Did inequality trigger the rise of populism?

Francis Fukuyama thinks so, but the evidence is mixed

In a 2019 Foreign Affairs piece, Francis Fukuyama threw his weight behind the standard economic story for the rise of populism over the past two decades and the onset of the democratic recession. Here’s how he put it:

I concur with the commonplace judgment that the rise of populism has been triggered by globalization and the consequent massive increase in inequality in many rich countries.

I really like the guy, especially his attitude toward liberalism and his distaste of illiberals and authoritarians. More importantly, I agree with several of his key ideas from The End of History and the Last Man and, of course, the two lesser known volumes on political order and development from the 2010s. But I think he’s wrong to agree with the common wisdom on the economic determinants of populism. The evidence we have is much more mixed than people, apparently including Fukuyama, tend to think.

(In all fairness, he qualifies his initial statement by adding that the causes surely can’t be purely economic and that we actually also have to look at culture and status, which I agree with. Moreover, he’s somewhat shifted to emphasizing technology and non-economic issues as key in recent years.)

I’ve already written about the link between economic globalization and populism, so I’ll skip that for now. Instead, I’d like to shift your attention to the alleged link between inequality and populism.

I understand how one would intuitively think there’s something to this hypothesis. You need just a few ingredients for a plausible-sounding explanation:

1. Voters have to roughly accurately perceive how high inequality in their country or region is and whether it’s rising.

2. Inequality then actually has to be relatively high or rising (on account of neoliberalism, or globalization, or whatever else).

3. Assume, also, that there is a near-universal human tendency to compare ourselves and to be unsatisfied when the distance between ourselves and the rest of society (or those at the top) is getting wider; that is, we’re more concerned with relative standing than with our absolute conditions.

4. Add in viral populist leaders and messages extoling the virtues of the common man and pouring vitriol on the rich, corrupt elites.

On first pass, all of that might sound reasonable, and so you start feeling you understand why populist political parties are electorally on the rise.

However, when you think through some of these points, issues start emerging. If I limit myself to just the two most obvious, it’s – first – actually not the case that inequality has been rising in most countries over the past two decades, at least not on standard measures of it. And second, most voters are notorious for not roughly correctly observing economic conditions, such as the unemployment rate or inequality.

So, it’s looking mixed right off the bat. Let’s turn to the empirical literature for possible salvation.

Country-level evidence on the inequality-populism link

There are quite a few recent papers that show no connection between inequality and populism at the country-level, either cross-sectionally or over time.

As one example, see Bergh and Kärnä (2022). In a panel sample of 26 European societies between 1980 and 2018, they find no evidence of inequality predicting either right-wing or left-wing populist vote shares. Similarly, in an expanded dataset of 49 European and Latin American countries, Stankov (2018) overall finds no evidence for the inequality hypothesis. This is especially so within Europe, while there is some weak, inconsistent evidence for it in the Latin American case. Then there’s another study by Bergh and Kärnä (2021), which likewise corroborates the absence of any relationship between income inequality and populism.

I’ll have a paper out soon myself, where I more or less follow the Bergh and Kärnä statistical setup, but focus on 31 European countries and the period between the mid-1990s and 2021. Cross-sectionally, looking at changes in the income Gini at the start and the end of this period, there’s no connection between inequality and populism. In some countries, like Sweden, Germany, Cyprus, Luxembourg, and Denmark, inequality rose quite a bit, but their populist vote shares increased just a tiny bit. In other European countries, where inequality stayed the same or even declined, populist vote shares have increased by the same amount as in the first set of countries (or even significantly more). There’s also nothing in my country-fixed effects panel regressions as far as inequality goes, regardless of how I slice the data. (Actually, I have not one but three papers of this kind, testing various sub-hypotheses on inequality and populism with various statistical methods. Whatever I look at, nothing is statistically significant.)

That’s not to say nobody finds anything. Engler and Weisstanner (2020) carried out an analysis on 14 OECD countries, where they get statistically significant results for the connection between the growth in the Gini coefficient and voting for the radical right over the medium and long term (nothing shows up over the short term).

Stoetzer et al. (2023) also find evidence for the inequality thesis. They limit themselves to 14 Western European countries and only the period between 2002 and 2017, but they show that a 1-point increase in Gini predicts a 1-percentage point increase in the support for populist parties. Given that in this period many countries didn’t see their Ginis rise, or they rose just by 1-2 points, this doesn’t really explain much. As they themselves report, Gini on average shifted from 28.8 in 2002 to 29.6 in 2016.

Region-level evidence

But perhaps looking at country-level associations is not granular enough or just wrong for other reasons. What if we turn to regional-level evidence?

To take a closer look myself, I loaded up the latest European Social Survey dataset (to get a populist voting proxy) and Eurostat’s regional inequality data. Eurostat provides data on the 80/20 ratio, or the distance between the 80th and 20th percentile of earners, not the Gini coefficient. That should do as a quick check (we’ll get back to Gini below).

Here’s a simple region-level scatter. No country-fixed effects here, so this is purely descriptive. I get the same perfectly null correlation if I use all the available regions (70) by dropping the requirement to have only regions with good respondent coverage. In this scatterplot below, where I’m using 44 European regions, there’s around 9,000 respondents.

And here is the coefficient plot of more informative models with country-fixed effects, different sets of control variables, and both broader and stricter operationalizations of populist voters. I find no statistically significant associations. It’s not the case that in European regions with higher inequality, there are more populist voters.

But these are just my quick-and-dirty analyses. What about the existing published and peer-reviewed research?

Rodríguez-Pose et al. (2023) looked at regions in both the US and Europe. They fail to find a statistically significant link between the Gini coefficient and support for populists. In the US, a 1-point increase in the Gini coefficient is associated with about a 0.25-point decrease in the Trump margin before adding controls. Once they add controls, the Gini turns non-significant. In Europe, they find more or less the same. Regions with higher current Gini levels are not the main far-right populist regions. In fact, there’s a negative relationship between the level of interpersonal inequality and far-right populist voting, and this remains after adding economic and demographic controls. As they say, “More unequal places have not rejected mainstream candidates.” They also toy around with change in Gini (instead of just levels), but mostly find non-significant results there. (Once they interact inequality with status/culture/race, however, things change quite a lot!)

Or take Georgiadou et al. (2018), who focus on hundreds of regions in 28 European countries in the 2000s and 2010s. They distinguish between the populist radical right (PRR) and the extremist right (ER). What they find is that regional income inequality is robustly associated with PRR voting, but not with ER voting; for ER it is zero or negative. Go figure.

So, can we say there’s no evidence that inequality and voting for populists go together? No. As usual, it’s just highly, highly mixed. So much so that I would absolutely refrain from positively making the Fukuyama case, at least if it’s cashed out in purely economic terms. He’s not necessarily wrong. But we don’t have nearly enough evidence to robustly (or even tentatively) claim he’s right.