The Third Reverse Wave

In the late 1990s there was considerably less optimism about the future of democracy than at the beginning of the decade. The third wave of democratization, by Samuel P. Huntington (1991), started with the democratic revolution in Portugal; it spread to Latin America in the 1980s; and it culminated in the early 1990s with the collapse of the Soviet Union. The mid 1990s saw the beginning of the third 'reverse wave', similar to the two counter-democratic reverse waves that followed the first and second waves of democratization earlier in the twentieth century, from 1922 to 1942 and from 1958 to 1975 (Huntington 1991, 16).

Particularly striking has been the rise of what are often called 'illiberal' or merely 'electoral' democracies, that is, countries that do have more or less free elections by universal suffrage but that lack some or most of the other requirements of democracy, like the freedom to form and join organizations, of expression, and fair competition among political alternatives 1971, 3). These are typically countries that the Freedom House Survey Team (1998) classifies in its 'partly free' category, in between the categories of 'free' and 'not free'. Such illiberal or electoral democracies' of course, not really democratic.

---

* Arend Lijphart, 4276 Caminito Terviso, San Diego, CA 92122, USA.

What are the causes of the third reverse wave? One plausible explanation is based on the up and down trends in democratization since the nineteenth century observed by Huntington. Every forward wave can be expected to be followed by a reverse wave and, because the third wave that started in 1974 was an especially powerful one, it is not surprising that the third reverse wave has been strong, too. However, this explanation is not necessarily reason for long-term pessimism about the future of democracy. If Huntington's trends extend into the twenty-first century, the third reverse wave may well be followed by a fourth forward wave.

A second explanation is the clash of two conflicting forces. The 1980s and 1990s appeared to be the age of democratization, but they were also the age of increased ethnic conflict - and ethnic divisions and conflicts constitute threats to stable democracy. Ethnic strife in many countries, especially those that are not or not yet democratic, is a serious reason for long-term pessimism because there are no signs at all that it will diminish in the twenty-first century.

My third explanation is the continuing popularity of two key democratic institutions that have negative effects for democracy generally and that are particularly unsuitable for ethnically divided societies: presidential government majoritarian election systems. If we look at the long-term consolidated the raw count is on the side of proportional representation (PR) and parliamentary government. For instance, of the 36 democracies that I analyzed in Patterns of Democracy (Lijphart 1999) - defined as countries with populations of at least 250,000 which were democratic the late 1990s and had been continuously democratic since 1977 or earlier - 22 used PR in the late 19905, compared with 13 that used majoritarian systems, and one that used a mixed system (Lijphart 1999,145). A different count was done by Mark P. Jones (personal communication, 25 August 1999). He considers 38 countries with populations of at least two million that were clearly democratic in 1998 (with average Freedom House scores on political rights and civil liberties of at least 2.0 on the scale ranging from a high of 1 to a low of 7 which is used by Freedom House). Jones classifies 26 as having PR systems, six as using majoritarian methods (in single-member districts), and six as mixed.

These numbers are rather deceptive, however. First, PR may be the norm for the established democracies, but majoritarian systems are much more common among countries that are not fully democratic and in Third World countries. Second, even among the well-established democracies, only a relatively small number use majoritarian election methods, but these few countries tend to be the larger countries: India, the United States, the United Kingdom, and France. Japan should be added to this group as well: it adopted a new electoral system in 1996 that has both majoritarian and PR elements, but the majoritarian component is the dominant one because

60 percent of the seats are allocated by plurality in single-member districts
without any proportional adjustments.

Similarly, among the established democracies there are many more parliamentary than presidential systems. Of the 36 democracies listed in Patterns of Democracy, 30 have parliamentary governments, compared with only five that have presidential governments (Lijphart 1999, 119). Switzerland is in an intermediate position but one that is closer to parliamentarism presidentialism, so the ratio between the two systems can be said to be a lopsided 31:5. However, these numbers are also unrepresentative the world at large. Presidential government is quite prevalent among the world's not fully democratic countries and in Third World countries generally. Moreover, the large countries again display a special pattern: Matthew S. Shugart's (1999, 78-81) tabulations of the world's less developed nations show that the more populous that nations are, the more likely they are to have presidential governments.

The Superiority of the Parliamentary-PR Combination

As I stated earlier, I regard the large numbers of countries that still use majoritarian electoral systems and/or presidentialism as a strongly negative factor for the future of democracy. Especially, but not only, in ethnically divided countries, PR and parliamentarism can work much more beneficially. fact, it is the combination of PR and parliamentarism which is optimal. There appears to be a growing scholarly consensus on this point. Moreover, this conclusion has been reached from a variety of quite different perspectives, four of which I shall highlight in this section of my paper: (1) Juan J. Linz's approach which takes the presidential-parliamentary contrast its point of departure, (2) my own distinction between majoritarian and consensus types of democracy, (3) Josep M. Colomer's public-choice argument, and (4) Bruce Ackerman's analysis from the perspective of constitutional

(1) Linz's indictment of presidentialism is so well known that I do not need to spell it out in great detail. He argues that presidential democracy is prone to failure because of serious institutional deficiencies such as the fixed terms of office which make the government very rigid, the tendency to executive-legislative and paralysis resulting from the coexistence of two branches that are separately elected and can both claim democratic legitimacy, the zero-sum or winner-take-all nature of presidential elections - a special problem in deeply divided countries - and the encouragement of the politics of personality instead of a politics of competing parties and party programs.

In his debate with Donald L. Horowitz in the Journal of Democracy, Linz
(1990a, 56) sharpens this conclusion by adding the element of PR to his

preferred parliamentary type of democracy. He recognizes that parliamentarysystems also have majoritarian and winner-take-all characteristics,and therefore adds, 'Although parliamentary elections can produce an absolute majority for a single party, they more often give representationto number of parties. Power-sharing and coalition-forming are fairly common, and incumbents are accordingly attentive to the demands and interests of even the smaller parties.' In his rebuttal, Horowitz (1990, 79) states that this line of reasoning means that 'Linz's thesis boils down to an argument not against the presidency but against plurality election, not in favor of parliamentary systems but in favor of parliamentary coalitions' - in other words, that Linz's contrast is not between the two categories of presidentialismand but between presidentialism and pluralityparliamentarismon hånd and PR-parliamentarism on the other.

Linz's (1990b, 85-6) reply to Horowitz's charge is very important, because it makes clear that he is thinking in terms of three rather than two categories. About plurality-parliamentarism Linz states that 'Mrs. Thatcher ••• probably has more power than an American chief executive [and] parliamentary in which a single disciplined party obtains the absolute majority of all seats find themselves in what is close to a "winner-take-all" situation'. But this does not mean that plurality-parliamentarism can be equated with presidentialism: 'While the actual situation of a powerful prime minister like Mrs. Thatcher might be comparable to that of a president with a legislative majority, the de jure difference is still significant. If Mrs. Thatcher were to falter or otherwise make herself a liability, for instance, the Conservative majority in the House of Commons could unseat her without creating a constitutional crisis' - an event that took place only about a month after these words were published! On the other hand, pluralityparliamentarism also be distinguished from parliamentary-PR government: winner-take-all situations like the British example are 'not the most frequent pattern in parliamentary systems, particularly when there is proportional representation'. Of the three main patterns of democracy, therefore, Linz's argument is that PR-parliamentarism is the most and presidentialism the least desirable form, with plurality-parliamentarism in between.

(2) My own approach has been to measure the majoritarian or consensual character of democracies on the basis of five variables: the degree of executivepower the power relationship between the executive and the legislature, the degree of multipartism, the proportionality of the electoral system, and the degree of corporatism of the interest group system. I find that consensus democracy is clearly superior to majoritarian democracy. There are no big differences in the effectiveness of decision making on macro-economic policy, although consensus democracies have a slightly better record in this respect. But consensus democracies perform a great

deal better with regard to many qualitative aspects of democracy: the representationof and minorities, income equality, voter participation, citizens' satisfaction with democracy, and the proximity between the governmentand median voter (Lijphart 1999, 258-93).

My next question is: which constitutional and electoral law characteristics associated with consensus democracy? The answer: parliamentarism and PR. Countries that have both parliamentary government and PR elections likely to be on the consensus side of the majoritarian-consensus continuum; countries that have presidential government or majority elections, both, are likely to be on the majoritarian side (Lijphart 1999, 303-4). My dependent variable differs from Linz's: my emphasis is not on the success or failure of democracies in terms of their survival, but on the performance record of democracies that have successfully survived. But our conclusion concerning the superiority of the parliamentary-PR alternative is identical.

(3) Spanish political scientist Colomer (2000) uses social choice theory to determine which are the most 'socially efficient' institutions, that is, which institutions maximize political satisfaction. He argues that this quality can be measured in terms of whether the party of the median voter is included among the winners and in the executive. According to his logic, the best systems are the parliamentary-PR ones; the worst are parliamentaryplurality and presidential and semi-presidential systems are intermediate these two. Colomer's rank order differs from Linz's, but both give the PR-parliamentary option their highest ranking. Colomer also tests his conclusions by means of quantitative analysis, which strongly supports theoretical arguments.

(4) Ackerman (2000), a professor in the Yale School of Law, presents a lengthy constitutional-legal analysis in which he first contrasts presidentialism the Westminster model of parliamentarism, and then introduces third model which he calls 'constrained parliamentarism'; he concludes that this third model is far superior to the other two. His primary empirical examples are, respectively, the United States, Great Britain, and Germany - which can also be used as the prototypical examples of presidentialism (the United States), plurality-parliamentarism (Britain), and PR-parliamentarism (Germany). The 'constraints' that Ackerman has in mind for his constrained parliamentarism are not so much PR - although he clearly favors PR - than judicial review and occasional referendums. parliamentarism therefore does not neatly coincide with PR-parliamentarism. Nevertheless, Ackerman's analysis can be read as at least an indirect and partial confirmation - from a fourth theoretical angle - of Linz's, Colomer's, and my own conclusions.

The only strong counter-argument is Shugart's (1999) contention that
presidentialism has important advantages for Third World countries because

political parties in developing countries tend to be weak, and hence poor interest aggregators. Presidents, he argues, may be produced by winner-takeallelections they are still much more broadly representative than the members of typically fragmented legislatures, and they can therefore assume the vital aggregative function.

This line of reasoning certainly has some merit, but I find it not entirely persuasive. For one thing, the development of strong parties is much less likely under presidentialism than under parliamentarism, because parliamentary needs cohesive and disciplined parties to support cabinets it therefore encourages the development of such parties. The classic example is the contrast between Canada and the United States; in Seymour Martin Lipset's (1990, 81) words, 'the difference between presidential parliamentary systems in comparable continent-spanning, federal polities results in two weak parties in the United States and multiple strong ones in Canada'. Moreover, it is too defeatist to simply accept as an unalterable given that parties in a particular country are too weak for parliamentary government and hence to look for alternatives to parliamentarism. is also possible - and much more constructive - to stimulate the growth of strong parties by direct measures; for instance, there are PR systems that make for cohesive parties, such as list PR with closed lists, and parties can also be strengthened by the regular receipt of state subsidies.

The Instructive Case of Bolivian 'Presidentialism'

The kind of quantitative analysis performed by Colomer, mentioned above, is similar to the many quantitative studies of the merits of parliamentary versus presidential government which do not take the second variable of the electoral system into consideration. Fred W. Riggs (1988) was the pioneering in this respect, and he found presidentialism to be extremely prone to failure and hence a highly 'problematic regime type'. Of the several later studies that came to the same conclusion, the one by Alfred Stepan and Cindy Skach (1993) is probably the best known. There have also been a few studies that have shown no significant differences between the two types (for instance, Power & Gasiorowski 1997). To my knowledge, however, not a single study has been produced that shows that presidentialism actually better than parliamentarism. This is in itself a significant finding that can make us more confident about recommending the parliamentary alternative.

However, these studies are beset by serious problems of definition and measurement. There is no consensus, for instance, on the exact definitions of presidentialism, parliamentarism, and semi-presidentialism, and thereforeno in classifying specific countries in these categories either. Another difficulty is how exactly to define and measure the 'success' of the

different types of democracies or how long they have to endure in order to qualify for 'survival'. Perhaps an even more serious problem is to control for other variables that may affect success and survival. In particular, presidential systems are concentrated in Latin America, and it has been argued that it is the political culture of the countries in this part of the world that can explain their sorry record of democracy, rather than their presidentialsystems government (Lipset 1990).

One Latin American case is of special interest: Bolivia. Since its return to democracy in the early 1980s, Bolivia has been one of the democratic bright spots in the region. This is particularly remarkable because Bolivia is a poor country and because it has one of the worst histories of military coups and dictatorships. What distinguishes Bolivia from other Latin American countries is that its political system is more parliamentary than presidential in spite of its formally presidential constitution; Bolivia also has PR elections (which is not unusual in Latin America).

One important parliamentary characteristic is that the president is normally by the legislature instead of directly by the voters. The first stage of the electoral process is still a popular election but if no presidential candidate receives an absolute majority of the votes the legislature selects the president from the top two candidates (the top three candidates prior to the 1997 election) - with no certainty, or even high probability, that the plurality winner of the popular contest will be selected. Because Bolivia has a multiparty system and multiple presidential candidates, no candidate has ever been elected by popular vote in the 1980s and 19905, and a legislative coalition has always needed to be formed to make the final choice. The second parliamentary feature is that this process has resulted in the appointment of a multiparty coalition cabinet with real power instead of a cabinet that is simply appointed by the president as a mainly advisory body. The Bolivian executive is therefore much closer to the collegial executive that is typical of parliamentary systems than to the one-person executive that is the hallmark of presidentialism.

In only one respect is Bolivia more presidential than parliamentary: the president has a fixed term of office and, once elected, cannot be removed by the legislature. Linz has coined the term 'parliamentarized presidentialism' for the Bolivian hybrid, which operates much more like a parliamentary system - and a PR-parliamentary system - than a presidential one (cited in Mayorga 1997, 143-4).

The recent success of Bolivian democracy can clearly not be explained in terms of a political culture that is different from that of the other Latin American countries. Nor is its economic condition more favorable than that of the neighboring countries; if anything, it is much less favorable. Nor does it have a more favorable democratic historical background than its neighbors in the region; here again, if anything, its history is less of a

positive factor. The conclusion seems inescapable, therefore, that it is the mainly parliamentary fashion in which its 'presidential' system works that has made the crucial difference. This conclusion is confirmed by the judgmentof Antonio Mayorga, Bolivia's best-known political scientist. Mayorga (1997, 143) attributes the country's exceptional democratic successto 'system of interparty bargaining, postelectoral coalitions, consensualpractices, congressional election of the chief executive', and he suggests that Bolivia may become an attractive 'model for imitation' by other Latin American countries.

Conclusion

My conclusion can be summarized in just a few sentences. I believe that a serious problem for the success and spread of democracy in the twenty-first century is the continued popularity, especially among political leaders in not yet or not fully democratic countries, of two major institutions that have negative consequences for democracy: presidential government and majoritarian election systems. This factor, together with the salience of ethnic tensions in many countries, necessitates serious pessimism about the future of democracy. However, politicians do have the ability to learn, and it is certainly not impossible that they will start to listen to the conclusions and recommendations of political scientists. Concrete examples like the highly significant and instructive Bolivian case of parliamentarized presidentialism also have an impact. For these reasons, I have added the words 'and also some optimism' to the title of this article.

REFERENCES

Ackerman, B. 2000. 'The New Separation of Powers', Harvard Law Review 113, 634-729.
Colomer, J. M. 2000. Political Institutions: Social Choice and Democracy. Oxford: Oxford
University Press (forthcoming)

Dahl, R. A. 1971. Polyarchy: Participation and Opposition. New Haven: Yale University
Press.

Freedom House Survey Team. 1998. Freedom in the World: The Annual Survey of Political
Rights and Civil Liberties, 1997-1998. New York: Freedom House.

Horowitz, D. L. 1990. 'Comparing Democratic Systems', Journal of Democracy 1, 73-9.
Huntington, S. P. 1991. The Third Wave: Democratization in the Late Twentieth Century.
Norman: University of Oklahoma Press.

Lijphart, A. 1999. Patterns of Democracy: Government Forms and Performance in Thirty-Six
Countries. New Haven: Yale University Press.

Linz, J. J. 1990a. 'The Perils of Presidentialism', Journal of Democracy 1, 51-69.
Linz, J. J. 1990b. 'The Virtues of Parliamentarism', Journal of Democracy 1, 84-91.
Lipset, S. M. 1990. 'The Centrality of Political Culture', Journal of Democracy 1, 80-3.
Mayorga, R. A. 1997. 'Bolivia's Silent Revolution', Journal of Democracy 8, 142-56.
Power, T. J. & Gasiorowski, M. J. 1997. 'Institutional Design and Democratic Consolidation
in the Third World', Comparative Political Studies 30, 123-55.

Riggs, F. W. 1988. 'The Survival of Presidentialism in America: Para-Constitutional
Practices', International Political Science Review 9, 247-78.

Shugart, M. S. 1999. 'Presidentialism, Parliamentarism, and the Provision of Collective
Goods in Less-Developed Countries', Consiitutional Political Economy 10, 53-88.

Stepan, A. & Skach, C. 1993. 'Constitutional Frameworks and Democratic Consolidation:
Parliamentarism versus Presidentialism', World Politics 46, 1-22.