Economic Performance, Institutional Intermediation, and Democratic Survival Michael Bernhard; Timothy Nordstrom; Christopher Reenock The Journal of Politics, Vol. 63, No. 3. (Aug., 2001), pp. 775-803. Stable URL: http://links.jstor.org/sici?sici=0022-3816%28200108%2963%3A3%3C775%3AEPIIAD%3E2.0.CO%3B2-W The Journal of Politics is currently published by Southern Political Science Association. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/spsa.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact email@example.com. http://www.jstor.org Mon Aug 27 11:38:42 2007 Economic Performance, Institutional Intermediation, and Democratic Survival Michael Bernhard Pennsylvania State University Timothy Nordstrom University of Mississippi Christopher Reenock University of Alabama The breakdown of democracies has long been associated with poor econon~icperformance. This study attempts to determine whether different configurations of democratic institutions can mediate the effects of poor economic performance. Using an original data set that includes all democracies from the period 1919 to 1995, we use continuous-time duration analysis to test hypotheses derived from the literature on democratization. Specifically, we test the interaction of party system and the configuration of legislative and executive power (parliamentarism and presidentialism) with economic performance to explain the likelihood of breakdown. Results suggest that majoritarian variants of democracy are more resistant to economic contraction than pluralist ones. Under conditions of economic growth, pluralist democracies outperform majoritarian ones. T h e question of why some states are democratic and some are not lies at the heart of contemporary comparative politics. Issues of regime type and regime change are likely to endure because of the important normative ramifications of democracy. Political scientists have approached regime change in two ways: (1) the transition to democracy, and (2) the breakdown of democratic regimes. A majority of recent work has focused on the former due to the preponderant direction of regime change in the last few decades. In this article we examine which factors determine whether democracies break down or survive. Within the literature, there are two predominant schools of thought. The first focuses on economic performance while the second emphaWe thank the following people for help in compiling the data set: Glenn Adler, Barry Ames, D. Scott Bennett, Thomas Callaghy, David Fleishman, Joshua B. Forest, John Martz, Jean Francois Mayer, Kurt van Mettenheim, Alex Motyl, David Myers, Karen Remner, Holly Reynolds, and Meredith Woo-Cumings. We also thank Eric Plutzer and D. Scott Bennett for helpful suggestions in the data analysis. THE JOURNAL OF POLITICS, Val. 63, No. 3, August 2001, Pp. 775-803 0 2001 Blackwell Publishers, 350 Main St., Malden, MA 02148, USA, and 108 Cowley Road, Oxford OX4 lJF, UK. 776 Michael Bernhard, Timothy Nordstrom, and Christopher Reenock sizes political institutions. To date, only a few large-n studies have attempted to integrate these two lines of thought. Absent from this literature has been a thorough investigation of the durability of different institutional patterns in varied economic contexts. Specifically, the interaction of economic and institutional factors has not been fully explored. In this article we examine whether certain types of democracy (majoritarian and pluralist) are better insulated from breakdown given different kinds of economic performance. We constructed these types taking into account two aspects of institutions highlighted in the literature on democratic survival-party systems and macro-institutional design. The key theoretical linkage between institutions and economic performance, we argue, lies in the qualities of dispersing or concentrating power in majoritarian and pluralist democracies. To test our hypotheses, we constructed an original data set that includes all democracies in the period 1919-1995. The data set consists of 115 countries with 156 episodes of democracy. We use continuous-time, event-history methods to analyze the data. We begin with a discussion of the relevant literature and a presentation of the theoretical positions from which we derive our hypotheses. Then we discuss our data and research design. Finally, we discuss our results and assess their significance. Economic Performance and the Survival of Democracy Works that link the fate of democracy to economic factors emphasize either the effects of economic performance (Gasiorowski 1995; Linz 1978; Przeworski and Limongi 1997), pace of development (Huntington 1968; Olson 1982), or level of development (Bollen 1979; Cutright 1963; Gasiorowski and Power 1998; Lipset 1959; Neubauer 1967; Przeworski et al. 1996).' These studies find that higher levels of economic development and positive economic performance enable democracies to endure. An exception, Olson's thesis that rapid economic growth threatens democracy, has found less empirical support. This literature argues that positive growth and higher levels of development encourage the survival of democratic regimes by alleviating socioeconomic conflict. Poor ecollomic performance and lower levels of development raise the stakes of socioeconomic conflict, raise levels of discontent, and increase the attractiveness of extra-systemic solutions, thus increasing the likelihood of regime breakdown (Dahl 197 1; Haggard and Kaufman 1995; Huntington 199 1; Im 1987; Linz 1978; Lipset 1959; Seligson and Muller 1987). 'While we are aware of the distinction between treatments of democratic "survival" and treatments of liberal democracy, the results of each of these streams of research have implications for the study of democratic survival. Negative movements on a scale of liberal democracy may be suggestive of eventual breakdown. Economics, Institutions, and Democratic Survival 777 Whereas there is a long tradition of linking level of development and regime type? recent scholarship has also linked level of development to democratic survival. Przeworski and Limongi conclude that "once established, democracies are likely to die in poor countries and certain to survive in wealthy ones" (1997, 167). They base this on the finding that democracies with a GNP per capita of less than $1,000 had an expected lifetime of eight years, those between $1,001 and $2,000, 18 years, and those above $6,055 were found to be essentially impregnable (Przeworski et al. 1996; Przeworski and Limongi 1997). Following this literature, we expect highly developed countries to be less prone to breakdown. Poor economic performance, conceptualized in terms of economic contraction, has also received substantial attention in the l i t e r a t ~ r e Downturns of this .~ nature, particularly if they are severe and prolonged, are strongly associated with breakdown in all democratic regimes, especially newly inaugurated ones (Haggard and Kaufman 1995). The impact of a year of negative growth on democratic survival is quite robust and consistent across various samples. Findings from South America between 1945 and 1988 (Przeworski and Limongi 1993), the postwar third world (Gasiorowski 1995), and the period 1950 to 1990 (Przeworski and Limongi 1997) show that a single year of negative growth is enough to significantly increase the probability of democratic b r e a k d ~ w n . ~ When negative growth lasts for more than one year, the effects on democratic survival are even bleaker (Przeworski and Limongi 1993). Therefore, we expect that negative growth will produce democratic breakdowns. Olson (1982) and Huntington (1968) suggested that rapid economic growth could also destabilize democracy. Accelerated growth is said to produce social upheaval by disrupting traditional social relations. Both the nouveau riches and 'Early empirical work identified a positive linear relationship between democracy and development (Cutright 1963; Lipset 1959). Subsequent work found a curvilinear relationship; democratic performance increased and then leveled off at a certain level of development (Jackman 1973; Neubauer 1967). Challenges to these findings have suggested that perhaps modernization theory is at best a weak predictor of democracy (Arat 1988; Gonick and Rash 1988). Improved measures and methodology have since confirmed modernization theory in its curvilinear form (Burkhart and Lewis-Beck 1994). Dissenting from this view, O'Donnell (1973) and Lipset, Seong, and Torres (1993) have argued that economic development does not always promote democracy. There are historical paths of development that may result in increased development leading to authoritarian political solutions (bureaucratic-authoritarianism being the most commonly cited). 3There is a literature that attempts to gauge the impact of inflation on democratic survival, although its findings have been modest. Moderate inflation actually seems to promote democratic stability. However, hyperinflation (over 30%) has strong deleterious effects, reducing a democracy's life expectancy from 71 years (with moderate inflation) to an average of 16 years (Przeworski et al. 1996). Gasiorowski (1995) and Power and Gasiorowski (1997) found inflationary pressures increase the risk of breakdown for democratic regimes, but only in the period prior to the early 1970s. "he effect of negative growth is greater for poor democracies. For the poorest nations experiencing negative growth, Przeworski and Limongi (1997) found an average democratic life expectancy of nine years. The comparable figures for mid-level and wealthy nations were 20 years and "everlasting life." 778 Michael Bernhard, Timothy Nordstrom, and Christopher Reenoclc the nouveau pauvres will develop new expectations and frustrations about their positions in society that may destabilize democracy. There is little evidence to support the contention that high positive growth rates destabilize democratic regimes. In fact, the research shows just the oppositethat positive rates of economic growth, measured in change in per capita GDP, deter coups d'etat (Londregan and Poole 1990) and promote stable democracy (Haggard and Kaufman 1995). Haggard and Kaufman report that in a collective 462 years of positive economic growth between 1960 and 1990, democracies survived 97% of the time, leading them to conclude that growth has beneficial effects by "reducing the frustrations and conflicts resulting from inequality or other social cleavages" (1995, 325-26). However, Olson's argument concerns the pace of development, not the existence of positive growth itself. Olson believed that high rates of positive growth could cause democratic breakdown. It is possible, given the right set of circumstances, that unusually high rates of growth will lead to democratic breakdown. Institutional Variation and Breakdown The literature on consolidation and stability of democratic regimes points to a number of different institutional variables that affect democratic survival. Both macro-institutional structure and party systems have received substantial attention. Macro-institutional structure is the way in which the relationship between the executive and legislature is configured. The main varieties are presidentialism and parliamentarism. Linz (1994) and Lijphart (1994a) have argued that the former contributes to the fragility of democracies by promoting a number of pathologies (e.g., winner-take-all electoral competition, the rigidity of fixed terms for the executive, and a propensity toward deadlock between the executive and legislature) and that the latter is a superior choice for democratic stability. Riggs (1993) and Stepan and Skatch (1994) have assembled supporting data from the developing world. Przeworski et al. (1996), in a survey of all democracies from 1950 to 1990, also provide strong evidence of this thesis. Shugart and Carey (1992), who include interwar Europe in their sample, do not present as clear a finding, showing the performance of both presidentialism and parliamentarism to be mixed. To the contrary, Power and Gasiorowski (1997) and Mainwaring and Shugart (1997) present evidence that raises doubts about the greater fragility of pre~identialism.~ There is also an older literature on democratic breakdown that focused on party and electoral systems rather than macro-institutional structure. Electoral 'The difference between Power and Gasiorowski's (1997) study and those of the others may be a product of how they define and operationalize their dependent variable, democratic consolidation. Of their three measures, the most rigorous considers democracies consolidated after 12 years of continuous operation. This includes some cases that subsequently broke down and are not considered durable democracies in other studies. In their subsequent study Gasiorowski and Power had no consistent findings on presidentialism and consolidation (1998, 758). Economics, Institutions, and Democratic Survival 779 systems have important ramifications for the composition of the legislature and the nature of the party system, including whether electoral victory translates into a legislative plurality or majority, as well as the total number of parties represented in the legislature. In systems where governments are formed on the basis of parliamentary support, this bears on whether governments are singleparty or coalitional in nature and how many parties are necessary to form a government. In the case of coalition governments, the number of parties can affect their potential for stability. In presidential systems, such features will determine whether government is divided or not, increasing the potential for interbranch conflict. In the literature, both two-party and moderate multiparty systems are described as being much more resistant to democratic breakdown than extreme multipartism (Linz 1978; Sartori 1976, 140). In this article, we do not consider the effects of the two dimensions in isolation from each other but in terms of their interaction. It is quite clear that different combinations of macro-institutional structure and electoral rules have different effects on the performance of the political system. Indeed, Horowitz (1990) pointed out that parts of Linz's original criticism of presidentialism were based on political outcomes that were, at least in part, attributable to the electoral system. Presidentialism, when combined with single-member district voting systems, is quite different from when it is combined with purely proportional representation. It certainly affects how often divided government occurs. The strongest arguments in this regard concern the interaction of presidentialism and proportional representation (particularly in the Latin American context). Most observers point to the regularity with which this combination breaks down (Mainwaring 1990; Przeworski et al. 1996; Shugart and Carey 1992; Valenzuela 1993). Here again, Power and Gasiorowski dissent from the consensus: "multipartism poses no apparent obstacles to the success of presidential rule" (1997, 146). While macro-institutional structure and party systems have been used separately to explore the stability of democracy, they can be combined to produce a typology of democratic regimes. It is precisely such a typology that we use to more systematically explore interaction effects as they relate to our dependent variable. In Figure 1, we distinguish between the two main types of macroinstitutional structure (parliamentary and presidential). Party systems are categorized according to patterns of representation. This is a product of the interaction of the political and socioeconomic cleavages in society and the fashion in which the electoral system translates votes into seats.6 6~rom an institutional perspective, the way in which an electoral system translates the percentage of votes received in an election to the percentage of seats received in a representative body, or what can be described as proportionality, is most important. In highly proportional systems, the percentage of votes received will be very close to the percentage of seats awarded. Disproportional systems are restrictive: they reward larger parties with a greater share of the mandates in the legislature than their percentage of votes. 780 Michael Bernhard, Timothy Nordstrorn, and Christopher Reenock FIGURE 1 Types of Democratic Systems According to Macro-institutional Structure and Party System Macro-institutional Structure Party System Predominant party Two Party Moderate Multiparty Extreme Multiparty We have divided party-systems in a quadripartite fashion based on distinctions in the qualitative literature (especially Sartori 1976. Chap. 6). Predominant party arrangements include those cases in which one party rules with a decisive legislative majority. The next demarcation. two-party, includes cases where one-party receives a majority, but one not so overwhelming as the predominant-party type. The next class in this dimension is moderate multiparty. This includes cases that require a small number of parties to form a governing coalition or pass legislation (corresponding to the cases that Sartori describes as moderate pluralism). The last class, extreme multipartism, represents a situation of high fractionalization in the legislature (corresponding to those systems that Sartori qualified as extreme pluralism (1976, 261). As systems move from left to right in the party dimension, we expect that they will give up governability at the expense of representativeness (i.e., a higher degree of multipartism will be more representative of society, while a legislature with fewer parties will facilitate the task of forming majorities). The interaction effects in which we are interested are those that Colomer (1995, 95) has described as majoritarian and pluralist patterns of democracy.' Colomer's categories clearly owe much to the notions of "major~tarian" and "consensus" democracy that Lijphart coined in h ~ long-term s study of democrat~cinstitut~ons(the latest iterat~on is Lijphart 1999). The atnbiguous position of presidentialism in Lijphart's formulation makes h ~ s conceptualization less useful than Colomer's for our purposes. It is important to note that Lijpllart's own contribution to the presidentialism/parliarnentarism debate (1991a) treats presidential~smas majoritarian because o f the winner-take-all nature of presidential elections. Economics, Institutions, and Democratic Survival 78 1 We are interested in these patterns because they represent the extremes in the ways in which democracies concentrate or diffuse power. Majoritarian systems concentrate power to the greatest degree by fusing executive and legislative power in parliamentary governments based on disciplined one-party rule. The latter is often promoted by restrictive electoral systems (majoritarian, SMD, or other variations that promote governability at the expense of representation).x These sorts of systems are concentrated in the upper left-hand corner, the darkest shaded area, of Figure 1. Those systems that diverge by one class in the party system dimension (parliamentary two-party systems) have been qualified as quasi-majoritarian. Pluralist systems disperse power to the greatest degree by separating executive and legislative functions through presidential government and have highly fractionalized legislatures. The latter is quite often the product of highly proportional electoral systems. These sorts of systems are concentrated in the lower right corner of Figure 1, in the white cell. As above, those systems that diverge by one class in the party dimension we have qualified as quasi-pluralist. Our residual category, "mixed" (denoted in light gray), includes systems that concentrate power in one dimension while dispersing it in a n ~ t h e r . ~ Putting Economics and Institutions Together It is the qualities of dispersing or concentrating power in majoritarian and pluralist democracies that we theorize as important for how democracies fare in relation to economic performance. Majoritarian systems should fare better under conditions of crisis because their ability to concentrate power should allow them to formulate politically coherent policy responses to crisis. First, governments will tend to be based on one or a small number of parties. Second, the parliamentary majorities on which governments are based will tend to be larger than the share of the votes on which they were elected (sometimes substantial majorities will even be based only on popular pluralities), reducing the diversity and range of actors that need to be incorporated into a policy 'There are exceptions to this. For instance, given a strongly bifurcated polity, a two-party system would not be out of the question under proportional representation. With a substantial number of locally concentrated minorities, multipartism is not incompatible with a single member district. "t may strike some readers as odd that we pair systems that seem very different together in the mixed category. Conventionally speaking, parliamentarisin with an extreme multiparty system is seen as very different from presidentialism with a predominant party system. However, both fall into our mixed category because they concentrate power in one dimension and disperse it in another. In a parliamentary extreme-multiparty system, parliamentarisin concentrates power while extreme multipartism disperses it. In a presidential predominant-party system, presidentialism disperses power while the predominant party system concentrates it. Thus, both of these con~binations concentrate power in one dimension and disperse it in another. though they do so in opposite dimensions. However, in the power concentrating/dispersing characteristic of systems we hypothesize as important to economic crisis management, both are mixed. 782 Michael Bernhard, Timothy Nordstronz, and Christopher Reenoclc consensus. Third, the fusion of executive and legislative power and the incentives that parliamentary systems pose for party loyalty within the prime minister's party also simplifies the task of framing a policy response to the c r i ~ i s . ' ~ Under conditions of economic crisis, pluralist systems with their propensity to disperse power are inuch more likely to deadlock, making them more prone to break down. Their ability to formulate responses to crisis is more limited because of the necessity of formulating a consensus or compromise between the legislative and executive branches. The highly representative nature of the electoral system almost always makes it difficult for the party of the president to achieve a majority in the legislature. This leads to an almost permanent condition of divided government, which under crisis conditions will strongly complicate the business of achieving legislative majorities. Further, the diversity of interests that any legislative majority will have to incorporate into a coalition to support a response to crisis will be broader, on average, than under majoritarian electoral systems. This is because strongly representative systems admit a broader variety of interests into the legislature. For the reasons outlined above, pluralist democracies are much more likely to deadlock under conditions of economic crisis than inajoritarian democracies and thus should exhibit a higher level of breakdown. This leads to our first hypothesis: H I : Under conditions of economic contraction, plur*alist systems shot~ldexperience shorter episodes of denzocracy than majoritarian systems. Following the same logic as above, we expect pluralist systems to fare better than majoritarian ones under conditions of economic expansion due to the more representative nature of their party systems. Under conditions of growth, it should be relatively easy to satisfy the broader constituencies that would underlie a majority under pluralism. This should create greater overall satisfaction with democracy. Majoritarianism, which produces legislative majorities on the basis of narrower social constituencies, has the potential to distribute the benefits in an unequal fashion. If majoritarian systems reward only the narrow interests on which they are based, they risk losing the advantages that economic growth may provide. This leads to our second hypothesis: H2: Under conditions of economic expansion, ~zajoritariansystems should experience shorter episodes of denzocracy compared to plz~ralistsystems. here needs to be a strong qualification introduced here. The parliamentary systems that Sartori calls "assembly government" would probably not function in this way during crisis because of the absence of a strong prime minister (1994, 110). This should not affect our findings because most parliamentary systems of this type rely on proportional representation. They should thus fall in the mixed boxes in the upper right-hand corner of Figure 1. Economics, Institutions, and Democratic Survival 783 Data and Measurement Our unit of analysis is the democratic episode, where episode represents a distinct period of democracy in a country's history. While previous data sets have been limited either to the post-World War I1 period or to the third world, we included all cases of democracy beginning with the interwar period.11 We began our coding of cases either in 191912 or with a new transition to democracy and ended them with either a democratic breakdown or a right censoring in 1995.13 We define as democratic any regime that approaches Dahl's (1971) minimum criteria for polyarchy, specifically those that permit a high level of contestation and include a large part of the adult population. We began our consideration of which cases to include by consulting the existing data sets that classify polyarchies or democratic regimes over a substantial period of timePolity 111(Jaggers and Gurr 1995), the Political Regime Change Dataset (Gasiorowski 1996), Freedom in the World (Freedom House), and Classzfiiizg Political Reginzes (Alvarez et al. 1997). Where there was substantial consensus between the data sets, we included these cases. However, because the time frames of these data sets were different, they differed on certain cases, and we were skeptical of particular codings, we consulted a wide variety of general works on democratization and political systems, as well as individual country histories, in order to make definitive decisions about which countries to include. Many data sets that attempt to operationalize polyarchy explicitly ignore the participation component of Dahl's conceptualization (Alvarez et al. 1997, 19; Bollen 1991, 6-7; Gasiorowski 1991). Since many of these studies only consider the post-World War I1 period, by which time most franchise restrictions had disappeared, this was not as important as it was for our study. We were compelled to pay greater attention to the participation component of Dahl's definition14 because of the even larger number of cases in which the political system was competitive but not fully inclusive in the interwar era. ' l While we have democracy-episode and institutional data for almost all the interwar cases, we were able to collect utilizable economic data in only 76.5% of these cases. "In the case of the several democracies whose origins predate 1919, we began their series in 1919, left censoring the earlier portion of their episodes (Australia, New Zealand, Norway). We treat the two other cases (France, Switzerland) that often are considered early democracies differently for reasons that will become clear later. '"n certain cases where democracy was interrupted by German conquest and occupation in WWII, we have omitted the wartime years and treated democracy as continuous. I4For instance, Przeworski et al. define democracy as "a regime in which governmental offices are filled as a consequence of contested elections. Only if the opposition is allowed to compete, win, and assume office is a regime democratic" (1996, 50). This definition ignores the inclusive participation component of Dahl's (1971) concept. Given that their study looks at democracy from 1950 to 1990, this is not a problem because most of their cases of new democracies evolved out of mass polities that were already inclusive. 784 Michael Bernhard, Timothy Nordstrovz, and Christopher Reenock Those studies that pay attention to the inclusiveness criterion, including Dahl's, do not provide clear-cut guidelines to operationalize the level of inclusiveness necessary for a polyarchy.'5 From a conceptual point of view, we found Coppedge and Reinecke's quadripartite demarcation ("universal adult suffrage," "suffrage with partial restrictions," "suffrage denied to large segments of the population," "no suffrage") the most useful (1991, 50). We chose a cut-off point of 50% for inclusiveness because it seemed to capture their two first distinctions. This also seemed to capture cases that would fall within the upper right-hand quadrant of the property space that Dahl developed to characterize polyarchy and other forms of rule (1971, 7). There were two other reasons why we chose this cutoff point. First, the quantitative literature on democratization and democratic survival is composed of data sets that already include similar cases.16 This is the literature that this article addresses, and we wanted our results to be comparable to those of the earlier studies. Second, in order to increase the size of our interwar sample, we engaged in a moderate degree of concept stretching to increase the number of cases in our study. This means that our data set includes both polyarchies and near polyarchies (notably a number of competitive Latin American regimes that posed literacy requirements on franchise, which has the effect of excluding a portion of the lower classes, and the United States prior to the inclusion of African Americans in the South with the passage of the Voting Rights ~ c t ) . " Because of the 50% cutoff, our operationalization excludes all cases where women were denied the franchise. We excluded countries for other reasons as well. In cases where we detected substantial voting fraud, enough to change the outcome of the election,Ix we did not consider a case polyarchic. We also paid careful attention to questions of sovereignty, not including in our sample many cases in which, despite competitive elections, full sovereignty was not formally achieved. We excluded from our set of democracies many countries in which internal wars and extensive civil disturbances were contemporary to elections. In cases in which there was either extensive or extreme violence connected to the electoral process that I 5 ~ a hbreaks l participation into three categories (under 20%, 20%-90%, and over 90%). Hadenius' study excludes the two countries (South Africa and Western Samoa) in which less than 20% of the population was allowed to vote (1992, 39-40). In his second data set, Gasiorowski elaborates a participation criterion: "a highly inclusive level of political participation exists in the selection of leaders and policies such that no major (adult) social group is excluded" (1997, 471). He does not, however, explicate how he operationalized this. 16For instance, both Alvarez et al. (1997) and Gasiorowski (1996) include Brazil and Chile in the 1950s when there still were substantial literacy restrictions on voting. Similarly, Bollen (1991) gives both countries very high democracy scores for 1960 (Brazil 90.5, Chile 99.7). 170bviously,it is not our intent to sanction such antidemocratic practices as democratic by including these cases in the data set. " ~ tyhe phrase "enough to change the outcome of the election," we mean the election of a different candidate to chief executive or to substantially alter the balance of party forces in the legislature. Economics, Institutions, and Democratic Survival 785 inhibited voters from expressing themselves openly, we excluded those cases. We also excluded cases in which political parties representing a large portion of the political spectrum were banned from parti~ipation.'~ However, in these instances we made allowances for "militant democracy."20 The cases and the years of their duration are included in Appendix 2. Dependent Variable The dependent variable is the duration (in years) of a country's democratic episode(s). Each individual episode of democracy is coded separately. The data set contains 2,693 country-year cases within which were 156 episodes of democracy. Democratic duration ranged from 1 to 77 years, with an average life span of 19.8 years. Given that we collected data in yearly increments, coding the initiation and termination of an episode required special attention. We adopted a first quarter cutoff decision rule for regime initiation: if a nation initiated democracy in the first three months of year t, the initiation was coded as having occurred in year t - 1. If the initiation occurred after the first three months of year t, we coded the initiation as having occurred in year t. The same decision rule applied for the breakdown of a regime. If an episode did not experience a breakdown by 1995, we right censored the case. The first quarter cutoff is justified for the following reason: given that economic performance is the main explanatory factor in our models and that the effects of a bad economy may lag political responses on the order of months rather than years, we expect fairly immediate economic conditions to affect political relations within a state. Independent Variables Our main independent variables measure economic performance and institutional configuration. We include several additional variables to control for factors that have been associated with democratic breakdown in the literature. We collected economic data from Monitoring the World Economy (1995), the International Financial Statistics Yearbook, UN publications such as the United Nations Statistical Yearbook, and country-specific sources. We collected data on party distributions within the legislature and executive-legislative relations from varI9As in the case of voting fraud, we omitted cases when we thought such exclusion was significant enough to change electoral outcomes. he notion of militant democracy (from the German Styeitbare Den~okratie)holds that democracies may ban parties that are con~mittedto their violent overthrow. In West Germany, it has been used to exclude antidemocratic political forces on the far right and far left. In cases where banned parties were committed to the violent overthrow of the political system, served as stalking horses for aggressive external powers (e.g., the communists in Finland in the interwar era), or were associated with former ruling authoritarian forces that committed crimes against humanity (e.g., neoNazis in Germany), we did not consider their banning sufficient condition for excluding a country as a democracy. 786 Michael Bernhard, Timothy Nordstrom, and Christopher Reenock ious sources, including The Political Handbook of the World, the CIA'S World Fact Book, The Journal of Democracy, The International Almanac of Electoral History (Mackie and Rose 1982), Electoral Studies, Encyclopedia Electoral Latinoamericano y del Caribe (Nohlen 1993), secondary accounts, statistical annuals, and consultations with specialists. We utilized data on religious and ethnic fractionalization from the Cultural Composition of Interstate System Members, Correlates of War Project (Singer 1997) that we supplemented with statistical annuals. ECONOMIC PERFORMANCE. We used the annual change in real GDP per capita for each case, measured as the proportion of change from the previous year in constant 1990 units.21 This allows us to investigate the unit effects of changes in growth, positive or negative, on the hazard rate. INSTITUTIONAL FORM. We construct institutional form from two components: macro-institutional structure and the number of parties within the legislature. The former is an ordinal-level variable ranking the separation of legislative and executive power. This variable takes on two possible values (Parliamentary = 0, and Presidential = 1). We classified executive-legislative relations according to Sartori (1994). Systems were classified as Presidential "if and only if, the head of state i) results from popular election, ii) during his or her pre-established tenure cannot be discharged by a parliamentary vote, and iii) heads or otherwise directs the governments that he or she appoints" (Sartori 1994, 84). Parliamentary systems are those in which parliament is sovereign. "Thus parliamentary systems do not permit a separation of power between parliament and government . . . all the systems we call parliamentary require governments to be appointed, supported22 and, as the case may be, discharged, by parliamentary vote" (Sartori 1994, 100). The other variable in this construct is the number of parties in the legislature. We attempt to capture the majority-building potential of a legislative body by considering the number of parties in the lower house. We used the LaaksoTaagepera index (1979) to calculate the effective number of parties existing in "In order to ensure the integrity of the time series, we never calculated change across sources. "This obviously does not exhaust the universe of potential patterns of executive/legislative configuration. Within the comparative politics literature on democratic institutions, semi-presidentialism (executive dyarchy between a popularly elected president and a prime minister selected by and responsible to parliament, see Duverger 1980) is considered a third major macroinstitutional type. There are also certain countries that have unique systems that do not easily fit into any category (e.g., Switzerland). Given the small number of cases of "semi-presidential" or "other" institutional forms, we left these cases out of the analysis. This would have meant the inclusion of several additional cells into the institutional matrix without a substantial population of cases to populate them. Consequently, we omitted several cases including Weimar Germany, France (it changed from parliamentarism to semi-presidentialism in 1958), contemporary Poland, Sri Lanka, and Switzerland from this analysis. Economics, Institutions, and Democratic Survival 787 the lower house.23 Figure 2 presents a histogram of the number of parties. In trying to determine significant cutoff points to classify party systems according to the effective number of parties, we found little overt discussion of this question in the l i t e r a t ~ r eWe . ~ ~have classified party-system according to four classes based on the distinctions in the qualitative literature. While we do not think there is a complete coincidence between our quantitative measures and 23 Where the effective number of parties, EP = 1/C:i, P:, where P,is the share of seats won by the ith party. We treat any "other" categories as Taagepera (1997) recommends. For each case that had a value in the "other" category, we calculated the number of parties' score for independents as both a single block of seats and as singular party seats. We then averaged these two fractionalization scores to produce the final number of parties We chose to use the Laakso-Taagepera measure of the effective number of parties rather than the measure developed by Molinar (1991). In assessing these measures and Wildgen's alternative, Lijphart notes that of the three systems, Molinar's gives greatest weight to the largest party, Wildgen favors the small parties to the greatest extent, and Laakso-Taagepera falls somewhere in the middle of the other two (1994b, 69-70). In looking at a number of cases in our data which we knew well, and holding Sartori's qualitative rules for counting parties (coalition or blackmail potential) as our standard (1976, 122-3), we were not convinced that either Molinar or Laakso-Taagepera is an ideal measure but, on the whole, were more conlfortable with Laakso-Taagepera's numbers. We will mention just a few examples. For instance, if we examine Germany in the 1960s we perceive that Molinar understates the number of parties. For instance, in 1961 three parties-the SPD, CDU/CSU, and FDP (190, 242, 67 out of 499 total)-won mandates. Eventually, a grand coalition of the two largest parties (SPD and CDU/CSU) was formed. However, any two of the three parties could have formed a viable coalition, and the excluded party would have constituted the only opposition in parliament. In this case, we found Laakso-Taagepera's figure of 2.51 more accurate than Molinar's 2.03. In looking at Weimar Germany, we also found Molinar's measures to be less accurate, often measuring a number of parties in the system fewer than the number of parties that sat in the government. In looking at Sweden in the 1960s, we also were more conlfortable with the numbers produced by Laakso-Taagepera, though not completely. In 1964, the social democrats took 113 mandates out of 232, just short of a majority. In order to form a government they needed the support of at least one party (or its abstinence during voting) to form a government. In theory, the social democrats could have turned either to the left or the right to get this support. In this situation Molinar gives the number of parties at 1.8 whereas Laakso-Taagepera puts the number at 3.25. Given the fluidity of the potential coalitions, Laakso-Taagepera seems to capture the situation better. However, if one looks at the next election (1968), in which the social-democrats took better then 50% of the mandates (125 of 233) and the strength of the other parties was dispersed (29, 37, 3, 32, 43), Molinar seems to do somewhat better in putting the number of parties at 1.5 to Laakso-Taagepera's 2.87. In these two examples, both measurements seemed to discriminate inadequately between a one-party government and a coalition government situation. Molinar seemed to do better in the one-party case, while Laakso-Taagepera did better in the coalition situation. Given the lack of a clear-cut verdict in the literature on relative merits of the counting systems, the more widespread use of Laakso-Taagepera, and our qualitative assessment of a number of cases, we opted to use LaaksoTaagepera over Molinar. 24Ahandful of studies have tried to incorporate the effective number of parties into their analysis. For instance, Stepan and Skatch draw a distinction between systems that have fewer than three parties and those that have three or more to reiterate the point on the problematic nature of multipartypresidentialism (1994, 121-2). Power and Gasiorowski take Stepan and Skatch's figure of three as the demarcation point for multiparty democracy. However, they demarcate their data in a quadripartite fashion (<1.8 parties, 2 1 . 8 <2.4, 2 2 . 4 <3.0, 2 3 . 0 ) to try to gauge the effect of party 788 Michael Bevnhavd, Tinzothy Novdstvom, and Christopher Reenoclc FIGURE 2 Histogram of the Effective Number of Parties (Lower House) (Laakso-Taagepera Index) 500 I Std. Dev = 1.60 Mean = 3.2 I '4?6c?Ic?Z?r??9?S~c3~8~v-fo-f6?c3?8~v~ N = 2742.00 % '7 '7 0.6'c3 '8 Effective Number of Parties The three reference lines included in the histogram mark the grouping cutoff points of 1.5, 2.5 and 4.0 from left to right. qualitative conceptions, we believe that the numbers get at the preponderant number of cases. We have designated the range below 1.5 as predominant party electoral outcomes (those cases in which one party holds a decisive majority in the legislature). The range from 1.5 to 2.49 parties coincides with a two-party electoral outcome (those cases in which in the largest party has a more modest majority).25The range from 2.5 to 3.99 captures those electoral results we have qual- systems and constitutional arrangernents on democratic consolidation (1997, 144-5). Neither explicitly explains how they arrived at these cutoff points, nor do Power and Gasiorowski explicate the meaning of their other three ranges. Mainwaring (1993), on the other hand, uses both a series of quantitative descriptors (predominant party, two-party, multiparty) that come from the qualitative literature on party systems, as well as Rae's fractional index and Laakso-Taagepera's effective number of parties to talk about the effects of different types of party systems on presidentialism. However, he does not explicitly connect the qualitative categories with either system of measurement (for an extension of his arguments, see Mainwaring and Shugart 1997). 25 Note that what we are qualifying is not the nature of the party system, for instance, characterizing a period in which one party rules for an extended period of time (like the social democrats in Sweden or the LDP in Japan) as predominant, but the results produced by a specific election. It is important to remember that systems that are categorized as two-party (like many SMD-plurality arrangernents) often have elections that produce fewer than two parties on this scale. The existence of two parties does not always coincide with the meaning of two-party system in the qualitative literature. Economics, Institutions, and Democratic Survival 789 ified as moderate multipartism (requiring a moderate number of parties to form a coalitional or legislative majority).26 Last, the range above 4.0 parties represents extreme multipartism (situations of high fractionalization). We then combined the executive-legislative variable with the effective number of parties to produce the eight-cell typology discussed above. These eight types were grouped as described in the theoretical section to produce an ordinal scale ranging from one through five (1 = Majoritarian Systems, 2 = Quasimajoritarian, 3 = Mixed, 4 = Quasi-pluralist, and 5 = Pluralist). Appendix 1 includes an enumeration of all our cases in terms of institutional form and year.27 RELIGIOUS AND ETHNIC FRACTIONALIZATION. Religious and ethnic fractionalization within a country is measured by Rae and Taylor's (1970) fractionalization index. This index produces the probability that two randomly selected persons from one country will not belong to the same social We calculated this index for both religion and ethnicity in each country. These data were, however, only available by decade and therefore are constant throughout each 10-year period (Singer 1997, supplemented by statistical annuals). We expect higher levels of fractionalization to lead to increased chances for breakdown. NUMBER OF DEMOCRATIC COUNTRIES IN THE REGION. We measure the "demonstration effect" (Gasiorowski 1995) of democracy on other countries as the total proportion of countries in a region that are democratic in any given year. The regions used to construct the "populations," modified from Gasiorowski (1995), were Latin America, Europe, sub-Saharan Africa, Northern Africa, the Middle East, South Asia (which runs from Pakistan and Afghanistan through 2 6 ~ h o u g Sh artori uses five as his cutoff point for effective parties to distinguish between moderate and extreme multipartism, this is but one dimension of his multidimensional typology. While we expect the kind of polarization, politics of outbidding, etc., that compose his type to be associated with a large number of parties (1976, 26), we realize there are cases where moderate pluralism is possible with a large nurnber of parties (e.g., interwar Czecho-Slovakia). We have chosen four rather than five as the cutoff point because in looking at a nurnber of cases in our data set, we observed that in higher multiparty outcomes, the systems of enumeration do not seem to count in the same way as Sartori. As the number of parties proliferates, the nurnber of parties fitting Sartori's criteria for effective (coalition or blackmail potential) seems to grow larger than the systems of counting admit. "Due to the high rnulticollinearity introduced into the model by including the Interaction term along with its component parts, we subtracted the respective means of both the economic performance variable and the institutional form variables before interacting them (Friedrich 1982). 2 8 ~used e Rae and Taylor's (1970) measure of fractionalization, glven as where F ranges from 0 to 1 790 Michael Bernhard, Timothy Nor*dstronz, and Christopher Reenock Myanmar), East Asia, North America, and Oceania. We expect that higher levels of regional democracy will promote democratic durability. NUMBER OF PAST DEMOCRATIC EXPERIENCES. Huntington (1991, 47) has speculated that past democratic failures lead to political learning that promotes future democratic success. We are not sure this is the case because multiple episodes of democracy may be indicative of inherent problems of instability. Nevertheless, we control for this contingency. To operationalize a country's experience with democracy and the possible effects of democratic learning, we codedwithin any given episode of democracy-the total number of previous experiences with democracy that a country had prior to the beginning of, but not including, the current episode. GDP PER CAPITA (LEVEL). To control for level of development, we include a measure of GDP per capita. We used the Penn World Tables (Summers and Heston 1995), which contain reliable data on a level consistent across cases only in the period after 1950 and interpolated where possible in the case of missing data.29 Methodology Given that we want to study the amount of time that passes between the installation of a democratic regime and its breakdown (or, alternatively, until 1995), we use event history methods to model this process. Specifically, we are interested in modeling the hazard rate, which is "the instantaneous probability that episodes in the interval [t, t At] are terminating provided that the event has not occurred before the beginning of th[e] interval" (Blossfeld, Hamerle, and Mayer 1989, 3 1). Thus, when using explanatory variables in a duration model, the point is to determine how the hazard rate is affected by the covariates highlighted by the theoretical model. Technically, the hazard rate is defined as: + In this study, the hazard rate refers to the risk3' that a country moves from a state of democracy to a state of nondemocracy. "Given the time range of our data, in particular the interwar cases, it was not possible to gather information on absolute levels of economic performance across countries. The lack of CPI conversion factors and the inability to locate interwar national income time-series data frustrated our efforts to gather reliable information on levels of economic performance. To ensure that the findings based upon the analysis of the maximurn number of cases in the data set were not related to economic level, we ran a secondary analysis (not reported here) in which we used regional dummies as proxy measures for economic level. Specifically, we included dummies for the most and least developed regions (Western Europe and sub-Saharan Africa) (Maddison 1995). The inclusion of these dummies did not alter our findings. 30Although it is easy to think of "risk" as a probability, it is actually a rate. It can take values greater than one, which, obviously, probabilities cannot. Economics, Institutions, and Democratic Survival 79 1 We utilize continuous-time hazard techniques for several reasons. First, like discrete-time techniques, continuous-time models allow for multivariate analysis. Second, many functional forms of continuous-time techniques allow for time-varying covariates. Several of our independent variables vary within a given democratic episode, and incorporating this intra-episode variation provides us with much more information. Third, continuous-time models allow for the explicit testing of duration dependence by estimating the duration dependence parameter, p . Of the several possible functional forms of the hazard rate, we have chosen to estimate fully parametric models with a Weibull form for three reasons: (1) It allows multivariate analysis. (2) Unlike the normal or log-normal forms, the Weibull allows for relatively trouble-free estimation with time-varying covariates. (3) It estimates a parameter indicating the presence of positive, negative, or no duration dependence. This last point is particularly important given that we are interested in the possibility that democracy may become institutionalized over time. With no covariates included in the Weibull model, the form of the hazard rate is h ( t )= hp(ht)"pl where h is a constant a n d p represents the nature of the duration dependence.31 The introduction of covariates into the model relaxes the assumption that h is a constant. Instead, = eP-'i' where the coefficients are estimated using maximum likelihood. In many subject areas where event history is applied, the destination state (in this case "nondemocracy") is absorbing. An absorbing destination state means that the transition to that state is the only transition that any given unit will experience. However, this does not hold for democracy. Although sometimes the destination state will be absorbing-as when a democracy breaks down and is never reinstated-it is possible for states to experience multiple democratic episodes. Ideally, we could capture the effects of the previous democratic episodes in statistical techniques. However, we are not aware of any statistical applications that allow for both multi-episode models and the incorporation of time-varying covariates. Given this tradeoff, we choose to treat all democratic episodes as statistically independent in order to capture the additional information provided by time-varying covariates. Results The results of the hazard analysis are provided in Table 1. Before moving to a discussion of the individual independent variables, we begin with a brief discussion of the overall performance of our models. An examination of the log311fp< 1, the dependent variable exhibits negative duration dependence. When p positive duration dependence. When p = 1 there is no duration dependence. > 1 there is TABLE 1 Hazard Model (Weibull) Failure Time Coefficients, Effects on Democratic Survival Model 2 Model 3 Model 4 Main Model with Interaction Term Post-1950 wlth interaction Term Post-1950 wlth GDP per caplta Level Model 1 Maln Model No Interaction Term Variable Coefficient Coefficient Standard Error 0.7856 0.6880** 3.4005*** 0.3706 0.1960 1.9006** 1.7858*** 0.7742 0.7000** 0.8760*** 0.3535 - - 0.2043 2.0046** Institutional Form Economic Growth Econ Growth X institutional Form Religious Fractionalization Ethnic Fractionalization Reglonal Democracies Past Attempts at Democracy GDP per capita Level Constant p (duration parameter) Log-Likelihood Log-Likelihood, Constant Only Number of Democratic Spells Number of Data Points (country-years as Democracy) Standard Error - 0.9498 -102.4 - 129.7 127 2184 ***p < 0.01, **p < 0.05, *p < 0.10, two-tailed tests Note: Standard errors clustered on country code. Coefficient 0.9799 - 100.3 - 129.7 127 2184 1.0442 Coefficient 117.1 124 1.4716 -67.4 -96.1 116 1857 1831 - 93 .0 - Standard Error Standard Error Economics, Institutions, and Democratic Survival 793 likelihoods in Table 1 shows that the full models we estimated performed better than the respective null models (constant only). The addition of our explanatory variables greatly improved the performance of the model, as the difference in log-likelihood values are statistically significant at y < ,001. Therefore, we are reasonably sure that our independent variables are adding explanatory power. It is also the case that the duration dependence parameter does not deviate significantly from 1.0 in our main results. This strongly suggests that there is, in fact, no duration dependence. This result has interesting substantive ramifications to which we will return. Table 1 includes four models. We are most interested in Model 2, but include Models 1, 3, and 4 to show the robustness of our findings. The first column contains the results of our main model with the economic change and macroinstitutional structure variables without the interaction term. This model displays the effects of the key variables in the additive manner often used in the existing literature. We see in this column that economic performance appears to matter more than institutional form. The GDP per capita change variable is highly significant and in the expected direction,32 while the institutional form variable does not achieve even the minimum standard level of significance. This basic model provides more evidence for those who argue that economic performance is crucial for democratic survival. Model 2 presents the results of our model using all cases in the data set. In this model the interaction of economic performance and institutional form emerges as ~ i g n i f i c a n tAs . ~ ~in Model 1, economic performance remains significant and institutional form remains insignificant. The comparison of the results of these two models leads us to believe that the effect of economic change and institutions is more than additive. It is the way that institutions mediate economic performance, as our theoretical argument suggests, that affects democratic survival. The substantive effects of this variable are discussed below. Both Model 3 and Model 4 contain results of analyses run on just the postWorld War I1 era (after 1950). We include these results in order to show that our main finding is robust, given that our use of interwar cases made it difficult to incorporate a measure of level of development. Model 3 addresses whether our findings are produced by the inclusion of the interwar years by eliminating those years from the sample. Model 4 adds GDP per capita to the postwar 32The coefficients here are represented in accelerated failure time form, which means that positive coefficients indicate "longer" democracy, while negative coefficients indicate "shorter" democracy. This should not be confused with hazard rate coefficients, in which the signs have the opposite meaning. 33 We also broke apart our institutional form variable into five dummies and interacted each of these with economic performance. This model supported our ordinal scaling of the institutional form variable with each of the components stacking above the other. In this variation, we continued to find that majoritarian regimes were more likely to break down under conditions of economic growth than under contraction. 794 Michael Bernhard, Timothy Nor*dstrom, and Christopher Reenock model to ensure that the omission of this important variable did not produce our main results.34 The findings in Models 3 and 4 show that the results concerning the interaction term are quite robust. In both models, the interaction term is significant at p < .01. The independent effect of change in GDP per capita change remains quite strong and in the expected direction. Also consistent is the performance of the institutional form variable, which never achieves statistical significance in any of the models. Model 4 replicates the findings in the literature on level of development. This variable is highly significant and suggests that more developed democracies (higher GDP per capita) can expect to last longer than less developed democracies. Table 2 and Figure 3 both display the estimated substantive effects of the interaction term by presenting the expected duration of democracy, given certain values of the independent variables in model 2. We estimated the expected duration using the procedure described in Greene ( 1 9 9 0 ) . ~In~ order to produce the expected durations, we assign differing values for the economic and institutional variables (with the interaction term adjusted accordingly) and set the control variables at their means. For the economic growth variable, we set the value at -4% to capture the expected duration under conditions of poor econolllic performance and to -2% growth to capture the effect of more moderate economic contraction. For positive growth, we present the effects of 2% and 4% growth on democratic duration. We also include zero growth for comparative purposes. We can assess our hypotheses about institutional form, economic performance, and democratic breakdown by examining the data in Table 2 and Figure 3. Upon examination of Figure 3, it is quite clear that institutions matter. We observe a range of diverse values for different democratic subtypes across economic performance. This is strong evidence that institutional form is impor34 We also ran models including several other independent variables, such as whether or not the country was a British colony, region dummy variables (Western Europe, sub-Saharan Africa, and Latin America), and a dummy variable for microstates. The introduction of these variables into the model did not affect the sign, magnitude, or significance of the interaction variable. In addition, none of these variables were significant in their own regard in the main model. We also ran a model to test whether the functional forrn of the interaction terrn varied across level of development. To do this, we created a dummy variable to code higher and lower levels of development. We included the durnrny variable as an additive terrn to test for an intercept effect and as a multiplicative terrn interacted with the "Economic Growth-Institutional Form (EG-IF)" variable in the model to test for a change in the functional form of the coefficient for the EG-IF variable. We did this for development cutoffs of both $6,000 (1985 ppp) and $4,000 (1985 ppp). In both of these models, the intercept effect was significant and positive as expected and the EG-IF remained significant. However, the interaction terms with level of development were not significant, suggesting that the EG-IF coefficient does not vary across level of development. 35Greene (1990) reports an incorrect specification of the expected duration for the Weibull model. See Bennett (1997) for a corrected version. Economics, Institutions, and Democratic Survival 795 TABLE 2 Marginal Effects of Institutional Form Variable on Democratic Survival Given Different Levels of Economic Performance, Measured in Years GDP per Capita Growth Rate Institutional Form 4 % 2 % 0% 2% 4% Majoritarian 156.70 142.34 129.30 117.45 106.69 Quasi-majoritarian 111.31 112.29 113.28 114.28 115.29 Mixed 79.07 88.59 99.25 111.20 124.58 Quasi-pluralist 56.17 69.88 86.96 108.19 134.63 Pluralist 39.90 54.86 76.19 105.27 145.48 Note: All control variables are set at their means. These values are as follows: Religious Fractionalization ,3709; Ethnic Fractionalization ,3312; Regional Democracies ,5683; Past Attempts at Democracy . I 718. tant for how democracies fare under different economic conditions. In the first column of Table 2, we see that the effect of severe negative growth on democratic survival is indeed modified by institutional form. With a contraction of -4% in GDP per capita, we see that pluralist democracies are expected to last 39.9 years. In contrast, majoritarian governments are expected to last nearly 156.7 years, over three times as long as pluralist regimes. Majoritarian governments continue to survive significantly longer than pluralist forms even given a more moderate (-2%) level of poor performance (majoritarian-142.34 years versus pluralist-54.86 years). The same pattern emerges for cases in which there is zero growth. Majoritarian regimes still perform somewhat better than pluralist regimes. These results provide strong support for Hypothesis 1-that pluralist institutions are far less durable in the face of poor economic performance. Under conditions of positive economic growth, Table 2 and Figure 3 also show that pluralist and quasi-pluralist regimes survive longer. Under conditions of strong positive growth (4%), pluralist democracies are expected to last over 145 years. In contrast, majoritarian regimes are expected to last 106.69 years. In other words, pluralist regimes are expected to last about 30% longer than majoritarian democracies under high rates of economic growth. This result provides solid confirmation of our second hypothesis. While this finding is consistent with Hypothesis 2, one aspect of this result was unanticipated. Majoritarian governments last longer under conditions of economic contraction than under conditions of positive growth (losing almost 50 years when growth goes from -4% to 4%). We did not expect this result and it is counterintuitive. Existing theory provides at least a partial potential explanation for this finding. Olson's hypothesis that high rates of economic 796 Michael Bernhard, Timothy Nordstrom, and Christopher Reenock FIGURE 3 Democratic Duration by Economic Performance and Democratic Subtype Democratic Subtypes H 1--Majoritarian e 2--Quasi-majoritarian -- 3--Mixed I I6-Quasi-pluralist a 5--Pluralist growth can be destabilizing may have some merit given the "correct" institutional configuration. However, this only explains why majoritarian regimes do better at 2% growth than at 4% growth. It does not explain why majoritarian regimes do better with no growth or economic contraction than with moderate positive growth. Although we do not show predicted durations from Models 3 or 4, it is important to note that this result is robust across all models that include the interaction term. Another interesting result is that quasi-majoritarian forms of democracy are stable across all ranges of economic performance. The results from Table 2 show that the change in expected duration for the quasi-majoritarian subtypes is negligible over the entire economic range. Specifically, we see that at the -4% level, quasi-majoritarian forms of democracy are expected to last 1 1 1.31 years, while at the 4% level the expected duration is 115.29 years. Thus, twoparty parliamentary regimes do reasonably well under both positive and negative economic conditions and, according to our findings, should be the most stable form of democracy. The consistent performance of two-party majoritarian regimes holds across all of our models. Of our four control variables, only two achieved statistical significance. First, ethnic fractionalization is significant and negative, which means that democ- Economics, Institutions, and Democratic Survival 797 racy is at greater risk in ethnically fractionalized societies in comparison to homogenous ones. Second, the coefficient of the variable that captures the proportion of democratic states that exist in a given country's region is positive and significant. This replicates Gasiorowski's finding (1995) on the existence of a "demonstration effect." However, both of these findings drop out when a variable for level of development is included in Model 4.36 Last, we address the issue of duration dependence from a substantive perspective. As was mentioned above, the duration parameter in our main model was not significantly different from 1.0. This leads us to conclude that there is no evidence to indicate the presence of positive or negative duration depenin Przeworski et al. (1996), this result has ramifications d e n ~ eAs . ~discussed ~ for the concept of democratic consolidation. Negative duration dependence would have suggested that as democracies last longer, their rate of breakdown would decline. However, our main results corroborate the findings of Przeworski et al. and do not provide any support for the notion of democratic consolidation. Democracies in our sample are no more likely to survive to time,, given that they have reached time,. , Conclusion Our findings show that in and of themselves institutions do not determine democratic survival. The impact of institutions on durability varies according to economic performance. Specific subtypes of democracy, majoritarian and pluralist, are more or less resistant to the effects of poor economic performance, depending upon the way they concentrate or disperse decision-making power. Unlike previous studies, we do not claim that macroinstitutional structure alone explains survival. In our model, parliamentarism and presidentialism have an effect on survival in combination with different party systems due to the extent that they concentrate or disperse decision-making power. This represents a step away from the Linz hypothesis on presidentialism that has so dominated the discussion 011 democratic survival. We now turn to the performance of the specific institutional forms under different economic conditions. Of the five, we find that quasi-majoritarianism, which corresponds to the Westminster model, does not vary with economic performance. The Westminster model thus effectively insulates democracy from 3"l~is suggests that the learning effect, as measured, may be capturing regional patterns of development and that people do not mobilize along ethnic lines in more developed countries. "The only indication that there is duration dependence of any kind conies froin Model 4. Unlike arguments for the consolidation of democracies, this model suggests that the longer democracies endure, the more likely it is that they will break down. However, the magnitude of the shape parameter indicates only a marginal impact on the expected duration. In addition, the introduction of level of development, which has a positive trend over time, is a likely cause for this duration dependence. Appendix 1 Cases by Institutional Form and Year Effcctne N u m h n of P a ~ t l r s(Ladkio-Txagepera) 800 Michael Bernhard, Timothy Nordstr*onz, and Christopher Reenock Appendix 2 Democratic Episodes in the Data Set Andorra 93-95 Antigua and Barbuda 81-95 Argentina 84-95 Australia 19-95 Austria 19-33, 55-95 Bahamas 73-95 Bangladesh 91-95 Barbados 66-95 Belgium 19-40, 46-95 Belize 8 1-95 Benin 60-62, 91-95 Bolivia 82-95 Botswana 66-95 Brazil 47-64, 86-95 Bulgaria 19-20, 90-95 Burkina Faso 78-80 Canada 19-95 Cape Verde 91-95 Chad 60-62 Chile 49-73, 90-95 Colombia 74-95 Congo 61-63, 92-93 Costa Rica 50-95 Czech Rep. 92-95 Czechoslovakia 20-39, 90-91 Denmark 19-40, 45-95 Dominica 78-95 Dominican Republic 6 3 , 78-95 Ecuador 48-52, 79-95 El Salvador 91-95 Estonia 19-34, 92-95 Finland 19-95 France 19-40, 45-95 Gambia 66-94 Germany 19-33, 49-95 Ghana 57-60, 69-72, 79-82 Greece 26-36, 75-95 Grenada 74-79, 84-95 Guatemala 46-54 Guyana 92-95 Haiti 95 Honduras 90-95 Hungary 90-95 Iceland 44-95 India 53-75, 77-95 Indonesia 55-57 Ireland 22-95 Israel 49-95 Italy 19-22, 48-95 Jamaica 63-95 Japan 52-95 Kenya 63-66 Latvia 22-34, 93-95 Lithuania 20-26, 91-95 Luxembourg 19-40 46-95 Macedonia 9 1-95 Madagascar 6 1-7 1 , 93-95 Malawi 94-95 Malaysia 59-69 Mali 92-95 Malta 64-95 Mauritius 67-95 Moldova 94-95 Mongolia 92-95 Mozambique 94-95 Myanmar 48-62 Namibia 90-95 Nepal 91-95 Netherlands 19-40, 46-95 N e w Zealand 19-95 Nicaragua 90-95 Niger 93-95 Nigeria 60-66, 79-83 Norway 19-40, 46-95 Papua N e w Guinea 77-95 Panama 94-95 Paraguay 91-95 Peru 80-92 Philippines 53-72, 87-95 Poland 19-26, 89-95 Portugal 19-26, 76-95 Romania 92-95 Russia 93-95 St. Kitts and Nevis 83-95 St. Lucia 79-95 St. Vincent 79-95 Sao T o m e 91-95 Seychelles 93-95 Sierra Leone 62-67 Slovakia 92-95 Slovenia 90-95 S o l o n ~ o nIslands 78-95 Somalia 60-69 South Africa 94-95 South Korea 61, 88-95 Spain 3 1-36, 78-95 Sri Lanka 48-83 Sudan 56-58, 65-69, 86-89 Suriname 75-79, 88-89, 91-95 Sweden 19-95 Switzerland 71-95 Taiwan 92-95 Tanzania 61-62, 95 Thailand 75-76 Trinidad 62-95 Turkey 61-71, 73-80, 83-95 Uganda 62-66 Ukraine 91-95 United States 20-95 United Kingdom 19-95 Uruguay 34-73, 85-95 Vanuatu 80-95 Venezuela 4 8 , 58-95 Zambia 9 1-95 Zimbabwe 90-95 the vagaries of economic performance. To our knowledge, this is the first study to report such a finding. We find our results on majoritarianism peculiar. We do not understand why this type fares better under economic contraction than under moderate levels of economic growth. Of all the types that we have identified, majoritarianism occurs with the least frequency. Therefore, we understand why this type has not received a place of prominence in the literature. Nevertheless, existing theory does not provide any ready-made explanation for this result. From a normative Economics, Institutions, and Democratic Survival 801 perspective, however, we find it reassuring that this most rare and most unrepresentative form of democracy performs differently from more inclusive variants given the centrality of representation and responsiveness to democracy. Last, our findings for pluralist and quasi-pluralist democracies are consistent with the literature on presidentialism in Latin America. These two subtypes, which combine presidentialism with representative party systems, explain the poor record of presidentialism in the Latin American context. Given the way these two subtypes perform under conditions of economic contraction and the periodic problems that many Latin American countries have experienced, this result now makes better sense. However, this does not mean that under conditions of sustained positive growth that these systems will fail with the regularity suggested by the literature on Latin American presidentialism. In fact, our model predicts that these systems would do quite well in a positive economic environment. Manuscript submitted 13 October 1999 Final mantiscript received 28 November 2000 References Alvarez, Mike, Jose Antonio Cheibub, Fernando Limongi, and Adam Przeworski. 1997. "Classifying Political Regimes." Studies in Con~pnrativeInterizntional Developnzent 3 l(2): 2-36. Arat, Zehra F. 1988. "Democracy and Economic Development: Modernization Theory Revisited." Conzpaiative Politics 2 1: 2 10-36. Bennett, D. Scott. 1997. "Testing Alternative Models of Alliance Duration: 1816-1984." Atilerican Jo~lrnalof Politicnl Science 41(3): 846-78. Blossfeld, Hans-Peter, Alfred Hamerle, and Karl-Ulrich Mayer. 1989. Event History Analysis. Hillsdale, NJ: Lawrence Erlbaum Associates. Bollen, Kenneth A. 1979. "Political Democracy and the Timing of Development." Ainericnn Sociological Review 44(4): 572-87. Bollen, Kenneth A. 1991. "Political Democracy: Conceptual and Measurement Traps." In On ~Measuring Denzocracy, ed. Alex Inkeles. New Brunswick: Transaction. Burkhart, Ross E., and Michael S. Lewis-Beck. 1994. "The Economic Development Thesis." Americon Political Science Review 88(4): 903-10. Colomer, Josep M. 1995. "Strategies and Outcomes in Eastern Europe." Jolrrtznl ofDemocracy 6(2): 74-85. Coppedge, Michael, and Wolfgang H. Reinecke. 1991. "Measuring Polyarchy." In On A4eastlr.itzg Deinocracy, ed. Alex Inkeles. New Brunswick: Transaction. Cutright, Phillips. 1963. "National Political Development: Measurement and Analysis." Ainerican Sociological Review 28(2): 253-64. Dahl, Robert. 1971. Polvarchy. New Haven: Yale University Press. Duverger, Maurice. [I9801 1992. "A New Political Systen~Model: Semi-presidential Government." In Parlianzentaly versus Presidetztial Govertznzetzt, ed. Arend Lijphart. New York: Oxford. Electoral Stlrdies. Various years. Guilford, England: Butterworths. Freedom in the World. Various years. New York: Freedom House. Friedrich, Robert J. 1982. "In Defense of Multiplicative Terms in Multiple Regression Equations." American Journal of Political Science 26(4): 797-833. Gasiorowski, Mark J. 1991. "The Political Regimes Project." In 011Meuszli.iizg Democmcy, ed. Alex Inkeles. New Brunswick: Transaction. 802 Michael Ber-nhar-d,Timothy Nordstr-om, and Christopher Reenoclc Gasiorowski, Mark J. 1995. "Economic Crisis and Political Regime Change: An Event History Analysis." Americaiz Political Science Review) 89(4): 882-97. Gasiorowski, Mark J. 1996. "An Overview o f the Political Regime Dataset." Coniparative Political Studies 29(4): 469-83. Gasiorowski, Mark J., and Timothy J. Power. 1998. "The Structural Determinants o f Democratic Consolidation, Evidence from the Third World." Con~paiutivePolitical Studies 3 l ( 6 ) : 740-71. Gonick, Lev S., and Robert M. Rosh. 1988. "The Structural Constraints o f the World Economy on National Political Development." Coi?iparative Political Studies 2 l ( 2 ) : 17 1-99. Greene, William H. 1990. Econometric Ana/j)sis. Englewood Cliffs,NJ: Prentice Hall. Hadenius, Axel. 1992. Democmcy and Developii~eizt.Cambridge: Cambridge University Press. Haggard, Steven, and Robert R. Kaufinan. 1995. The Political Ecoizoniy ofDemocmtic Transitions. Princeton: Princeton University Press. Horowitz, Donald. 1990. "Comparing Democratic Systems." Joz~riialofDemocracy 1 : 73-9. Huntington, Samuel B. 1968. Political Order in Changing Societies. New Haven: Yale University Press. Huntington, Samuel B. 1991. The Third Wave: Democmtization in the Late nveiztieth Centuly. Norman: Oklahoma University Press. Im, Hyug Baeg. 1987. "The Rise o f Bureaucratic Authoritarianism in South Korea." World Politics 39(2): 231-57. Jackman, Robert W . 1973. "On the Relation o f Economic Development to Democratic Performance." American Journal qf Political Science 17(3): 61 1-21. Jaggers, Keith, and Ted Robert Gurr. 1995. "Tracking Democracy's Third Wave with the Polity 111 Data." Journal qf Peace Research 32(4):469-82. Jo~irrzalqfDenzoc~ucy.Various years. Washington, DC: National Endowment for Democracy. Laakso, Markuu, and Rein Taagepera. 1979. "EffectiveNumber o f Parties: A Measure with Application to Western Europe." Comparative Political Silldies 12(1):3-27. Lijphart, Arend. 1994a. "Presidential and Majoritarian Democracy: Theoretical Observations." In The Failz~reof Presidential Democmcy, eds. Juan J. Linz and Arturo Valenzuela. Baltimore: The Johns Hopkins Press. Lijphart, Arend. 1994b. Electoml Systems and Para) Systems. New York: Oxford University Press. Lijphart, Arend. 1999. Patterns of Denzocracy, New Haven: Yale University Press. Linz, Juan J. 1978. "Crisis, Breakdown, and Reequalibration." The Breakdown of'Deniocri~ticRegimes, eds. Juan Linz and Alfred Stepan. Baltimore: The Johns Hopkins University Press. Linz, Juan J. 1994. "Presidential or Parliamentary Democracy: Does it Make a Difference?" In The Faillwe of Presiderztial Delllocracj~,eds. Juan J. Linz and Arh~roValenzuela. Baltimore: The Johns Hopkins Press. Lipset, Seymour Martin. 1959. "Soine Social Requisites o f Democracy: Economic Development and Political Legitimacy." A~izericarzPolitico1 Science re vie^^ 53(1): 69-105. Lipset, Seymour Martin, Kyoung-Ryung Seong, and John Charles Torres. 1993. " A Comparative Analysis o f the Social Requisites o f Democracy." Internatioi~alSocial Science Jozcrrial 45(2): 155-75. Londregan, John B., and Keith T. Poole. 1990. "Poverty, the Coup Trap, and the Seizure o f Executive Power." World Politics 42(2): 151-83. Mackie, Thomas, and Richard Rose. 1982. The Interiiational Almanac ofElectora1 Histoi:~.New York: Facts on File. Maddison, Angus. 1995. Monitoring the World Econom)~,1820-1992. Paris: Development Centre o f the Organisation for Economic Co-operation and Development. Mainwaring, Scott. 1990. "Presidentialism in Latin America." Latin America12 Research Revie111 25(1): 157-79. Mainwaring, Scott. 1993. "Presidentialism, Multipartism, and Democracy: The Difficult Combination." Coriiptrrative Political Studies 26(2): 198-228. Mainwaring, Scott, and Matthew Soberg Shugart. 1997. Presidentialisnz and Democizrcj~in Latiiz America. Cambridge, U K : Cambridge University Press. Molinar, Juan. 1991. "Counting the Number o f Parties: An Alternative Index." American Political Science Review) 85(4): 1383-9 1 . Economics, Institutions, and Democratic Survival 803 Neubauer, Deane. 1967. "Some Conditions of Democracy." American Political Science Review 61(4): 1002-9. Nohlen, Dieter, ed. 1993. Ei~cyclopediaElectorzrl Latinoatwericatza y del Caribe. San Jose, Costa Rica: Instituto Interamericano de Derechos Humanos. O'Donnell, Guillermo. 1973. Modernizatiotz and Bi~reaucraticAuthoritarianism. Berkeley: Iustitute of International Studies. h, and Social Olson, Mancur. 1982. The Rise and Decline ofNations: Ecotzomic G r o ~ ~ tStagJlatio~z Rigidities. New Haven: Yale University Press. Political Hann'book of the World. Various years. New York: CSA Publications. Power, Timothy, and Mark J. Gasiorowski. 1997. "Institutional Design and Democratic Consolidation in the Third World." Cort~parativePolitical Studies 30(2): 123-55. Przeworski, Adam, Michael Alvarez, Jose Antonio Cheibub, and Fernando Limongi. 1996. "What Makes Democracy Endure?" Journal of Denzocrucy 7(1): 39-55. Przeworski, Adam, and Fernando Limongi. 1993. "Political Regimes and Economic Growth." Jozlrnu1 of Econoinic Perspectives 7(3): 5 1-69. Przeworski, Adam, and Fernando Limongi. 1997. "Modernization, Theories and Facts." TVorld Politics 49(2): 155-83. Rae, Douglas W., and Michael Taylor. 1970. The Analysis ofPolitica1 Cleavages. New Haven: Yale University Press. Riggs, Fred. 1993. "Fragility of the Third World's Regimes." International Social Science Joiirtial 45(2): 199-243. Sartori, Giovanni. 1976. Parties and Pcirr), Sj~steins.Cambridge: Cambridge University Press. Sartori, Giovanni. 1994. Coinpamtive Constit~itiotzalEngineering, 2nd ed. New York: New York University Press. Seligson, Mitchell A,, and Edward N. Muller. 1987. "Democratic Stability and Economic Crisis: Costa Rica, 1978-1983." Itzternational Studies QuarterLv 31(3): 301-26. Shugart, Matthew Soberg, and John M. Carey. 1992. Presidents and Assetnblies: Constitiitiotzal Design and Electoral Dynartzics. Cambridge: Cambridge University Press. Singer, J. D. 1997. Cziltliral Composition of Itztei.state Sj~sternMeilzbers. Correlates of War Project Data. Ann Arbor: University of Michigan Press. Stepan, Alfred, and Cindy Skatch. 1994. "Presidentialism and Parliamentarism in Comparative Perspective." In The Fail~rreof Presidential Democracy, eds. Juan J. Linz and Arturo Valenzuela. Baltimore: The Johns Hopkins Press. Summers, Robert, and Alan Heston. 1995. Petzn World Tables, version 5.6. Philadelphia, Center for International Comparisons. Taagepera, Rein. 1997. "Effective Number of Parties for Incomplete Data." Electorul Studies 16(2): 145-51. United Natiotzs Statistical Yearbook. Various years. New York: United Nations. Valenzuela, Arturo. 1993. "Latin America: Presidentialism in Crisis." Jorrl.rla1 of Denzocracy 4(4): 3-16. Wildgen, John K. 1971. "The Measurement of Hyperfractionalization." Compalative Political Studies 4:233-43. World Fact Book. Various years. Washington, DC: Central Intelligence Agency. Michael Bernhard is an associate professor of political science, The Pennsylvania State University, University Park, PA 16802. Timothy Nordstrom is an assistant professor of political science, University of Mississippi, Oxford, MS 38677. Christopher Reenock is an assistant professor of political science, University of Alabama, Tuscaloosa, AL 35487. http://www.jstor.org LINKED CITATIONS - Page 1 of 5 - You have printed the following article: Economic Performance, Institutional Intermediation, and Democratic Survival Michael Bernhard; Timothy Nordstrom; Christopher Reenock The Journal of Politics, Vol. 63, No. 3. (Aug., 2001), pp. 775-803. Stable URL: http://links.jstor.org/sici?sici=0022-3816%28200108%2963%3A3%3C775%3AEPIIAD%3E2.0.CO%3B2-W This article references the following linked citations. If you are trying to access articles from an off-campus location, you may be required to first logon via your library web site to access JSTOR. Please visit your library's website or contact a librarian to learn about options for remote access to JSTOR. [Footnotes] 2 National Political Development: Measurement and Analysis Phillips Cutright American Sociological Review, Vol. 28, No. 2. (Apr., 1963), pp. 253-264. Stable URL: http://links.jstor.org/sici?sici=0003-1224%28196304%2928%3A2%3C253%3ANPDMAA%3E2.0.CO%3B2-U 2 Some Social Requisites of Democracy: Economic Development and Political Legitimacy Seymour Martin Lipset The American Political Science Review, Vol. 53, No. 1. (Mar., 1959), pp. 69-105. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28195903%2953%3A1%3C69%3ASSRODE%3E2.0.CO%3B2-D 2 On the Relation of Economic Development to Democratic Performance Robert W. Jackman American Journal of Political Science, Vol. 17, No. 3. (Aug., 1973), pp. 611-621. Stable URL: http://links.jstor.org/sici?sici=0092-5853%28197308%2917%3A3%3C611%3AOTROED%3E2.0.CO%3B2-Q 2 Some Conditions of Democracy Deane E. Neubauer The American Political Science Review, Vol. 61, No. 4. (Dec., 1967), pp. 1002-1009. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28196712%2961%3A4%3C1002%3ASCOD%3E2.0.CO%3B2-J NOTE: The reference numbering from the original has been maintained in this citation list. http://www.jstor.org LINKED CITATIONS - Page 2 of 5 - 2 Democracy and Economic Development: Modernization Theory Revisited Zehra F. Arat Comparative Politics, Vol. 21, No. 1. (Oct., 1988), pp. 21-36. Stable URL: http://links.jstor.org/sici?sici=0010-4159%28198810%2921%3A1%3C21%3ADAEDMT%3E2.0.CO%3B2-E 2 Comparative Democracy: The Economic Development Thesis Ross E. Burkhart; Michael S. Lewis-Beck The American Political Science Review, Vol. 88, No. 4. (Dec., 1994), pp. 903-910. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28199412%2988%3A4%3C903%3ACDTEDT%3E2.0.CO%3B2-3 3 Economic Crisis and Political Regime Change: An Event History Analysis Mark J. Gasiorowski The American Political Science Review, Vol. 89, No. 4. (Dec., 1995), pp. 882-897. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28199512%2989%3A4%3C882%3AECAPRC%3E2.0.CO%3B2-1 23 Counting the Number of Parties: An Alternative Index Juan Molinar The American Political Science Review, Vol. 85, No. 4. (Dec., 1991), pp. 1383-1391. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28199112%2985%3A4%3C1383%3ACTNOPA%3E2.0.CO%3B2-0 27 In Defense of Multiplicative Terms in Multiple Regression Equations Robert J. Friedrich American Journal of Political Science, Vol. 26, No. 4. (Nov., 1982), pp. 797-833. Stable URL: http://links.jstor.org/sici?sici=0092-5853%28198211%2926%3A4%3C797%3AIDOMTI%3E2.0.CO%3B2-A 35 Testing Alternative Models of Alliance Duration, 1816-1984 D. Scott Bennett American Journal of Political Science, Vol. 41, No. 3. (Jul., 1997), pp. 846-878. Stable URL: http://links.jstor.org/sici?sici=0092-5853%28199707%2941%3A3%3C846%3ATAMOAD%3E2.0.CO%3B2-K NOTE: The reference numbering from the original has been maintained in this citation list. http://www.jstor.org LINKED CITATIONS - Page 3 of 5 - References Democracy and Economic Development: Modernization Theory Revisited Zehra F. Arat Comparative Politics, Vol. 21, No. 1. (Oct., 1988), pp. 21-36. Stable URL: http://links.jstor.org/sici?sici=0010-4159%28198810%2921%3A1%3C21%3ADAEDMT%3E2.0.CO%3B2-E Testing Alternative Models of Alliance Duration, 1816-1984 D. Scott Bennett American Journal of Political Science, Vol. 41, No. 3. (Jul., 1997), pp. 846-878. Stable URL: http://links.jstor.org/sici?sici=0092-5853%28199707%2941%3A3%3C846%3ATAMOAD%3E2.0.CO%3B2-K Political Democracy and the Timing of Development Kenneth A. Bollen American Sociological Review, Vol. 44, No. 4. (Aug., 1979), pp. 572-587. Stable URL: http://links.jstor.org/sici?sici=0003-1224%28197908%2944%3A4%3C572%3APDATTO%3E2.0.CO%3B2-G Comparative Democracy: The Economic Development Thesis Ross E. Burkhart; Michael S. Lewis-Beck The American Political Science Review, Vol. 88, No. 4. (Dec., 1994), pp. 903-910. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28199412%2988%3A4%3C903%3ACDTEDT%3E2.0.CO%3B2-3 National Political Development: Measurement and Analysis Phillips Cutright American Sociological Review, Vol. 28, No. 2. (Apr., 1963), pp. 253-264. Stable URL: http://links.jstor.org/sici?sici=0003-1224%28196304%2928%3A2%3C253%3ANPDMAA%3E2.0.CO%3B2-U In Defense of Multiplicative Terms in Multiple Regression Equations Robert J. Friedrich American Journal of Political Science, Vol. 26, No. 4. (Nov., 1982), pp. 797-833. Stable URL: http://links.jstor.org/sici?sici=0092-5853%28198211%2926%3A4%3C797%3AIDOMTI%3E2.0.CO%3B2-A NOTE: The reference numbering from the original has been maintained in this citation list. http://www.jstor.org LINKED CITATIONS - Page 4 of 5 - Economic Crisis and Political Regime Change: An Event History Analysis Mark J. Gasiorowski The American Political Science Review, Vol. 89, No. 4. (Dec., 1995), pp. 882-897. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28199512%2989%3A4%3C882%3AECAPRC%3E2.0.CO%3B2-1 The Rise of Bureaucratic Authoritarianism in South Korea Hyug Baeg Im World Politics, Vol. 39, No. 2. (Jan., 1987), pp. 231-257. Stable URL: http://links.jstor.org/sici?sici=0043-8871%28198701%2939%3A2%3C231%3ATROBAI%3E2.0.CO%3B2-9 On the Relation of Economic Development to Democratic Performance Robert W. Jackman American Journal of Political Science, Vol. 17, No. 3. (Aug., 1973), pp. 611-621. Stable URL: http://links.jstor.org/sici?sici=0092-5853%28197308%2917%3A3%3C611%3AOTROED%3E2.0.CO%3B2-Q Tracking Democracy's Third Wave with the Polity III Data Keith Jaggers; Ted Robert Gurr Journal of Peace Research, Vol. 32, No. 4. (Nov., 1995), pp. 469-482. Stable URL: http://links.jstor.org/sici?sici=0022-3433%28199511%2932%3A4%3C469%3ATDTWWT%3E2.0.CO%3B2-Q Some Social Requisites of Democracy: Economic Development and Political Legitimacy Seymour Martin Lipset The American Political Science Review, Vol. 53, No. 1. (Mar., 1959), pp. 69-105. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28195903%2953%3A1%3C69%3ASSRODE%3E2.0.CO%3B2-D Poverty, The Coup Trap, and the Seizure of Executive Power John B. Londregan; Keith T. Poole World Politics, Vol. 42, No. 2. (Jan., 1990), pp. 151-183. Stable URL: http://links.jstor.org/sici?sici=0043-8871%28199001%2942%3A2%3C151%3APTCTAT%3E2.0.CO%3B2-P NOTE: The reference numbering from the original has been maintained in this citation list. http://www.jstor.org LINKED CITATIONS - Page 5 of 5 - Presidentialism in Latin America Scott Mainwaring Latin American Research Review, Vol. 25, No. 1. (1990), pp. 157-179. Stable URL: http://links.jstor.org/sici?sici=0023-8791%281990%2925%3A1%3C157%3APILA%3E2.0.CO%3B2-G Counting the Number of Parties: An Alternative Index Juan Molinar The American Political Science Review, Vol. 85, No. 4. (Dec., 1991), pp. 1383-1391. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28199112%2985%3A4%3C1383%3ACTNOPA%3E2.0.CO%3B2-0 Some Conditions of Democracy Deane E. Neubauer The American Political Science Review, Vol. 61, No. 4. (Dec., 1967), pp. 1002-1009. Stable URL: http://links.jstor.org/sici?sici=0003-0554%28196712%2961%3A4%3C1002%3ASCOD%3E2.0.CO%3B2-J Political Regimes and Economic Growth Adam Przeworski; Fernando Limongi The Journal of Economic Perspectives, Vol. 7, No. 3. (Summer, 1993), pp. 51-69. Stable URL: http://links.jstor.org/sici?sici=0895-3309%28199322%297%3A3%3C51%3APRAEG%3E2.0.CO%3B2-L Democratic Stability and Economic Crisis: Costa Rica, 1978-1983 Mitchell A. Seligson; Edward N. Muller International Studies Quarterly, Vol. 31, No. 3. (Sep., 1987), pp. 301-326. Stable URL: http://links.jstor.org/sici?sici=0020-8833%28198709%2931%3A3%3C301%3ADSAECC%3E2.0.CO%3B2-I NOTE: The reference numbering from the original has been maintained in this citation list.
Link or Click Back
Here will be a configuration form