by A Przeworski · Cited by 266 — Adam Przeworski. Department of Politics. New York University. I examine the mutual relation between political regimes and economic development. An analysis.
43 KB – 27 Pages
PAGE – 1 ============
Democracy and Economic Development* Adam Przeworski Department of Politics New York University I examine the mutual relation between political regimes and economic development. An analysis of regime dynamics shows that while the paths to democracy are varied, once established for whatever reasons, democracies survive in developed countries. Contrary to long-standing arguments, political regimes do not affect the rate of investment and of the growth of total income. But since population grows faster under dictatorships, per capita incomes increase more rapidly under democracies. In the end, there is not a single reason to sacrifice democracy at the altar of development. * Published in Edward D. Mansfield and Richard Sisson (eds.), Political Science and the Public Interest (Columbus: Ohio State University Press). This is a revised version of a paper originally written for the United Nations Development Program, with whose permission it is published here.
PAGE – 2 ============
2 Introduction I examine the mutual relation between political regimes and economic growth. Two questions are discussed: (1) Whether economic development affects the emergence and the survival of political regimes and (2) Whether political regimes affect economic performance. These two questions are inextricably connected. To determine whether political regimes affect economic performance, we must first ask how political regimes emerge and endure. Unless this question is posed first, we will be unable to distinguish the effect of the conditions under which political regimes find themselves from the effect of regimes. Suppose that you were to observe that in 1985 per capita income of Mali, dictatorship, grew at the rate of 5.35 percent. Would the rate of growth of Mali in 1985 been different had it been a democracy? This is what we want to know when we ask about the impact of political regimes on growth. But we do not observe 1985-Mali as a democracy, only as a dictatorship. True, we could look for a case that was like 1985-Mali in every aspect other than its political regime. But what are we to do if we cannot find a democracy like Mali-in-1985? As you probably already know, and will soon learn again, democracies are very rare in poor countries, such as Mali, which in 1985 had a per capita income of $532. 1 In turn, we may observe that 1985 France, which was a democracy and had per capita income of $12,206, grew at the rate of 1.43 percent. Was its growth slow because it was a democracy? Again, we may try to find a dictatorship that would look in all respects like France. But the wealthiest dictatorship we observed between 1951 and 1990, Singapore, had per capita income of $11,698. Hence, we will not find a single case of a dictatorship as wealthy as France during the same time span. Why does it matter? Suppose that poor countries in general grow slower than wealthy countries. Since most poor countries are dictatorships and all wealthy countries are democracies, we will conclude that economic growth is faster under democracies. But this will be an invalid conclusion: the difference will be due to conditions under which these regimes exist, not to anything they do. As another example, consider the possibility that democracies are vulnerable to economic crises, while dictatorships survive them. Again, if we were to just compare the growth rates observed under the two regimes, we would conclude that democracies grow faster. And, again, this conclusion would be erroneous: we will have observed this difference only because democracies died when they encountered bad economic conditions and became dictatorships capable of surviving survived under these conditions. Finally, consider the possibility that there is some factor which cannot be observed systematically and which affects both the political regime and the rate of growth. Enlightened leaders, for example, may opt for democracy and well manage the economy. If we rely on comparisons of the observed cases, we will–yet again erroneously–conclude that faster growth is due to democracy, rather than to the enlightened leadership. 1 All income figures are in 1985 purchasing power parity dollars. For the period 1951-1990, they are based on Penn World Tables, release 5.6. For the period 1991-1999, they are taken from World Development Indicators 2001. These series overlap between 1970 and 1990 and the correlation between per capita income figures derived from these two series is 0.998.
PAGE – 3 ============
3 To distinguish the conditions under which regimes find themselves from the effects of these regimes, therefore, we need to understand under which circumstances these regimes come into being and under which circumstances they survive and die. Only then can we isolate the impact of regimes on economic performance. Statistical methods for doing this were developed by Heckman (1976, 1988) and the reader can find their explanation in Przeworski et al.(2000). The plan of this chapter is thus the following. In the next section, I briefly state what I mean by democracy and dictatorship. Then I summarize results concerning regime dynamics. Finally, I proceed to the central question, namely, the impact of political regimes on economic performance. To keep the discussion free of technical issues, I present some descriptive tables and only report conclusions of statistical findings. The reader who wishes to know how these results were derived is directed to Przeworski et al.(2000), from which this discussion borrows. Most of the results summarized below are based on observing 135 countries (excluding major oil producers) between 1950 and either 1990 or 1999.2 Democracies and Dictatorships The prototype of democratic politics that underlies this analysis reflects Schumpeter’s (1942) focus on filling governmental offices by elections. Democracy is a political regime in which rulers are selected through free and contested elections. Operationally, democracy is a regime in which incumbents lose elections and leave office if they do. Hence, my definition of democracy is Schumpeterian or fielectoralist.fl Dictatorships are treated as a residual category, finon-democracies.fl Obviously, one can argue about definitions, and indeed there is a vast literature attempting to define what we should deem to constitute a democratic regime. But while one can, and some do, engage in endless hair splitting, most people readily agree whether a particular country is a democracy. Inkeless (1990: 5-6) found that the scales of democracy developed independently by Gastil (1990), Coppedge and Reinicke (1990), Bollen (1980), and Gurr (1990) are all highly correlated. Przeworski et al. (2000: 56) reported that their dichotomous classification of regimes is almost perfectly predicted by all these scales. Any two informed persons asked whether a particular country has a democratic regime during a particular year would almost certainly say the same, even if no criteria were specified in advance to influence their judgments. Hence, the results presented below do not depend on the particular classification of political regimes. Economic Development and Regime Dynamics Introduction 2 The analyses concerning regime dynamics cover the period 1950-1999, so they update Przeworski et al. (2000). Unfortunately, some of the economic analyses could not be updated because the required data are not available.
PAGE – 4 ============
4Any casual observer will note that democracies are rare in poor countries and frequent in the affluent ones. Between 1950 and 1999, of the 1,335 annual observations of countries with per capita income under $1,000, we observed only 142 years of democracy. Of the 880 annual observations of countries with incomes above $8,000, only 147 years were spent under dictatorships. Indeed, if one takes per capita income alone, one will correctly predict 75 percent of the 5179 annual observations of regimes. There are two potential reasons for this pattern. One, advanced by the modernization theory, is that democracies are more likely to emerge as countries become economically developed. Yet it is also possible that democracies emerge without any relation to economic development, but that once established for whatever reasons, they survive in developed countries. As Lipset (1959: 56) put it, fiThe more well-to-do a nation, the greater the chances it will sustain democracy.fl To determine which of these explanations is valid, therefore, we need to study separately why democracies emerge and why they survive once established. Note that the trajectories of particular countries have been quite different. Among the 135 countries considered here, forty-four countries remained under dictatorships during the entire period during which they were independent between 1950 and 1999, while thirty-four countries had been democratic during this entire period. Twenty-nine countries experienced one regime transition: two from democracy to dictatorship, twenty-seven from dictatorship to democracy. Twelve countries experienced two transitions, five went through three, five countries experienced four transition, three experienced five transitions, two went through six, and one, Argentina, had eight. To understand why a country has a particular regime this year, therefore, we need to know which regime it had and what conditions it had encountered during the preceding year. This is what I mean by firegime dynamicsfl: the process by which the two regimes emerge and die. Since we are dealing with only two types of regimes – democracies and dictatorships – one is born when the other one dies. Hence, the question remains the same whether we ask about the survival of dictatorships or about the emergence of democracies, about the deaths of democracies or the births of dictatorships. I will couch the discussion in terms of the emergence and survival of democracies. The most general observation is that transitions to democracy occur under a wide variety of conditions, while transitions to dictatorships exhibit well-defined patterns. Another way to put it is that it is easy to predict statistically whether a democracy will survive and it is next to impossible to predict whether one will be established. Transitions to Democracy Here is a summary of historical patterns of collapse of dictatorships that led to transitions to democracy:
PAGE – 5 ============
5(1) Consider first the column pAD of Table 1, giving the probabilities that a democracy will emerge, given per capita income. Democracies are unlikely to be established in poor countries, they are more likely to emerge in countries at middle-income levels, and again less likely if a dictatorship exists in a country with a high level of per capita income. Hence, it is not true that economic development breeds democracy: if a dictatorship survives in a country that became sufficiently wealthy, transition to democracy is less likely. The income levels at which democracies emerge vary widely: there is no threshold beyond which one could be almost certain that a dictatorship would die. (2) The same is true with regard to the rate of economic growth. On the average, dictatorships are almost as likely to survive when their economies grow as when they decline during one, two, or three consecutive years. Some dictatorships fell after several years of continuous growth while some other died after several years of economic decline. (3) The impact of income distribution is difficult to assess, given the paucity of data, but it appears that dictatorships are more vulnerable when inequality of functional distribution of income is large (labor share in manufacturing is low) and when household incomes are becoming more concentrated. (4) Considering a variety of other factors – including past political history, ethnolinguistic heterogeneity, distribution of three religions, and the colonial heritage – adds little predictive power. In sum, it seems that dictatorships die under a broad variety of conditions. Some disappear in the midst of economic crises but others vanish after a long period of prosperity. Some die when the founding dictator dies. Some collapse as a result of defeat in foreign wars. Some crumble under international pressure. Dictatorships are subject to multiple risks. The reason transitions to democracy are so difficult to predict by observing the conditions under which dictatorships are that conditions determine only the possibility that a transition to democracy would occur, but it is actions of people under these conditions which shape the outcomes. The interplay between long-term, macro-processes and short-term, micro-analysis is a complicated issue, one that still remains unclear. The hundreds of dissertations that are written on the history of different transitions divide between those that see the process as fostered by transformations of social structure, boiling up through the creation of civil society, and coming to fruition almost of itself, and those that start with “hardliners,” “reformers,” “moderates,” and “radicals” playing a strategic game and reaching a bargain under conditions taken as a datum. The tens of literature reviews pit “sociological” against “strategic” perspectives and find both of them wanting. Yet these approaches are not mutually exclusive and, educated by Wantchekon (1996), I think that this is one useful way to think about them. Suppose for simplicity that at some moment, t = 0,1,,T,, the strategic situation facing two political forces, called R and C, is the following: C
PAGE – 6 ============
6 Fight Talk Fight 2, 1 d, 0 R Talk 0, d 4, 3 Suppose that the strategic interaction takes place once, at t = 0, and that d0 > 4. The game is then a prisoners’ dilemma, with a unique equilibrium {Fight, Fight}, which establishes the dictatorship of R with associated payoffs of {2,1}. Now history begins and various processes shape the value of dt. At some moment, t = T, dt falls to a value dT < 2. Now the game has two equilibria, either {Fight, Fight}, with the status quo payoffs, or {Talk, Talk}, with Pareto superior payoffs. Hence, long-term economic, social, or political processes alter at some moment the strategic situation, and trigger a search for a compromise. If this is the framework of analysis we adopt, our attention should focus on the determinants of the benefits that each of the political forces derives from enjoying being a dictator without any active opposition. This is what dt stands for: the value to a dictatorship of being fully in control. Hence, whether we think that the key to the long-term transformations lies in economic evolution, cultural change, the rise of civil society, or what not, we should be analyzing these processes with the view toward their implications for the grip that the dictatorships can hold over the society and the associated benefits, material or symbolic. What then shapes the value of dt? In my view, it can be practically anything. Processes that change the value of dt can be exogenous or endogenous, macro or micro. In Lipset's account, or at least the "modernization" version of it (Przeworski and Limongi 1996), democracy is secreted by exogenously occurring economic development. A story told about country after a country is that as it develops, social structure becomes complex, new groups emerge and organize, labor processes begin to require an active cooperation of employees, and, as a result, the system can no longer be effectively ran by command: the society is too complex, technological change endows the direct producers with some autonomy and private information, civil society emerges, and dictatorial forms of control lose their effectiveness. But one can also tell a story in which the factors that render dictatorship ineffective are still exogenous but of a micro order: say the imminent death of a founding dictator, a Franco, uniquely capable of maintaining the dictatorial order. In turn, other processes may be of a macro-order and endogenous: in one story, the "goulash" compromise -- "your stomachs will be full if you keep quiet" -- fell apart as communists ran their economies into the ground. Even more obviously endogenous are the effects of a prolonged civil war, say in El Salvador, which at some time made military victory no longer worth pursuing. And the defeat in the war over the Malvinas was endogenous but a just a single event. Moreover, the same factors may work in one direction in some countries and in the opposite direction in others. Economic growth may weaken the authoritarian regime by making the economy too complex to be ran by command but economic decline may also bring a
PAGE - 8 ============
8terms of their welfare (or of the groups they represent5). They anticipate that a federal system will promote trade but may lead to domination of large states over small ones, and they associate some measure of welfare with these consequences of federalism. And if these institutions have distributional effects -- as they do -- then the particular groups have preferences over the equilibria among which they can choose. Some people anticipate that they would be better off under the separation of Church and State than under official religion, under a presidential system rather than under a parliamentary one, under a bicameral legislature rather than a unitary one, etc. Suppose that there are two possible democratic systems, D1 and D2. The structure of the games engendered by each constitution may be one of coordination but it may be also a prisoners' dilemma or even a pure conflict. Indeed, had there been no conflicts, democracy would not have been necessary: democracy is a method for resolving conflicts. For example, the game induced by D1 may be a prisoners' dilemma: C A B A 3, 2 5, 1 R B 1, 5 4, 4 while the game induced by D2 may be a pure conflict C A B A 3, 2 2, 3 R B 1, 4 0, 5 with respective equilibria at {A,A} and {A,B} and their associated payoffs {3,2} and {2,3}. The generic situation confronting political forces in conflict over the choice of institutions is then the following: C 5 I am skirting over complications: as Schmitter (1984) observed some time ago, politicians may have interests of their own and may collude against those they are supposed to represent. Indeed, they may agree to democracy because they expect to be able to pursue interests of their own rather than those of their constituents. But again, whether the pay-offs in the table are to the representatives or to the represented, the analysis remains the same as long as the former can control the latter.
PAGE - 9 ============
9 Fight D1 D2 Fight 1, 1 d, 0 d , 0 R D1 0, d 3, 2 1, 1 D2 0, d 1, 1 2 , 3 where D1 and D2 are two alternative systems of democratic institutions, and 3>2>1>0. We already know that the “constitutional moment” arises when d changes, for whatever reasons, from d>3 to d<2. If d>maxR,C U(D1,D2), the entire game is a prisoners’ dilemma, with a unique equilibrium of {Fight, Fight}. If d