Does economic hardship really explain populism?

Let's go through the only existing meta-analysis of causal estimates

There’s a widespread belief that economic hardship drives populism. The story has different layers and versions, but the overall idea is fairly straightforward, and has been popular since at least around 2015-2016 with the coming of Donald Trump (pun intended). Allegedly, globalization leaves many workers behind, factories close, communities crumble, and angry voters turn to populist leaders who promise to shake up the system. We liberals might not like populism, but our globalized neoliberal order is what actually brought it about.

A new meta-analysis of 36 causal studies (and 144 estimates) seems to confirm this narrative on a superficial reading. But you’re likely here for a nuanced take, so let’s discuss crucial details.

Naively, one third of the surge can be explained

First, the magnitude of the effect. The review reports a rough rule of thumb emerging from the analyzed studies, which is that “economic hardship” (operationalized differently across studies) explains about 1/3 of the recent populist surge. That’s definitely something. Last year, before I read the study, I’d been skeptical of economic explanations for populism and would have been somewhat surprised to find out the actual effect is of such magnitude. But because this meta-analysis is the best evidence I think we have right now, I’ve updated accordingly. Populism is partly an economic phenomenon. Fair enough.

However, I fear that when people hear (or say) that economic hardship “explains” populism, they assume it’s either the only cause or the overwhelming cause. Apparently (and not surprisingly), neither is true. If economic factors explain a third, that means two-thirds is explained by other factors, such as cultural anxieties, political supply-side dynamics, sheer ignorance and irrationality, media narratives, identity politics, and so on. So those people should update as well.

Austerity and subjective hardship, not import/labor-market shocks

Second, what exactly is economic hardship? I think many people assume we’re talking about the China shock or general import competition that ostensibly devastated manufacturing communities. Or perhaps they have widespread unemployment and objective labor-market uncertainty in mind. Yet the evidence for these specific causes is actually quite mixed, judging from the paper.

Both import exposure and labor market uncertainty show inconsistent effects across studies. A simple vote-counting exercise conducted by the authors tells us that, for import exposure, of the 47 causal estimates, 28 are positive and 19 show no (or negative) effect. With labor-market uncertainty, it’s 24 (positive) versus 18 (no link or negative). If these factors matter at all, their effects appear fairly small and/or non-robust, at least in light of how much has been made of them in the media and online.

So what does clearly matter? Two things stand out unequivocally to my mind: austerity and subjective economic hardship. Housing demand shocks also show consistent positive effects, though we have only two estimates for them, so I think we can leave this aside for now.

How do we interpret the former two? Austerity is straightforward enough. When governments raise taxes on ordinary people or slash spending on services they depend on, it disrupts lives and generates anger. I’m not surprised by that and even if you’re a libertarian who likes austerity, I think you shouldn’t be surprised either. Austerity is just bad, politically. So, okay, austerity clearly contributes to populism.

What about “subjective economic hardship”? Here, it’s much trickier. I don’t think this is a clean corroboration of the popular story. That’s because subjective economic hardship is not necessarily (or even likely) aligned with objective economic hardship (emphasized by the common story). People might be voting for populists not because of real, objective material deprivation, but because of their vibes that things are falling apart, even when they objectively aren’t. This distinction matters enormously for how we think about addressing populism and who we blame.

Actual magnitude is smaller

I want to expand on that, but let me first just emphasize another complication demonstrated by the paper: publication bias. Unsurprisingly, the meta-analysis finds substantial evidence that researchers are more likely to publish studies showing strong positive effects. The naive, averaged, raw estimate suggests a partial correlation coefficient of 0.058, but after correcting for publication bias, this shrinks to 0.016, which is a reduction of more than 70%.

No confidence intervals for the unadjusted estimate are reported, so I can’t show them.

Now, we shouldn’t directly translate this into claiming the one-third rule shrinks to merely one-tenth, because partial correlation coefficients don’t work quite that way. But the exercise reveals that we shouldn’t simply trust the effect sizes typically reported in individual studies on the topic. The true causal impact of economic factors on populism is likely considerably smaller than the literature suggests at first glance.

Voter ignorance and social media-driven panics

Back to the point about subjective hardship. I’d like to marshal two pieces of external evidence for interpreting this finding. First, it’s well known in the research that citizens routinely misperceive the national economy (even when they feel reasonably clear about their own finances). Polls have repeatedly found majorities who incorrectly believe the country is in recession even when GDP is growing and unemployment is low. Cross-national evidence shows similarly large misperceptions about core facts like inequality and unemployment. People systematically under- or over-estimate macro conditions, and those perceptions reflect media tone and partisanship.

Second, pair that with the classic finding from the economic-voting literature, which is that people tend to vote sociotropically (on their reading of the nation’s economy) more than egotropically, i.e., based on their own pocketbook. The foundational studies (and later reviews) show that assessments of national economic performance are much more predictive of vote choice than self-reported personal finances.

Put together, this implies that subjective economic hardship can be far removed from actual, objective economic hardship. If voters mostly use national economic perceptions to guide their choices, and those perceptions are often parti­sanship-colored and media-negativity-skewed, then measures of subjective hardship risk capturing vibes rather than real macro shocks or the alleged savagery of late-stage neoliberalism.

So, how should we think about the economic causes of populism? I think it’s become clear that they exist and are related both to objective and subjective hardship. However, (1) the magnitude is likely much smaller than you think, (2) the favorite measures of objective hardship (import/labor-market shocks) have only mixed support, and (3) apart from austerity (which really is bad), the economic hardship that causes populism might in important part consist of people’s (misguided, social media-driven) perceptions rather than reality.