The Populist Harm to Democracy: An Empirical Assessment
Jordan Kyle (Senior Fellow and Chief of Data and Analytics, Renewing the Centre) & Yascha Mounk (Executive Director, Renewing the Centre), Tony Blair Institut for Global Change, 26th December 2018
EXECUTIVE SUMMARY
Many academics and commentators are sounding the alarm about the threat that rising populism poses to the stability of liberal democracies. Others respond that populism is, on the contrary, a sign of democratic resilience, providing a necessary corrective that will help address popular grievances, curtail the excessive power of elites and make political systems more democratic.
To resolve this important debate on the basis of sound empirical evidence, this paper measures the impact that past populist governments have had on democracy by drawing on a first-of-its-kind global database of populist rule. It looks at the effect of populist government on three aspects of democratic institutions: the quality of democracy broadly, checks and balances on executive power, and political participation. The paper finds that populist rule—whether from the right or the left—has a highly negative effect on political systems and leads to a significant risk of democratic erosion.
KEY FINDINGS
- Populists last longer in office. On average, populist leaders stay in office twice as long as democratically elected leaders who are not populist. Populists are also nearly five times more likely than non-populists to survive in office for over ten years.
- Populists often leave office in dramatic circumstances. Only 34 per cent of populist leaders leave office after free and fair elections or because they respect term limits. A much larger number are forced to resign or are impeached, or do not leave office at all.
- Populists are far more likely to damage democracy. Overall, 23 per cent of populists cause significant democratic backsliding, compared with 6 per cent of non-populist democratically elected leaders. In other words, populist governments are about four times more likely than non-populist ones to harm democratic institutions.
- Populists frequently erode checks and balances on the executive.Over 50 per cent of populist leaders amend or rewrite their countries’ constitutions, and many of these changes extend term limits or weaken checks on executive power. The evidence also suggests that populists’ attacks on the rule of law open the way to greater corruption: 40 per cent of populist leaders are indicted on corruption charges, and the countries they lead experience significant drops in international corruption rankings.
- Populists attack individual rights. Under populist rule, freedom of the press falls by some 7 per cent, civil liberties by 8 per cent and political rights by 13 per cent.
APPENDIX: METHODOLOGY
CODING POPULISM
The data in this paper rely on the database “Populists in Power, 1990–2018” developed by Jordan Kyle and Limor Gultchin. Full details of the methodology can be found in the methodological appendix to their report Populists in Power Around the World. 34 A summary is offered below.
The database supports efforts to build a systematic understanding of how populists govern, including how they reshape state institutions, how they may or may not erode the quality of liberal democracy, and the economic policies they implement. To understand these questions across a wide range of social, economic and political contexts, a global accounting of populism in power is necessary.
To make the project cross-regional, the focus of this project is on both leaders and parties that can be classified as populist. While parliamentary systems tend to give precedence to political parties, presidential systems favour individual leaders. This analysis focuses on populist parties and leaders who attained executive office in at least minimally democratic countries between 1990 and 2016. 35 This includes only those populists who reached the presidency or prime ministership (or the equivalent executive office), and not those who governed as minority partners in a coalition government. 36 Specifically, we used the Archigos database of political leaders, which identifies the effective leader of every country in every year going back to 1875.
Requiring that countries have attained a certain level of democracy to be included leaves off many instances of populism that have risen in semi-democratic or authoritarian settings. This omits, for example, many cases of African and Middle Eastern populism. Similarly, requiring that the populist leader or party has attained the highest executive office ignores many instances where populism has been highly influential yet has never risen to the level of controlling the executive branch. In this sense, the database conservatively undercounts the global incidence and influence of populism.
Classifying particular parties and leaders as populist is a fraught exercise, due to the many disagreements on the definition of populism and the fact that populism is hardly a binary phenomenon that is either fully present or fully absent. Some leaders may be readily identifiable as full-blown populists, yet many sit on the boundary. Moreover, to the extent that populism is a political strategy that can be adopted to different degrees by different actors over time (rather than a strict political doctrine that actors either subscribe to or not), the presence or absence of populism is a matter of degree that can vary over time.
Given the difficulty of this exercise, a reasonable place to start is the extensive scientific literature on populism and the deep well of subject matter and case-study expertise that can be found there. Even though the literature famously disagrees on the exact definition of populism, there is, according to political scientist Benjamin Moffitt, “at least some (mild) consensus regarding the actual cases of actors that are usually called ‘populist’”. 37 This can be seen in the fact that scholars of populism tend to reference the same set of cases over and over.
To identify leaders associated with populism, we developed a three-step process. First, we identified 66 leading academic journals in political science, sociology and area studies that commonly publish articles on populism, as well as the new Oxford Handbook of Populism. From these sources, we queried all articles containing the keyword “populist” or “populism” in their title or abstract and scanned the texts using natural language processing technology that can identify names. These names emerged as the potential list of populist leaders.
Second, from this potential list, we carefully read each source to ensure that we included only those with substantive discussion of why the leader in question qualified as populist. We reviewed the sources for each case to verify that the leader in question met both of the elements of the definition of populism set out in the report.
Third, we sent the list of potential populist leaders that emerged from this exercise to several populism experts, to verify both whether the leaders from their region of expertise met their understanding of populism and whether there were any additional leaders whom we may have missed. To investigate these additional leaders, we often reached beyond the initial list of leading academic journals and books to other peer-reviewed specialist journals and case-specific academic books. In short, for every potential case of populism that emerged either from our initial text searches or from our consultations with experts, we consulted as many credible sources as possible to ascertain whether the case in question met our definition of populism.
POPULISTS’ LENGTH OF TIME IN OFFICE
One indicator of whether populists are eroding democratic norms and institutions is whether they tend to stay in office longer than democratically elected non-populist elected leaders. Duration models estimate the effect of populism on the ‘risk’ that a leader leaves office during a given period, conditional on the length of his or her tenure until that point. If the leader did not leave office by 31 December 2015, the data are considered censored.
Table 4 reports hazard ratios from Cox proportional hazards models rather than coefficients for each independent variable. Hazard ratios are interpreted relative to 1: a hazard ratio greater than 1 means that high values of that variable increase the risk that a leader leaves office; hazard ratios less than 1 indicate variables that decrease the risk that a leader leaves office. 38 For example, a hazard ratio of 0.67 suggests that a one-unit change in the independent variable is associated with the risk of a leader losing office dropping by one-third.
Column 1 reports results from the simplest possible duration model, with no co-variates that vary with time. In this case, the dependent variable is the number of months that the leader spends in office, and per capita income and the number of years that a country has been a democracy are measured at the point in time that a leader enters office. 39 Populism is linked with a 57 per cent reduction in the risk that a leader leaves office in any given year.
Table 4: How Long Leaders Stay in Office, 1990–2014
Dependent Variable | Time in Office (1) |
---|---|
Populist rule | 0.432 (0.094)** |
Per capita income (log) | 1.046 (0.067) |
Presidential system | 0.714 (0.148) |
# years as a democracy (log) | 0.999 (0.004) |
Civil conflict | 0.957 (0.142) |
N | 527 |
** p < 0.01. Standard errors clustered by country in parentheses.
DEMOCRATIC BACKSLIDING
Democracies that elect populists are at a far greater risk of democratic backsliding than democracies that elect non-populist leaders. However, the types of democracies that elect populists may be less consolidated and more likely to backslide in the first place. The extent to which democracy has consolidated in a country is about much more than the level of democracy in that country today. Rather, we considered the overall stock of democracy in a country: a country’s history with democratic institutions. 40
We additionally controlled for other factors that may contribute to democratic backsliding in a country: per capita income (logged), whether a country has a presidential system, average growth rates during the past three years and whether the country has an ongoing civil conflict. The models include a lagged dependent variable (that is, the democracy score of the country in the previous year), so the model estimates the effect of populism on the change in the democracy score. We also include regional and year fixed effects to account for any systematic differences between regions or over time.
Given the importance of estimating changes in the democracy score—rather than the level—we also estimate country fixed effects models. Country fixed effects control for any background characteristics of particular countries that do not change over time. Rather than examining why countries have different democracy scores overall, then, the fixed effects model focuses on why democracy scores in individual countries change over time. Variables that do not vary in countries over time—such as whether the system is presidential or parliamentary—cannot be included in the fixed effects models, as the fixed effects already account for any factors that are constant over time.
Finally, we looked at models that consider whether the length of time that a populist holds office shapes the likelihood of backsliding. These models interact populist rule with the amount of time a leader has been in office. Coefficients from three different variables are reported: “populist rule” is equal to one in any year in which a populist is in power in a given country; “time in office” is equal to the number of years the leader of a country has been in power; and “# years populist has been in office” multiplies these two terms together, so it is equal to the zero if the leader is not a populist and equal to time in office if the leader is a populist.
The coefficient on “populist rule” in this case represents the effect of populist rule if the populist has been in power for zero years. Thus, it is a rather meaningless coefficient. The coefficient on “time in office” represents the relationship between the time a non-populist leader spends in office and the likelihood of democratic erosion. The coefficient on “# years populist has been in office” is the real coefficient of interest in these models, and can tell us how the likelihood of democratic backsliding evolves as the populist is in power for longer and longer.
Table 5: Effect of Populist Rule on Democratic Backsliding, 1990–2014
Dependent Variable | Polity IV Democracy Score | ||
---|---|---|---|
(1) | (2) | (3) | |
Populist rule | -0.153 (0.077)* | -0.797 (0.429)† | 0.096 (0.290) |
Leaders’ time in office | -0.044 (0.013)** | ||
# years populist has been in power | -0.303 (0.088)** | ||
Per capita income (log) | 0.047 (0.021)* | 1.696 (0.545)** | 2.221 (0.503)** |
Growth rate (three-year average) | 0.018 (0.010)† | -0.088 (0.050)† | -0.064 (0.041) |
Presidential system | 0.054 (0.035) | ||
# years as a democracy (log) | -0.015 (0.022) | 0.455 (0.175)* | 0.396 (0.146)** |
Civil conflict | -0.050 (0.067) | -0.744 (0.502) | -0.695 (0.463) |
Lagged dependent variable | YES | NO | NO |
Regional fixed effects | YES | NO | NO |
Year fixed effects | YES | YES | YES |
Country fixed effects | NO | YES | YES |
N | 2021 | 2129 | 2103 |
** p < 0.01; * p < 0.05; † p < 0.10. Standard errors clustered by country in parentheses.
Even controlling for many of the key ways in which countries with populist rule may differ from countries without populist rule, such rule is linked with democratic backsliding: populist rule is associated with a 10 per cent drop relative to the mean democracy score. Compare this with the effect of per capita income on democratic backsliding: a country would have to drop from $16,000 to $10,000 in per capita income (a fall of 38 per cent) to see a similar risk of democratic backsliding.
What is particularly striking about the democratic backsliding overseen by populists is the strong correlation between the length of their tenure and the extent to which they damage institutions. Indeed, the first four years that a populist is in office are associated with a 14 per cent drop in a country’s Polity score (relative to the sample mean), while being in office for eight years is associated with a 29 per cent drop in the Polity score. Since many democracies run on a four-year electoral cycle, this implies that the stakes for democratic survival increase each time populists seek to renew their mandate. While most countries can contain populists’ attacks on democratic institutions while they are first in office, the system’s capacity to keep populists’ authoritarian instincts in check is significantly weakened each time a populist wins re-election.
Another way to assess whether populism is leading to democratic backsliding is to use duration analysis. In this case, we evaluated whether populist rule affects the risk that a democracy breaks down altogether (that is, drops to a Polity score below 6). Table 6 reports the results from a Cox proportional hazard model, taking the time until a democracy breaks down as the dependent variable. Countries that retained democratic institutions until 31 December 2015 are considered censored. Note that this model includes time-varying co-variates. Populist rule is associated with a 13 per cent increase in the risk that a democracy breaks down in any given year.
Table 6: Populism and the Duration of Democracy, 1990–2014
Dependent Variable | Time to Democratic Breakdown (1) |
---|---|
Populist rule | 1.132 (0.038)** |
Per capita income (log) | 0.919 (0.032)* |
Growth rate (three-year average) | 0.977 (0.020) |
Presidential system | 0.479 (0.332) |
# years as a democracy (log) | 0.971 (0.033) |
Civil conflict | 1.118 (0.033)* |
N | 1820 |
Standard errors clustered by country in parentheses.
POPULISM AND CHECKS AND BALANCES ON THE EXECUTIVE
We also tested specifically whether populists are more likely than non-populists to reduce institutional checks and balances on the executive. We measured checks and balances using the Constraints on the Executive variable from Polity IV, which measures the extent of institutionalised constraints on the decision-making powers of chief executives. In democracies, this involves having an independent, empowered legislature and a strong, independent judiciary.
Table 7 reports the results from the analysis. Controlling for per capita income (logged), whether a country has a presidential system, average growth rates during the past three years, the number of years that a country has been democratic (logged), whether the country has an ongoing civil conflict, regional and year fixed effects, and the country’s level of executive constraints in the previous year, we found that populist rule is associated with a 4 per cent drop in the country’s score on the executive-constraints indicator relative to the sample mean. A four-year term is associated with a 6 per cent drop in the country’s executive constraints indicator (relative to the sample mean).
Table 7: Effect of Populist Rule on Constraints on the Executive, 1990–2014
Dependent Variable | Constraints on the Executive | ||
---|---|---|---|
(1) | (2) | (3) | |
Populist rule | -0.053 (0.028)† | -0.270 (0.202) | -0.007 (0.149) |
Leaders’ time in office | -0.031 (0.013)* | ||
# years populist has been in power | -0.090 (0.038)* | ||
Per capita income (log) | 0.018 (0.008)* | 0.553 (0.217)* | 0.638 (0.197)** |
Growth rate (three-year average) | 0.009 (0.004)* | -0.0270 (0.017) | -0.021 (0.015) |
Presidential system | 0.023 (0.016) | ||
# years as a democracy (log) | 0.001 (0.010) | 0.166 (0.076)* | 0.189 (0.078)* |
Civil conflict | -0.026 (0.027) | -0.287 (0.241) | -0.280 (0.227) |
Lagged dependent variable | YES | NO | NO |
Regional fixed effects | YES | NO | NO |
Year fixed effects | YES | YES | YES |
Country fixed effects | NO | YES | YES |
N | 2021 | 2129 | 2103 |
** p < 0.01; * p < 0.05; † p < 0.10. Standard errors clustered by country in parentheses.
We also looked at whether populism affects rule of law in a country. We used the rule-of-law indicator of the World Bank Governance Indicators, which captures the extent to which agents have confidence in and abide by the rules of society. In particular, it looks at the quality of contract enforcement, property rights, the police and the courts. 41 Table 8 reports the results.
Table 8: Effect of Populist Rule on the Rule of Law, 1990–2014
Dependent Variable | Rule of Law | ||
---|---|---|---|
(1) | (2) | (3) | |
Populist rule | -0.018 (0.007)** | -0.059 (0.041) | -0.047 (0.041) |
Leaders’ time in office | -0.004 (0.002)† | ||
# years populist has been in power | -0.003 (0.010) | ||
Per capita income (log) | 0.0134 (0.005)** | 0.420 (0.175)* | 0.444 (0.175)* |
Growth rate (three-year average) | 0.003 (0.001)* | -0.010 (0.006)† | -0.011 (0.006)† |
Presidential system | 0.005 (0.005) | ||
# years as a democracy (log) | 0.006 (0.004) | 0.029 (0.036) | 0.029 (0.036) |
Civil conflict | -0.007 (0.009) | -0.042 (0.071) | -0.040 (0.072) |
Lagged dependent variable | YES | NO | NO |
Regional fixed effects | YES | NO | NO |
Year fixed effects | YES | YES | YES |
Country fixed effects | NO | YES | YES |
N | 1573 | 1687 | 1668 |
** p < 0.01; * p < 0.05; † p < 0.10. Standard errors clustered by country in parentheses.
We found that, as predicted, populist rule is associated with a deterioration in the rule of law on this metric. However, the effect size is substantively small and not robust in all models.
POPULISM, POLITICAL RIGHTS AND CIVIL LIBERTIES
Replicating the same models with the same controls used throughout this paper, we found that populist rule is associated with less freedom of the press, fewer civil liberties and fewer political rights (see tables 9, 10 and 11). We measured each of these concepts using data from Freedom House, which makes annual assessments of a country’s press freedom, civil liberties and political rights. For each of these variables, we reversed the scales so that higher values indicate more freedom.
Populist rule is associated with a 7 per cent decline in freedom of the press, an 8 per cent fall in civil liberties and a 13 per cent drop in political rights (relative to sample means).
Table 9: Effect of Populist Rule on Press Freedom, 1990–2014
Dependent Variable | Freedom House Press Freedom | ||
---|---|---|---|
(1) | (2) | (3) | |
Populist rule | -0.945 (0.314)** | -2.570 (1.477)† | 0.081 (1.572) |
Leaders’ time in office | -0.118 (0.094) | ||
# years populist has been in power | -0.953 (0.500)† | ||
Per capita income (log) | 0.207 (0.099)* | 11.80 (3.348)** | 12.50 (3.093)** |
Growth rate (three-year average) | 0.132 (0.028)** | -0.397 (0.121)** | -0.373 (0.114)** |
Presidential system | 0.109 (0.171) | ||
# years as a democracy (log) | 0.011 (0.122) | 1.085 (1.146) | 1.121 (1.182) |
Civil conflict | -0.307 (0.365) | -6.408 (1.751)** | -6.171 (1.681)** |
Lagged dependent variable | YES | NO | NO |
Regional fixed effects | YES | NO | NO |
Year fixed effects | YES | YES | YES |
Country fixed effects | NO | YES | YES |
N | 1748 | 1892 | 1870 |
** p < 0.01; * p < 0.05; † p < 0.10. Standard errors clustered by country in parentheses.
Table 10: Effect of Populist Rule on Civil Liberties, 1990–2014
Dependent Variable | Freedom House Civil Liberties | ||
---|---|---|---|
(1) | (2) | (3) | |
Populist rule | -0.071 (0.030)* | -0.187 (0.093)* | -0.059 (0.086) |
Leaders’ time in office | |||
# years populist has been in power | |||
Per capita income (log) | 0.062 (0.012)** | 0.686 (0.242)** | 0.792 (0.224)** |
Growth rate (three-year average) | 0.004 (0.004) | -0.019 (0.011) | -0.015 (0.010) |
Presidential system | 0.0491 (0.016)** | ||
# years as a democracy (log) | 0.004 (0.012) | 0.142 (0.075)† | 0.144 (0.074)† |
Civil conflict | -0.086 (0.037)* | -0.150 (0.114) | -0.148 (0.113) |
Lagged dependent variable | YES | NO | NO |
Regional fixed effects | YES | NO | NO |
Year fixed effects | YES | YES | YES |
Country fixed effects | NO | YES | YES |
N | 1970 | 2077 | 2052 |
** p < 0.01; * p < 0.05; † p < 0.10. Standard errors clustered by country in parentheses.
Table 11: Effect of Populist Rule on Political Rights, 1990–2014
Dependent Variable | Freedom House Political Rights | ||
---|---|---|---|
(1) | (2) | (3) | |
Populist rule | -0.085 (0.026)** | -0.280 (0.151)† | -0.030 (0.125) |
Leaders’ time in office | -0.015 (0.007)* | ||
# years populist has been in power | -0.084 (0.030)** | ||
Per capita income (log) | 0.043 (0.011)** | 0.784 (0.379)* | 0.946 (0.351)** |
Growth rate (three-year average) | 0.001 (0.004) | -0.022 (0.012)† | -0.015 (0.010) |
Presidential system | 0.042 (0.019)* | ||
# years as a democracy (log) | 0.008 (0.013) | 0.060 (0.090) | 0.045 (0.084) |
Civil conflict | -0.053 (0.033) | -0.222 (0.212) | -0.211 (0.199) |
Lagged dependent variable | YES | NO | NO |
Regional fixed effects | YES | NO | NO |
Year fixed effects | YES | YES | YES |
Country fixed effects | NO | YES | YES |
N | 1970 | 2077 | 2052 |
** p < 0.01; * p < 0.05; † p < 0.10. Standard errors clustered by country in parentheses.
ACKNOWLEDGEMENTS
We thank Louise Paulsen and Limor Gultchin for excellent research assistance.
N&R
- Jared Diamond, “Facing up to the democratic recession”, Journal of Democracy 26, no. 1 (2015): 141–155.
- Steven Levitsky and Daniel Ziblatt, How Democracies Die (New York: Crown, 2018); Yascha Mounk, The People vs. Democracy: Why Our Freedom Is In Danger and How to Save It (Cambridge, Massachusetts: Harvard University Press, 2018).
- Timothy Snyder, On Tyranny (New York: Tim Duggan Books, 2017); Madeleine Albright, Fascism: A Warning (New York: HarperCollins, 2018).
- Martin Eiermann, Yascha Mounk and Limor Gultchin, European Populism: Trends, Threats and Future Prospects, Tony Blair Institute for Global Change, 29 December 2017, https://institute.global/insight/renewing-centre/european-populism-trends-threats-and-future-prospects; Jordan Kyle and Limor Gultchin, Populists in Power Around the World, Tony Blair Institute for Global Change, 7 November 2018, https://institute.global/insight/renewing-centre/populists-power-around-world.
- Dalibor Rohac, “Hungary and Poland aren’t democratic. They’re authoritarian”, Foreign Policy, 5 February 2018, https://foreignpolicy.com/2018/02/05/hungary-and-poland-arent-democratic-theyre-authoritarian/; Diego Cupolo, “The decline and fall of Turkish democracy”, Atlantic, 13 April 2017, https://www.theatlantic.com/international/archive/2017/04/turkey-referendum-erdogan-kurds/522894/.
- Freedom House, “Freedom in the World 2018: Democracy in Crisis”, 2018, https://freedomhouse.org/report/freedom-world/freedom-world-2018.
- Samuel Huntington, The third wave: Democratization in the late twentieth century (Norman, Oklahoma, University of Oklahoma Press, 1993).
- See, for example, Chantal Mouffe, “Populists are on the rise but this can be a moment for progressive too”, Guardian, 10 September 2018, https://www.theguardian.com/commentisfree/2018/sep/10/populists-rise-progressives-radical-right.
- On the ambiguous relationship between populism and democracy, see Cas Mudde and Cristóbal Rovira Kaltwasser, Populism in Europe and the Americas: Threat or corrective for democracy? (Cambridge: Cambridge University Press, 2012); Cristóbal Rovira Kaltwasser, “The response of populism to Dahl’s democratic dilemmas”, Political Studies 62, no. 3 (2014): 470–487; Cristóbal Rovira Kaltwasser, “The ambivalence of populism: threat and corrective for democracy”, Democratization 19, no. 2 (2012): 184–208.
- Kyle and Gultchin, Populists in Power Around the World.
- Ibid.
- The exceptions are studies on Latin America that ask whether populists in public office have had a negative effect on democratic institutions in the region. Broadly, they find that populist rule is associated with lower estimates of the quality of a country’s democracy. See, for example, Christian Houle and Paul Kenny, “The political and economic consequences of populist rule in Latin America”, Government and Opposition 53, no. 2 (2018): 256–287; Robert Huber and Christian H. Schimpf, “Friend or foe? Testing the influence of populism on democratic quality in Latin America”, Political Studies 64, no. 4 (2016): 872–889; Steven Levitsky and James Loxton, “Populism and competitive authoritarianism in the Andes”, Democratization 20, no. 1 (2013): 107–136; Saskia Pauline Ruth, “Populism and the Erosion of Horizontal Accountability in Latin America”, Political Studies 66, no. 2 (2018): 356–375. For work looking at both Europe and Latin America, see Mudde and Rovira Kaltwasser, Populism in Europe and the Americas.
- Authors’ calculations using data on democratic breakdowns from Ko Maeda, “Two modes of democratic breakdown: A competing risks analysis of democratic durability”, Journal of Politics 72, no. 4 (2010): 1129–1143.
- Levitsky and Ziblatt, How Democracies Die.
- Authors’ calculations using data on democratic breakdowns from Maeda, “Two modes of democratic breakdown”, and Kyle and Gultchin, Populists in Power Around the World.
- Levitsky and Ziblatt, How Democracies Die, 3. According to most definitions, full democracies must meet at least four minimum criteria: executives and legislatures are selected through free and fair elections; virtually all adults have the right to vote; political rights and civil liberties are respected; and elected authorities can truly govern. Even in democracies that occasionally violate one of these criteria, the opposition retains the power to challenge incumbents in democratic elections that are mostly free and fair. However, once leaders begin trampling on electoral rules, the independence of institutions as well as political rights and civil liberties, a country can cease to be fully democratic. Once the violations become both frequent and serious, it fundamentally alters the playing field between incumbent leaders and opposition challengers; a once-democratic regime then ceases to be a democracy in anything but name.
- On how democracies die slowly, and in plain view, under populist governments, see Cas Mudde, “Don’t blame democracy’s decline on ignorance. The problem lies deeper”, Guardian, 15 December 2018, https://www.theguardian.com/commentisfree/2018/dec/15/democracy-authoritarianism-media-spotlight-viktor-orban.
- H.E. Goemans, Kristian Skrede Gleditsch and Giacomo Chiozza, “Introducing Archigos: A Data Set of Political Leaders”, Journal of Peace Research 46, no. 2 (2009): 269–283. Only countries with a score of at least 6—the traditional cut-off for measuring democracy—on the Polity IV index are counted as democracies. For the Polity IV database, see http://www.systemicpeace.org/inscrdata.html.
- For an excellent introduction to duration analysis in political research, see Janet Box-Steffensmeier, Dan Reiter and Christopher Zorn, “Nonproportional hazards and event history analysis in international relations”, Journal of Conflict Resolution 47, no. 1 (2003): 33–53.
- Some populist leaders leave power and later assume office again. We counted each populist term separately for this analysis.
- Calculated using Polity IV database.
- For the Polity IV database, see http://www.systemicpeace.org/inscrdata.html.
- John B. Judis, “Us v Them: the birth of populism”, Guardian, 13 October 2016, https://www.theguardian.com/politics/2016/oct/13/birth-of-populism-donald-trump.
- Cas Mudde, “The Populist Radical Right: A Pathological Normalcy”, West European Politics 33, no. 6 (2010): 1173.
- See, for example, Mouffe, “Populists are on the rise but this can be a moment for progressive too”.
- Nadia Urbinati, “Democracy and Populism”, Constellations 5, no. 1 (2008): 110–124, https://doi.org/10.1111/1467-8675.00080.
- Cases of constitutional change during populist rule include Argentina (1993), Belarus (1996, 2004), Bolivia (2009), Brazil (1992–1993), Bulgaria (2015), the Czech Republic (1998, 2002), Ecuador (2008), Georgia (2004, 2010), Hungary (2011), India (2015, 2017), Italy (2011), Macedonia (2014, 2015), Nicaragua (2013), Paraguay (2011), Peru (1993), Poland (1992), Russia (2008), Slovakia (1998), Slovakia (2015), South Africa (2013), Sri Lanka (2010), Taiwan (2004–2005), Turkey (2007, 2010, 2017) and Venezuela (1994, 2000, 2009).
- For information on how the rule-of-law indicator is constructed, see “Rule of Law”, World Bank, http://info.worldbank.org/governance/wgi/pdf/rl.pdf.
- Barry Eichengreen, “Populism’s common denominator”, Project Syndicate, 9 November 2018, https://www.project-syndicate.org/commentary/populism-common-denominator-political-corruption-by-barry-eichengreen-2018-11.
- Ernesto Laclau, On Populist Reason (London: Verso, 2005).
- On the ambiguous relationship between populism and democracy, see Cas Mudde and Cristóbal Rovira Kaltwasser, Populism in Europe and the Americas: Threat or corrective for democracy? (Cambridge: Cambridge University Press, 2012); Cristóbal Rovira Kaltwasser, “The response of populism to Dahl’s democratic dilemmas”, Political Studies 62, no. 3 (2014): 470–487; Cristóbal Rovira Kaltwasser, “The ambivalence of populism: threat and corrective for democracy”, Democratization 19, no. 2 (2012): 184–208.
- Kirk Hawkins, “Who mobilizes? Participatory democracy in Chavez’s Bolivarian revolution”, Latin American Politics and Society 52, no. 3 (2010): 31–66.
- Houle and Kenny, “The political and economic consequences of populist rule in Latin America”.
- Kyle and Gultchin, Populists in Power Around the World.
- Only countries with a score of at least 6—the traditional cut-off for measuring democracy—on the Polity IV index are included. Venezuela is a bit of an odd case. When Hugo Chávez attained office in 1999, Venezuela was a democracy. By the time he died in office in 2013, Venezuela had backslid into autocracy. However, we include the Maduro regime in the database as it is really one long spell of populism in the country. For the Polity IV database, see http://www.systemicpeace.org/inscrdata.html.
- H.E. Goemans, Kristian Skrede Gleditsch and Giacomo Chiozza, “Introducing Archigos: A Data Set of Political Leaders”, Journal of Peace Research 46, no. 2 (2009): 269–283.
- Moffitt, The Global Rise of Populism, 41.
- We clustered standard errors by country. However, because the tables report hazard ratios rather than coefficients, the standard rule of thumb of two-to-one for comparing coefficients with standard errors does not apply.
- An essential assumption of Cox proportional hazards models is that the ratio of the hazards between observations is constant over time. Figure 2, which shows the Kaplan-Meier survival curves for populist vs. non-populist leaders, illustrates quite clearly that the assumption of proportional hazards is appropriate for these data: the ratio between the curves remains relatively constant over time. We also test the proportional hazards assumption formally using the Schoenfeld residuals; these tests reveal that the proportional hazards assumption is appropriate for these data. Results are available on request.
- On why we should measure the stock of democratic history in a country rather than the contemporary level of democracy to assess the strength of democratic consolidation, see John Gerring, Philip Bond, William Barndt and Carola Moreno, “Democracy and economic growth: A historical perspective”, World Politics 57, no. 3 (2005): 323–364.
- For information on how the rule-of-law indicator is constructed, see “Rule of Law”, World Bank, http://info.worldbank.org/governance/wgi/pdf/rl.pdf