Is Democray Stable? Compared to What?

A Preliminary Exploration

with Robert J. Mundt

introduction

The empirical study of democratic regimes in the last fifty years has focused on the question of what makes for stable democracies.[1] Various hypotheses have been put forward and tested about the social and political conditions under which democratic regimes come to be or to endure. A presupposition of most of this research is that democratic regimes are particularly fragile. The supposition that democracies are fragile probably has a number of sources.  The frightening experience of the descent of European democracies into fascism and communism is perhaps the most important. But we can also find support for this presupposition in the evident fragility of democratic regimes in the less developed world.  And, standing behind these events, is the long standing tradition in political philosophy—and especially, in pre-modern political thought—of disparaging democracy and warning that it is likely to lead to tyranny.

We do not dispute the notion that democratic regimes are fragile.  But we observe that all political regimes are fragile.  Political stability is by no means the norm in human history.  The question thus becomes whether democratic regimes are more fragile than authoritarian regimes.  This, we believe, remains a much ignored and thus open question. 

The aim of this paper is to present a preliminary exploration of this issue. We present some initial empirical data that address the relative stability of authoritarian and democratic regimes. But, before we begin to attempt to test the hypothesis that democratic regimes are at least as stable than authoritarian ones, if not more so, we must first answer some preliminary questions about the conceptual and operational definitions of the notions of democracy, authoritarianism and stability. This is the task of parts II and III of the paper. We more briefly discuss our data and statistical methods in parts IV and V of the paper and present some initial results in part VI. We begin, in part I, with some theoretical reasons for thinking that democratic regimes might be quite as stable as authoritarian ones.

i. democracy, autocracy and stability

Doubts About Democracy

Contemporary doubts about democracy follow the long tradition of pre-modern political thought, a tradition that pointed both to the disadvantages of democratic regimes and the advantages of non-democratic regimes.

Democratic regimes allow for the expression of the range of views held by the members of a political community. And it has often been held that, in a number of ways, this makes government difficult. For one thing, the people are likely to be resistant to the demands of government, and especially, to taxes and military service. But effective government requires that governments pay their bills and mobilize armies. Critics of democracy suppose that monarchs and aristocrats, who have greater experience with and training for political matters will, be better able to grasp and deal with these necessities of politics. For another thing, the people are likely to be divided about the proper direction of government. This leads to two possible problems. If political circumstances allow one group and then another to triumph over another, a government might adopt a series of radical changes in direction. If, on the other hand, political circumstances make it difficult for one group to gain power, the result is likely to be stalemate. Either way, it will be difficult for democratic regimes to adopt consistent and effective public policies.

Under favorable conditions, conflict in or paralysis of democratic regimes may not be too serious. But, under unfavorable conditions, these regimes will, it is held, be unable either to  protect themselves or serve the common good. The result, then, is likely to be dissatisfaction, dissent and, eventually instability. Once opposition to the regime arises, a democracy is, again, likely to find it difficult to respond in ways that preserves itself. Democratic regimes are often reluctant to use force against their own population. And, even when they do so, they are not as likely to be as brutally decisive and potent as a dictator who is not constrained by the rule of law or popular opinion. That, we all think, is to the good. But in our preference for benign government, we should not assume that a good government is always the most stable government.

Why Democracies Are Likely to Be Stable

There are, then, a number of reasons to think that authoritarian regimes are likely to be effective and / or stable. But a number of arguments have been advanced by modern and contemporary political theorists in support of the notion that democracies are likely to be especially resistant to the problems that afflict all government. Most of these arguments go back to Machiavelli (1970), who long ago pointed to the advantages of republics over principalities.

First, democratic governments have a better mechanism for handling the transitions from one leader to another than authoritarian regimes.  If they accomplish nothing else, elections are a practical—and often peaceful—way for choosing between rival political leaders.

Second, democratic regimes allow for a variety of leaders with different kinds of skills to come to power in the appropriate circumstances.  Democracies are not always guaranteed a Washington or Lincoln when times get tough. But, if one leader clearly fails, it is possible to replace him with someone with very different qualities.

Third, the freedom of speech that comes with democratic government gives political leaders early warnings of serious problems, including threats to the regime. Albert Hirschman (1972) has pointed out that the complaints of citizens give leaders, including those in countries too poor to hire political scientists to do survey research, knowledge of where and how their policies run into trouble. And Amartya Sen (1984) has plausibly argued that, precisely because they give citizens a means to express  themselves, no democratic regimes have suffered from the kinds of massive famines that have afflicted authoritarian regimes such as China.

Fourth, democratic regimes allow conflicting ideas to be put forward about how to deal with public problems. Democratic debates are often indecisive and lead more often to compromise solutions than striking innovations. But, because new and varied ideas can ultimately reach political leaders, democracies can avoid the worst outcomes. And they may find that “muddling through” encourages the creation of public policies that are tolerably effective and minimally appealing to citizens of many different views. The democratic inclination to take a middle way between two extremes stifles change. But it is often the wisest course.

Fifth, the political liberty of democratic regimes allow citizens to express their resentment of the demands of government. Expressions of distaste for government often have little impact on the policies that regimes follow—as we mentioned, all governments need taxes and soldiers. But political officials must humble themselves before the people in order to attain and retain public office. And many are humbled even more by being rejected by the public. As Machiavelli pointed out, this process helps reconcile the people to the demands of government while, at the same time, reminding these officials of who is ultimately in charge. The public accusations of Machiavelli’s time have been replaced by both Congressional hearings and the political humor of Jay Leno. But they are no less effective in helping political leaders and the people live with one another.

Sixth, and finally, democratic regimes are likely to be freer and thus more dynamic economically. Economic growth can take the hard edges off political conflict. For it allows governments to meet the demands of citizens without raising taxes. And, under particularly good conditions, governments can redistribute from the rich to the poor without reducing the level of income to which the rich have already become accustomed.

In all these ways, then, democratic regimes are likely to be better than authoritarian regimes at developing policies that are effective and, as a result, that satisfy the desires of the people. A more satisfied people is not likely to be good tinder for revolutionaries. And thus democratic regimes are likely to be especially stable.

ii. conceptual issues: stability

To conduct a comparative study of the stability of democratic and non-democratic regimes requires us to make a number of difficult choices of research methods. These involve more than the usual question of how to operationally define certain variables. For, before operationally defining variables, we must have  a clear conceptual understanding of them And this is not easy. Difficulties arise with regard to both the notions of stability and democracy.

There are a number of ways in which we can think about political stability. The most common is to define the stability of a regime as its durability—the length of time it was, or has been, in existence. Thus, Mainwaring defined a stable democracy as a democratic regime that has been in existence for an uninterrupted period of 25 years (Mainwaring 1993: 204; see Umez 1993: 39). While this is a plausible way to think of stability, a long-lasting regime might still be subject to a great deal of political violence and, especially, violent opposition. So, another way to think of stability is as the ability of a regime to, as Machiavelli might have put it, pacify its people. We will consider each of these approaches in turn.

Stability as Durability: Regimes and Regime Change

If we would like to define stability as durability, a fundamental question will arise: should be concerned with the stability of a particular regime or the stability of a certain kind of regime? To address this issue, we must first clarify the notion of a regime and regime change.

A regime consists of a certain order of governmental offices that specifies the powers granted to the government as a whole, the powers granted to those who hold each office; and the procedures by which these offices are to be filled. A regime may be formal or informal. A formal regime is often embodied in a written constitution that is a higher law, that is, one meant to constrain the normal legislative process. Or it may be embodies is statutory law that can be changed like all other legislation. Some regimes are, on the other hand, quite informal. The consist of conventional institutions, practices and procedures that are broadly understood and followed yet remain unwritten. In some cases informal the informal institutions and practices are of greater importance than the formal rules. This was true both in the USSR and in Chicago under the first Mayor Daley. We can call the institutions and practices of a regime its constitution, recognizing that not all constitutions are written higher law as in the Constitution of the United States but may be statutory law, as in the Basic Law of Israel or largely informal as in the unwritten constitution of the United Kingdom.

Regimes may be highly elaborate or minimal. Some authoritarian regimes are quite minimal in nature. One dictator or a small body of rulers is given the authority to take all governmental action. That small body is granted—or takes—the power to fill vacancies or a particular person is appointed the successor to the dictator. On the other hand many regimes, including authoritarian regimes, are quite complex, having a wide variety of offices that are granted particular powers.

One regime comes to an end, and another begins, under one of two circumstances. Either there is a transformation from one type of regime to another.[2] Or there is a non-constitutional change in the ordering of governmental offices or the procedures by which they are filled. Most formal or informal constitutions have some procedure by which some or all of the provisions of the constitution can be changed or amended. Constitutional changes in the ordering of offices or in the procedures by which they are filled must follow the constitutional procedures for constitutional amendment or change. Non-constitutional changes occur when people come to hold government offices in ways that violate the procedures of the established constitution or when the ordering of offices is changed in a way that violates the procedures of the established constitution.

The most apparent and dramatic cases of non-constitutional change can be found in transitions from authoritarian to democratic regimes and from democratic to authoritarian regimes. When a King is stripped of power and then executed by an newly elected assembly, the change of regime is quite obvious. It is also apparent that a regime has been overthrown when a military coup sends a prime minister and the parliament that elected him or her packing. But regime change can also occur when one authoritarian or democratic regime gives way to another of the same type. Authoritarian governments that are created by a coup often die in the same way. And democratic regimes sometimes replace other democratic regimes when a new constitution is adopted in ways that contravene the established procedure for constitutional amendment. This often occurs when elites propose a new constitution that is then adopted by referendum as was done, for example, when the current (and second) Constitution of the United States was adopted, replacing the Articles of Confederation or when the Fifth Republic of France replaced the Fourth Republic. Democratic government can also use non- or questionably constitutional procedures to amend their constitution. This frequently occurs in the aftermath of domestic strife or civil war. For example, one might question whether the adoption of 13th, 14th, and 15th amendments to the US Constitution, strictly followed the procedures for amendment.[3]

Stability as Duration of Regime Type

Given this account of a regime, one way to define stability is as the length of time a particular type of regime has survived. On this view, we would ignore changes from one regime to another when the type of regime has not changed. That is, if an authoritarian regime followed another authoritarian regime—as frequently happens when a series of military coups occur—we would not re-start the clock measuring the survival of the authoritarian regime. And the same would be true when one democratic regime follows another. This is the approach taken by Gasiorowski (1996).

Why is it plausible to define stability in terms of the durability of a type of regime? Largely because citizens are often indifferent to changes in regime when these changes do not involve a change in the type of regime. Rulers may come and go, but the difference isn't felt by ordinary people—rather like the old stereotypes of the Mexican peasant asking at the palace door "Who's President Today?" To the political elites, of course, a coup that replaces one authoritarian regime with another is quite important. But, because the effects on policy of this sort of change in regime are often minimal, most people who live in a country—and who are excluded from any power—will not be much concerned or troubled by such changes in regime. One such regime is generally as bad as another. And that, of course, is why political scientists have typically been most concerned with changes from authoritarian to democratic regime and back again. Given the great importance of this kind of change, it would seem to matter much less how long a particular democratic or authoritarian regimes lasts, so long as it is followed by a regime of the same type.

Another, somewhat paradoxical, reason not to be concerned about the transition from one authoritarian regime to another is that, under many circumstances, it should not surprise us that authoritarian regimes replace one another rapidly. Many authoritarian regimes are minimal regimes, dependent upon the political skill of a single person. And it is difficult for a single individuals to rule for long, particularly against the combination of ambitious politicians who long to take his place. Few individuals are skilled enough to maintain power for long.

That authoritarian regimes have a tendency to instability is, as we saw above, contradicted by the long tradition in political thought of emphasizing the great efficiency of authoritarian regimes. But, we can perhaps reconcile the empirical reality of frequent changes in authoritarian with the view that emphasizes the stability of authoritarian regimes by noting that stable political authoritarian regimes typically occur when a politically skilled leader comes to the fore. Such a person may be rare. But when he comes along, it is possible for him to pacify a country and hold power for a long time

Stability as Durability of Regime

While there are good reasons why a study of political stability should focus on changes from one type of regime to another, there are also good reasons for us to be concerned with the durability of democratic and authoritarian regimes even when they are succeeded by regimes of the same type.[4]

We suggested above that the replacement of one authoritarian regime with another does not matter much to those who must live under such regimes. But while this is often true, the same point cannot be so confidently made about the transition from one democratic regime to another. Precisely because such regimes are democratic, they are more likely to respond to popular opinion and to encourage political participation. And, for that reasons, changes from one democratic regime to another will often have a dramatic effect on the on the ability of different citizens to make their influence felt and to receive the kinds of public policies they desire.

The importance of the durability of a particular regime can also be seen if we ask why people are concerned about stability at all. For the stability of constitutional institutions, practices and procedures—whether formal or informal—is generally valued. The reason is that formalities are important in politics. A settled constitution and regular procedures, of whatever type, give people the ability to make decisions about how best to attain their ends. And this is true, even if their main end is to avoid all dangerous involvement in politics. For as many who lived under Stalin found out, when politics and policy is unsettled, it can be very hard to know how to stay out of the way of trouble. On-going constitutional arrangements also encourage peace, even where peace is sustained by a modus vivendi rather than by widespread support for a particular regime.  Once contestants for power reach some settled understandings between them, the threat of active warfare is reduced. For each side know what to expect and what they can demand from the other without risking an offense that would lead to war. And, of course, one of the most important part of such agreements is an procedure for handling the succession of power, when rulers come to the end of a term or die prematurely. These are dangerous moments for all polities, and are particularly troublesome when no constitutional settlement is in place. Extra-constitutional change, then, raises the possibility of civil war and, especially for small countries, foreign intervention as well. A constitutional regime, then is highly desired, on all sides.

Stable government is also advantageous because it is more likely to be able to keep peace and civic order. In a stable regime, political office holders can focus on something besides maintaining themselves in power. Stable governments are also more likely to carry out other policies generally desired by a people. All this is not to say that instability always has bad effects. People do go on and live their lives as well as they can in unstable polities. And stability does not by any means guarantee good government. Yet if one must live under an authoritarian regime, it is generally better if that regime is stable rather than unstable. Of course, on the assumption that democracies are better than authoritarian regimes, it would be nice if democracies were always stable and authoritarian regimes were always not. Then, if democracies had an equal chance to replace failed authoritarian regimes, the trend would be towards more and more democratic regimes. We may have some hopes that this trend will, over the very long term, be realized. But, in some times and places, such expectations are highly unrealistic. In those places, stable authoritarian regimes are the best that can be hoped for.

Given the importance of political stability, then, it is important for us to look at the durability of individual regimes. For if democratic regimes are short-lived, then they are likely to be problematic, even if they are often replaced by other democratic regimes. A series of short-lived democratic regimes will create the uncertainties and other problems we pointed to above. And, of course, a series of short-lived democratic regimes will create opportunities for the replacement of a democratic regime by a authoritarian regime. It is also important for us to note changes from one authoritarian regime to another, for this sort of instability creates difficulties and uncertainties as well. In addition, the instability that results from the recurring replacement of one authoritarian regime by another creates an opening for democratic government. Democracy has, after all, often evolved out of the frustration of conflicting political elites who could find no other way to settle their disputes.

Stability and Political Violence

To this point, we have looked at two different ways of defining political stability in terms of the durability of regimes. There is, however, another way to look at political stability. A moment ago, we suggested that one of the difficulties of frequent regime change is that it can lead to political violence. This suggests that we might understand political stability in terms of opposition to the regime and that this, in turn, can be measured by the level of political violence. This approach has been used by Umez, who suggests deaths from political violence is an useful measure of political stability, because political violence tends to affect all levels of society (Umez 1993: 40). 

The advantage of this way of thinking about political stability is that it allows us to take into the account the possibility that a regime might teeter for a very long time even though it does not fall. Some regimes endure for a long time even if the face of continuous internal conflict and on and off again civil war. The mere survival of such a regime would not, however, lead us to think it is terribly stable. A regime that must put down a coup attempt every year for twenty years should not be considered stable due to the incompetence of the coup plotters. Even though the high potential for collapse is unrealized for many years, that potential helps create the serious problems of instability we considered above.

The disadvantage of this third way of understanding stability is that a regime that represses violent political opposition by brutal and effective means is not necessarily all that stable either. Repression might allow a regime to survive for a long time despite a large opposition. And then, when the repression lightens just a bit or the repressors grow over-confident, this regime might quickly be over thrown. Such a regime might have as high, or almost as high, a potential for collapse as one that is not as effective in repressing violent opposition. On the other hand, a regime that effectively represses the opposition would have some of the advantages of stable government to which we have pointed.

There are then, at least, three ways in which to think about and then try to measure political stability. On the whole, we believe that that the second way of thinking about stability makes the most sense. That is, all things considered, it seems that we should measure stability by the durability of individual regimes. And we are particularly dubious about the first way of looking at stability. Changes from regime to another seem to us to be an important phenomena, one that raises the concerns of those who worry about stable government. For the purposes of this paper, however, we will present some evidence that draws upon all three views of stability.

iii. conceptual issues: democracy

Through out this paper, we have been using the terms “democratic” and “authoritarian” with impunity. While these two notions are more or less well understood by all political scientists, we should say a bit more about precisely what we mean by them here..

Representative Democracy

When we speak of democracy we mean, of course, the modern notion of representative democracy. And this notion is very different from the ancient idea of democracy, which we today call direct or participatory democracy. A representative democracy is a Republican government in Madison's sense: a government that

derives all of its powers directly or indirectly from the great body of the people and is administered by persons holding office during pleasure for a limited period or during good behavior...It is sufficient for such government that the persons administering it be appointed, either directly or indirectly by the people; and that they hold their appointments by either of the tenures just specified.[5]

There is a fairly broad agreement about the basic requirements of representative democracy. To be a full representative democracy then, a regime must meet the following criteria: [6]

1. It must be Republican in Madison’s sense: all government decisions must be made by officials who are elected by the citizenry or appointed by those elected by the citizenry.

2. There must be elections for office holders that are:

a. Regular in that that they are held frequently at regular intervals;

b. Fair in that there is no coercion of voters or corruption in counting them;

c. Open in that all adults are allowed to run for elective office;

d. Inclusive in that all adults are allowed to vote.

3. The civil liberties necessary to democratic political life must be fully protected.

a. There must be freedom of expression with regard to political matters.

b. A variety of sources of political information must be available from other than official government agencies.

c. There must be freedom of association, all adults must have the right to form independent associations, organizations, parties and interest groups.

While these criteria define an ideal representative democracy, we have modified them slightly in conducting empirical research. We will call a regime a democracy even if there are, by our standards, some inappropriate limits on freedom of expression or if elections are, to some degree unfair, so long as we have good reason to believe that these restrictions do not usually influence the outcome of elections. We will also call regimes democratic even though some political offices are held by hereditary monarchs or aristocrats, provided that the democratic parts of the regime can, with some minimal effort or persistence, overcome those the non-democratic parts of the regime. These are fairly small departures from the ideal of representative democracy. We will also allow one very large exception to the criteria of an ideal representative democracy. We will call a regime democratic even if women, those with limited economic resources and many members of ethnic minorities do not have the franchise. The reason for this exclusion is, in part, simply a reflection of the practice of most of the regimes in the last few hundred years that were otherwise more or less democratic. If we were to insist on an inclusive franchise as a criteria for democracy, there would have been very few democracies until recently. Excluding all but these few regimes from the democratic category would seem, for two reasons, to be a mistake.

First, while the extension of the suffrage undoubtedly brings about important changes in the character of democratic government, once government officials are selected directly or indirectly by a substantial number of citizens, a decisive step has already been taken. For the dynamics of government are very different when government officials are responsible to public opinion. In other words, in our view a democratic government with a limited franchise is, in decisive respects, closer to an inclusive democratic government than to an authoritarian regime.

Second, while we are not moral relativists, we do believe that if we are to conduct a serious study of polities different from our own, we must take into account the ideas and ideals of those who live in those political communities. And, on the views current at the time, there were plausible reasons for the limited franchise. Women were presumed to be represented by men and to be lacking in capacities to take part in politics. And the poor were thought to have less of a stake in the political and social life of their country. If one holds these views, then a limited franchise is entirely compatible with democratic government. While we would not accept these arguments for a limited franchise, we believe that the character of governments must be understood, at least in part, in the way they were understood at the time.[7]

Alternatives to Democracy

There are a wide range of alternatives to democratic government. We shall call regimes that have little or no element of democracy, authoritarian or autocratic governments. There are, of course, many kinds of authoritarian regimes including traditional monarchies and aristocracies; non-traditional dictatorships and military juntas; and totalitarian regimes. For the purposes of this paper, we will ignore the important differences between these different authoritarian regimes.

Another alternative to democracy is what we will call semi-democratic regimes. These are regimes that have some of the elements of full fledged democracies but in which democratic rule is severely constrained. All of these regimes have a legislative body that is elected by some substantial part of the people and that has some share of political power. But semi-democratic regimes do not meet the criteria for democracy in one or more of three ways:

1.      Important decisions are made by government officials who are not selected directly or indirectly by the people. Many semi-democracies can be found in late 18th and 19th Europe in which ministers were responsible not to the parliament but to the monarch. In many of these regimes parliament had to approve certain actions of the ministers, especially with regard to budgets and taxes. But, on other matters, ministers could act on the authority of the monarch. The attainment of parliamentary sovereignty over government ministers was, in most of these countries, the decisive moment at which a democratic government was attained.

2.      Elections are not entirely open or fair. In many semi-democracies, governments set limits on who can stand for office. And, by monopolizing the mass media, massively using government resources in political campaigns or committing outright fraud, elections never surprise anyone. For the ruling party or clique is never in danger of losing an election to the most important offices. Other parties are likely to win some votes and seats in the national legislature. And they may, therefore, have some influence over both public opinion and the government. But, until the regime changes, they are destined to play a small role in the political life of their country. The decisive moment at which democratic government is attained (or will be attained) in these countries is when people have a good reason to wait up for the election returns. Mexico and Singapore are contemporary examples of this second kind of semi-democracy.

3.      While some civil liberties are protected in semi-democratic regimes, these liberties are, in important ways, constrained. Governments may set limits on criticism of the regime or on the expression of certain ideological views. And they may prevent the formation of some associations, interest groups and parties. Many of the regimes that are semi-democracies by virtue of the second criteria, meet this third criteria as well.

Many regimes today, and in the past, qualify as semi-democracies. Indeed, in an analysis with a purpose similar to ours, William F. Case used the cases of Singapore, Thailand, and Malaysia to make the argument that “semidemocracies” are not merely transitional regimes, but that semidemocracy may indeed be a separate form of stable regime.” [8] 

iv. methods

In this section we will describe our methods of analysis, and some of the inherent limitations in this preliminary exploration of the stability of democratic and authoritarian regimes.

Democracy as a Progressive Phenomena

One difficulty with this kind of research is in determining the period of time to study. Two problems arise here. The first is, how far back in time should we study the relative stability of democratic and authoritarian regimes? The difficulty for our research is that representative democracy is distinctly a modern phenomena, one that does not arise until the late 18th century. It obviously makes little sense to conduct our study by examining regimes from the 15th to 18th centuries. That this makes no sense raises some important questions about the way our analysis should proceed.

Certain pre-conditions had to occur before representative democracies could arise. These include social pre-conditions—such as the creation of what we may alternately call a market or individualistic or capitalist or organic or urban polity and society—and ideological preconditions—such as the development of liberal political thought.[9] So it seems sensible to begin our analysis in the late 18th century with the emergence of democratic government in the United States. This is how we intend to proceed. We will include in our analysis of the duration of regimes all regimes that came into existence in independent countries after 1776.

While this choice makes a great deal of sense, it does raise a problem which we will have to deal with in future research. The preconditions for democratic government did not arise everywhere in the world at the same time. This would not be a problem for our analysis if these preconditions always arose together or at one moment. Then we would find that, in some parts of the world, authoritarian regimes continued much later than others. The real difficulty for this research is that, while the social preconditions arose along with the ideological preconditions in the earliest democracies, this did not occur in more recent democracies. Ideas about politics can gain influence with elites even in countries that are far from meeting the social preconditions of democracy. We can have nothing but admiration for the hardy souls in such countries as Greece, Albania, Bulgaria,  and Romania who tried—and sometimes succeeded—in bringing democratic government to largely agricultural, illiterate countries that, in some cases, could measure the time since the serfs were freed in years rather than decades. But it is hardly surprising that democratic government was not terribly stable in these countries. It is also true that, in many of these countries, authoritarian government was equally unstable. Indeed, it may well be that the instability of both democracy and autocracy is, to some extent, the product of  conflicting ideals of politics as well as social conditions that encourage some groups to hold democratic ideals but are not sufficiently favorable to democratic government.[10] In any event, this problem suggests that is important to control for the social conditions when comparing the stability of democratic and authoritarian regimes. In some cases, we do report separate results for more and less developed countries. But this is not that good a proxy for the social pre-conditions of democracy. We hope to pursue a full fledged multivariate analysis in the future.

Censorship of Data

We will, then, begin our analysis with regimes coming in existence from 1776. But what shall be the end point of our analysis? Our data extends up to 1992. But many regimes in existence in 1992 exist today and will continue to exist for some unknown time into the future. Political and social scientists doing this sort of research typically exclude or censor any regimes that were in place prior to the beginning date or that continue to exist at the ending date. This approach has an important advantage. If our measure of stability is a function of the length of time a regime has survived, it seems that we should do our best to accurately measure this  length of time. To shorten this period by either setting an arbitrary beginning point for a regime or by, in essence, treating a regime as if it had come to an end in the present, is thus seriously misleading. Thus the first data we will present—which we will call the censored data—will exclude all regimes in existence prior to 1776 and after 1992.

While censoring the data set in this way has some advantages, it has disadvantages as well. For it excludes many long lived regimes, and, in particular, many of the most long lived democratic regimes. At the same time, because of the fall of the Soviet Union, a number of fairly long lived authoritarian regimes came to an end by 1992 and thus are included in the analysis.[11] To see if this creates any biases in our data set, we shall present a second set of data, in which we do not exclude regimes in existence in 1992. We shall call this the uncensored data.

 

Statistical methods

For this preliminary analysis, we will present basic univariate measures of the duration of regimes and types of regimes and of levels of political violence found in different kinds of regimes. Some form of survival analysis is, of course, most appropriate in studying phenomena of this sort. And it is necessary in order to conduct the multivariate analyses of the type we hope to do in the future. But it is unnecessary in this preliminary exploration of the data.

v. data

Regimes and Regime Change

The kind of comparative research we propose to do is impossible in the absence of substantial datasets. Fortunately, we could help ourselves to data collected by other political scientists. Unfortunately, when it came to regimes and regime change no one, to our knowledge, has put together precisely the right dataset for our purposes.

Our starting point in the study of the duration of regimes and types of regimes is the Political Regime Change Dataset developed by Gasiorowski (1996). This dataset gives dates of transition among democratic, semi-democratic, authoritarian, and transitional regimes in the 97 largest Third World countries from independence to 1992.  While very useful, this dataset has two limitations. First, it only includes regimes in Africa, Asia and Latin America. And second, it defines a change in regime as a change from one type of regime to another. That is, the Gasiorowski coding doesn't show a shift from one form of authoritarianism to another, but only changes among the four possibilities of democratic, semi‑democratic, authoritarian, and transitional regimes.  Thus Ethiopia was coded authoritarian from 1871‑present. The Political Regime Change Dataset, then, is satisfactory if we adopt the first approach to understanding stability and define stability as the duration of a particular kind of regime. By itself, however, this dataset will not allow us to define stability as the duration of a particular regime.

We solved the first problem by creating our own dataset for the largest countries of Europe, North America and Australasia.[12] We consulted a number of standard history texts to determine the beginning and ending dates for each regime and to determine whether each regime was a democracy, semi-democracy or autocracy.[13] Like Gasiorowski, we did not attempt to fix any quantitative criteria for coding the type of any particular regime. On our view, determining whether a regime is an autocracy, semidemocracy or democracy is, and can only be, a matter of political and historical judgment. The criteria for different regime types we established above guided our coding decisions. In most cases, coding decisions were easy. Some cases, however, required difficult judgments about, say, whether elections were usually so unfair that a particular regime should be considered a semi-democracy rather than a democracy. Or, in some cases, we had difficulties deciding whether an elected legislature had gained enough power that a semi-democratic regime had replaced an autocracy or whether the remaining power entrusted to a hereditary monarchy and aristocracy did not disqualify a regime for inclusion as a democracy.[14] While additional reading of and reflection on the historical record might lead us to change some of these coding decisions, we are prepared to defend our particular coding decisions. All we can say in defense of our decisions in general, however, is that the were made in the only way possible, by a careful reading and analysis of the record of history.

We solved the second problem by merging the Gasiorowski data with the Bienen / Van De Walle (1991) dataset on leadership duration. Bienen and Van De Walle have carefully examined transitions from one leader to another in almost all countries in this century. By consulting works of political history and contemporary journalism, they have tried to distinguish those leaders who have entered power by constitutional means from those who have entered power by extra-constitutional means. This distinction neatly fits with our second way of thinking about political stability. For, on that account, one regime comes to an end, and another begins, under one of two circumstances. Either there is a transformation from one type of regime to another.[15] Or the procedures of a particular constitution are dispensed with entirely as a new leader overthrows an old one. The Gasiorowski dataset can tell us the period of time in which a particular type of regime is found in a country. And the Bienen / Van De Walle dataset can tell us when leaders come to power through extra-constitutional means. By putting them together, we can create a dataset that allows us to determine the duration of any particular regime.[16]

Political Violence

We used data from the World Handbook of Political and Social Indicators (Third Edition) when looking at the relationship between regime type and stability defined in terms of levels of political violence. That data compilation covered 136 countries with populations of at least one million in 1965, plus all other states that had become members of the United Nations by 1968.  The same criteria applied in 1975 produced a population of 155 countries.  We used the political rights variable in order to operationally define regime type. Based on the annual Freedom House ranking of countries, it “measures the extent to which people are able to play an active and critical role in the choice of their leaders.”[17] Countries with a low political rights score are more democratic while those with a high score are more authoritarian. In this study, the measurements as of 1979 are used, as this is the date closest to that at which other important variables were measured.  In any event, measurements of political rights changed rather infrequently through the period covered by this data set.

Stability is operationalized by what we judged to be the most direct of the various measures of political protest in this data set, the number of politically related deaths.  Data on protests and demonstrations are less useful in comparing democracies and dictatorships, because they might well be a normal part of democratic decision making.  Deaths from political violence, on the other hand, cannot be considered compatible with stability in any regime type.  The political deaths variable here is the mean number per year over the years in the study.

vi. results

 

When we define political stability as the duration of regimes or the level of violence, our results tend to support the hypothesis that democratic reimes are no more unstable than authoritarian regimes. When we define political stability as the duration of types of regimes, however, the data suggests that authoritarian regimes are more stable than democratic regimes. We will argue below, however, that the data confirms our view that this is an unsatisfactory way to define political stability.

Duration of Regimes

Tables 1 to 6 display the mean and median duration of regimes for all countries; for Europe, North America, Australasia; and for Africa, Asia and Latin America. In each case we present both censored and uncensored data. In all but three cases, the mean and median duration of democratic regimes was longer than that for authoritarian regimes. When we look at censored data the median duration for democratic regimes is less than that for authoritarian regimes for all countries and for North America, Europe and Australasia. Clearly this result occurs because censoring the data eliminates the many, long-lived, democratic regimes in North America, Europe and Australasia. As we indicated above, it hardly makes sense to exclude these long lived regimes when it seems quite clear that, no only do they have a long past behind them, but they also have a long future in front of them. The mean duration of authoritarian regimes in Africa, Latin America and Asia is slightly larger than that for democratic regimes, at least when we look at uncensored data. The difference is quite small, however. Rather than refute our hypothesis, this small difference tends to confirm our claim that democratic regimes are not significantly more unstable than authoritarian regimes. In all other cases the means and median duration of democratic regimes is higher, and, in some cases, far higher than that for authoritarian regimes.

Table 1

Duration of Regimes in Years

Censored Data

All Countries

Regime Type (N)

Mean Duration

Median Duration

Democratic (76)

11.53

4.0

Semi-democratic (87)

12.76

5.0

Authoritarian (505)

8.41

5.0

 

 

Table 2

Duration of Regimes in Years

Uncensored Data

All Countries

Regime Type (N)

Mean Duration

Median Duration

Democratic (126)

20.46

7.00

Semi-democratic (105)

12.82

5.00

Authoritarian (568)

10.14

5.00

 

 

Table 3

Duration of Regimes in Years

Censored Data

Europe, North America, Australasia

Regime Type (N)

Mean Duration

Median Duration

Democratic (36)

15.22

4.00

Semi-democratic (30)

20.83

12.00

Authoritarian (46)

13.11

6.00

Table 4

Duration of Regimes in Years

Uncensored Data

Europe, North America, Australasia

Regime Type (N)

Mean Duration

Median Duration

Democratic (57)

33.77

12.0

Semi-democratic (31)

23.90

12.0

Authoritarian (48)

15.69

7.0

 

 

 

 

 

 

 

 

 

Table 5

Duration of Regimes in Years

Censored Data

Africa, Asia and Latin America

Regime Type (N)

Mean Duration

Median Duration

Democratic (4)

8.21

5.00

Semi-democratic (58)

8.72

4.00

Authoritarian (460)

7.95

4.00

 

 

 

 

 

 

 

 

 

 

 

Table 6

Duration of Regimes in Years

Uncensored Data

Africa, Asia and Latin America

Regime Type (N)

Mean Duration

Median Duration

Democratic (69)

9.46

6.00

Semi-democratic (74)

8.18

4.00

Authoritarian (520)

9.62

5.00

 

 

 

 

 

 

 

 

 

 

Duration of Types of Regimes

When we repeat the analysis using the duration of a type of regime as the definition of political stability, we get a very different result (Tables 7 to 12). In all but one case the mean and median duration of regime type is longer for authoritarian than democratic regimes. The one exception occurs when we look at uncensored data from Europe, North American, and Australasia. The most telling piece of evidence here, however, is the number of cases, N. Whereas we had  a total of 505 (censored) and 568 (uncensored) cases of authoritarian regimes under the first different definition of political stability, now we have only 112 and 173 cases respectively. But the drop in the number of democratic regimes is not so precipitous, from 76 (censored ) and 126 (uncensored cases) to 64 and 125 cases. What accounts for this pattern of change, of course, is that authoritarian regimes typically follow other authoritarian regimes whereas the collapse of democratic regimes often leads to their replacement by an authoritarian regime. This suggests that what is right about the conventional wisdom is this: it is difficult for a political community to break free from authoritarianism and establish democratic regimes. That is, although authoritarian regimes are highly unstable, authoritarianism is likely to survive for a long time, as one authoritarian regime replaces another. The collapse of authoritarian regimes does not always lead to the establishment of democracy. When democratic regimes collapse, however, authoritarian regimes are as likely as not to replace them.[18]

A closer look at the data helps us understand why this pattern results. In Africa, Asia and Latin America, authoritarianism is much longer lived than democracy. But in Europe, North America and Australasia, democratic regimes tend to last as long as authoritarian regimes. The reason for this pattern, we would suspect is that the social and ideological pre-conditions for democratic government are much more likely to be met in Europe, North America and Australasia than in Africa, Asia and Latin America. As a result, it is difficult to establish democratic regimes in Africa, Asia and Latin America, even when authoritarian regimes collapse. Again, the problem for democracy is not that, once established, democratic regimes are likely to fail sooner than authoritarian regimes, even in Africa, Asia and Latin America. Tables five and six rather strikingly show us that democracies last quite as long as authoritarian regimes on these continents. Rather, the difficult seems to be in getting democracy off the ground. And the logical explanation for this difficulty. is that the pre-conditions for democracy are not found to a very high degree. Indeed, if we are right to think that the preconditions for democracy are met to a much lower degree in Africa, Asia and Latin America, the results of tables five and six are even more striking. For, it would seem that democracy must be a fairly robust form of government to survive under difficult circumstances.

Our proposed explanation for the data found when we define stability in these first two ways deserves further examination. We have made what we think is a plausible claim, that the pre-conditions for democracy are found to different degrees in the different regions of the world. But, as we suggested above, region is not a perfect proxy for these pre-conditions. For our data set goes back three centuries to a time when many European countries did not meet these pre-conditions either. So a better test would be to directly control for as many of these pre-conditions as possible.

Table 7

Duration of Type of Regime in Years

Censored Data

All Countries

Regime Type (N)

Mean Duration

Median Duration

Democratic (64)

10.45

4

Semi-democratic (77)

16.16

5

Authoritarian (112)

29.35

10

 

Table 8

Duration of Type of Regime in Years

Uncensored Data

All Countries

Regime Type (N)

Mean Duration

Median Duration

Democratic (115)

22.47

7

Semi-democratic (92)

14.46

5

Authoritarian (173)

33.33

19

 

 

 

 

 

 

 

 

 

 

Table 9

Duration of Type of Regime in Years

Censored Data

Europe, North America, Australasia

Regime Type (N)

Mean Duration

Median Duration

Democratic (26)

13.8

4.0

Semi-democratic (27)

27.48

12.0

Authoritarian (32)

16.75

7.0

 

 

 

 

 

 

 

 

 

 

Table 10

Duration of Type of Regime in Years

Uncensored Data

Europe, North America, Australasia

Regime Type (N)

Mean Duration

Median Duration

Democratic (48)

40.23

16

Semi-democratic (27)

27.48

12

Authoritarian (35)

19.97

7

 

 

 

 

 

 

 

 

 

 

Table 11

Duration of Type of Regime in Years

Censored Data

Africa, Asia and Latin America

Regime Type (N)

Mean Duration

Median Duration

Democratic (38)

8.16

3.5

Semi-democratic (50)

10.05

4.0

Authoritarian (80)

34.39

11.5

 

 

 

 

 

 

 

 

 

 

 

 

Table 12

Duration of Type of Regime in Years

Uncensored Data

Africa, Asia and Latin America

Regime Type (N)

Mean Duration

Median Duration

Democratic (67)

9.75

6.00

Semi-democratic (65)

9.05

4.0

Authoritarian (138)

36.72

24.5

 

 

 

 

 

 

 

 

 

 

Political Violence

The relationship between Deaths from Political Violence and Political Rights is shown in figure 1. Although the relationship is not linear, it is clear that democracies, identified as values 1 and 2 on the political rights scale, had much lower levels of deaths from political violence than did more authoritarian systems, which are identified as values 6 and 7. Again, by this measure of political stability, democratic regimes are more stable than authoritarian regimes. Moreover, these data provides some indirect support for our view that the political instability of authoritarian regimes creates serious problems for those who must live under these regimes. For there is no problem more important for governments to solve than death from political violence.

 

Conclusion

The collapse of the Soviet Union has lead to what some observers have called the third wave of democratization, both in Europe and in what once was called the third world. Along with this third wave, however, has come great concern over the stability of newly emerging democracies. This is an issue well worth worrying over. For, as the evidence of this paper suggests, forms of government of all sorts are difficult to sustain. So it is important for us to figure out how, when, and why democracies and authoritarian government survive and flourish. This knowledge will help politicians in newly emerging democracies preserve their fledgling regimes. And it will help those of us in established democracies figure out how we might aid aspiring politicians in newly emerging democracies. Too often in the past, however, we have been reluctant to provide all the aid we should to emerging democracies because we had little faith in their ability to survive. The message of this paper is thus quite important to contemporary politics. For we will be more likely to provide aid to newly emerging democracies if we recognize that democratic regimes can be as stable as authoritarian regimes. All regimes are fragile. All need attention, care and prudence. But democratic regimes are not more fragile than non-democratic regimes.

 

References

 

Eric C. Browne, John P. Frendreis, and Dennis W. Gleiber.  1986.  The Process of Cabinet Dissolution: An Exponential Model of Duration and Stability in Western Democracies.”  American Journal of Political Science 30, pp. 628-650.

Bienen, Henry and Nicolas Van De Walle. 1991. Of Time and Power: Leadership Duration in the Modern World. Stanford: Stanford University Press.

Burkhart, Ross E. and Michael S. Lewis-Beck.  1994.  Comparative Democracy: The Economic Development Thesis.  American Political Science Review 88 (4, December), pp. 903-910.

Case, William F.  1996.  Can the “Halfway House” Stand?  Semidemocracy and Elite Theory in Three Southeast Asian Countries. Comparative Politics (July), pp. 437-464.

Dahl, Robert. 1989. Democracy and Its Critics. New Haven: Yale University Press.

de Haan, Jakob and Clemens L.J. Siermann.  1996.  Political Instability, Freedom, and Economic Growth: Some Further Evidence.  Economic Development and Cultural Change (), pp. 339-350.

Gasiorowski, Mark J.  1996.  An Overview of the Political Regime Change Dataset.  Comparative Political Studies 29 (4, August), pp. 469-483.

Hirschman, Albert O. 1972 Exit Voice and Loyalty: Responses to Decline in Firms, Organizations, and States. Cambridge: Harvard University Press

Linz, Juan and Alfred Stepan. 1978. The Breakdown of Democratic Regimes. Baltimore: Johns Hopkins University Press.

Lowi, Theodore J. 1979. The End of Liberalism. New York: Norton.

Machiavelli, Niccolo. 1970. The Discourses on Livy, edited by Bernard Crick and translated by Leslie J. Walker, S.J. with revisions by Brian Richardson. Harmondsworth: Penguin Books.

Mainwaring, Scott.  1993.  Presidentialism, Multipartism, and Democracy: The Difficult Combination. Comparative Political Studies 26 (2, July), pp. 198-228.

Rabushka, Alvin and Kenneth A. Shepsle.  1972.  Politics in Plural Societies.  Columbus,

Sen, Amartya. 1984. Poverty and Famines : An Essay on Entitlement and Deprivation. Oxford: Oxford University Press.

Taylor, Charles Lewis and David A Jodice.  1983.  World Handbook of Political and Social Indicators.  New Haven, Yale University Press (2 vols.).

Umez, Bedford N.  1993.  Has Social Mobilization Caused Political Instability in Africa?  A Granger-Causality Test.  The Review of Black Political Economy 22 (1, Summer), pp. 33-54.

Weingast, Barry R.  1997.  The Political Foundations of Democracy and the Rule of Law, American Political Science Review 91 (June, 2), pp. 245-263.

 

 



[1] See, for example, Linz and Stepan (1978).

[2] This transformation may be constitutional in that the procedures of the old constitution are followed in adopting a new one. This was done, for example, in the gradual transformation of Great Britain into a democratic regime as well as in the transformation of South Africa into a democratic regime. In other cases, old constitutions are simply put aside as a new constitution is promulgated by a Monarch or constituent assembly or a referendum of the people. Whenever the type of regime changes, we will say that a new regime has come into existence, even if this transformation was constitutional in nature. So non-constitutional changes in leadership are only important when there is no change in the type of regime.

[3] Another situation in which one democratic regime might be said to replace another is when the provisions of a written constitution are dramatically reinterpreted. Thus, sometimes the United States is said to have adopted a third regime beginning with the New Deal in 1932 (or, perhaps, in 1937 when the Supreme Court changed direction and finally accepted the New Deal.) Theodore Lowi makes such a claim in The End of Liberalism (1972). For the present, we will not consider such changes in government powers as a change in regime.

[4] At least one study assessed the stability of 97 countries over time: De Haan and Siermann (1996) define  instability as the total number of government changes or “transfers” (as determined in the  World Handbook of Social and Political Indicators) in the period 1963-1988. 

[5] The Federalist, no. 39

[6] A number of political philosophers have set out somewhat different criteria for a regime to be a representative democracy. This list is our own, although it has been influenced by previous writers and, especially, by Robert Dahl, 1989.

[7] There are, of course, important limits to this principle. Thus we do not think of the People’s Republic of China as a democratic government.

[8]Case, 1996, p. 440. Case’s definition of stability appears to be the absence of violent regime turnover. Case finds that Singapore, Malaysia, and Thailand have achieved a significant level of stability and are not in transition toward democracy. He concludes that this set of semidemocracies is “more stable than the new, ‘fuller’ democracies in the region.” However, in the absence of an explicit comparison to a set of democracies, the grounds for his conclusion remain unclear.

[9] For a good, recent survey of these conditions see Dahl (1989) chapter 18.

[10] This would suggest that, over time, democratic governments are likely to emerge when the political and social preconditions for such government are finally in place. It is unfashionable in academia today to adopt such a progressive view of history (along with the Marxist notion that material conditions determine the shape of political and social life.). But we are inclined to think that this story is, at least partly, true. It is not the whole truth, however. We do not think that material political and social conditions, by themselves, determine the shape of political regimes. Ideology and the shape of political arrangements exert an independent influence.

[11] One could plausibly argue that Eastern European communist states should be excluded, on the grounds that they were kept in power by the Soviet Union. It is possible, we suppose, that Communist regimes would have come to power in some of these states without the presence of the Red army. But it is highly unlikely that they would have survived until 1992. On the other hand, it is also true that some Western European regimes were kept in power by large does of American aid. For now, we have decided to include these regimes in our analysis, largely because they are among the longest lived authoritarian regimes in Europe. Thus our data set has been chosen so as to make it less, rather than more, likely to find support for our own hypothesis that democratic regime are as, if not more, stable than authoritarian regimes.

[12] At present, however, the dataset is not entirely complete. No data has been collected for Belgium, Hungary, Poland, Portugal and for the Austro-Hungarian empire prior to World I. Iceland, Cyprus, Liechtenstein, Luxembourg, Malta are excluded due to their small size.

[13] Following Gasiorowski, we also coded some regimes as transitional. A number of kinds of regimes are included in this category. Most of them are regimes that are in the process of creating a transition to democratic government. Some of these regimes were authoritarian regimes that put themselves out of business. Others were regimes formed by elites after a civil or foreign war, such as DeGaulle’s immediate post-war  government in France. Some regimes are considered transitional if they have only a limited hold on the country, due to civil or foreign war. All transitional regimes have been excluded from the statistical analyses.

[14] Thus, in our view, the United Kingdom was a full fledged democracy long before the final emasculation of the House of Lords in 1911.

[15] This transformation may be constitutional in that the procedures of the old constitution are followed in adopting a new one. This was done, for example, in the gradual transformation of Great Britain into a democratic regime as well as in the transformation of South Africa into a democratic regime. In other cases, old constitutions are simply put aside as a new constitution is promulgated by a Monarch or constituent assembly or a referendum of the people. Whenever the type of regime changes, we will say that a new regime has come into existence, even if this transformation was constitutional in nature. So non-constitutional changes in leadership are only important when there is no change in the type of regime.

[16] We have not used either the Political Regime Change Dataset or the Bienen / Van De Walle uncritically. In a few cases we have re-coded the regime type as found in the Political Regime Change Dataset. For example, on our view, Mexico in the last fifty years has had a semi-democratic, rather than democratic regime. And we have changed a few dates for the beginning and endings of regimes as found in the Bienen / Van De Walle data. Our aim, as yet unattained, is to independently check all of the data contained in these two datasets.

[17]Taylor and Jodice, 1983, p. 50.  Burkhart and Lewis-Beck (1994) are some of scholars who have continued the use of this indicator as their measure of democracy, although their scale combines the scores on the political rights and the civil rights indexes.

[18] It would, of course, be interesting to look more closely at these patterns of change from one type of regime to another. We have not yet performed this analysis, however.