Michael Coppedge is a Professor of Political Science at the University of Notre Dame, a principal investigator of the Varieties of Democracy project, and a faculty fellow at the Kellogg Institute for International Studies. He is a coeditor (along with Amanda Edgell, Carl Henrik Knutsen, and Staffan Lindberg) of Why Democracies Develop and Decline.
Democracy is a complex concept. It has to do with elections. It has to do with legislatures. It has to do with civil society organizations and courts and political styles of politicians. There's a lot packed into the concept and it's multidimensional, because some of these components don't move together.
- Democracy as a multidimensional concept
- How the conditions for democratization differ from those for backsliding
- Ways researchers use information from V-Dem to discover new insights about democracy
- New findings from V-Dem research regarding presidentialism, party system institutionalization, and anti-system parties
- How has V-Dem changed research about democracy
Thank you for listening to the Democracy Paradox: A podcast on democracy, democratization, and world affairs. Each week we talk about big picture insights to better understand political issues and events. These are complex ideas that might be unfamiliar, so I always provide a complete transcript at democracyparadox.com.
Today’s guest is Michael Coppedge. Michael is a Professor of Political Science at the University of Notre Dame, a principal investigator of the Varieties of Democracy project, and a faculty fellow at the Kellogg Institute for International Studies. Recently, he coedited a new book with Amanda Edgell, Carl Henrik Knutsen, and Staffan Lindberg called Why Democracies Develop and Decline.
Our conversation touches on a lot of complex topics surrounding democracy including democratization and democratic backsliding. We talk about some of the book’s findings, but also some of the big picture ways to think about democracy. Michael also explains how the Varieties of Democracy project has changed how we think about democracy. For those unfamiliar with the Varieties of Democracy, I recommend checking them out at v-dem.net. It’s features the most extensive attempt to measure different aspects of democracy.
If you want to learn more about the history of the V-Dem project, there is a short bonus episode with additional material from my conversation with Michael for patrons at Patreon. Monthly contributors can access additional bonus material like this for as little as $5/month. There is a link in the show notes or you can just look up Democracy Paradox at patreon.com. Like always feel free to email me questions or comments to email@example.com. Here is my conversation with Michael Coppedge…
Michael Coppedge, welcome to the Democracy Paradox.
Well, Michael, I was really caught off guard by this new book that your team has put together, Why Democracies Develop and Decline. Because when I came across it, I thought it was going to be a basic overview of some of the findings from the Varieties of Democracy project and wasn't going to really say anything that caught me by surprise. I thought it was going to say a lot of things that I would expect somebody like Varieties of Democracy or even Freedom House or Polity to be able to say about some of the basic findings of the direction of democracy. But I was incredibly impressed that you guys really went above and beyond to use that same type of ambition that Varieties of Democracy has been known for to put together this book to consider ways in which democracies really do develop and decline.
So, I want to start with a phrase that you wrote along with your coeditors Amanda Edgell, Carl Henrik Knutsen, and Staffan Lindberg. In the book you right, “Democracy is a multidimensional concept.” And I personally think that this really gets at the heart of what the V-DEM project is. But I'd like to hear it from you. What does that phrase mean to you when you say democracy is a multidimensional concept?
Well, what it means to me is that first off democracy is a complex concept. It has to do with elections. It has to do with legislatures. It has to do with civil society organizations and courts and political styles of politicians. There's a lot packed into the concept and it's multidimensional, because some of these components don't move together. That is if one component has a really high, very democratic value, other components might not have a higher democratic value in that particular country. So, like all good things don't necessarily go together.
An example of that would be suffrage and party competition. So, if suffrage and party competition were unidimensional with each other, that is they aligned on the same dimension, then all the countries that have high party competition would also have high suffrage and vice versa. And all the countries that have low suffrage would also have very little party competition. But that's really not the way the world works, because we have historical cases like the Soviet Union which had full adult suffrage. They had elections. Everybody could vote. In fact, everyone had to vote and yet there was no party competition. You could vote for the Communist Party or you could vote for the Communist Party. That's it.
A similar kind of thing happened in Hong Kong during the big protest. The Beijing led administration was saying, ‘Well, we can provide you with full suffrage. Don't you want to have full suffrage.’ But what they were promising was full suffrage without real party competition, because there would be proscriptions of opposition legislators and also elections would be to a body that wouldn't really have ultimate authority to make the laws. So, full suffrage doesn't necessarily go with other things that we mean by democracy.
But, of course, this is a matter of degree and it's basically a question of how much variance do you lose when you combine indicators into a summary measure. If components are unidimensional, then you lose little information when you combine them into a measure of a more general concept. But if it's multidimensional, it's still kind of meaningful at the extremes. Like only the cases that are high on everything get the high scores and all the cases that are low on everything get the low scores. But in the middle, you get a mixture of cases that are high on some things and low on others. They're different from case to case which ones are high and which ones are low. So, there's a meaningless mess in the middle if it's multidimensional in that way.
So, the implication of this is it's not very useful to combine measures that are multidimensional, unless you really know what you're doing. Some of the components that we consider in our democracy measures of the hundreds of components or specific variables we consider are not unidimensional. Even the pretty high level five components like the liberal component index, the egalitarian component index, participatory, deliberative, and electoral, those component indices probably should not be combined into a single index of what we sometimes call big D democracy, like democracy writ large. Too much useful information would be lost in the process of doing that, because those different components don't go together perfectly well. Some of them go together better than others.
So, electoral and liberal components go together pretty well and they go together pretty well with participatory, but the egalitarian and deliberative components don't go together all that well. They can still be combined if they're multidimensional, but it requires a really carefully thought-out rationale that leads to a mathematical formula that tells you how to combine these measures into a single index. Our deliberations yielded some formulas for doing that. It's certainly not the only possible way to do it. We deliberated for about a year over how to do this and we were kind of exhausted at the end of it.
It kind of ended up being a compromise between people – well, compromise, not between people, but between necessary conditions. A kind of logic that says that everything matters. The other things don't matter unless this is here and a substitutive principle that says that a high score on one thing can compensate for a low score in another thing. The first principle leads to a multiplicative formula. The second leads to an additive formula or a weighted average and our final indices are mostly an average of a multiplicative formula and an additive formula.
But anyway, that made some sense to us because there seems to be a rationale for both ways of aggregating things. Also, the resulting index seems to us to have some pretty high face validity. That is, it gives high ratings to countries that we think are very democratic and very low ratings to countries that we think are almost completely undemocratic and rates the things in the middle pretty well.
Do you find that some, some of the measurements when they're combined with others become exponentially more important? Like if you combine one part of free speech with electoral rights that they become more than the sum of their parts.
Yeah, I think that's true. Because, I mean, if you're trying to measure some kind of democracy, it's necessarily a complex, multi-component, multi-dimensional concept and you can't measure that concept unless you have all of its parts in there somewhere somehow. On the other hand, for some purposes, I think we focus too much on a big concept of democracy and we might be better served by narrowing our focus down to components of democracy that give us more detailed information about what's going on.
Yeah. I think that gets back to the idea of it being multidimensional once again.
Yeah. One of the solutions to multidimensionality is not to aggregate things. It's just to keep them separate.
So, one of the interesting points that you bring up in the book and I find this idea kind of comes up in some of the other chapters as well. You write, “Conditions that explain levels of democracy are often different from those that explain change and that explanations for advances in democracy are probably different from those that explain setbacks.” Now it makes intuitive sense to me in a lot of ways, but I don't know that I've heard anybody say that directly before. How do the conditions that help produce democratization possibly differ from the conditions that produce democratic backsliding?
Well, if you look at a lot of democracy data, I think it quickly becomes obvious that there is sort of a punctuated equilibrium. Most countries don't change their democracy scores very much most of the time. So, there are long periods of relative stasis that happen in every country's democracy time series. So, the levels of democracy reflect mostly these long periods of stasis and because they are static most of the time, the best predictors for these levels of democracy are the more static or slowly moving kinds of variables like geography, patterns of European settlement, or levels of literacy or income or state capacity which are things that tend to change slowly and not really dramatically. They're more like stocks than flows.
So, they're relatively stable, but there are changes that happen in democracy scores that tend to cluster into short periods of more dramatic change. That's the punctuation of the equilibrium and to some extent, these short periods of rapid change can result from an explosion of pent-up long-term forces. But I think more often they're triggered by more proximate variables, causally approximate variables that change more dynamically. So, things like civil society campaigns and national economic growth rates or episodes of violence that are kind of sporadic and sudden. So, you know, if it's here this year, it may not be there five years from now. Those things are probably naturally better suited for explaining periods of change.
But as for upturns versus downturns, that's just change in two different directions. I think the first article that I saw that hypothesized that improvements might have different explanations than declines was Dankwart Rustow’s article in 1970 about democratization. That's one of the things that he said. Nevertheless, a lot of the quantitative models to explain democratization ignored that and didn't test for that hypothesis. There are a few that did including a nice 2010 book by Jan Teorell who's one of our principal investigators and we're following in his mold there very explicitly. But there are some different factors that matter. So, for downturns, some things that matter are presence of anti-system movements, poor state capacity, the national economic growth rate and presidentialism and the health of the civil society environment matters a lot for promoting upturns.
So, there's some different factors that matter and in fact, there's some that can have some different effects like economic growth and decline. Economic growth tends to stabilize a regime whether it's a democratic or a non-democratic regime and economic decline tends to promote more change. So, an economic decline tends to push democracies in the direction of less democracy, but it tends to push less democratic regimes in the direction of more democracy. So, it can have an opposite effect. That's the only variable I can think of that really has an opposite effect on change, but they are different models.
So, when we think about factors that are different for the process of democratization and ones that are different for backsliding, is there ever some overlap where the same exact factors that produce democratization can sometimes produce backsliding at different points in its trajectory? I mean, I guess one of them you've already mentioned is GDP. If there's GDP decline, when it's authoritarian, it can produce democratization. If there's GDP decline, when there's an established democracy it can produce backsliding. Are there any others that have overlap like that that both produce opportunities for democratization in some aspects, but can also produce backsliding in other aspects?
Well, this is not part of this book, but I had a paper once, it's not published, but it's based on looking at the relationship between per capita GDP and levels of democracy. You know, that's one of the best-established findings in comparative politic. High income is associated with higher levels of democracy. But if you look at the country level and estimate how much of an impact income has on democracy in each country you find that, yes, on average across many countries, there's a positive association. It's very significant and robust. However, if you break it down at the country level, some countries have a really high positive association between income and democracy and others have a negative association between income and democracy over a long period of years.
So, in this paper we found that one possible explanation for that is what is the source of economic growth going on in the country. It turns out that if a country derives a lot of its income from drug trafficking, then economic growth is negatively associated with democracy both because it leads to human rights violations and abuse and corruption. There are all kinds of things about that path of economic growth. So, more generally, this is just a hypothesis that is not part of this book, but I think that there are probably conditional relationships in which some factors that tend to improve democracy might in some situations be detrimental to democracy.
That's interesting. Obviously, I can think of a few different examples. One would be oil revenues tend to work in that opposite direction. But another one that caught me off guard until I read her book was Bryn Rosenfeld's argument about the state dependent middle class that came out about two years ago. She made the case that the people that work in autocratic regimes for the government are more likely to want to keep that stable regime type and not want to go out on the street and protest. So, it kind of breaks down that expectation that a growing middle class will rise up against the autocratic regime, because if that middle class is dependent on the autocratic regime, their interests diverge. So, it reminds me of that argument.
Yeah and I think this is a good occasion to acknowledge one of the limitations of this book that we have which is that we're making generalizations about a huge sample. We're covering almost.. well, covering more than a hundred years of history and not differentiating too much and in most of the chapters among different regions of the world.
For the most part, we haven’t allowed the relationships to be different in different historical periods like the Interwar Period and the immediate post-World War II period, and after the fall of the Berlin Wall. It could be that the relationships are different in different periods and different in different regions of the world. We decided not to take that on in this book, because we already had enough on our plate. But I think that would be a great opportunity for future work. So, there could be situations where there's a state dependent middle class that's a special syndrome that would run counter to some of our findings.
Well, like you said, your book already challenges and brings up a lot of different ideas within political science. But you also bring up an interesting point which is that a lot of qualitative work oftentimes identifies some of the nuances that quantitative approaches can sometimes miss when you're taking these broad-brush approaches. In fact, quantitative work is oftentimes using the ideas from qualitative work and approaching them from a different methodology. Do you feel that there's always going to be some of those nuances that quantitative approaches might miss that qualitative approaches can capture?
Yeah, sure. I mean, quantitative work almost necessarily overlooks the proper nouns. You know, instead of talking about the Duma, we talk about legislatures. Instead of talking about the Prime Minister, we talk about a head of government. So, prime ministers can have really different missions and roles in political systems around the world depending on the context. So, we miss that and we also miss some idiosyncratic institutions that some countries have that are not generally very relevant and so it was not practical to try to measure them.
For example, one of my friends strongly urged me to include participatory municipal budgeting among the things that we measure. But as far as I was aware at that time, this was a phenomenon that was mostly limited to Southern Brazil and while it might be really important for local level democracy in part of Brazil and it might spread to other parts of Brazil and to other countries, and it has to some extent, I was just looking ahead and thinking, do I want to have a question about participatory municipal budgeting in which most of our thousands of country experts are just going to say no. That doesn't exist here and it just didn't seem worth.
Another example is the US electoral college which is pretty unique in the world these days. There have been a few other countries that had an electoral college like Argentina, but don't anymore. Obviously, that has a big impact on democracy in the United States, but it's not something that we really distinguish or include in our measures. Like I said before, we haven't in this book tested many hypotheses about how these relationships might differ in time and space.
So, earlier you mentioned how for most countries there's a long stasis where democracy generally remains the same and there's really short periods where there's a lot of dynamism where things might drop or might increase.
Yeah. That's the tendency. There are a few exceptions to that where there have been slow improvements like in South Korea since the 1980s. But most countries are relatively static punctuated by periods of rapid change. Some, of course, are changing a lot all the time. Argentina, Peru, Thailand, they've been really unstable.
No, of course. So, when we think about democratic backsliding, because that's really a big focus of a lot of researchers these days, what Larry Diamond calls the democratic recession, others call it the third wave of democratization. Do we get too caught up in these short periods of backsliding within countries rather than thinking about much longer periods that might take 50 years, 25 years or whatever kind of timeframe that you think is appropriate.
I wouldn't say that we get too caught up in short periods of change. But I think what we sometimes lose sight of is that these periods of change are defined by relatively few countries that happen to move in the same direction around the same time when there are always a lot of countries that are not changing during these, you know, so-called waves of democratization or autocratization. There are countries that also move in the contrary direction to the wave. You know, it's not like in a wave most countries are becoming less democratic around the same time. It's a result of a small number of countries. I think last year, by our account there was something like 33 countries that declined and only… Well, I forget the exact number, but between 10 and 15 that improved.
But then there were lots more that didn't change significantly in either direction. I always want to remind people about the big picture. You know, what's worrisome about the current wave of autocratization is that it's been happening in some very populous countries (India, the United States, and Brazil). So, it affects many more people.
And sometimes those are countries that have importance beyond just the number of people that live there. Countries like the United States and India have symbolic importance as well as the importance just based on their size economically or demographically.
Yeah. I think what happens in the United States especially has a lot of consequences for democracy in other countries around the world. So, if we're suffering as we are right now, then I think that sends a signal to other countries that democracy may not necessarily be the best way to go.
So, Michael, one of the things from V-DEM is that there's five key indicators or five different perspectives to think about democracy. You've talked about it already, but at the same time some of those indicators kind of track with one another. You already mentioned how liberal democracy and electoral democracy track with one another. In the book, Carl Henrik Knutsen and Svend-Erik Skaaning write, “The countries that perform well on one aspect of democracy tend to perform well also on other aspects and vice versa.” So, when we're looking at these multiple aspects of democracy what do they really tell us about the differences within these countries, if oftentimes they're tracking well on multiple dimensions or if they're doing poorly, they're doing poorly on multiple dimensions as well. It seems like they track very well with one another.
Yeah, the highest-level indices, what we call the high-level indices of electoral democracy, liberal democracy, participatory democracy, et cetera, they are all very highly correlated. I think in every case it's 0.9 or higher. So, for the most part that really high level of correlation is because each of those contains the electoral democracy index. So, they literally measure the same thing in part with an extra component for the special variety of democracy that they're trying to measure. But the more you disaggregate. The more you focus on the less general, more specific components of democracy, the less these things are associated with one another.
So, there's less of an association between the liberal component, egalitarian component, participatory component, et cetera and you can break those things down into other indicators. Like for the liberal component, there is individual liberties and constitutional guarantees. It's not the exact phrase, but it's basically a civil liberties indicator. Legislative checks on the executive and judicial checks on the executive and those things have some association with one another, but less than the higher-level indices. So, basically the more you disaggregate the less these indicators have to do with each other. In fact, if you took the approximately 60 specific indicators that go into our high-level democracy indices, and compared the variance that they contain with the variance that's contained in just the five components near the top, you lose about 36% of the information that was originally contained in there.
So, if you tried to combine them all the way up to a big D democracy, you'd lose almost half of the information that was contained in the very specific indicators. So, I think that's a really important lesson that we lose valuable information, if we focus just on the most general ideas of democracy. In many cases, it's more useful to focus on the more specific indicators or even the components that are halfway in between. And if it's appropriate to do that, like if we're really interested in legislative checks on the executive, we should use that component rather than using democracy in an analysis.
So, I mentioned early on one of the things I was impressed with the book was that you guys were not afraid to challenge some very controversial or some very big picture ideas within political science. A few of those were a finding made in the book that argues against the perils of presidential thesis. There's an argument against the idea of party system institutionalization having a significant impact on the health of democracy. There's also a very important finding that anti-system movements contribute to downturns. I was really surprised at some of the things that you guys found and that you guys really kind of put into this model. Was there anything that really caught you off guard that you expected to have more significance, but didn't have as much or anything that you thought maybe was irrelevant, that you were surprised to find was highly significant?
Well, the presidentialism finding was surprising to me. I mean, Juan Linz was my dissertation advisor. So, I was steeped in that literature as a graduate student and I know that the subsequent literature has found very little support for it. I'm kind of agnostic about it. But in chapter eight, the model says that there is a significant impact in I believe the institution's chapter that it actually helps. Frankly, I just don't know what to make of that. I'm not sure whether to trust it or not. One of my co-authors suggests that we do need to take it seriously and try to make sense of it.
But I'm just not trusting it yet. I think one possibility is that anytime you put a hundred variables in a model, some of them are going to turn out to be significant even if they're not really causal. So, I'm just kind of holding off in believing that particular one. On party institutionalization, yes, the institution's chapter didn't find a significant impact of that, at least not in the expected direction. The concluding chapter does say that there's an important role for party institutionalization. That it's one of the three components that are part of what we're calling a protective belt around democracy and in fact, around non-democratic regimes as well. That institutionalized political parties together with vibrant civil society and the rule of law help preserve the regime, whatever it is, and kind of buffer the impact of more distant factors.
They tend to be a stabilizing influence and they help explain why democracy is a punctuated equilibrium. They're part of the reason for the equilibrium and why things tend not to change very much from year to year, because these three factors are a buffer for the level of democracy. They are themselves fairly stable and they reinforce one another and they are caused by other things that tend to be fairly stable trends. So, I really like that finding.
You also mentioned that the most important finding was how anti-system movements contribute to downturns. It wasn't surprising, but it was definitely interesting that you emphasized how important it was within the model that you had for democratic backsliding.
Yeah, I think we called it the most important finding, not necessarily because it was the most influential, immediate cause of downturns, but because this had not been well established in the literature before. This variable came out of the chapter on social movements, civil society, and in that chapter, Michael Bernhard and Amanda Edgell were using measures of civil society that come out of V-Dem that weren't available to civil society researchers before including a new set of measures from the NAVCO project that differentiate among the different kinds of organizations including peaceful and violent organizations and anti-system and pro-democracy movements and things like that.
So, it was striking that this measure of anti-system movements had a very clear impact in the direction that you would expect it to be. So, it is an important finding, not because it's counterintuitive, but because it had not been well studied or documented before.
So, as we kind of wrap up, you've mentioned that you studied under Robert Dahl and you studied under Juan Linz and those are both absolute giants in political science. I mean, they are people that we all continue to read and we all continue to learn from even though that they've passed away and their writings are older, but we continue to learn from those texts over and over again. At the same time, there's so much more information that we have today. V-Dem is a demonstration of how much farther that we've gone and have tools that they didn't have when they did their research. What do you feel is the most important finding or the most important thing that you've learned about democracy that maybe Robert Dahl or Juan Linz didn't realize because they didn't have access to the tools or maybe they just didn't have access to the scholarship that exists today?
Well, I think the thing that I have learned that goes beyond what I learned from Dahl and Linz and other people I studied with is just the complexity and the detail that we're getting now that we have V-Dem. Because I think this is what distinguishes V-Dem from other democracy measures more than anything is the number of indicators that we produce. Hundreds of indicators. So, it gives you a really detailed x-ray of what is going on in every country in every year. So, we don't have to stay confined to big generalizations about regimes much less categories of regimes or regime types. Linz wrote a lot about types of non-democratic regimes. We can really look at things in detail and track them as they change sometimes by small increments from year to year. It sort of brings things to life
It's as though we used to have to look at still snapshots of the political world and learned a little bit about them at scattered points in time. Now we have sort of a continuous movie of what's going on in democracy around the world every year. I think it's just fascinating to watch this develop, at least to watch it in retrospect, as what has happened in every country on many different levels on a year to year basis.
Well, Michael, thank you so much for joining me. I want to plug the book one more time for you. It's Why Democracies Develop and Decline. I thought it was a great book. For anybody who hasn't looked into the Varieties of Democracy project, I recommend checking out the website as well. Thank you so much for joining me today.
You're welcome. My pleasure.
Learn more about the Varieties of Democracy Project
Follow the V-Dem Institute on Twitter @vdeminstitute
Why Democracies Develop and Decline edited by Michael Coppedge, Amanda B. Edgell, Carl Henrik Knutsen and Staffan I. Lindberg
Democracy Paradox Podcast
Email the show at firstname.lastname@example.org
Follow on Twitter @DemParadox