, the dictator will open. Result: the dictator opens, the opposition enters, and you get a broader dictatorship. You don’t get transition.

Suppose that the dictator cannot repress, for one reason or another. What happens then? If the dictator opens, the opposition mobilizes; it mobilizes into transition, which it prefers to broader dictatorship. If the opposition is going to go to transition, and since the dictator prefers status quo to transition, the dictator will never open. What happens then? As you see, if everybody knows everything, transition never occurs. Something is obviously wrong. Two facts, I think, are illuminating.

(1) Dictators tend to believe that the genie can be put back into the bottle. They always somehow think that they can open and yet repress, that they will not lose control. They always believe that they’re going to leave the door ajar, let whoever they want in, and then close it whenever they want. It turns out that they can rarely do that. We had periods which were euphemistically called normalization—Poland in 1957 or 1981, and the Dominican Republic in 1965—but they are rare. Once you get the cork out of the champagne bottle the dictatorship is not going to hold.

(2) Once the dictatorship is already at the point of holding elections, they are persuaded they are going to win. Pinochet, when the votes were being counted in the plebiscite of 1989, still thought he was going to win. The Sandinistas in 1990 certainly thought they were going to win until the last moment. In June 1989, right before the Polish elections, I was assured by a friend who was a reformed communist that the communists were going to win more than half the seats in the competitive elections for the Polish senate. Half of the 100 seats. You know how many they won? Zero.

“I’m persuaded that the fall of communism was to a large extent due to a succession of small mistakes.”

This is obviously not the stuff of which rational choice is made. It’s a peculiar kind of situation in which game theory is very useful because it tells you that you cannot have a situation in which everybody behaves rationally or strategically and everybody knows what everybody else is going to do. Whether you are speaking of the French monarchy or of communism, the fall of the regime seems inconceivable. If you’re the king of the dynasty which rules by the grace of God for four centuries, you don’t think that if you make this move it may endanger the regime. Or if you represent the force which is the future of mankind, you don’t think that making a little mistake and choosing this political strategy or that political strategy is going to eventually reach a point where you have no choice but to abdicate. And yet that is exactly what happens. I’m persuaded that the fall of communism was to a large extent due to a succession of small mistakes.

Transitions to democracy

We still don’t have either a good long-term explanation of transitions to democracy, or a good short-term story about how it really happens, about the dynamic of these processes. But more than that is at stake. Look at that literature now and you’ll be struck by the fact that other than in the phrase “transitions to,” the word democracy seldom appears. And the reason is that we were not thinking about it. These were extraordinarily tense times. Lives were at stake and what happens is that your perspective is shortened. What’s most important is what has just happened, because what has just happened portends the future. The “current conjuncture” is the operative concept in these kinds of situations. I remember many times landing somewhere at an airport and being picked up by a friend and told, “Look what happened yesterday, General so-and-so just said that…” And then I went to conferences and people presented papers where the main theoretical problem was always what happened yesterday. They wrote papers, the papers were eventually published as an acquittal of academic responsibility, and nobody ever read them because three months later the most theoretical problem of the day was different, because yesterday was different already.

So we were not looking forward, we were chasing the tail of events rather than looking to see where they would end. And note that the title of those Wilson Center volumes is Transitions from Authoritarian Rule. Guillermo O’Donnell has a couple of recent papers in which he attacks the notion of “transitions to” as teleological. But I think he confuses the fact that we may not know where we are going with the fact that we knew where we wanted to go. Nothing tells us that where Latin American or Eastern European countries are eventually going to end up will be the same place where the United States, or France, or England were. What may happen, and I am going to argue that perhaps it is happening, is the formation of hybrid political systems which are unprecedented. Nevertheless, we should not forget that what animated most people was some vision of democracy. In understanding the dynamic of these processes we have to take into account the fact that people wanted to go in a particular direction, and that itself is an element that structures the process. The result of this myopia was that we were eventually—and still are, perhaps—surprised by what, in Spanish, is called the desencanto, disenchantment. The progression seems to go from liberalization, to transition, to consolidation, to disenchantment.

Disenchantment with democracy. Why? Because as an ideal it seems more attractive than the reality, but also because all kinds of new problems are raised. Indeed, the “democratization” crowd has a new catchword and the new catchword is the “quality of democracy.” That’s what the conferences are about now—no longer liberalization, nor transition, nor consolidation; now they’re about the quality of democracy. And I think they should be. Not only in Latin America and Eastern Europe, but in the developed democracies as well. I think this is an agenda that is being pursued and will be pursued in years to come. We don’t know whether contested elections with all their attendant conditions (civil liberties, political freedom, freedom of the press) lead to decisions that are rational, to allocations that are in any sense just, to governments that are representative.

We don’t know whether they give citizens control of government, whether they lead to accountability, or whether they lead to equality in the social and economic realm which, for about 80 percent of Eastern Europeans and about 70 percent of Brazilians and Chileans is what democracy means. What we suspect is that different institutional arrangements matter. A research program needs to deal with the impact of different institutional arrangements in all systems that we recognize as democracies whose rulers are elected by contested elections.

Democracy is a system of positive rights, both negative and positive rights in the political realm, but there’s nothing about democracy that guarantees that individuals have the effective capacity to exercise those rights. John Stuart Mill says somewhere that without decent wages and universal reading no government of public opinion is possible. But there is nothing about democracy per se which guarantees decent wages and universal reading. What was the 19th century solution to this problem? It was to restrict citizenship to people who had the effective capacity to exercise these rights. We don’t do that. We give these positive rights to everybody. Yet we don’t necessarily create the conditions for the effective exercise of those rights. And the result may be a democracy without effective citizenship. If these conditions are to be created they need an active and extensive role of the state.

Role of the state

Let me briefly go over what I think is the current state of political economy or the debates on the proper role of the state. The historical sketch of this problematic starts with the notion of a world of perfect markets. Markets allocate all resources efficiently. Distributional problems are not serious because if you don’t like a particular allocation by the market you can simply redistribute resources and make markets work. You then arrive at a solution which is going to be both efficient and just according to whatever criteria we use in distributing income. In this world there is no role for the state to play since markets do everything perfectly. Or, the state is assumed to have created whatever conditions are necessary for the markets to operate.

Markets work most of the time, but not all of the time; sometimes markets fail, markets have imperfections. If markets fail, then there is something for the state to do to correct market imperfections. In 1959 the German Social Democratic Party abandoned its commitment to Marxism under the slogan “markets whenever possible, the state whenever necessary.” So there is the notion that markets do what they do well; whenever markets fail, the state comes in. This worked until the mid-1970s when my former colleague from Chicago, the late George Steigler, asked why we would assume that, if markets failed, states would do any better. Why wouldn’t the state fail as well? Moreover, why would we assume that when people act in the economic realm they act in self-interest, but that the same people, once they become public officials, suddenly become these benevolent public servants? Why wouldn’t we assume that once in government they would also act selfishly? And indeed this is the cornerstone of the neoliberal theory, this distrust of the state.

“The state doesn’t know any more than private agents and yet there are all kinds of state intervention which improves welfare in a Pareto way, that is, make anybody better off without anybody being worse off.”

Information provides a newer round of the debate. Once you abandon the assumption that everybody knows everything and the assumption that there are markets for all commodities now and in the future—and both of these assumptions are untenable—then you don’t even have a presumption that markets work. There is a brilliant, important recent book by Joseph Stiglitz, Whither Socialism, 2 which is the best summary of this whole argument. It becomes very clear that there is a role for the state to play. Moreover, it’s not the kind of role we used to dream about before, in the sense that in many of the capitalist-socialist debates it was assumed that the state knew what private economic agents didn’t know. Now, the state doesn’t know any more than private agents and yet there are all kinds of state intervention which improves welfare in a Pareto way, that is, make anybody better off without anybody being worse off.

It is very clear in the economic literature today that there is a very strong presumption of a positive role of the state, that is, there are things that governments can do to improve welfare. Nevertheless, the effects of the neoliberal punch linger, because whatever the potential role of the state is, the question still remains: will governments do what they should and not do what they should not? The current round of this debate is about the reform of the state. The answer to this question is to start thinking about how to design political institutions in such a way as to enable the state to do what it should and prevent it from doing what it should not.


“The quality of state intervention depends on the mechanism by which the state is made accountable to those on behalf of whom it is supposed to function.”

The hardest item on that agenda is reform of the state. Country after country is asking that question. My own spin on it is that states will intervene effectively if they’re subjected to external popular control. The quality of state intervention depends on the mechanism by which the state is made accountable to those on behalf of whom it is supposed to function. There are at least two dimensions involved. One is mechanisms for oversight by politicians of bureaucrats. By politicians I mean those who have been elected and by bureaucrats I only mean public officials who are not elected. How can the elected politicians control the bureaucrats? A second question is, how can voters control politicians? Both of these are highly problematic. Moreover, politicians don’t necessarily know how well the bureaucracy is functioning. They don’t know how long you’ve waited in the post office, whether the teacher was absent or present, whether the cop took bribes. We the citizens are the recipients of these services and we have that knowledge. For that reason, one direction the reform of the state can take, and is taking, is in building institutions that allow and organize and aggregate information from citizens about the functioning of bureaucracy. There are all kinds of alternative arrangements being discussed. But they are all of this generic nature: how to organize forms of popular control and at least popular aggregations of citizens’ information about the function of bureaucracy.

The second problem is accountability by politicians. If voters know everything, then they know they can set up a system in which they say to the politicians: if you do this and this, we will vote for you; otherwise we will not vote for you. And indeed you can show that this can be an effective instrument of control. But obviously, citizens don’t learn everything, they can’t know everything. They just don’t know enough. Therefore, the direction of reform should strongly emphasize independent sources of information. I’ll give you one example. Every time anybody makes a contribution to a politician it appears on the first page of the newspaper. You can get a list of who gets paid by whom. Or you have an independent electoral commission with its own investigatory powers. For example, the Argentines just passed a constitutional amendment establishing an independent accounting office like the General Accounting Office, but one that is independent, the head of which is to be appointed by the largest minority in the senate. This gives the opposition an investigatory arm to see what the government is doing.

We’ve had very little institutional innovation during the past two hundred years. Think about the major institutional innovations during the past two hundred years in the political realm, where somebody thought of some kind of institutional setup and it eventually became accepted somewhere and utilized. Perhaps the most recent of them is the system of proportional representation which was developed in England and Denmark in the 1860s. Otherwise, if you think of major innovations of the past, there haven’t been that many. If you look at the experience of new democracies you will find that, typically, if they had a democratic regime in the past they just adopted the old constitution, even if it didn’t work in the past, like the Argentines. They used the 1853 constitution which led to eight coups d’état.

Now that we have a global world with global institutions, the question of institutional reform takes on a larger meaning. We know very little and understand next to nothing about “globalization.” All we have so far is slogans and anecdotes. But we do know that the supra-national question is alarming from the point of view of democratic theory. We have these bodies that are not accountable to anybody, anywhere. This is the sort of thing we are going to be thinking about.

Adam Przeworski is professor of politics and economics professor as well as the Carroll and Milton Petrie Professor of European Studies at New York University. He has written extensively on democracy. In addition, he is coauthor of the SSRC’s much-used pamphlet On the Art of Writing Proposals.

This essay originally appeared in Items/Items & Issues Vol. 51, No. 1 in the Spring of 1997. Visit our archives to view the original as it first appeared in the print editions of Items.

Posted on February 7, 2017