I’d say that’s actually good news overall, because certain features of geography are hard or impossible to change, while institutions are at least in principle malleable. If the strong institutionalist theory is correct, countries that drew a bad lot are not destined to poverty. It’s not easy to change your own institutions, but it’s at least in principle possible. In any case, this isn’t just Ferguson’s own superficial gloss on the book. Acemoglu and Robinson are themselves very clear that their aim is to reject “the geography hypothesis,” as they call it.

Subscribe now

They start by admitting that poverty and prosperity around the world are geographically concentrated, which “gives a superficial appeal to the geography hypothesis.” They then go on to say that world inequality nevertheless “cannot be explained by climate or disease, or any version of the geography hypothesis.” This is so for many reasons.

Some reasons have to do with anecdotes or, to put it in nicer terms, individual cases like Nogales. There are actually two cities of Nogales: one in Arizona and the other in Sonora (Mexico). The two cities aren’t separated by geography or climate, but simply by the US-Mexico border. It’s the differences in institutions, mostly laws, that separate the wealthy part of Nogales from the poor one. That’s an evocative anecdote, and it’s not the only one they have, but it unfortunately tells us nothing about general patterns regarding the causes of development.

Other reasons they offer are somewhat more systematic, like their “reversal of fortune” point. Here it is from one of their earlier studies:

Among countries colonized by European powers during the past 500 years, those that were relatively rich in 1500 are now relatively poor. We document this reversal using data on urbanization patterns and population density, which, we argue, proxy for economic prosperity. This reversal weighs against a view that links economic development to geographic factors.

Instead, we argue that the reversal reflects changes in the institutions resulting from European colonialism. The European intervention appears to have created an “institutional reversal” among these societies, meaning that Europeans were more likely to introduce institutions encouraging investment in regions that were previously poor. This institutional reversal accounts for the reversal in relative incomes.

This is also less convincing than seems at first glance, as far as rebutting the geography hypothesis is concerned. But it at least has some quantitative power behind it and so is more systematic. Hold that thought.

Poverty and prosperity are geographically concentrated

Below is a world map I made with a composite score of geographic advantage for individual nation states. Higher values of the index combine ostensibly negative geographic and climatic conditions – namely, heat, malaria, remoteness, landlockedness, terrain ruggedness and low farm suitability. You can clearly see that the wealthiest regions of the world (Europe, the US, South Korea, Japan) have low geographic disadvantage. The poorest regions (Africa, parts of Central Asia) have the highest geographic disadvantage.

Wealth really is geographically concentrated, as Acemoglu and Robinson admit. Here’s a simple correlation between income per capita and geographic disadvantage. The more disadvantage a country has, the worse its level of development.

The question, of course, is what explains this relationship. It’s unlikely to only be geography, because there are many confounders we would like to control for. For instance, it’s very likely that economic and political institutions are related to both geography and development, so that’s one thing we’d like to account for, right?

Actually, not really!

Think about what confounders are. They are common causes or deeper variables/factors that mutually determine our independent variable (geography) and dependent variable (development) of interest. Institutions sure help determine development. But do they also determine geography? Depending on what we mean by geography, no. Take latitude or terrain ruggedness for example. Those typically can’t be caused by institutions; instead, if anything, they (the geographic variables) help shape institutions. In other words, institutions are not confounders here but mediators. And you don’t want to control away mediators from an analysis that aims to estimate the total effect of geography on development – some of that will be direct, unmediated, and some of it will occur through a thing like an institution. The direct and indirect effect together make up the total effect.

Of course, other aspects of the geography, like malaria or agricultural productivity can to an extent be affected by institutions and so controlling for institutions in this case can be more sensible. (Even here, there’s a difference between malaria prevalence and malaria ecology or agricultural productivity and low farm suitability; the latter of the two pairs is largely exogenous.)

What I’m getting at is recognized by Acemoglu and Robinson, but typically forgotten in wider discussions. Why Nations Fail doesn’t reject geographic determinants tout court. Instead, it (1) mostly disregards their direct effect and it (2) frames their indirect effect as something only belonging to the institutional account. I’ll get back to (1), but first let’s focus on (2). Why Nations Fail and Acemoglu et al.’s famous early studies constantly refer to geographic factors. In the Colonial Origins paper, good and bad institutions are set up in response to the ecological disease burden encountered by settlers. In the book, predetermined population density (plausibly a “geographic” and at any rate non-institutional variable) is a key factor in explaining why some places got saddled with extractive colonial institutions, while other places got good institutions.

Direct and indirect effects of geography

So, in one important sense, Acemoglu and Robinson are totally on board that geography matters both for institutions themselves and (through them) for development. They wouldn’t say geography is the only thing that matters, or even that it is just one thing that matters on its own, but they do see it as important in conjunction with other stuff (namely, institutions).

But I think there’s more to it than that. Geography very plausibly matters on its own as well – not as a totalizing cause or something that cannot be outweighed by good institutions, of course. But it does contribute to the equation, and it does so in its own right (not merely through institutions).

As an illustration of what this looks like, see the coefficient plot below. First up is geography’s relationship to development on its own, without controlling for anything (grey). As said, this isn’t as faulty an exercise as it typically is because in this case, geography is to a large extent exogenous to development. Now, there are yet deeper geological variables that might be common causes of both my geographic-disadvantage variables and development, so those need to be controlled for. The coefficient in the second row (black) shows you how the relationship shrinks after controlling for geological confounders, i.e., true predetermined controls.

So, with controls for confounders or without them, geographic disadvantage seems to matter for development. But how much of that is a direct effect versus an indirect effect that geography has through determining which kinds of human-devised institutions are likely to appear in different places? The smaller coefficients you see in blue, green, and purple give us some rough idea. The coefficient shrinks (but doesn’t get insignificant) when we control for plausible institutional mediators. That is, geographic disadvantage makes it less likely that state capacity will get built up or that a society will have liberal democracy. But because state capacity and liberal democracy themselves plausibly matter for development, geography in that sense indirectly influences development through these mediators.

Here’s another way of showing this. Below, you see a (descriptive) mediation graph that shows how much geography matters for development “on its own” (holding institutions constant). That’s the dark blue arrow that goes directly from geography to development. The coefficient is not large, but it is negative and significant. The other lines first show you how geographic disadvantage plausibly causes various institutional bundles. For instance, it seems geographic disadvantage hurts liberal democracy and state capacity (though it doesn’t do anything to economic freedom). Then, the outgoing lines tell you how institutions relate to development. In this illustrative descriptive case we have here, it seems state capacity is positively and significantly related to development (see the “b” coefficient below the green line), while the other two institutional bundles are not significant. In the colored boxes, there’s another set of coefficients that tell us what the indirect effect of geographic disadvantage on development is through the individual mediators. The only sizable and significant negative effect occurs through state capacity (-0.134*).

Don’t take this too seriously. It’s just an illustration of a general point (though based on real data). For one thing, this isn’t true causal mediation, because the paths from institutions to development are endogenous (i.e., institutions cause development, but development also causes institutions, and there are confounders in this reciprocal relationship as well). For another, I have just some randomly picked institutional variables as mediators here, not everything that plausibly matters. But the point is that geography probably matters for development, and it does so both directly and indirectly.

I’m not the only or the first person to say something like this with systematic data. Employing similar cross-sectional designs, the nowadays unfortunately disgraced economist Jeffrey Sachs showed already in the early 2000s that malaria is significantly related to development, net of institutional controls. In the same period, Dani Rodrik and colleagues, though favoring the institutional approach that is also dear to my heart, showed geography matters for development – its direct effect tends to be weak and its indirect effect (through institutions) is strong, but both are there.

There’s also a large recent review of a bunch of different kinds of accumulated evidence by Pascaline Dupas, which reaffirms the notion that “geographical factors have clear causal effects on development outcomes, both directly and indirectly”:

The literature now provides compelling causal evidence that a poor draw in the (rigged) geographic endowment lottery constrains economic output and wellbeing.

This is not to say that geography is destiny: geographic constraints can be overcome through appropriate investments and institutional arrangements, but doing so often requires sustained effort and substantial resources. In other words, a good geography is neither a necessary nor sufficient condition for economic prosperity, but it is more conducive to it.

I’ve written a few critical-yet-appreciative/defensive posts on institutionalist theory and Daron Acemoglu. This one is perhaps the most critical yet, as he and his colleagues simply wrongly deny that geography is important on its own, and to the extent that they accept its indirect power and work it into their own theory, they nominally reject it. Remember, as they say in the book, they reject “any version of the geography hypothesis.” But their own theory is a (soft) version of the geography thesis, is it not?

This is sometimes criticized for a variety of reasons. Some are very misguided, like saying that Fraser’s index can’t be trusted because the institute has a libertarian identity and is thus biased. The conclusion might apply to the Heritage Foundation’s construct, but doesn’t follow in the case of Fraser’s index, because it’s not made in-house but has all of its components sourced from third parties, like the World Bank, V-Dem, and so on. There are also no special, proprietary weightings applied to components or subcomponents when deriving the overall score. It’s all completely open and transparent.

Other critiques are more sensible. The index is made up of 5 components – namely, size of government, legal system/property rights, sound money, freedom of trade, and regulation. But are policing quality, the rule of law, impartial courts (all parts of the legal system component), and inflation (part of sound money) really strong indicators of neoliberalism (or even just “capitalism” in general)? Probably not. This, however, is not a fatal problem because one can just focus on the most plausibly neoliberal components of the index, and we’re good to go.

Who can deny, as a first approximation at least, that large and heavily redistributive governments, strongly tariffed and closed trading systems, and strict labor and business market regulation are not the quintessential ways of defining/measuring to what extent neoliberalism is present or absent in a country? So, throw out the legal system/property rights component and sound money if you want, and focus just on what’s undeniable.

“Carrying the neoliberal banner”

Though many revere him, I’m not too crazy about Quinn Slobodian or his research on neoliberalism. That’s again for at least a couple of reasons that need not detain us here too long. For one thing, I’m sorry to say I just don’t find intellectual history that interesting, because it says nothing about whether an idea is correct or not or how a social phenomenon actually works on the ground. I’m interested in what the empirical correlates and effects of neoliberalism are, not how people wrote about it in books 50 or 100 years ago. For another, though I couldn’t care less about von Mises, and I actively dislike how acerbic Phil Magness is on social media, Slobodian seems to have done some sloppy work.

But what I do appreciate about him is how he characterizes economic freedom indexes. Here’s what he said a few years ago in a Guardian piece titled Democracy doesn’t matter to the defenders of ‘economic freedom’:

If Hong Kong were a country, it would be in the world’s top 10 most unequal.1 Observers often use the word neoliberalism to describe the policies behind this inequality. The term can seem vague, but the ideas behind the economic freedom index help to bring it into focus.

… The ideal world described by these indexes is one where property rights and security of contract are the highest values, inflation is the chief enemy of liberty, capital flight is a human right and democratic elections may work actively against the maintenance of economic freedom.

…

[E]conomic freedom indexes carry the neoliberal banner by deeming the goals of social justice forever illegitimate and pushing states to regard themselves solely as guardians of economic power.

To Slobodian, it seems completely obvious and damning that economic freedom indexes like Fraser’s are soaked through and through with neoliberalism. That’s both because they were invented by people like Milton Friedman and because countries get dinged if they tax and redistribute too much. Moreover, economic freedom indexes don’t put actually democratic, nice countries at the top of the list. Instead, authoritarian places like Singapore, Hong Kong, and (in the distant past) the Guatemalan dictatorship can all rank high.

Slobodian is of course outraged by this. “True” economic freedom by definition means more democratic power to the people. “True” economic freedom means lots of redistribution and business regulation (not deregulation!), so that workers are protected and free from the clutches of Capital. That all of this is missed by the actual economic freedom indexes we have, Slobodian suggests, is a clear sign that what they’re concerned with is mere sham economic freedom – freedom for the capitalists.

Now, I’m happy to accept all of that. Neoliberalism is sham economic freedom. Worker power and democracy is the true, non-neoliberal economic freedom. I don’t mind the outrage, and we can always define words however we want, so I wouldn’t want to insist on what the only correct definition of economic freedom is. I only care about whether the existing indexes of “economic freedom” help us measure neoliberalism. It seems I have Slobodian’s approval in that department. He even chucks in inflation for good measure, which as he sarcastically characterizes it, neoliberals see as the “chief enemy of liberty.” (I was willing to give up on the sound money component of Fraser’s economic freedom index as a bad proxy for a neoliberal economy, but he has made me rethink my position.)

The more neoliberal we are, the more equality there is…

Okay, armed with the permission (from one of the greatest researchers of neoliberalism today) to use the index to measure this phenomenon, let’s look at what it tells us about inequality. Focusing on Hong Kong, Slobodian suspects that a strong positive relationship between the economic freedom index and inequality (meaning that the one increases the other) is an undeniable pattern. Where there’s more neoliberalism, there’s not more democracy or quality of life, but there will be exploding inequality.

At first glance, that’s not only incorrect but actually the opposite of reality. Where there’s more of Friedmanite sham economic freedom – remember, “economic freedom indexes carry the neoliberal banner” – there’s significantly less inequality. See below. This is not only true of the standard Gini measure of income inequality, but it also shows up with income inequality measured through the top 1% income share, as well as wealth inequality!

An obvious catch with this is that countries don’t differ from each other only on inequality and economic freedom. There are myriad other factors that might be confounding the relationship. One candidate could be economic development. At high levels of development, inequality usually drops (for many reasons). But highly developed countries are also likely to be more neoliberal (or are they?!). If that’s the case, we might want to control for development or investigate the relationship separately in developed and developing societies.

You’ve heard of the Simpson’s paradox, right? Sometimes, you have a strong relationship in a larger group of societies (or individuals), which completely disappears or even reverses when you break that larger group of observations down into smaller groups.

… Or not really

When we do so by developed/developing status, the picture changes dramatically. In the developing world, nothing happens to inequality as we go from least neoliberal to most neoliberal societies. In the developed world, there are inklings of a positive relationship, though it’s uncertain with only ~35 societies. All developed societies are neoliberal-leaning, but when you lean stronger toward neoliberalism, you also have higher income and wealth inequality.

Up to this point, we were looking at the aggregate economic freedom index. What if we disaggregated it and focused solely on the most neoliberalism-relevant aspects? This is what we get. (The results are virtually identical for wealth inequality.)

That’s some interesting heterogeneity we got here.

(1) Countries with leaner governments in the developed world are more unequal. The same isn’t really true for the developing world.

(2) Virtually all developed societies have very liberalized trade, so there’s not much variation to exploit there, as you can see. However, interestingly, both in the developed and developing world, as trade gets more liberalized, inequality seems not to budge.

(3) Deregulation in the developed world goes hand in hand with more inequality, while in the developing world, having more deregulation means being somewhat lower on the inequality spectrum.

As you can also see, we basically have big inchoate blobs of observations, so it’s not like these are some kinds of tight and very informative relationships. Perhaps the biggest takeaway so far is that, in simple cross-sections, neoliberalism and inequality don’t seem to have a very robust and clear relationship either way.

Meta-analytic results are mostly null too

So much for simple, descriptive relationships. What does the more systematic research literature tell us? More or less the same thing.

Bennett surveyed 16 peer-reviewed studies in 2024, and found an overall mixed result. 4 studies reported more economic freedom to be associated with less inequality, 5 with more inequality, 2 found a nonlinear/U-shaped relationship, and 5 found no significant relationship.

Lawson and colleagues did a larger quantitative review (26 papers) and arrived at a similar conclusion in 2024, though noted some limited evidence that moves over time towards more economic freedom may correlate with inequality. Here’s how they conclude (note the weird result with regards to lean government):

The best estimate is that overall economic freedom is simply unrelated to inequality, though liberalization (i.e., moving toward more economic freedom) may engender some additional inequality. Though the estimates were mostly insignificant, it appears that (counterintuitively) smaller government size and stronger property rights might be associated with more equality.

And then there’s the narrower 2026 review again by Lawson and colleagues, in which they intentionally restricted their inequality measure solely to the Gini coefficient of income inequality to make the evidence more comparable across studies. Here are the headline results:

The literature suggests that stronger property rights and rule of law are associated with greater equality, while sound money, freer trade, and lighter regulation show weak correlations with inequality. Government size appears unrelated. Overall, greater economic freedom tends to correspond with modestly higher inequality: a 1.25-point increase in EFW is associated with about a 1.2-point increase in the Gini coefficient.

Put slightly differently, when looking at the aggregate economic freedom index, 58% of all estimates are positive and statistically significant. More economic freedom, more income inequality, at least judging by the (slight) majority of findings. As they said in the cited passage, the actual strength of the relationship (“a 1.25-point increase in EFW is associated with about a 1.2-point increase in the Gini coefficient”) is very modest: note that economic freedom is on a scale from 0 to 10, while Gini goes from 0 to 100.

Interestingly, leaner governments don’t mean more inequality (only 35% of estimates are positive and significant, while a near-majority are negative and significant). If anything, leaner governments tend to go together with lower inequality. Trade liberalization and deregulation lean positive (i.e., boosting inequality), though only a minority of estimates are actually significant (23% and 37%, respectively). And relationship size is miniscule (standardized effects of 0.07 and 0.01, respectively). Anything smaller than 0.1 is typically said to be negligible.

It wouldn’t surprise me in the slightest if neoliberalism was not just strongly related to inequality but actively causing it. From the (admittedly not impressive) observational evidence we have, I’m not at all convinced this is so. Very weird, but the evidence is what it is. Those critical of neoliberalism on the grounds of inequality should take note.

Shouldn’t there be pervasive collective-action problems and monitoring difficulties in such cases, so much so that generalized trust never gets off the ground, which in turn gives people no good conscious or unconscious reason to cooperate with each other? Cheating and free-riding should be pervasive under such conditions and cooperation impossible.

In the remit of cultural evolutionary theory, one famous candidate explanation has to do with the idea of Big Gods and subsequent group selection. The argument is fairly simple.

1. A random social group invented the belief in moralizing Big Gods (i.e., gods “who watch, intervene, and demand hard-to-fake loyalty displays”).

2. As a consequence, its members became more pro-social and trustworthy because of the belief. That is, even if a person finds themself in a situation with no terrestrial monitoring, rewards, and punishment, where cheating and exploiting is tempting, they will fear supernatural retribution and seek supernatural approval.

3. This same group can now become larger, in terms of population size, without succumbing to standard collective action problems, which should emerge after some scaling threshold is passed. Relatedly, this group will be able to sustain larger cooperative networks and reap economies-of-scale benefits.

4. If there is competition with other groups, the Big Gods group will, for those reasons, more likely prevail in conflict or be able to absorb the other groups.

5. Over time, through group selection, only (or mostly) Big-Gods societies remain and prevail.

6. Big Gods birthed social complexity; they enabled the transition from small communities to large-scale societies, states, and civilizations.

This, of course, is just a story. It might be true or it might not be. We need empirical evidence to tell whether the individual components laid out above are true. Moreover, we need empirical evidence to check whether the overarching claim (Big Gods —> complexity) holds true. There are also alternative stories we can tell, and so on.

The two key empirical pillars on which the Big Gods story classically rests are: (a) experimental priming studies and (b) correlational cross-cultural research. This is how Ara Norenzayan, the most salient proponent of the Big Gods hypothesis, described the first pillar about 10 years ago:

Let us discuss the religious priming evidence that allows us to make causal inferences. Since the book was written, my colleagues and I have completed a comprehensive meta-analysis of all known religious priming experiments with a wide range of outcome measures (Shariff, Willard, Andersen, & Norenzayan, 2014).

We found an overall religious priming effect across a broad range of relevant dependent measures, summarizing 66 studies (Hedges’ g = 0.36). The religious priming effect emerged also for prosocial behaviors only, summarizing 25 studies (Hedges’ g = 0.27). The effect in both analyses gets stronger for self-reported believers and is on average not statistically different from zero for nonbelievers.

The effect is reduced but stays robust and statistically significant when we correct for the estimated publication bias in psychology (that non-significant results are less likely to be published).

The idea is that if religious priming nudges people toward prosocial behavior, it is plausible that the emergence of similar religious ideas millennia ago had a similar effect on our ancestors.

The second pillar has to do with cross-cultural correlations. Typically, the idea here is to use older ethnographic samples such as the Ethnographic Atlas or the Standard Cross-Cultural Sample (or SCCS) by Murdock and White and regress presence/absence of Big Gods on various social traits. Obvious traits include measure of social and cultural complexity and size. If you find a positive correlation between the presence of Big Gods in a cultural unit and the same unit’s social complexity/size, net of confounders, that’s clear evidence that the two go together. Moreover, if small-scale societies really turn out to typically not have Big Gods, while large-scale societies do have them, you might then be tempted to suggest that one caused the other, just as the story says. (Though you can’t really do that due to endogeneity - i.e., reverse causality and unobserved confounding.) Historically, this type of evidence was very influential in anthropology and sociology.

The evidence crumbles

Unfortunately for the Big Gods theory, both types of evidence have considerably weakened over time, even if we set endogeneity issues aside for the moment.

Norenzayan was able to confidently point to his meta-analysis of all known religious priming experiments back in 2014 as supporting the theory. But newer research goes against the thesis. For instance, van Elk et al. (2015) reanalyzed the same literature using different publication-bias corrections. They show the religious-priming effect is essentially zero after the correction. At the same time, they also tried their hand at a Bayesian bias correction method, which didn’t erase the effect, so they simply concluded:

These contradictory statistical results demonstrate that meta-analytic techniques alone may not be sufficiently robust to firmly establish the presence or absence of an effect. We argue that a conclusive resolution of the debate about the effect of religious priming on prosocial behavior – and about theoretically disputed effects more generally – requires a large-scale, preregistered replication project, which we consider to be the sole remedy for the adverse effects of experimenter bias and publication bias.

This is just what Gomes and McCullough (2015) set out to do. They collected a large sample (n = 650) and preregistered their study. The result was a null: they didn’t find any statistically significant effect for religious priming.

In 2018, additional preregistered replications followed and found no support/small effect, depending on the conditions, and no support for moralizing Gods in particular:

In both experiments, we sought to activate religious cognition implicitly and explicitly, and we investigated the possibility that religious priming depends on the extent to which subjects view God as a punishing, authoritarian figure.

Results indicated that in both experiments, religious subjects transferred more money on average than did non-religious subjects. Bayesian analyses supported the null hypothesis that implicit religious priming did not increase Dictator Game transfers in either experiment, even among religious subjects. Collectively, the two experiments furnished support for a small but reliable effect of explicit priming, though among religious subjects only.

Neither experiment supported the hypothesis that the effect of religious priming depends on viewing God as a punishing figure. Finally, in a meta-analysis of relevant studies, we found that the overall effect of implicit religious priming on Dictator Game transfers was small and did not statistically differ from zero.

A very recent 2024 meta-analysis of almost 60 years of research on the topic finds a vanishing correlation between religiosity and directly measured prosocial behavior (r = 0.06). The correlation turns from vanishing to small (r = 0.15) if prosociality is measured with self-reports, which you’d of course expect but should also discount (in favor of the direct measure).

Galton’s problem, the omnipresence of supernatural punishment, and reverse causality

What about the more macro cross-cultural evidence? There are a couple things worth distinguishing here, but virtually all of it goes against the Big Gods theory.

One thing is that older research usually didn’t bother with Galton’s problem, which prevents drawing valid statistical inference from cross-cultural data when the units studied aren’t statistically independent. This is basically spatial autocorrelation applied to culture. Standard statistical tests assume that every datapoint represents an independent observation, which is of course not the case with societies which are geographically clustered together or share the same cultural ancestry. There are ways of getting around the problem, but in the past, the problem was almost always ignored. Hence, the older studies are bad.

Here’s what you get from these faulty analyses:

A study that explicitly addresses this problem, as well as reverse causality (I say more about this below), is by Watts and colleagues (2015). In a sample of Austronesian cultures, they use the phylogenetic comparative method to guard against cultural autocorrelation, and they show that moralizing high gods (Big Gods) in fact do show some correlation with social complexity. Importantly, however, the same research finds that Big Gods don’t actually precede the evolution of political complexity. What does precede it, at least in their sample, is just ordinary, run-of-the-mill “broad supernatural punishment.”

One might think that’s okay, because even if the specific Big Gods thesis is unsupported, perhaps all we need is just broad supernatural punishment to scale up societies and cooperation. In this sense, religion and group selection are still crucial, aren’t they?

No, that doesn’t work. Here’s how Watts et al. qualify their result:

ancestral spirits are suboptimal supernatural punishers as they generally have restricted abilities to know about and influence the world, are only important to their own lineage and are biased to favour this lineage. Ancestral spirits may be suited to facilitating cooperation within small scale lineage based groups, but fail to maintain cooperation in larger more complex societies in which there are multiple lineages.

So, Big Gods seem causally weak, because they come after complexity, not before it. Broad supernatural punishment seems more apposite, because it does come before complexity, and so might facilitate it. But even if that’s the case, it isn’t able to sustain it, so in that sense it’s irrelevant.

That’s not all. We can say more. The most recent studies we have from the 2020s (see also this) argue that even such broad supernatural punishment is not really associated with social complexity in larger samples! Setting the issue of Big Gods aside once more, this more ordinary presence of belief in moralizing and punishing entities seems to be everywhere, socially speakin. See the figure below (it shows the traditional correlation for Big Gods and the absence of one for moralizing gods in general).

Broadly moralizing and punitive gods are present at all levels of complexity, not just higher up. So they definitely can’t be the differentia specifica for human civilization.

Back to Big Gods. Reverse causality really is a strong possibility and a plausible explanation for the correlation between Big Gods and social complexity. It seems that societies first grow more complex over time for unrelated reasons (like warfare and resource increase). Only later do they invent or adopt Big Gods. See the two panels in the figure below from a study by Whitehouse et al. (2023), which tried to do as much as possible to tease apart causality.

If there’s any positive connection between Big Gods and social complexity, it’s there because the latter precedes the former, so Big Gods aren’t able to cause complexity. Moreover, after Big Gods finally appear, social complexity doesn’t subsequently show a further increase! That’s pretty damning for the thesis.

Alternative explanations

But if not Big Gods, what then? How is large-scale human cooperation possible? I have no idea, though the traditional explanations offered by Lenfesty and Schloss, including the indirect reciprocity explanation I mentioned back in January in another post on cooperation, seem very convincing to me. As they say, there are indications that (1) nonreligious costly signaling can help with trust, (2) status and reputation, or indirect reciprocity, expands cooperation beyond kin and friends, (3) and ordinary state-based policing in hierarchical societies can underpin large-scale collective action. Here are all three explanations in more detail:

1. Theoretical proposals for dyadic exchanges suggest and empirical studies demonstrate that nonreligious, hard-to-fake nonverbal cues can reliably signal commitment to strangers.

2. Trading networks may be facilitated by 2° or higher-order indirect reciprocity along with the social norms and cognitive/affective systems for managing moral reputation. Interestingly, the motivational salience of feeling “guilt” is mentioned just once in Big Gods; “shame” and “conscience” not at all.

3. Many large-scale collective endeavors – Norenzayan cites Hume’s account of a 1000-person project – require a sophisticated social infrastructure for coordination. Once that exists, it may also serve a policing function. It is not clear that Big Gods can facilitate such collective action in the absence of social hierarchy; and in the presence of organizing social structures, it is not entirely clear that Big Gods are necessary.

Apart from that, I’m of course open in principle to other cultural-evolutionary arguments.

To conclude, let me return to Norenzayan. In the 2014 book symposium on Big Gods, he sensibly proposed the following:

it is important to distinguish between two versions of [the] argument. The weak version is that Big Gods are not only a cause, but also a consequence of large-scale cooperation – the effects are bidirectional, an autocatalytic process of mutual galvanization. The strong version is that Big Gods are a byproduct of societal complexity but they have no causal power in the scaling-up of human cooperation.

It seemed to him that the strong version must be false: Big Gods can’t have been a mere byproduct of social complexity with no causal power of their own. And it seemed obvious to him that the weak version must be true: that is, yes, sure, Big Gods themselves get a boost from social complexity, but they were also an important cause of it.

The evidence that accumulated since 2014 and was reviewed here suggests to me the reverse is true. Namely, the weak version that Norenzayan prefers isn’t supported, while the strong (critical) version seems ever more likely. It could very well be that Big Gods didn’t play any role in the scaling-up of human cooperation.

Ward bills himself a Marxist of sorts or is at least interested in “Marxism,” judging from his Substack tag. That’s cool, because my past self shares that with him (see also this by me). He’s written a critical essay in which he tries to demonstrate a series of claims that undermine the optimistic story about capitalism and poverty told by Steven Pinker and others like him. That’s cool too, because I’ve also been critical of Pinker on both poverty and war.

He says a lot in the piece, and I won’t be able to cover everything, lest my essay become too long. So, for this installment (perhaps there’ll be another one), I’ll just focus on what are to my mind the most important and most biasedly presented claims, those that need the most correcting and additional context.

Here they are:

1. The World Bank and other organizations are (perhaps nefariously) changing the poverty line, so that the overall poverty trend looks better. This is, as Ward puts it, “deeply disingenuous and amoral if not immoral.”

2. Even if they made no changes to the line, the extreme- or international-poverty threshold is simply not enough – it’s just about mere survival, he says, which shows its “utter unseriousness.”

3. Relatedly, the international poverty line according to him “severely underestimates poverty and associated deprivation.”

4. Capitalism actually began earlier than in the 1800s, the century after which we have poverty data available. It began already in the 15th and 16th centuries, and in the following centuries (up to the 19th), “rates of poverty (lack of access to basic needs for subsistence) sharply increased.” Capitalism goes up, colonialism goes up, poverty goes up as well!

5. It wasn’t so much capitalism but rather anti-capitalist social justice movements that reduced poverty. In his words, global poverty decreases “since the Industrial Revolution correspond[] with the rise of organized labor movements and progressive and anti-colonial efforts against capitalism.”

The nefarious World Bank

It’s of course true that the World Bank raises (“shifts”) the extreme poverty threshold over time in nominal terms. It was $1 a day under the 1985 purchasing-power parity (PPP) anchor, then it nudged up to $1.08 in 1993 PPP, and so on over the years. It stood at $1.9 in 2011 PPP, and it’s now $3 (in 2021 PPP).

Why would they do that? Ward is clear that, for him, it’s just “statistical jiggery pokery,” an arbitrary “click of a button.” In reality, you have to do this because you want to take account of changing prices through time, so that your dollar/poverty numbers capture the real and changing cost of living in different countries. For instance, the jump from $1.25 (2005 PPP) to $1.90 (2011 PPP) happened when the World Bank took the same national poverty lines of 15 poor countries, updated them into 2011 local prices, converted them using the new PPPs, and obtained an average of $1.88, rounded to $1.90.

They did so because they were following these three rules:

1. Use the most accurate and recent set of prices available to compare the real standards of living across countries.

2. Minimize changes to the goalposts: keep the definition of the line unchanged, and its new value as close as possible to the $1.25 line in real terms.

3. When defining “real terms”, the price levels that matter most for measuring global poverty are those faced by the world’s poorest people.

Nefarious indeed. If you’re an epistemically curious person and just want to find things out, it’s easy to do so. Oh, and by the way, when the World Bank did this PPP updating, global poverty remained “basically unchanged.” So much for them shifting the threshold with the intention of obtaining a better picture.

Now, sometimes it’s not just about the PPP updating. Some changes to the extreme poverty threshold have to do with the country-sample that’s used for benchmarking. When we got the jump from $1.90 to $2.15, this happened because the World Bank used the national poverty lines of a new sample of 28 low-income countries for its calculation. In part, this decision was made because the previous sample of 15 countries became outdated (some of the countries stopped being poor, and the poverty lines were just no longer comparable). Again, you can just read about these things from the horse’s mouth. It’s all out there, publicly available and transparent.

The World Bank has perfectly legitimate reasons for revising the numbers. It’s simply a conspiracy theory that the revisions are made so that capitalism looks good. I’ve seen no evidence presented for that conspiracy theory. However, if Ward’s criticism was simply aimed at the fact that PPP updating and sampling can be methodologically fraught (whether we’re talking about poverty or GDP per capita or anything else for that matter), I’d wholeheartedly accept it in principle. This is hard and uncertain work.

(To foreshadow what’s coming, Ward states that in 2010, the Mexican government reported a whopping 46% poverty rate, while the World Bank estimated it to be around 5%. A serious underestimate, he charges. What he doesn’t say is that the two measures are completely incommensurable. The World Bank figure refers to the share living in extreme poverty, while the Mexican government figure refers to a much broader “multidimensional” poverty measure, which deliberately has nothing to do with extreme poverty. So we could just as well report that in 2018, in Guinea, 12% of the population was in extreme poverty, yet almost 70% experienced multidimensional poverty. And you can do the reverse too: in South Africa, there was only 6% of the population in multidimensional poverty in 2016, while almost a third of the population was extremely poor in 2014. It’s of course good to know both measures for every country – they’re measuring different sorts of deprivation. But this has nothing to do with the World Bank being nefarious or it underestimating anything.)

Who cares about extreme poverty? It’s only about survival…

When lay people come to me with their misunderstandings of how PPP works and why the nominal poverty threshold changes, I can somewhat understand them. I’m less understanding of the next critique: being lifted out of extreme poverty isn’t that important or salutary, because even when you’re no longer extremely poor, you still aren’t living a decent life.

Like, it’s in the name of the thing: extreme poverty. Yes, people who are no longer extremely poor are still … very poor. Shocker. They’re not living it up. It’s of course better that they’re no longer extremely poor, because now they’re at least no longer on the verge of starvation. But it’s still not good enough. Absolute poverty of all kinds should be reduced as much as possible. But for that to happen, you must first abolish extreme poverty. After that’s done, we shouldn’t stop. That’s just the first (and very important) step. Other absolute thresholds of poverty must fall to zero subsequently. And even then we shouldn’t stop. Everyone’s income must rise, so that comfortable living standards are achieved. The mere absence of extreme (or any other absolute) poverty isn’t enough! Hence the utmost importance of economic growth, but that’s for another day.

Ward asks:

If those who live just underneath or above the IPL do not have a realistic chance of having a normal human life, what good does lifting anyone above this line do in real terms?

What good does it do in real terms? Apart from reducing chronic hunger, resulting in fewer skipped meals, and allowing for more vegetables and meat, and so on, moving from $1 a day to ~$3 a day for a rural household literally means almost a doubling (descriptively speaking) of high school attendance share (from around 38% to 65%). So, yeah, nothing really, I guess?

But that’s not all. The other thing about extreme poverty is that you can just express it (or measure it) more directly in terms of basic needs instead of abstract dollars. There are at least two upshots of doing so: it becomes more transparent to the lay person what we’re measuring and how important it is for a person to be above (not below) the threshold, and we no longer have a problem of relying on a singular poverty line for all countries. As Michail Moatsos, whom Ward cites a lot, explains:

The most important difference between the cost of basic needs and the dollar-a-day methodologies is that while the later [sic] uses one poverty line value for years and countries, the former allows us to estimate the poverty line separately for each country and year. At the same time the contents of the basket are explicitly defined, while in the dollar-a-day approach those are only implicit in nature due to the statistical nature of its definition.

In the figure below, you can see what’s been happening to extreme poverty over the past 40 years according to a variety of measures.

There are two key “cost of basic needs” measures. First, the black BBB line shows the Bare Bones Basket aiming at subsistence level conditions. It includes the following as a basket of goods (locally measured and focused on the absolute cheapest sources):

In terms of food the BBB matches the calories required by the MDER (minimum dietary energy requirement) and provides for at least 40 g of proteins per day.

In addition, the BBB basket, following the tradition by Allen of the real wages literature, includes (per year):

3 kg of meat or 6 kg of fish (depending which one is cheaper),

3 kg of fat,

2 kg of sugar.

It also includes

1.3 L of lamp oil,

1.3 kg of soap,

1.3 kg of candles (all per year) and

a 8% allowance is allocated to clothing expenses.

Housing costs are assumed to be 5% of the BBB cost.

Finally, the costs in fuel are included (a) heating fuel (based on the heating needs of a room shared by the household and the given meteorological conditions), and (b) cooking fuel (depending on the kcal to be consumed cooked).

Truly, absolute bare bones. As you can see on the graph, over the past 40 years, this measure has remained virtually flat and has been hovering at about 5%.

Moving beyond just bare-bones survival, second, the dotted BCS line shows what happens to extreme poverty when we stop focusing only on the absolute cheapest sources (so that “some minimal variety is introduced in the dietary consumption”) and when we add:

at least 40 kg of beans, peas, or legumes, and 12 kg of meat or 24 kg of fish (all per year), depending which one is cheaper. The costs for clothing, lighting, personal hygiene, heating, cooking and housing are now based on the consumption shares provided by the World Bank Consumption database.

Measured in this very clear and substantive way, extreme poverty has more than halved from just over 70% in the mid-1980s to almost 30% in the mid-2010s. Why is this important? Because, as you’ve seen, people above the extreme poverty threshold can at least heat themselves and eat some decent, humane amount of nutritious food that’s required for normal biological functioning. Sounds kinda important.

Finally, the grey and red lines represent the more standard dollar-a-day measures ($1.25 in 2005 PPP and $1.90 in 2011 PPP). As you can see, the levels are different, but the trends are virtually exactly the same – huge reduction over time – if we compare them to the dotted BCS (“cost of basic needs high”) headcount. Even in terms of levels, the intermediate measure of cost of basic needs (dashed line) is on par with the traditional World Bank measure.

Note, by the way, that with the exception of the red line, all other series show what’s been happening to the developing world, not the world as a whole. In that sense, as Moatsos warns, you should not focus too much on the levels of the red line (just its trend). It’s somewhat low in the present context solely because it also includes the developed world – it is, after all, meant to measure the whole world, including developing and developed countries.

But okay, we don’t need to concern ourselves with extreme poverty, however measured (i.e., in terms of dollars or cost of basic needs). We can just look at various other, lower and higher, thresholds of absolute poverty to see if decreases in extreme poverty are misleading. Do decreases in this allegedly unimportant (because low) threshold mask stagnation (or perhaps even an increase) in higher-threshold and apparently more meaningful poverty lines? See figure below – it’s clear that for decades now, all absolute poverty lines have been falling, not just the “unserious” one. Note that $5 a day is slightly higher (so more stringent) than the World Bank’s poverty line for lower-middle-income countries, while $10 a day is higher (again, more stringent) than the poverty line for upper-middle-income countries.

Capitalism began in the 16th century in England, and it only spread to Europe in the 19th century

In the last graph, you can see the extreme poverty line being extended back into the 19th century. If you don’t like the dollar-a-day measure, you can look at the more rigorous and direct cost of basic needs measure of extreme poverty for the 19th century as well. See the black line below (“Moatsos, 2021”).

People have used the decline you see to argue that extreme poverty was pervasive before the Industrial Revolution around 1800, but then started falling with the emergence and spread of capitalism throughout the world, especially during the 20th century and after 1950. I say “people,” because funnily enough, they can be, depending on the context, either anti-capitalist Marxists or pro-capitalists and anti-Marxists.

Already in the 1980s, Baumol called it as it is:

Both vulgar Marxists and vulgar opponents of Marx have propounded two associated myths: that he believed wages under capitalism are inevitably driven near some physical subsistence level …

Marx himself argued that before capitalism and modern growth, all you had was poverty. Moreover, at least in his mature works, he was quite sure that capitalism will rapidly do away with material scarcity and will thus eradicate extreme poverty. This was my view of Marx back when I was a well-read Marxist, and it was also my own Marxist view. I grew up on Marxists like Andrew Kliman, Ellen Wood, Robert Brenner, Erik Wright, and others, all of whom were quite adamant that Marx didn’t view capitalism as materially immiserating the majority of people (that’s a silly caricature of him, they’d say), and they all outlined several theoretical reasons and mechanisms that explain why capitalism can do away with absolute poverty.

So, if you show these graphs of declining extreme poverty to contemporary Marxists, many of them will scoff and say, “Of course, that’s what Marx himself expected. What, did you think I’d be surprised and shocked because capitalism reduced poverty? It was never our (or Marx’s) critique of capitalism that the system would generate extreme poverty for the typical person. You’re thinking of feudalism and tributary societies! We’re well aware that capitalism generates immense growth, which can benefit the ordinary person.” The problem with capitalism is exploitation and inequality, we’d say, not absolute poverty.

At the same time, there are many (vulgar) Marxists around today who wouldn’t say that. Instead, they would be surprised if capitalism were credited with reducing poverty. So, they’d want to deny it, perhaps by stating that these poverty measures are wrong, or tampered-with, or in any case unimportant for what really matters to human beings. I’ve covered those objections already, so let me now say a few words on what we actually know about the relation between capitalism and extreme poverty.

Citing Sullivan and Hickel’s work on the topic, Ward self-assuredly proclaims that, in fact, “rates of poverty (lack of access to basic needs for subsistence) sharply increased with the rise of global capitalism and colonialism.” Instead, what has reduced poverty or is at least consistent with its reduction (temporally), is “the rise of organized labor movements and progressive and anti-colonial efforts against capitalism.” What can be said about this?

I’m well-acquainted with Sullivan and Hickel’s World Development paper. In fact, I’ve written and published (in the International Review of Sociology) research of my own, directly contradicting key parts of their story (while also corroborating some of what they say against Steven Pinker). My paper is 35 pages long with another 20 page appendix, so I won’t bother you too much with it. I also have an accessible writeup of it published here.

Here are the most important things I find:

First, the standard narrative is wrong to suggest that extreme poverty was nearly universal before capitalism. In many pre-capitalist societies, particularly in Europe, most people likely lived above the extreme poverty threshold most of the time [though just barely above it!]. The claim that 90% of humanity lived in extreme poverty before 1820 simply isn’t supported by the non-GDP evidence we have. I mean, it might be true, but we simply don’t know. And the more direct data on laborer wages we do have seem to go against the idea.

At the same time, second, pre-capitalist life was pretty shitty: absolute poverty for the majority, bodily stunting, high likelihood of dying as a kid, high likelihood of dying early as an adult, relatively frequent famines, high rates of violence, and all the rest of it.

Third, the transition to capitalism, at least as measured as the start of “endless growth,” i.e., by the onset of sustained GDP per capita growth, was generally associated with improving living standards, not deteriorating ones. Famines virtually disappeared, heights eventually (though veeery slowly and unevenly) began to increase, and extreme poverty rates declined. The regions that transitioned to capitalism earliest (England, possibly the Low Countries) unsurprisingly saw these improvements earliest.

You might ask: why define capitalism with reference to “endless growth?” In part, because this is how Sullivan and Hickel, using Wallerstein, define it. As they say, “According to world-systems theorists, capitalism is a system predicated on the ‘constant accumulation of capital,’ or endless economic growth …” But I also have other reasons for focusing in on that key aspect of capitalism.

As I review in my paper:

What distinguishes a modern capitalist economy from a premodern, precapitalist economy is the breaking out of Malthusian cycles of extensive growth, stagnation, and decline that have characterized human societies for millennia since the Neolithic revolution, and the emergence of systematic economic dynamism and intensive growth exhibited in the famous ‘hockey stick’ graph, which depicts a sudden, almost completely vertical increase in the GDP per capita curve after millennia of virtual stagnation.

General agreement about this exists among many contemporary neoclassical economists (see textbooks such as Cowen & Tabarrok, 2018; Bowles et al., 2017), institutional economists (North et al., 2009; Acemoglu & Robinson, 2012), notable Marxist historians and historical sociologists (Wood, 2017; Brenner & Isett, 2002; Teschke, 2003; Brenner, 2007; Žmolek, 2014; Dimmock, 2015; Isett & Miller, 2017; Lafrance & Post, 2019; Clegg, 2020; Wickham, 2021), libertarian philosophers and economists (Brennan and van der Vossen, 2018; McCloskey, 2017), and economic historians (Allen, 2011).

The economic institutions or structures which this diverse literature points out as principally responsible for generating the modern, intensive type of long-run, self-sustaining economic growth are foremost market competition between economic actors (not just markets in themselves), widespread market dependence of these economic actors (instead of the majority of producers being shielded both from market competition and market dependence through the institution of small-scale subsistence farming), and secure rights of private property.

These structural characteristics induce or incentivize economic actors to systematically invest in, and uncover, technological innovations; they also systematically pressure them to cut production costs so as to be able to outcompete rivals. This, in turn, results in systematic improvements in labor productivity, which unleashes ‘endless growth’ or sustained increases in GDP per capita decade-on-decade, century-on-century.

In this key sense, capitalism most definitely hadn’t existed during the 15th or 16th centuries, except in England and perhaps the Netherlands. Remember that Germany still had serfdom in 1806! For Europe, the rise of capitalism is a decidedly 19th century phenomenon. So, to the extent that we believe the historical reconstructions of extreme poverty during the 19th and 20th century (which, again, I’m iffy about), the documented decline very well coincides with the emergence of capitalism. That’s no causality, but the correlation is there. So you can’t just point to the rise of organized labor in the late-19th century and take that as the preferred explanation for the decline in poverty. There’s much more work you need to do to rebut the “capitalism did it” theory, and you also need to do much more to rule in your “anti-capitalist struggle did it” theory.

What about the link between capitalism and colonialism? We can’t just skip that, you might say! Indeed, check out this post of mine below.

Remember also that Japan offers a fairly pristine case study, where there was no capitalism or European colonialism in the pre-1853 era, yet subsistence-ratio data show extreme poverty. Moreover, as capitalism was introduced to Japan during the late 1800s with the Meiji restoration, subsistence ratios concomitantly jumped above extreme poverty levels. See the two blue series from my paper below. Note that organized labor in Japan was very weak for the first few decades after its rise in 1890 – yet subsistence ratios jumped. It’s almost as if what you need is simply explosive economic growth for the initial spell to be broken.

Last thing. Allen (2020) reports that every fourth person in England lived below basic-needs poverty at the end of the 13th century, when the country was clearly pre-capitalist. Then, around 1688, as England was well on the road to capitalism, less than 10% of the population lived in extreme poverty according to Allen’s social tables. During the Industrial Revolution, “absolute poverty largely ended.” Again, capitalism and the breaking out of Malthusian stagnation turns out to be key, not so much contemporary progressive movements.

Motte-and-bailey

I have no objections to Ward’s postscript, which he appended to his piece 10 days after its release. There, he states that we can question how reliable poverty measures are and whether they all show the same decrease. He adds that merely claiming that capitalism decreased poverty is overstated and that there’s nuance to the story. Moreover, he recognizes that free-market economies have nevertheless contributed to economic growth and to poverty reduction, and that denying this would be foolish. Globalization, too, he says, has played a role.

All of that is well said, and it’s good that he updated in light of criticism and further reading. However, some of the key strong statements from the piece itself are either wrong, undocumented, misguided, or biased in the sense that too many legitimate arguments and empirical facts are overlooked and not reported. There’s nuance, indeed, but the reader would scarcely get that impression from reading the initial piece (without the postscript).

Oh, and by the way, it’s not true that, as Ward says at the beginning of his piece, “there are reams of data suggesting that many facets of human existence, wealth inequality for starters, have gotten much worse under capitalism.” In fact, the very data he refers to there (the World Inequality Database) show the opposite. Read my piece below.

To the extent that the economy play a role, it does so subjectively. Citizens who (wrongly) feel they’re low on the income distribution ladder and (wrongly) think their national economy is doing bad, are more attracted to populism. I’m tempted to say that it’s precisely because Piker, Uygur, and many other right-wing pundits are falsely telling people that the economic sky is falling, these voters then develop subjective economic grievances, and are thus more attracted to populists. The same pundits, of course, use this fact to confirm to themselves that they were right all along – it is the economy! How ironic.

Aside from the social-media-generated subjective economic grievances, it seems that cultural issues are a much more salient explanatory factor for populism. If you worry about your nation’s finances (though you yourself are doing just fine), you’re somewhat more likely to vote for populists. But if you worry about stuff like immigration, you’re much more likely to succumb to the populist temptation.

There’s a new paper that shows this for Germany and the radical-right party AfD.

When voters start fearing migration more, they also vote for the AfD more

Using eleven waves of the German Socio-Economic Panel from 2014 to 2024, Martin Schröder and colleagues compare the explanatory power of concern about immigration with a wide range of variables intended to capture economic deprivation (including objective and subjective deprivation, and fear of future deprivation). They find that concern about immigration is by far the strongest predictor of AfD support both across individuals and “within” them (over time).

Cross-sectionally, a one-standard-deviation (SD) increase in concern about immigration is associated with roughly ten times higher odds of supporting the AfD. When they control for socio-economic variables, like income, life satisfaction, concern about the national economy, and so on, the migration-populism association remains almost as large as it was initially without these controls. Immigration concern by itself explains about 3/4 of the variation explained by the fuller model containing the mentioned deprivation measures. Depending on the specification they run, immigration sometimes explains even 90% or more of the variation explained compared to a model that also includes deprivation.

Because the 2014-2024 data they use are true panel data, they were able to run even more informative analyses. Do changes about immigration concern (or anything else) within the same person over time predict changes in AfD support? They of course do. As the same individual becomes more concerned about immigration, that individual becomes substantially more likely to support the AfD. The economic variables, in contrast, mostly show nothing – except for the expected subjective grievance about the national economy, though that’s comparatively tiny.

We would also want to know whether various forms of economic deprivation might still be important for populist voting by making concerns over immigration more politically consequential. That is, instead of treating concerns over migration and economic deprivation as two separate explanatory forces, it could be that the effect of migration concerns depends on how economically deprived people are. Put simpler, globalization and neoliberalism might be impoverishing voters, who for that reason are especially likely to react to worries about migrants stealing their jobs or putting pressure on scarce resources by supporting the populists more.

This isn’t the case. In the between-person models, some of the interactions actually go in the opposite direction, which would suggest that more educated and relatively affluent people who are also concerned about immigration are, if anything, especially likely to support the AfD!

The authors further illustrate how weakly deprivation matters (relative to immigration) with two hypothetical respondents. One is maximally deprived according to the variables in the model but not concerned about immigration. The other is maximally advantaged but very concerned about immigration. It turns out that the latter has more than twice the predicted probability of supporting the AfD. The exact probabilities are not especially important, since they refer to extreme and statistically constructed cases. The key thing is that the model doesn’t behave as one would expect if deprivation were the main force underlying immigration politics. Knowing how somebody feels about immigration tells us much more about whether that person supports the AfD than knowing whether that person is economically secure or insecure.

Are liberals wrong about migrants?

Okay, so at least in Germany but probably elsewhere too, populism is to a significant extent about migration. But this finding is compatible with several hypotheses about how migration works. The first, obvious one would look as follows: migrants are objectively something to be feared, German voters are rational people, hence they correctly perceive the objective, real danger of rising migrant influx, and so they vote for the party which will get rid of this migrant danger.

What can we say about this? Well, for starters, there’s some clear evidence that some groups of migrants have higher offending rates (and are thus plausibly “more dangerous”). As I reported a few months back:

[T]ake actual conviction rates in Denmark. The overall male population stood at 0.8% convicted in 2023. For male immigrants from the Middle East and North Africa, the share was 1.8%. Their male descendants were higher still, at 4.3%.

In Sweden, after adjusting for age, sex, income, education, and municipality type, relative risks fall from 2.5x to 1.8x and from 3.2x to 1.7x, respectively. Note that even after demographic adjustments, rates can remain elevated compared to the native population.

At the same time, these are not extreme numbers. And we don’t have clear causal evidence that aggregate migrant influxes generate more violence and crime. In simple descriptive terms, it’s also true that as the developed world has seen rising migrant shares over the past few decade, the national homicide rate has not concomitantly increased – instead, it’s fallen. Moreover, it is an undisputable fact that people on social media routinely go berserk when it comes to migrants and the alleged downfall of the Western world due to viral but misleading or outright fake content.

It would be foolish of me to say there’s nothing to this first hypothesis about why some Germans are (and have increasing become) worried about immigration. More knowledgeable people than me, like Sheri Berman, think the German state has lost the plot on immigration. I agree Germany has some answering to do. And sure, it’s true that some migrants in some places are a real threat that affects some people and makes them (justifiably) more worried. This is true of many social issues.

But I’m not sure we have good evidence this is all there is or that this is the predominant explanation of migration worries. Just think of the Springfield pet-eating hoax and how many people (especially populist Trump voters) fell for it immediately. It’s possible to be legitimately concerned about migration, and it’s very possible to be illegitimately concerned about migration, especially in today’s social media-soaked world.

You can make up your own mind, but to me, an important datapoint in this regard is how many people who worry about immigration are not actually the people who might have something to lose from immigrants. As both the authors and the study mentioned above find and as I did back in April, it’s simply not true that migrant worries are most concentrated in the cities, where most (ostensibly dangerous) migrants live. They’re also not concentrated in country-regions where migrants are most salient, nor is it true to say that migrant fears come primarily from people who are objectively economically threatened by them – the poorer and the unemployed, say.

(It’s also not true that, in Germany, living near a refugee facility and thus being exposed to potential everyday dangers from these foreigners has any effect on locals’ attitudes toward refugees. Anti-immigrant sentiments are “unaffected by the presence of refugees in respondents’ hometowns.”)

Instead, as I report in the piece below, people with right-wing ideology and subjective economic grievances, the less-educated, and the less-trusting folks are those who are most worried about immigration. That’s at least consistent with if not outright suggestive of the possibility that people misperceive the dangers of migration.

Obviously, none of this proves migrants aren’t an objective danger, and it doesn’t rule out that many German citizens develop worries about migration based on their real-world contact with problematic migrants. But to me, it makes it at least a bit less plausible – or, it increases my belief that some (if not many) populist voters are actually in large part concerned for false, invented reasons. Might also be both: some people justified, others less so; some worries justified, others not.

What is to be done?

Okay, but set aside the debate over whether migration worries are justified or not. You might simply say that, justified or not, this is what Germans worry about. And so if a liberal like me hopes the AfD and similar populists would go away, shouldn’t I support severe crack-downs on immigration? Additionally, isn’t Germany a democracy? Suppose, for a second, that it’s hypothetically true that migration has no downsides or is on net beneficial. Despite that, if the majority of voters, for whatever reason, normatively value that Germany’s migrant share be reduced or new immigration stopped, isn’t that completely fair? Immigration doesn’t need to be objectively harmful for people to legitimately democratically reject it. Values are not matters of fact, so you can’t say someone’s values are right or wrong. Values just are. And in a democracy, shouldn’t the values of the majority be implemented?

I take that line of reasoning seriously. And if it really turns out that the AfD takes over and destroys a liberal society just because people’s (justified or false) migration concerns weren’t respected, then it would have been better to severely restrict migration, thus cutting off AfD’s oxygen supply, and preserving a future liberal Germany (albeit one with sad illiberal migration restrictions).

But apart from that extreme, emergency scenario, I don’t think I wholly buy the “values objection.” That’s because I’m not for democracy simpliciter: I’m for liberal democracy, where certain rights and social issues are taken off the electoral menu and are non-democratically protected. For instance, it’s good that voters can’t democratically decide to revert women’s rights and private property, or elect judges, or have a say over central banking – despite what their values are on these issues. Having at least a somewhat-to-moderately open border (with proper regulation, of course) seems to me to be one of these fundamental liberal, non-democratic aspects that should typically be preserved regardless of what the majority wants in terms of their values. (Note, of course, that I’m not advocating for completely open borders here.)

Like, suppose that even though private property and free international trade pose no danger to the of voters or are on net beneficial, the majority for whatever reason becomes fearful of these capitalist institutions and wants to reject them. (*Ahem.*) Would we just say: yeah, let them, it’s a democracy after all. Or would we hope our liberal institutions manage to prevent the majority from implementing their illiberal, anti-capitalist values?

Now, to the extent that migration has real fiscal, criminal, cultural, and other downsides, something should be done to ameliorate those. To the extent that a serious and threatening political backlash is generated by worried voters, their worries should of course also be addressed somehow. But here, compromises will have to be made. Yes, just continuing with the status quo seems unworkable (for justified and unjustified reasons), but swinging to the other extreme and enacting illiberal and inhumane closed-borders measures is even worse. How the compromise that lies somewhere on the continuum between these two possibilities looks like, I have no idea.

He says they identify good institutions more or less solely with private property protection, which is in his opinion highly problematic and reductionist. In later popular works, he adds, they do expand the definition significantly to encompass the rule of law, provision of public services, free entry of new businesses, and freedom of people to choose their careers. He’s thinking of Why Nations Fail from 2012 here. But this, Reddy charges, “confusingly, run[s] together … the proclaimed effects and the descriptive features of such institutions”. Additionally, in this later period, according to him they also appended inclusive political institutions to inclusive economic institutions, but this again seems, in his words, “opportunistic and unsupported.”

So, the first ostensible problem is that Acemoglu et al. have a reductionist view of good institutions; and their later attempts at being non-reductionist are for some reason faulty.

Next, according to Reddy, the key claim by Acemoglu et al. is that good institutions appeared only in certain places, which explains why only certain places got economically developed. Here, Reddy naturally asks what explains why good institutions were successfully transplanted only in certain places, and he settles for this interpretation:

It came down to another explanatory factor, namely whether a sufficient number of colonizers had arrived on the scene and had been able to establish themselves as a dominant force, by beating back natives and disease.

Somewhat strangely, he immediately adds that Acemoglu et al. also used an additional, “correlative” finding related to the identity of the colonizers. His gloss is that they say (in his words): “The British had the best institutions, with the French, Spanish and Belgians having increasingly inferior ones.” Which place the colonizers originated from helped explain why some colonies got better or worse institutions.

Reddy then notes that, actually, Acemoglu et al. explicitly contrast their own perspective with that of other authors, and quotes them directly as follows:

Similarly, North, Summerhill and Weingast (1998) argue that former British colonies prospered relative to former French, Spanish and Portuguese colonies because of the good economic and political institutions they inherited from Britain, and Landes (1998, Chapters 19 and 20) stresses the importance of the culture inherited from Britain in these colonies.

In contrast to this approach, which focuses on the identity of the colonizer, we emphasize the conditions in the colonies.

Reddy provides his gloss on this statement: to him, Acemoglu et al. are again confirming here that they are concerned only with property rights protections (which were “enabled by the presence and dominance of Europeans”), not “the relative superiority of specific European institutions.” He additionally muses that even this version of the argument, where colonizers’ identity is unimportant, might be just as chauvinistic as the other one, namely, the one which focuses on the alleged superiority of domestic institutions and culture of the colonizers.

A not wholly fair critique

In all, the issues Reddy outlines are:

1. Reducing good institutions to secure private property.

   1. Expanding good institutions to mean other things as well later on, but doing so unsatisfactorily.

2. Explaining diverging development merely by pointing to the presence of European colonizers and to whether natives are numerous and strong or not.

3. Explaining diverging development by pointing to the identity and domestic institutions of the colonizers, but also saying that this isn’t their argument but other authors’ argument.

4. Both explanations are inadequate and also perhaps chauvinist.

There’s a lot to unpack here. Schematically, here’s my rendering of Reddy’s reconstruction of Acemoglu et al.’s chain of reasoning:

I think this is both somewhat confused and uncharitable or reductionist to how Acemoglu et al. really think about the matter.

My more charitable gloss on Acemoglu et al.’s institutionalist theory of diverging development would look as follows:

Now, it might be said that I’m drawing too much on Acemoglu et al.’s “later” work here, like Why Nations Fail. (Though why would that be bad? They got the Nobel for all the work they did, not just a few early papers). But the rich institutionalist theory in fact comes much before that, already in their celebrated 2005 Handbook of Economic Growth chapter. Interestingly, Reddy cites the 2005 chapter in his critique, but doesn’t engage with it and its theoretical substance at all.

This famous scheme below comes from the 2005 Handbook. Where is the ostensible reduction of everything to “property rights?” What is ad hoc about the inclusion of political institutions in the theory here? They fit in quite elegantly and are nicely theoretically elaborated in the piece. It also makes a whole lot of empirical sense to include them.

Notice also that the overarching argument about development (as I’ve laid it out above) has nothing to do with the identity of the settlers. Acemoglu et al. are (or, were1) rational choice and game theorists, for whom everyone primarily maximizes material interests like power and wealth, given environmental incentives and strategic interactions. The upper classes try to pump surplus out of subordinated classes. It’s simply that in some places, for contingent reasons, the best way of going about that was building more inclusive institutions, while elsewhere extractive institutions made more sense (in light of the material circumstances on the ground).

Whatever the merits of this argument, it’s not chauvinist, it’s not normative, it’s got nothing to do with British people being culturally superior or having superior institutions at home, it’s not logically wedded to only secure property rights mattering (to the exclusion of any other institution), and it’s a fairly plausible argument at an abstract level at least.

The famous comparison between Jamestown and Latin America from the 2012 book is explicitly about colonizers encountering different (and complex) possibilities for coercion. Both Spanish and English colonizers initially wanted gold and labor. But in densely populated societies with centralized political and tributary structures, the Spanish could capture the indigenous elite and appropriate existing systems such as the encomienda and mita. In Virginia, by contrast, neither indigenous people nor European settlers could easily be forced to work because land was plentiful and population density was low (with non-elite settlers having credible opportunities to escape if coerced by the elite settlers). So, ostensibly, the Virginia Company had to abandon its coercive system and offer land and economic incentives, and eventually even political representation to keep the non-elite settlers in place and work them.

The English were, on the Acemoglu et al. view, just as brutish and non-enlightened as the Spanish or anyone else. They go out of their way in the book to show how English elites repeatedly attempted to create hierarchical and coercive colonial systems. Where’s the chauvinism in that? Also, the theory which postulates that colonizers from different metropoles acted differently due to the culture they brought over is an alternative to the standard Acemoglu et al. story.

What works better: inadequate proxy measures and the “Kith and Kin” alternative to good institutions

Now, where I think Reddy is on firmer ground is in his questioning the empirical proxy for good institutions Acemoglu et al. used initially in their work. He says they were mostly reliant on “protection against expropriation” and “constraints on the executive.” This is still not “just secure private property,” like Reddy charges, but it’s definitely lacking as a broad measure of good institutions. Still, this was work being done in the 1990s and early 2000s, when better and broader measures of institutional quality weren’t readily available. With some charitability, Acemoglu et al. could be excused for focusing on relatively narrow measures of institutional quality and using them as proxies for something broader. Moreover, if we obtain more or less the same result with better data and measures, the critique definitely loses some of its core force and substance. And even Acemoglu et al. in that original paper did robustness tests with additional measures, such as law and order, rule of law, and judicial efficiency. This is not just “private property.”

I also feel his “Kith and Kin” alternative explanation for diverging patterns of development after colonialization is in principle fine and hasn’t been ruled out. His argument is that colonies containing large populations descended from the colonizing country enjoyed relationships with the metropole that other colonies didn’t, like preferential access to markets and capital and such. Colonies populated primarily by Indigenous peoples were more often treated as territories to be taxed and exploited. So, because a large population of European descent was itself correlated with low settler mortality and low precolonial population density, the apparent effect of settler mortality on institutions and development might partly capture the advantages produced by these preferential external relationships.

This is fair in principle, but again Reddy should know that Acemoglu et al. obviously don’t just ignore possibilities like this. For instance, they include dummies for British and French colonial origin and a control for French legal origin, after which the estimated institutional effect changes very little. They also control for the contemporary proportion of the population of European descent precisely because, as they put it, a large European-descended population might capture European culture or “special relations with Europe.” The control is statistically insignificant and the estimated effect of institutions remains significant, so at least a simplistic test of Reddy’s objection shows things might be fine. (They also show that their results are not driven solely by the United States, Canada, Australia and New Zealand.)

Yes, all of this is fairly crude, so Reddy’s critique still has bite, but Acemoglu et al. weren’t simply oblivious to such confounding, as is wrongly intimated by him.

Also, Reddy does nothing to empirically establish this alternative. And even if he were to do so, the alternative explanation is not necessarily completely incompatible with the original settler mortality/institutions findings. Why not have both? The way in which places were settled and which attendant institutions sprang up might have mattered alongside (not in contrast to) how colonies and metropoles were linked together. We even have some (relatively crude) indication that this might have been so.

Easterly and Levine’s 2016 study is relevant here. They construct a direct measure of the European share of the population during the colonial period and find that it is strongly positively associated with present-day income. The relationship survives controls for a plethora of confounders, and it persists when the “Neo-Europes” are removed and when attention is restricted to colonies in which Europeans formed only a small minority. To the extent that the colonial European population share can be taken as a proxy for the density of “Kith and Kin” connections (which it obviously can be only and at most in part), the Easterly and Levine paper provides some evidence that’s congenial to Reddy’s objection. European settlement appears to predict later development through channels not exhausted by Acemoglu et al.’s institutional measures.

(Of course, this really is crude. The Euro share finding doesn’t allow us to disentangle European human capital/culture from preferential colonial treatment.)

Importantly, Easterly and Levine rightly say their story is still compatible with Acemoglu et al.’s. That’s because when current government quality and colonial European share are entered together in the models, government quality (as a proxy for institutional quality) remains strongly associated with current income. It doesn’t collapse. That’s no proof of the Acemoglu et al. story, but it’s at least consistent with it.

Curb your critique

Reddy opens Why Nobelists Fail with this:

Readers of the work leading to the 2024 Nobel Prize in Economics, awarded to Daron Acemoglu, Simon Johnson and James Robinson, generally fall into two camps. One perspective sees their work as helpfully drawing attention to the role of institutions, and of colonial legacies in particular, in shaping long-run economic outcomes. Another sees their framework as being neither original nor revelatory, considering it excessively reductionist, starkly ideological and wholly inadequate for understanding a complex global reality. It will be argued here that despite the imprimatur these ideas have now received, these charges remain pertinent …

I’m not sure that most readers fall just in the one camp or the other, but if we’re framing the debate in this polarized way, then I’d say the first camp is clearly closer to the truth than the second one. According to Reddy, the first camp just says Acemoglu et al. were helpful in drawing attention to institutions for long-run economic outcomes. The first camp is not, in Reddy’s presentation, wedded to the idea that they did so perfectly or even very well. We don’t have to insist that their measures were very good or even satisfactory. Ditto for methods. If this is really the position of the first camp, then it’s very secure. No overstatements are made.

(Relatedly, see my more critical but also appreciative piece on the “original sin” of Why Nations Fail.)

But the second camp, as Reddy describes it, clearly overstates the criticism. The second camp is committed to seeing the Acemoglu et al. work not merely as imperfect, somewhat problematic, or even outright reductionist, but as “excessively reductionist, starkly ideological and wholly [wholly!] inadequate.” That’s a very silly thing to say, especially when we consider the decades of research by Acemoglu et al. and all of the other institutionalists doing similar and good work.

Even though he calls himself “an inequality economist” – scratch that, he actually styles himself “one of the best, if not the best, inequality economists in the world” – I pointed out in that piece that he’s not really an economist at all. For instance, he is not a professor of economics, he doesn’t have a PhD in economics, and he doesn’t conduct or publish economics research. That’s not necessarily a problem or a put-down, of course, but it is a fair description of reality. So, he’s not an economist, but his ideas about economics might in principle still be relevant, interesting, and even true – regardless of his credentials and (absence of) research.

The real issue is that when we turn to how he talks about inequality economics, Stevenson seems to be, far from “the best,” profoundly confused about both economics and inequality.

What’s his main claim to fame? If you go to his website, he says that “Living standards are dropping because of wealth inequality.”

He explains in more detail what “The Problem” is. As he states:

The 50s, 60s, 70s and 80s were the only time in history when society successfully created decent living conditions for ordinary people. This was done by aggressively redistributing wealth.

We are now losing this.

Poverty is rising, health is declining, public services are crumbling. People are losing trust in the government and each other. Far right politics is surging.

Okay, so there are a couple of claims here that need disentangling and checking. He basically repeats this in a very recent interview titled “Is this Gary Stevenson’s last EVER interview?” on PoliticsJOE (~800k subscribers):

We have economies where living standards are being aggressively destroyed by growing inequality of wealth but we have a broad academics economics class of people who do not believe that’s true, have never studied inequality anyway, so have no way of understanding even if it would be true …

Inequality is growing, living standards are in turn being destroyed by it, and (an additional claim now) economists are oblivious to it all. In fact, as he emphasizes:

When the government goes to its economists, in academia or in think tanks or in civil service, they have never ever studied inequality.

So after three years of studying economics, two years at Oxford, I was specifically looking for courses where we could study inequality. Those courses did not exist on a two-year economics Master’s.

Fact-checking Stevenson

Let’s start with the 1950s-1980s claim. Were those decades “the only time in history when society successfully created decent living conditions for ordinary people?” Well, no. Here’s just the data for the United States, which is often singled out by critics as the clearest example of a developed society that is “broken” for the ordinary person, and data on the United Kingdom, of which Stevenson typically speaks as a case in point.

We’re looking at four things here:

(1) Income shares across social groups (including the bottom 50% and the middle 40%),

(2) wealth inequality (top 1% share of wealth),

(3) a high-threshold absolute poverty rate (life with $10 or less per day, adjusted for inflation),

(4) and general trust.

Stevenson is saying that various measures of living standards are being aggressively destroyed while inequality of wealth is growing. Only when aggressive redistribution of wealth had been in effect were living conditions good. But that’s not true at all.

First, the middle 40% of income earners command the same share of total societal income today as they did in the 1970s in both countries. Second, the bottom 50% in the UK are, in recent years at least, just a tad below where they were in the 1970s. In the US, the declining trend is a bit clearer (though still very modest), which means some corroboration of what Stevenson is saying there; maybe? Third, don’t forget that the same or slightly declining relative income share over a decades-long period where absolute incomes are going up fast (even when corrected for inflation) means people are materially or absolutely better off than in the past. See the figure in the footnote, in which income is of course adjusted for inflation.1

Fourth, during the 1960s and 1970s, wealth inequality in the UK was actually higher than it is today (see panel A in the first figure above). The same, though, is not true of the US. There, wealth inequality has grown. Fifth, poverty was much higher in the 1960s in both countries than it is today (panel C in the first figure above). Sixth, trust in other people (panel D) is on par with where it was in the allegedly glorious 1980s.

So, Stevenson is quite far off the mark. But he also claimed that “health is declining.” Is this true? According to one key metric, no. In both countries, Disability-Adjusted Life Years (DALYs) per 100,000 individuals from all causes have declined if you standardize for aging. This is a measure of the total burden of disease, and it registers both premature death and years lived with a disability. So on this important aggregate measure of health (“One DALY equals one lost year of healthy life”), Stevenson is quite wrong.

To be fair to him, it is absolutely true that trust in government is going down, and that right-wing populism is surging – which were his other two claims. No doubt about that. However, he’s done nothing to show that declining trust in government and the surging far-right are caused by, or are at least related to, ostensibly declining living standards (which are not actually declining) and rising inequality. In fact, my research on whether income and/or wealth inequality are in any sense related to populist electoral victories explicitly finds no relation (still in R&R). I’ve also written about some of the inequality-populism link in one of my previous posts (though this is only on income inequality). See below.

Economists and the state of inequality knowledge

What about those dastardly economists, who according to Stevenson are either hiding inequality from us or are oblivious to it? I think there’s virtually nothing to that claim.

For at least the past two decades, inequality has been an extensively studied topic in economics. Here’s how I put it in my book Capitalism for Realists 5 years ago: Especially in recent decades, inequality has been a primary research agenda for mainstream economics. Thomas Piketty wrote a bestselling, 800-page book on the topic in 2014. His colleagues Gabriel Zucman and Emmanuel Saez are superstar economists (with over 65,000 citations between them at the time of my writing) whose claim to fame was precisely the study of inequality.

Additionally, Saez received the John Bates Clark Medal in 2009. Zucman got it in 2023. This is basically the Nobel for young economists.

In just these last few years after writing my book, Zucman and Saez’s combined citations have move significantly north of 100,000. Thomas Piketty’s 2014 bestseller on inequality has already amassed almost 40,000 citations. Inequality might be the hottest thing in econ today.

Don’t take it from me, though. Here’s how Anthony Atkinson, who “virtually single-handedly established the modern British field of inequality and poverty studies,” characterized it already in 1997 (my emphases):

The title of this Presidential Address is chosen to highlight the way in which the subject of income distribution has in the past been marginalised. For much of this century, it has been very much out in the cold. There are signs that in the 1990s it is being welcomed back …

On wealth inequality specifically, Zucman was able to say almost a decade ago now in the Annual Review of Economics that “there has been an explosion of research on wealth inequality” during the 2010s. That’s right around (actually, a bit before) when Stevenson did his MPhil, when on his account no economist was talking about either wealth or income inequality.

But there’s more. When you poll economists, they overwhelmingly agree with the statement that distribution of income in America should be more equal (they’ve done so at least since 1990). Why do they say this if “they have never ever studied inequality?” Already in 2000, there was a moderate consensus among economists that inequality can have deleterious effects on growth and stability; after 2011, this consensus has become “strong.” Again, how is this possible on the hypothesis that economists are uneducated on inequality? You literally have the well-known Journal of Economic Inequality, which is currently in its 24th volume (publication started a bit more than 2 decades back) and ranks in the 1st quartile by citations (Scopus).

Even an explicitly conservative, Bush-supporting economist like Gregory Mankiw has a whole standalone chapter on inequality in his widely-taught textbook from the mid-2000s. He opens it by saying “The gap between rich-and poor is a fascinating and important topic of study—for the comfortable rich, for the struggling poor, and for the aspiring and worried middle class.” But economists disregard and never study inequality? Their students never hear about it? Baffling.

If for some reason you want to learn more about Stevenson, check out how Decoding the Gurus have covered him. Here’s a preview:

In this episode, Matt and Chris take a look at one of the UK’s most compelling economic crusaders: Gary Stevenson, aka Gary’s Economics. A millionaire trader turned YouTube firebrand, Gary’s message is simple and potent: wealth inequality is spiralling, the ultra-rich are hoarding everything, and economists and politicians are either complicit or clueless.

Gary’s story has all the trappings of a mythic arc: from humble East London roots to Citibank’s trading floor, where he made millions betting against the poor during the financial crisis. Now he claims the system is so broken that only someone like him, working class and mathematically gifted, someone who entered the high-power world of financial trading and took on the system, could see it. As Gary puts it, a sort of economic Copernicus, who brought a revolutionary message that was dismissed by a stultifying orthodoxy.

With his righteous critique comes a hefty dose of swagger, whether it is in considering himself like a Usain Bolt of trading or in the frequent laments about how exhausting it is to be a lone voice of truth facing bad-faith hit pieces. Gary straddles an odd tension: self-effacing underdog one moment, saviour-on-a-soapbox the next. He rails against academia, dismisses journalists as clickbait merchants, and urges people not to heed critics, due to their ulterior motives.

1

In doing so, I’m afraid sometimes the aim is simply to support your own pet theory of what’s wrong with the world. Suppose you hate capitalism or the Western world more generally. The fundamental issue you have is that it’s hard to point to a convincing metric which would makes sense of your hatred. Capitalism and Western societies score high on virtually every conceivable social measure, so that won’t work. But perhaps you could point to a social phenomenon universally reviled by liberals, like the rise of populism and the far-right over recent decades, and say that capitalism or Western societies are clearly pathological at least in the sense that they spawned this unfortunate event. You could even claim that there’s clearly something deeply and objectively wrong with these societies, which is ostensibly why increasing shares of (rational, perceptive) voters are rebelling against them by electing populists. Check-mate, liberals!

I don’t buy this particular move, for many reasons. Moreover, my general materialist charitability (= people’s attitudes and behaviors generally reflect the real happenings and incentives in their surrounding environment) is, for good or ill, wearing increasingly thin when it comes to voters, especially populist ones.

This is so in part because there’s a lot of research showing any straightforward materialist explanation for the rise of populism is either wrong or highly mixed in terms of the evidence we have for it. I’ve written about how many studies fail to support the inequality thesis for the rise of populism, and how not only national but also regional inequality doesn’t seem to matter. I also reviewed a causal meta-analysis, which reveals only mixed evidence that import exposure (globalization) and objective labor-market exposure have anything to do with populism, and additionally reveals that there’s overall lots of publication bias. Once publication bias is corrected for, the average partial correlation coefficient on measures of economic hardship shrinks by more than 70%. Pundits should start updating on this.

The same meta-analysis suggests that subjective economic hardship stands out as a more robust cause of populism. That makes a lot of sense to me, given everything else we know about widespread voter ignorance and social media-driven moral panics. For various (non-objective) reasons, many people have the mistaken impression that their world is falling apart, economically, which is why they’re frustrated, which in turn is why they are throwing their lot in with the populists. At least that’s my speculation.

As Kevin Bryan puts it in a related context:

Actual inequality doesn’t matter, but inequality misperceptions do

So, instead of only (or primarily) looking at how reality ostensibly forces rational people to vote for populists, we should also be investigating how irrational people’s misperceptions about the world make them vote for populists. Here’s a quick economic illustration of this very fact.

Statistically, I’m unable to predict from actual inequality which European countries’ voters are more or less attracted to populists. But I have no problem predicting populist voting based on perceived inequality. See below.

I used a recent ISSP module on inequality to construct a country-level Gini measure of perceived inequality, inspired by this paper, and relied on the latest ESS wave for individual-level data on populist voting. Importantly, it turns out that actual and perceived inequality show no statistically significant correlation with each other. That is, people think inequality is high or low in their country regardless of whether it’s in fact high or low! Misperception.

As you can see in the coefficient plot, I try to mess with the data in different ways, but the results stay the same. Now, this is just illustrative and correlational, and I don’t intend for it to be taken too seriously. However, just the other day, a much more serious preprint was published on the topic of relative income misperception and populist attitudes. It links up nicely with, and significantly boosts, what I have argued and presented so far.

When the well-off feel left behind

Thilo Albers and colleagues ask why people who are doing perfectly well economically sometimes behave politically as though they’ve been screwed over. To answer the question, they use German survey data in which they observe both a household’s actual position in the national income distribution and where respondents think they rank. As you’d figure, people are not especially good at locating themselves. Folks toward the bottom tend to imagine themselves further up, while those toward the top think they’re much closer to the middle. A substantial number of affluent Germans are walking around under the mistaken impression that they are considerably less affluent than they really are.

I doubt this is only true in Germany; just think of how most Americans really live and how different the economic nonsense they say on X is.

In any case, what the authors find is that, when holding people’s true income position constant, respondents who underestimate their relative standing express more populist attitudes. The relationship isn’t particularly large, but it’s notable: a 25-percentile-point underestimate of one’s rank (which represents about a 1 standard deviation in the sample) is associated with roughly 0.15 standard deviations more populist sentiment. Not a large relationship, but it’s very robust and even potentially causal in one of their approaches.

Also important is the fact that the relationship is asymmetric. That is, people who falsely believe they are poorer than they really are tend to be more populist, but erroneously believing yourself richer doesn’t produce an equivalent movement away from populism! Makes total sense and goes to show that this is not just about being generally bad at estimating. See the coefficients on “Misperception” below in the total sample and the subsamples.

What are some other things we can read off this regression table? There are a few that make me secure in my increasing skepticism about materialist explanations for populism, but there are also some results that push against it. I don’t want to wave those away.

First, higher local inequality (Local P90/P50) doesn’t predict more populism. If anything, it does the reverse! More locally unequal contexts see less widespread populist attitudes. Additionally and somewhat relatedly, how locally poor or rich you are (in relative terms) doesn’t matter at all.

Second, however, where you are in the income distribution nationally does seem to matter. The richer you are, the less you are attracted to populism, holding all else (including misperception) constant. At first glance, this even seems to matter more than misperceptions - the coefficient in, say, column 2 is higher for the actual percentile compared to misperception. But for that to be informative, the original decomposition has to be recovered, which shows that both are of comparable size.

Fair enough, income misperception and reality might both matter in this case. (See also the coefficient on local median income.)

An additional neat and somewhat ironic fact about the analysis and the data is that the respondents who underestimated their income rank were, descriptively speaking, the more economically comfortable group. See below. On the right-hand side are the “pessimists”, or citizens who think they’re doing worse than they actually are. If you compare them to the “optimists” (people who are also misguided but in the other direction), these pessimists aren’t just doing better than they think they are but are also overall the better-off group. They have more income, are more likely to own their homes, have savings accounts, be employed full time, and even hold financial investments.

This is how the authors summarize these funny descriptive differences:

pessimistic misperception is common among economically secure respondents and is not obviously reducible to prior far-right identification, prior anti-establishment attachments, or a generalized pessimistic disposition.

The other interesting part of the study has to do with an experiment they ran. The authors wanted to see what happens when citizens are informed of the truth about their income positioning. In a preregistered experiment with roughly 5,000 Germans, respondents were randomly shown where their household really sits in the income distribution. The intervention worked exactly as an information treatment should, in that is that people corrected their beliefs to a notable degree. (Learning that they were better off than they’d thought also produced somewhat greater satisfaction with their income, but this isn’t statistically significant.)

However, correcting people’s economic misconception had essentially zero effect on the populist attitudes they expressed.

So, it seems that false beliefs about one’s material standing are involved in the formation/maintenance of political grievance. But then once that grievance has already been incorporated into a broader populist worldview (i.e., politicians don’t listen to people like me, elites have rigged the system, and so on), probably via social media and simply the political supply of populism, correcting the factual premise does not dissolve the misguided political conclusion. They’re locked in now.

Succinctly put, the thesis states that a peculiar, violence-inducing code of conduct seems to develop where people cannot rely on the state to protect them. Historically, this is most likely in societies built around easily stolen property such as livestock (not fixed assets like land or even somewhat stealable crops). In those settings, a reputation for answering insults or threats with force can deter future aggression. So, people have a strong incentive to invest upfront in being known as ruthless and unforgiving, and reactively violent. The thesis is more or less costly signaling theory applied to lawlessness and herding.

In the book, Pinker uses the famous Nisbett-Cohen psych experiments comparing Northern and Southern American men to illustrate this. Southern participants reacted more strongly when insulted, showing greater stress, dominance, and unwillingness to back down. The canonical experiment was fairly simple: while walking through a narrow hallway, a confederate bumped into the participant and called them an “asshole.”

Here’s how the researchers unpack the results in a bit more detail:

Compared with northerners—who were relatively unaffected by the insult—southerners were (a) more likely to think their masculine reputation was threatened, (b) more upset (as shown by a rise in cortisol levels), (c) more physiologically primed for aggression (as shown by a rise in testosterone levels), (d) more cognitively primed for aggression, and (e) more likely to engage in aggressive and dominant behavior.

Why Southerners in particular? Partly, because they are presumably more likely to be descendants of people from hilly/rocky parts of Ireland and Scotland where herding predominated (see Albion’s Seed). And partly because the South itself has a longer history of lawlessness and weaker state capacity due to the longer frontier experience. (The idea, of course, is not that the behavior is genetic but that it got transmitted through the generations via cultural evolution and socialization.) So, we can plausibly interpret this as evidence that honor cultures create a conditional readiness for violence, especially when status or reputation is challenged.

As already intimated, the background logic in all of this is that once such a culture is established (presumably for originally ecologically rational, material reasons), it can persist even after its original conditions disappear. This can be so for cultural-evolutionary reasons or just general cultural persistence. To an extent, the tendency might persist to today, as implied by the experiments.

Now, as you can imagine, this seems to be a story that’s too good/neat to be true. And sure enough, there are some important dissenters. For instance, Chu and colleagues find that a certain macro-level extension of the psychological findings doesn’t replicate:

We attempted to replicate and extend the empirical test of the herding-culture-of-honor proposition proposed by Nisbett and Reaves. We find that when differences in white poverty are controlled, there is either no discernible contemporary (1976- 83) relationship between the white male homicide rates and environmental regions or a difference that is inconsistent with the predictions of the theory. The differences reported by Nisbett [and colleagues] are artifacts of unreliable estimates of homicide rates, skewed distributions, and the failure to control for differences in the distribution of white poverty.

Another study by Berthelot and colleagues also found the opposite of a generalized culture-of-violence prediction. In their paper “Scots-Irish Women and the Southern Culture of Violence,” they found that counties scoring higher on that index had lower white female homicide rates, both before and after their controls. So there might not be any Scots-Irish cultural ancestry that simply raises violence among everyone, though it’s still possible (and I’d say quite probable) that men specifically are the ones expected to defend reputation and honor.

New global evidence for the culture of honor and reactive violence

Though I’m not happy about it, I think that the culture of honor explanation for violence is probably correct - though only as a very partial account. The original individual-level experiments seem convincing, and there’s a “new” (yet to be published) paper by Cao and colleagues that tests the proposition in many different ways and with the use of different data sources at the world level. As you’ll see, the “effect” size is not impressive at all, so we shouldn’t be overstating anything. But as a potential partial explanation of why violence differs both across the world and within countries, it seems to be quite robust.

The paper begins with information on 1,100+ pre-industrial ethnic groups in the Ethnographic Atlas. For each group, the authors estimate how much of ordinary subsistence came from animals that generally have to be taken out to pasture (and so can get stolen). The average group obtained about 14% of its subsistence from herding. Around a whole third obtained less than 5%, while some at the very upper end of the distribution got more than 50% of sustenance from herding. As you’d predict, high levels of herding appear in several parts of Africa, the Middle East, Central Asia, and northern Europe.

They then connect these historical groups to modern populations on the basis of language as a proxy, and finally match everything to conflict data.

The key finding is that language groups with a stronger historical connection to herding experience more conflict. This is true for events involving the state and for conflicts among non-state actors. A one-standard-deviation (SD) increase in historical herding corresponds to about one-tenth of an SD more conflict. As I said, this is a quite small “effect.” Perhaps more intuitively, they have a separate count model that produces this: an increase of 0.1 in the herding index is associated with about 40% more conflict events.

I say “effect” because this is basically a correlation. Obviously, they have loads of controls, and reverse causality is somewhat ruled out because of the significant time delay between the measures. Regardless of the controls they throw at the result (settlement patterns, terrain, development, religion, urbanization, kinship patterns, and so on), the original link persists. They also separately focus solely on Africa and use a more fine-grained conflicts dataset, where they find the same correlation (11-14% of an SD).

But the paper is much more impressive than this.

The central question with respect to the culture of honor is why the extra violence occurs. If the mechanism is revenge, as is argued by the theory, the difference between historically pastoral and non-pastoral groups should be most visible primarily after some earlier injury has created a reason to retaliate. And this is exactly what we see.

Following a year with no conflict, the historical herding measure has almost no relationship with violence, but following a year in which conflict did occur, the relationship becomes larger and statistically significant! More detailed estimates show that the difference is greatest soon after an event and then declines, just as you’d expect. (In the monthly analysis, it becomes negligible after about 4 months, and in an analysis using 10-day intervals, most of it has disappeared after roughly 40 days.) Next, using African data, they specifically look at revenge-related conflict events. They find that events that aren’t described in retaliatory terms show virtually no relationship with herding, while revenge-coded ones do.

Then they turn away from conflict records and look for evidence of the proposed attitudes themselves, using the Global Preferences Survey. The survey asks whether people are prepared to punish someone who treats them unfairly, or if they would seek revenge after serious injustice, and so on. And again, regions with greater ancestral herding score higher on the measure derived from these questions. A one-SD increase in the regional herding measure is associated with 7.3% of an SD more negative reciprocity. See the binned scatters below.

There’s still more! They have a historical folklore catalogue covering more than 1,100 societies that shows if traditional stories contain themes connected to violence, toughness, and revenge. Of course, groups with more herding history turn out to have a higher share of these themes in their recorded stories. (“Effect” size around 15% of an SD.) A smaller historical dataset gives us information on whether violence was considered acceptable. This is just 60 societies but the link is again there and is of comparable size to everything that has come before.

Crazy what they were able to do with cleverly linking already existing data together. To my mind, this is overwhelming enough to make me update in favor of the culture of honor thesis as one small piece in the puzzle about violence.

Several more or less Hollywood-style films have been made about Zimbardo’s “experiment,” including the not especially imaginatively titled Das Experiment (2001), the American film The Experiment (2010), and The Stanford Prison Experiment (2015). The impression one gets from these films, or from older psychology textbooks, is that Zimbardo really proved human personality to be strongly, or even entirely, determined by the social situation and symbolic role assigned to a person by others. The participants given the role of guards supposedly transformed rapidly and spontaneously (without outside encouragement or instructions from the experimenters) into disrespectful aggressors and even sadists, while the prisoners became passive, obedient figures stripped of their self-respect.

More careful researchers of human behavior have, however, long understood that, despite its evocative cinematic appeal, Zimbardo’s “experiment” is in fact devoid of scientific value. His undertaking has never been successfully replicated, its sample was unrepresentative, and Zimbardo himself recently acknowledged, in response to criticism, that it was not really a scientific experiment. It was more an “exploratory investigation,” you see. It’s also significant that two-thirds of the “guards” did not behave aggressively or sadistically at all. Thus, soon after the study was published in the 1970s, Erich Fromm sensibly pointed out that, if the experiment demonstrated anything, it was precisely that people are not simply and immediately transformed into sadists by how others define their identity.

What’s more, new evidence clearly reveals that the participants behaved as they did precisely because of the implicit and explicit instructions given by Zimbardo and his colleagues – an external influence Zimbardo denied. They were told, in effect, that the experiment would not work unless they behaved in particular ways. The expectation of receiving a financial reward after the successful completion of the experiment also could have played an important role in their behavior.

Solomon Asch and rampant conformity

But even if Zimbardo’s study falls apart, did Solomon Asch, a pioneer of social psychology, not demonstrate something similar through his entirely legitimate experiments on conformity and group pressure?

Asch first showed a small group of participants a card bearing a single line, followed by another card containing three lines labelled A, B, and C. They were then asked which of the three lines on the second card was the same length as the line on the first. One line was clearly shorter than the isolated line on the first card, another was clearly longer, and the last one was the same length. We would expect almost everyone to recognize what was the correct answer.

However, Asch instructed his confederates in the experiment – actors whom the genuine participants did not realize were part of the study – to give the correct answer aloud during the opening rounds, but later to insist unanimously on an incorrect one. He wanted to find out whether the unsuspecting participants would succumb to group pressure and give an obviously incorrect answer themselves simply because the actors supported it, even though it should have been perfectly clear that the answer was wrong.

The casual impression many people now have of this experiment is that most of the genuine participants gave in to the example set by the group and conformistically repeated the incorrect answer, even when the first line was plainly much longer or shorter than the line selected by the others.

Sheeple, right? People will support any absurdity, provided someone else says or does it first. They cannot think independently and quickly succumb even to the most foolish opinions and beliefs expressed by others. No wonder the world is supposedly as terrible as it is when society contains no Critical Thinkers™. I remember musing in exactly this way when I first heard about Asch and his experiment sometime in my late teens.

In reality, matters were somewhat more complicated. Asch found that, in most rounds of the experiment, most participants’ responses were correct even when the actors insisted on the wrong one. In the rounds in which the actors deliberately gave incorrect answers, only 37% of the responses were incorrect. A slightly smaller share, 33%, gave the same answer as the actors. A substantial majority (63%) continued to provide correct, independent answers.

And although it’s true that as many as 75% of the subjects answered incorrectly, or conformistically, at least once during the experiment, it is also true that 95% of them gave a correct, independent answer at least once.

It is hardly surprising that these and other nuances reported by Asch are absent from popular culture. What is surprising is that the same nuances are missing from a whole range of both older and newer psychology textbooks. In a 1990 study published in the European Journal of Social Psychology, researchers found that, in the decades following Asch’s experiment, textbooks increasingly emphasized the 37% of incorrect responses while increasingly omitting any mention of the 63% correct answers. Textbooks were also increasingly likely to report that 75% of the subjects conformed at least once, while failing to emphasize that 95% answered independently at least once.

A more recent study, published in Teaching of Psychology in 2015, found a similar pattern in contemporary textbooks. 14 of the 20 textbooks examined mentioned the conformist responses (the 37%) but only one mentioned those remaining independent, the 63%. 16 of the 20 textbooks highlighted the 75% who conformed at least once, while not one mentioned the even larger share (the 95%) who acted independently at least once.

This kind of selective reporting is problematic because it can help create a false and already excessively popular image of human beings as remarkably naive, passive, and obedient. It is also somewhat ironic, since Asch himself presented the results of his experiments as confirmation that the facts – not group pressure – were the decisive force guiding human judgment, and that independence, rather than conformity, had prevailed.

Subscribe now

Stanley Milgram and the shock machine

There’s one more famous psychological experiment, similarly mythologized in popular culture, that supposedly demonstrates just how desperately obedient human beings are.

Stanley Milgram, the renowned Yale psychologist, was operating in a completely different and methodologically far more sophisticated league than Zimbardo. Through a series of experiments, Milgram supposedly demonstrated that most people (65% of the participants in his first study) are entirely prone to blind obedience and will obey authority even in situations in which they ought to be skeptical of it.

Most participants supposedly administered extremely powerful electric shocks to an unknown person in another room merely because an authority figure instructed them to do so. They did not stop even when they heard the other person moaning and screaming. Unlike Zimbardo’s experiment, Milgram’s work has been successfully replicated on several occasions, making its findings considerably more robust.

Yet even with Milgram, we must pay attention to nuances conveniently excluded from the story of “people as sheep.” Gina Perry documented several of these nuances in her book Behind the Shock Machine.

First, the figure of 65% obedient subjects came from only the first of 24 different versions of Milgram’s experiment. In some of the remaining versions, the figure was much lower: 10 percent, 20 percent, 30 percent, or 47.5 percent. Second, approximately half of the subjects doubted that the experiment was genuine – that is, they doubted that their actions were actually delivering electric shocks to another person. Among those who were convinced that the shocks were real, the majority (two thirds) refused the experimenter’s commands. Third, the first version of Milgram’s experiment shows, at most, that most people are willing to trust authority when it is presented as science, not that they will trust any form of authority whatsoever. The subjects received a whole series of otherwise credible environmental signals suggesting that the situation was safe. They had been invited into the premises of the prestigious Yale University and were received by a scientist wearing a laboratory coat.

This may still be problematic, of course! But it does not demonstrate, as some people appear to believe, a blind willingness to trust authority in general.

People can sometimes be successfully manipulated. We can be excessively deferential to authority and susceptible to conformity. These tendencies can lead us to behave thoughtlessly and even harmfully. All of that is true. But the general idea that almost all people are typically sheep – apart, of course, from the enlightened person making the accusation – and that Zimbardo, Asch, and Milgram teach us unequivocally just how blindly conformist and dangerously manipulable human beings are is a mythology that is not grounded in reality.

For obvious reasons, you can read this post as a companion piece to my other post on the falsity of the (Marxist) “dominant ideology thesis.” Read it here:

Let me just say that a $30 minimum wage is about what it was in the '60s.

Robinson approvingly replied:

Just to be clear, you’re not talking crazy here.

Using my best movie narrator voice: As it turns out, Hanauer definitely was talking crazy here, and elsewhere as well. In 1968, the federal minimum wage reached $1.60, which today (adjusted for inflation) would stand in the mid-teens, so almost half of what Hanauer is suggesting.

But that’s the least we should worry about in this interview. The title of the piece is Neoliberal Economics Was a Con From The Start, and Robinson boasts how his two interviewees show “free-market fundamentalism” to have been “empirically discredited.” In actual fact, the interview doesn’t at all manage to show this, but it does inadvertently reveal a series of economic errors, misstatements, and empirical falsities on the part of the three people involved. I sure hope other Current Affairs resources are not as sloppy, especially when they’re making similarly brash claims to overturn whole economics fields and existing economic systems.

I should say I’m not the only one who’s noticed. Here’s how a guy a bit more ideologically aligned with Robinson than me put it.

Conceptual confusion

One of the absolute economics basics you should understand before opining on how the discipline/theory works, is the concept of economic equilibrium. It’s not a hard thing to get right, and it’s at the core of econ thinking. Very simply put, an equilibrium is a condition or state in which the relevant decisions and forces in a model are mutually consistent, and there is no tendency to change. That’s all.

Like, consider a car-dependent city. In it, people drive because public transport is infrequent and unreliable, and governments invest little in public transport because few people use it. Moreover, businesses build around cars because most customers arrive by car! Each decision reinforces the others. This is an unfortunate stable equilibrium, which no single commuter or firm can easily change alone, even though many people might prefer a city with better transit and less traffic.

Here’s what was said in the interview:

the contemporary sort of orthodox understanding of the economy is to conceive of it as an equilibrium system. That is to say, a system within which if one thing goes up, another thing has to come down. And so if you believe that, then it is effectively true by definition that if you raise the minimum wage, for example, there has to be corresponding job loss.

Impressively, this is wrong in three or four ways, not just one. First, it’s just a wrong description of how the “orthodox understanding of the economy” thinks of equilibrium. Second, a system in which if one thing goes up, another thing must come down kind of, maybe, resembles the logic of zero-sum. But why would you conflate that with an equilibrium? Third, economics doesn’t state “by definition” that if you raise the minimum wage, there has to be corresponding job loss. It’s all about the model: the perfect competition model does lead to that conclusion, but the monopsony model doesn’t (at least not if the wage increase is not extreme)! Fourth, economics has given us a lot of empirical research, a majority of which shows minimum wage increases don’t lead to corresponding job losses. (Depending on what exactly you mean, I’ll come back to this issue.)

Unfortunately, it doesn’t seem this is a one-off lapse in judgement. Look at this:

For 50 years or so, we in the West, and particularly in the United States, have adopted a framework of economics that we variously call ‘neoliberalism’ or ‘neoclassical economics’ or ‘trickle-down economics’ that structured the rules in ways that advantage capital above all else and disadvantage everything else.

If I squint really hard, maybe this is somewhat passable? But not really? I mean, how can you identify neoclassical economics with neoliberalism? Neoliberalism is of course itself a notoriously hard phenomenon to define and probably an essentially contested concept, but speaking broadly, we typically mean “deregulation, privatization, liberalization” when invoking it. That’s not completely unrelated to neoclassical economics, but the latter is a field of study, the former a system or set of real-world institutions and policies.

Moreover, neoclassical economics has a lot to say about externalities and market failure in general, offers loads of different models (with some being quite inimical to neoliberal policies), and the attitudes and preferences of its practitioners - neoclassical economists - are far removed from the trickle-down, neoliberal caricatures they are made out to be by socialists, anti-capitalists, and post-liberals of today. Can we, please, not treat neoliberalism and neoclassical econ as mere synonyms, especially in a piece ostensibly devoted to an in-depth analysis of these very issues?

This further silliness below requires no comment, so let me just highlight it:

[T]he interesting thing about the game of Monopoly is, no matter how many times you play it and who is playing it, it always ends up in the same result—a plutocrat.

…

Exactly. And what we point out is that a market economy is a game like Monopoly. That’s the takeaway. … A thriving middle class does not emerge naturally from a market economy.

Empirical falsities

It’s one thing to make a relatively minor mistake about how much the 1960s minimum wage would represent in today’s dollars. You’d be an unreasonable stickler and probably an ideologue if you made too much out of that. But the relatively minor mistake becomes more reasonably irksome when you consider all the other, sweeping mistakes made in the interview.

The biggest one is probably the claim about income stagnation and life becoming worse off over time. Both interviewees make this inexcusable error:

Incomes stagnated, social mobility dropped, and the middle class hollowed out.

…

[F]amilies have been waking up every day for 50 years, no matter who was in power, to a circumstance where they were worse off, while a few people at the very top were better off.

This is spectacularly wrong. Incomes in the United States have not stagnated in recent decades (or in the past half-century). Real, inflation-adjusted median (not average) personal income has increased by 70% between 1980 and 2024. Okay, but that’s the median person. What about someone poor, like a person in the 3rd decile by income? The increase has been a bit more than 50% over the same time period. Also relevant, according to BLS consumer expenditure surveys, between 1985 and 2023, the share of income Americans spent on clothing, food, and housing has dropped from 48% to 37%.

Social mobility and changing dynamics of the middle class are a complicated topic due to compositional issues, women entering the work force, and radical shifts in family dynamics. But even when looking at it simply, naively, the story told in the interview is not corroborated. The middle class has shrunk, yes, and the share at the bottom has even increased a bit, but there’s been movement at the top as well, and it accounts for an even larger share of the middle class decrease.

That is to say, the middle class has gotten smaller, but the upper-middle and highest class have gotten significantly bigger. I’m afraid people imagine “the hollowing of the middle class” to mean exclusively or almost exclusively downward mobility, especially when the statement is paired with explicit claims that “social mobility dropped.” In fact, both sides of the distribution increased, with the bottom going from 25% to 29% of adults, while the top climbed from a mere 14% of adults in 1971 to 21% today! Every fifth American is in the upper-middle or highest class nowadays. See below.

What else? One of the interviewees says this:

[T]he economists doing the empirical work kind of carried on, and they addressed the questions about the studies and did more studies and more studies and more studies, and they kept getting the same results that basically the minimum wage has no real negative impact on jobs, but has significant positive impacts on the wages of low-income workers, not just those on the minimum wage, but it also transfers up to other low-wage workers.

I appreciate the results of modern minimum-wage research very much myself, and have mentioned in the past that the old economics consensus (before the ground-breaking quasi-experimental research from the 1990s) on deleterious minimum wage effects has rightly shifted. But let’s not overstate what has been shown in recent decades. This is how I characterized the issues last year:

It’s sometimes said that the shift in consensus on this issue since the credibility revolution has been dramatic. For instance, the story goes that in 1978, surveys found that around 90% of economists thought minimum wages markedly increase unemployment among low-paid workers. In 2015, the share of such economists was only 26%. In reality, it’s more complicated than that. As a case in point, a more recent survey (done in 2021) shows 45% of economists still agreeing that the minimum wage can have negative effects for low-wage workers in many states. No doubt, there’s been a clear and to my mind justifiable shift of opinion over the long term, but definitely not as much as some proponents of the minimum wage imagine.

Also, just going off the big recent Dube and Lindner review chapter, I think the least we can say is that:

1. Point estimates for particular narrow groups of workers do lean mildly negative.

2. Yes, estimates covering the broad population of affected workers are centered more toward zero, and I tend to believe that result. But it’s not the only conclusion worth emphasizing.

3. Additionally, there are studies that report larger negative effects, which we also shouldn’t overlook, at least not in highly specific contexts, though these tend to be more scarce.

You can’t just paper over all of that.

Here’s another howler.

It [i.e., the minimum wage] also increases spending in local communities and has no negative impact on prices that anybody’s been able to find, and so on.

Huh? There are literal meta-analyses, like this one done on hundreds of estimates that suggests “a 10% increase in the minimum wage leads to a 0.3 to 1.1% rise in prices.” These are not just one-off studies, and even minimum-wage proponents like Dube and Lindner say that “higher prices may be one of the margins through which firms offset higher labor costs and not reduce employment in response to higher minimum wages.” There are of course cool thing to talk about here, like how better studies (utilizing quasi-experimental techniques) find much smaller increases in prices than standard regressions. Moreover, there’s interesting and expected heterogeneity, like the price elasticity being larger for labor-intensive goods. But good god, man, “no negative impact of prices that anybody’s been able to find?” What are you on about?

Same with this:

[T]he other thing you often hear is, "Well, it's technology change and globalization—can't do anything about those. Tough luck, workers." We have this thing called skill bias technical change. But when you look at the data, these [inequality] trends started back in the mid '70s and early '80s, and the trade and technology stories are very real and have had a powerful effect on the economy, but they don't really start until the 1990s. So something else was going on during this period, and that something else, we believe, was the changes in policy wrought by the neoliberal economic revolution.

Trade and technology stories don’t really start explaining inequality until the 1990s? That’s not at all true. Research on computerization found unusually rapid growth in demand for college-educated workers over 1970–1995 (with skill upgrading especially pronounced in computer-intensive industries). So, skill-biased technological change was already one of the main proposed explanations for the sharp rise in wage inequality during the 1980s. And it’s similar with globalization. International outsourcing and import competition were already affecting American production and relative wages during the 1980s. Feenstra and Hanson explicitly studied the decline in the relative employment and wages of less-skilled workers in that decade and identified rising imports and outsourcing as a contributing factor.

Another self-evidently crazy claim:

When the middle class thrives, that’s what creates economic growth. The thriving middle class is the cause of economic growth, not its consequence.

There are still other substantial errors in the piece, but I’ve had enough, and you probably have as well. Don’t get me wrong, you can make good critiques of some economics reasoning (see my recent piece on the Invisible Hand Theorem), question aspects of neoliberalism and free markets or undue opposition to state ownership, and you should argue in favor of social democracy and the welfare state. But what these guys are doing is not the way.

This was a particularly unserious read from Current Affairs.

The first is that the 6 points below weren’t at all intuitive and self-evident to me as a student, nor did I learn them in my courses, at least not in the sense of substantively grasping them. That’s not necessarily my professors’ fault. I wasn’t that motivated to learn as an undergrad. In fact, I didn’t even study social science initially. I did my undergrad in journalism (with a few introductory Gen Ed courses and electives in sociology and some other social sciences) and then moved on to a political theory MA, which was basically a humanities education, not political science. There was lots of Foucault, Deleuze and Guattari, Negri and Hardt, and other stuff like that. Not really my jam, as it turned out.

Now, my PhD was in fact in sociology, but funnily enough the 3-year program, which I started in 2013, didn’t really offer any kind of structured classes or coursework where you’d have to learn the fundamentals of social science or grasp empirical methods of analysis. I also had extremely limited contact with my supervisor, who in any case was a Marxist theorist and psychoanalyst, not a practicing empirical social scientist.

Think of it what you will, but I learned probably 95%+ of the social science I now know outside the lecture room and without any coursework. Starting in 2012 and really ramping my learning up in 2014, with various additional spurts and broadenings in 2017 and virtually every year since, I learned almost everything I know “on my own.”

How did I do it? I don’t think I’m anything special as far as smarts go – literally, not at all. But I think I can say I became very unusually motivated to learn at some point in time, and because I had access to the internet and every digitally published journal article and book out there, I was able to browse and closely follow myriad actual syllabi from universities abroad (mostly the US).

One early endeavor that stuck with me was when I “attended” every single one of Erik Olin Wright’s courses and seminars (undergrad and above) through his personal website. It’s still available, check it out here (and see the image below). Other than that, I just read up voraciously on anything and everything that interested me. I had no hobbies and didn’t have to be employed for the first 5 years of this process (up until 2017), so all I had was time. In 2017, I got hired as a sociology professor. This was nice, because now it was literally my job to read and learn as much as possible.

When I say I read a lot at the time, I mean something like deeply reading every or almost every day for around 6 hours (conservatively speaking), scattered throughout the day, every single year since 2012 up until around 2024. (Over the past few years, I significantly ramped up the amount of empirical research I do and so I read somewhat less now.) I don’t know, perhaps that’s actually not a lot, but it was enough for me to catch up on what I’d missed during my undergrad and Master’s.

I should also say my initial lack of social science knowledge didn’t just extend to me being theoretically and empirically less than well-informed. I was also methodologically almost completely innocent. Again, I mean this quite literally. Though I vaguely remember one intro stats course during undergrad, I mostly recall us deriving some (simple) math by hand and opening up SPSS for a minute or two. No shade intended towards my professors; I just wasn’t personally that interested.

So, even after I finished my PhD and became an academic ~10 years ago, I didn’t really know what statistical significance meant and what p-values were all about. If I’m being painfully honest, even just 5 years ago I didn’t know how to read a multivariate regression table on my own, let alone how to generate one in SPSS, Stata, or R. (My strong motivation for knowledge existed since 2012, like I said, but my love for statistics in particular developed pretty late.)

Okay, that’s a long-winded way of explaining the first reason why I now hold dear many “social science fundamentals,” which I realize I still haven’t directly broached at all in this post! Sorry, getting to it.

The second reason has to do with my experience as a professor teaching undergrads and Master’s students. Not only did I personally miss out on key social science basics as an undergrad, but I see the same thing happening to my own students. That’s to say, when I try to teach them these fundamentals, the majority don’t find them intuitive, obvious, or easy to understand at all. I like my students and some of them are brilliant – smarter, harder working, and more motivated than I was initially. But I feel they have a hard time really internalizing the lessons below.

Still, perhaps this helps someone out there somehow.

1. Meta-lesson: Forget about your petty political identity and personal pride (“I’m a libertarian! A socialist! …”), be curious, ready to change your mind, and read widely.

This probably cannot be taught, especially not to adolescent students, but god do I wish I’d somehow internalized this lesson already as an undergrad.

The problem is that humans are intensely social creatures who, by default, tend to pursue non-epistemic, social goals like fitting in, attaining status, and being (or coming across as) at least a bit special in their circle. So, when you adopt a certain political goal or identity by virtue of these higher-level social motivations, like becoming a Marxist socialist or a Friedmanite libertarian, you become locked-in, epistemically speaking. You’re less open to freely encountering new information, changing your mind in light of evidence, and just generally pursuing the truth/going by the actual evidence.

That’s bad if we’re doing social science and practicing epistemic hygiene. Of course, I’m not advocating literally going against and completely overriding our “nature.” That would be a fool’s errand. Nothing is gained by saying, “You should be enlightened and not pursue social goals and not nurture your special personal identity.” No, but what I am saying is that it might pay to try and hack yourself by striving to craft a very peculiar meta-identity and reputation. Instead of becoming a proud and notorious socialist or libertarian in your circles, try to become a proud “person who always complicates things and is unpredictable in their opinions on any single political or non-political issue.”

That can still bring you joy in establishing a special and perhaps envied social status for yourself (just like becoming known as “the special socialist guy who knows everything about Marx and is well read in Theory”), which I guess is human nature. But it does so without the epistemic drawback of needing to contort and/or discount every new bit of information you encounter such that you continue reaffirming the concrete political identity you formed as an edgy know-nothing teenager, who just happened to fall neck-deep in some Randian novel or whatever! Best of both worlds.

It’s so freeing and mind-opening, not being a slave to a very specific and detailed political identity. It enables you to read and absorb anything and everything, not just the ideology-approved works that your political identity and tribe demand. Instead of forming a specific identity, or fruitlessly trying to resist creating and enjoying one, just form a meta-identity.

A good book to read on these issues is Julia Galef’s The Scout Mindset.

2. Another meta-lesson: Forget about disciplinary boundaries and read widely.

Like with specific political identities, nothing is really gained when you box yourself in by pledging allegiance to a given social science discipline. Don’t think of yourself as “a proud sociologist,” “a humanities person,” “an economist,” and so on. Load up on every legitimate method with a good track record, every intersubjectively verifiable source of systematic data, and every solid paper, no matter its disciplinary provenance.

The same goes for theories. You’re shortchanging yourself if you think of yourself solely as a “rational choice theorist,” “a functionalist,” “a cultural evolutionist,” and so on. Again, there’s something radically human and pleasing in building yourself up as a person who only does theory X. There might even be some epistemic benefits to being that monomaniacal, at least at first. But with time, this becomes a burdensome straight-jacket that makes you dumber.

Instead, familiarize yourself with many widely-used contemporary social science theories. They’re all so interesting. Some are more applicable, others less so, but there are usually at least some phenomena and circumstances under which one or the other applies. Perhaps it’s lamentable, but there simply isn’t a grand social science theory out there to rule them all, despite what Peter Leeson claimed in response to my take on rational choice (which I otherwise love).

So, if you’re truly interested in social science and how the social world works, I suggest you read widely and without much regard for disciplinary boundaries. Even if you’re interested only in economics, or sociology, or political science, you will benefit tremendously by familiarizing yourself with, in no particular order, all of the following (speaking from my experience):

• at least non-technical intros to game theory, rational choice, public choice, economics, political science, and sociology;

• cultural evolution, (quantitative, empirical) anthropology;

• analytical sociology, the basics of philosophy of (social) science;

• statistics and econometrics/causal inference (at least the basics);

• non-popularizing big history overviews;

• evolutionary psychology, behavioral genetics, intelligence research.

Random examples for some of these with which to start include (following the order above):

• Elster’s Explaining Social Behavior, Yoeli and Hoffman’s Hidden Games; Wheelan’s Naked Economics, Rojas’ Theory for the Working Sociologist, Clark et al.’s Principles of Comparative Politics, Jackson and Sørensen’s Introduction to International Relations, see also Singh et al.’s 2017 paper “Self-Interest and the Design of Rules”;

• Henrich’s The Secret of Our Success, Boyd and Richerson’s Culture and the Evolutionary Process, Ember and Ember’s Anthropology;

• Hedström’s Dissecting the Social, Little’s Varieties of Social Explanation and Microfoundations, Method, and Causation, Searle’s The Construction of Social Reality;

• I have no good suggestions for non-painful intros to stats and causal inference, but you could try browsing Cunningham’s Causal Inference: The Mixtape and Angrist and Pischke’s Mostly Harmless Econometrics;

• Mann’s The Sources of Social Power (all four volumes), Fukuyama’s The Origins of Political Order and Political Order and Political Decay, North et al.’s Violence and Social Orders, Giddens’ A Contemporary Critique of Historical Materialism and The Nation State and Violence;

• Barkow et al.’s The Adapted Mind, Sapolsky’s Behave, Raihani’s The Social Instinct; Pinker’s The Blank Slate, Buss’ Evolutionary Psychology, Warne’s In the Know.

3. It’s not so much about whether particular people are good or bad but what incentives and institutions they’re facing.

Despite what I just said, starting your social science journey specifically with rational choice theory is not a bad way to go. Just don’t stop there!

Obviously, there are personality differences between people, and some tiny percentage in every population are even outright psychopathic. Still, most people are cognitively and motivationally ordinary, so explanations for social phenomena that invoke peculiar behavioral motivations (like being outright “evil” or deluded) should not be the first thing you reach for when accounting for what is typically happening in societies. Even when it comes to some extraordinary but mass-based phenomena, like the Rwandan genocide or electoral support for Hitler and stuff like that, you should at first try to account for them with reference to ordinary human psychology under non-ordinary (and perhaps only temporary) incentive structures.

In other words, the naive, pre-scientific explanation for why bad, evil things happen is that there are bad, evil people out there. But social science teaches you to first look for the possibly bad institutions and incentives that tempt good (or, well, normal) people to do bad things. The stylized examples that can be given are myriad. Here’s a common one that illustrates the basic contrast.

King Leopold II of Belgium ruled very differently simultaneously in two very different institutional settings. In Belgium, political constraints and interlocking domestic social forces made predation costly, so he ruled relatively pro-socially. In the Congo, the absence of accountability made destruction and predation profitable. The same psychology (he didn’t like democracy and cared little for others both in Belgium and abroad) was channeled into different behavior under different institutional pressures. Check out Bruce Bueno de Mesquita and Alastair Smith’s The Dictator’s Handbook for more.

There’s also this fun example by Leeson of how seemingly very psychologically peculiar pirates (who were outlaws, criminals, crazies) actually ruled democratically aboard their floating societies because they faced very different incentives than legitimate merchant ships, which were ruled over autocratically. Piratical criminals literally came up with constitutional democracy and the division of powers 50+ years before Enlightenment thinkers like the founding fathers came up with the US constitution (see the figure below). Check out Leeson’s The Invisible Hook to get why it made sense, materially, for pirates to practice democracy, while dictatorship was the right choice for legitimate vessels.

Often, the following example is given when it comes to “incentives mattering.” Cowen and Tabarrok relate it in one of the editions of their econ textbook (I think the 5th), and I admit I’ve used it on my students. In the late 18th century, Britain transported prisoners to its Australian penal colonies on privately operated ships. The problem was that on some early voyages, prisoners suffered severe malnutrition, scurvy, and just overall high mortality. A particularly bad case was the Second Fleet of 1790, on which about one-quarter of the prisoners died at sea. Why so? Was it because the contractors were evil? No, it’s more that they were paid mainly according to the number of prisoners who boarded in Britain. They thus had virtually no financial incentives to provide sufficient food and medical care. There were also allegations that rations were withheld and unused provisions sold after arrival for more profit, which again makes sense given the stupid rule they operated under.

How do you fix this and get to a more optimal social outcome? You might want to try hiring more humane captains, whatever that means, or putting doctors on board to help heal diseased passengers. But the real solution was incredibly simple: change the main institution (or rule), namely, link payment to the number of prisoners who arrived alive and in good health instead of the number who embarked. As a result, the incentives (and behavior) totally changed.

That sounds awesome and convincing, but it was much more complicated in reality. Historical narratives are cool, but when we get our hands on quantitative data, those have to be consulted (see next lesson!). As one relatively recent study puts it:

The statistical analysis supports the view that this series of interlocking reforms – the keeping of detailed journals, the payment of financial bonuses based on the production of certificates, and the issuance of detailed instructions prescribing how surgeons were to behave – contributed to the collapse of convict mortality rates around 1800.

…

The finding that contract price and financial incentives are not significant will also come as a surprise to historians who have studied the early convict fleets, since it has been assumed that the fixed price contract for the Second and Third Fleets, and the lack of any financial incentives to land the convicts in good health, contributed to the high mortality on those voyages … A number of different contractual and financial arrangements were tried in this early period, and it seems there were too many other factors involved for the contractual incentives to make a statistically significant difference.

Still, the general point about incentives, constraints, and monitoring mattering clearly stands. For more fascinating historical examples of how environmental incentives can work, read Hugo Mercier’s Not Born Yesterday, Leeson’s WTF?!: An Economic Tour of the Weird, and Marvin Harris’ Cows, Pigs, Wars, and Witches.

Incentives and institutions are not everything, but they’re very important for understanding the peculiarities of social life.

4. Empirical evidence is king (even when we don’t have causality).

You can’t do anything without theory, but theory also always lets you do too much! I mean this in the sense that internal consistency is the only real constraint on pure theory, and that’s just not enough. If you’re smart and creative, virtually anything can be rationalized with theory. You can come up with almost any theoretical mechanism. Does democracy reduce the likelihood of war or increase it? There’s a logically coherent account for either. Does capitalism enrich or impoverish workers? Depending on the starting assumptions, you can argue for both. And so on.

At the end of the day, what you need is empirical evidence. You have to look at how things work in the real world, outside your theoretical constructions. Theory should be in service of the facts, not vice versa.

So, even if it’s not as glorious, get ready to look at, and learn to love, a lot of numbers. And be prepared to be humbled. Numbers are hard to get right, they can conflict with each other, and new stuff tends to get uncovered as time goes by.

You don’t necessarily have to get fancy with it. Forget about identifying causes and effects. Just browsing and soaking in descriptive facts from an aggregator like Our World in Data can be tremendously useful. Without data, you have absolutely no idea how social life actually is in your own country let alone abroad and in the world generally.

5. It’s hard to get at causality in the social sciences, but not impossible.

I recently wrote a whole post on this. The two lessons here are that it’s very hard to infer what’s causing what in the social world, but at the same time you can get plausibly causal estimates from observational (“correlational”) evidence sometimes.

See more here:

6. It’s not so much about whether a thing sounds nice (like rent control or democracy or the welfare state or socialism) but what outcomes it reliably produces.

I still quite vividly remember how I felt back in spring 2017 (or was it summer?) when I encountered the Against Democracy podcast episode of Rationally Speaking with Julia Galef and Jason Brennan. I just couldn’t wrap my mind around the title. How could someone write a book titled Against Democracy? I no longer remember if I ended up listening to the episode or not, but even if I did, I must have blacked out. I felt visceral disgust for the thought that one could be against, or critique, democracy. Obviously, democracy is by definition good. Or so I imagined. (For some weird reason, I had no problem entertaining a democracy-critical attitude, and both reading and appreciating the whole book, just a year later! Funny how that works.)

More generally, I’m afraid we humans tend to instinctively think about social processes, phenomena, and systems primarily based off their label or the vibes surrounding them. Capitalism seems bad, just from the sound of it. Rent control sounds nice. Who could be against lowering the price (poor) people have to pay to put a roof over their head? Moreover, you’re a monster if you are against the welfare state. It’s right in the name, isn’t it? “A state that generates human welfare, wellbeing.” Who doesn’t love that? Socialism might be more iffy due to its problematic historical incarnations, but it’s definitely better than capitalism. Again, it’s in the name: socialism. Who’s against being social and caring for others?

These naive, knee-jerk reactions should be resisted if you’re serious about social science, understanding the social world, and trying to contribute to improving it. It doesn’t matter how things sound. It doesn’t matter what their intended goal is (like rent control striving to make housing affordable). Virtually all that matters is how things are likely to work in the real world with the common constraints we face and the awkward quirks of human psychology we’re all well aware of. So, if the evidence unequivocally shows that rent control doesn’t work, and that is also the consensus among social scientists, you shouldn’t support the policy. Same with capitalism: if evidence shows the system works as one would desire (at least in certain important aspects), don’t get mad at the word.

I can also flip this around. If you’re a “hard libertarian” who got convinced after reading Rothbard that capitalism is desirable and just (or non-exploitative) a priori, almost by definition, then that’s completely useless in my view. You’re being ideological and haven’t said anything worthwhile. A good example of the silliness that ensues when you ask definitions to do too much work is this discussion between Walter Block and Bryan Caplan, with the silliness being wholly due to Block.

The right way to think about social institutions and policies is to ask, usually at least, whether they generate desirable consequences; not whether they are symbolically appealing, good-intentioned, and pleasingly sounding. There are probably important exceptions, but this is typically how we should want to think about things. Here’s how Jason Brennan puts it with respect to democracy:

When we ask what makes a hammer valuable, we usually ask whether it is functional for us, as we are. Hammers have a purpose—to pound in nails—and good hammers serve that purpose. Hammers primarily have instrumental value.

When we ask what makes a painting valuable, we generally look to its symbolic value. We ask whether the painting is sublime, whether it evokes various feelings or ideas. We also value some paintings more highly because of how they were made and who made them.

When we ask what makes human beings valuable, we will often say that they are ends in themselves. Sure, people can also have instrumental value—the person who makes you coffee serves a purpose— but they also have intrinsic value. People have a dignity, not a price.

….

I will argue that democracy’s value is purely instrumental; the only reason to favor democracy over any other political system is that it is more effective at producing just results, according to procedure-independent standards of justice. Democracy is nothing more than a hammer. If we can find a better hammer, we should use it.

Again, you can go overboard with this kind of reasoning. But it’s a useful corrective to how people tend to think about things. You might also say I’m cheating a bit now, because here we are moving outside the territory of social science proper and into moral philosophy, and I promised you social science fundamentals.

Fair enough, I’ll stop.

The Marxist world-systems theory was contradicted by the empirical facts in that period. The theory doesn’t preclude any one peripheral (i.e., developing) country sometimes catching up a bit, but it does say general catching-up is either impossible or highly unlikely. The hyper-globalization years demonstrated this is much too pessimistic. True, when there was less trade and economic freedom in the world (like in the initial protectionist decades of the post-war period), convergence was not at all happening. The gap between the rich and the poor was instead just growing and growing. But as the Soviet experiment collapsed, as China and India liberalized, and as more economic openness and market integration swept the world, the poor finally started outpacing the rich, world-systems theory be damned. Here’s the famous convergence graph by Sandefur and Subramanian.

Unfortunately, as you can see, convergence is coming to a close.

In a new paper, Jason Hickel and Dylan Sullivan go one step further and call catch-up development a “myth.” As they say:

Even in relative terms, convergence is not occurring; for most regions and most countries in the periphery their relative position vis-à-vis the core has deteriorated. Core–periphery inequality worsened particularly during the period of market liberalisation in the 1980s and 1990s. … While some peripheral countries have been integrated into the core for geopolitical reasons, an increasing majority of the world population is peripheralised.

These results support the insights of world-systems analysts who argue that convergence is unlikely to occur within the existing structure of the capitalist world economy. Real development in the South will require strategies of industrial policy and planning to increase economic sovereignty, develop South-South trade, and delink from the imperial core.

Lots could be said about this framing. As I’ve already pointed out, there exists a substantial period of time when convergence was happening. If you look at WDI data on GDP per capita, you can easily see that the periphery-core ratio rose from 11.5% in 1990 to 23.7% in 2023. Excluding China (though why would you exclude it?), the ratio rose from 14.4% to 20.4%. If you use Maddison or PWT data, you get similar results. At the country level, 63 of 106 peripheral countries improved their relative position, while world-systems predicts no general improvement or even additionally widening gaps.

Contra Hickel and Sullivan, this convergence period started precisely after economic freedom significantly improved around the world. In 1975, 1980, and 1985, the global combined score for size of government, freedom of international trade, and regulation was treading water at around 5.3 (down from 5.7 in 1970). The score then skyrocketed during the 1990s to 6.7 in 2000 (out of 10). Unprecedented convergence began in the mid-1990s. Of course, this is just an association, but we also have plausibly causal evidence that liberalization helps with development.

Coincidentally, the era of unconditional convergence is coming to a close just as the neoliberal system is crumbling and talk of de-globalization is omnipresent (to the perverse delight of many on both the right and left).

Finally, it’s too dismissive and ideologically convenient to characterize the 18 former peripheral states that have become part of the core as simply “some peripheral countries,” which “have been integrated into the core for geopolitical reasons.” And so on.

But let’s leave that aside. Let’s grant that economic convergence has mostly not been happening, or that it’s been very slow. Is it true, as Hickel and Sullivan claim, that this then “supports” world-systems theory? It’s clearly consistent with it; but does it actively support it, rule in favor of it? That is, if there’s been virtually no convergence, does it follow that this is so because the core is consciously underdeveloping, and appropriating surplus from or exploiting and plundering, the periphery?

An alternative explanation: historical path dependency

I think not. One interesting alternative explanation, which is also pessimistic but in a different way, concerns the Deep Roots of development. This strand of literature claims that countries and populations which entered the modern period with different productive capacities might remain on persistently different developmental trajectories – even in the complete absence of any international appropriation and contemporary unequal exchange.

There seem to exist persistent historical differences in geography (obviously) but also technology, state formation, agricultural development, human capital, and just culture in general (kin-based vs. non-kin based, for instance).

Probably the most relevant for present purposes is the Comin et al. (2010) study. They construct measures of technology adoption in 1000 BC, 0 AD, and 1500 AD and relate them to contemporary technology and income. As folks might remember from the various Garett Jones debates (including, if I remember correctly, the two Caplan-Jones discussions), Comin and colleagues use a migration-adjusted measure that assigns historical technology according to the ancestral origins of a country’s present population. They find that migration-adjusted technology in 1500 is strongly associated with income in 2002 and accounts for a substantial portion of its cross-country variation, net of many controls. Their results for 1000 BC and 0 AD also become stronger after making this ancestry adjustment, although the 1500 measure remains the key result.

Here’s my remade version of Comin et al.’s key scatterplot (their journal-ready, published version is surprisingly ugly and unreadable).

Now, ideologically, I don’t want this to be true. This is partly for idiosyncratic personal reasons of mine but also just because we would obviously want the world to be a place where poor countries could quickly get out of poverty and significantly develop through, say, applying just a few fixes to their institutions. We wouldn’t want history to be tending toward the deterministic and inflexible. Instead, we’d want to say to the developing world: just adopt good institutions like open international trade, slash subsidies and inefficient regulations, and educate the population – and you’re on your way to the top!

But perhaps history really is much more path dependent. Sure, good institutions are a key part of the story, but maybe it’s much more difficult for them to truly spread across the world, and they’re probably just one part of the story.

In any case, the point is that the slow overall pace of convergence could very well be due to Deep Roots. If measures recorded before the formation of the modern capitalist world economy predict a considerable amount of contemporary development, it’s difficult to argue that the present hierarchy was created entirely after the 16th century. Some relevant differences evidently preceded European colonial expansion and the modern international division of labor.

Could also be both. Pre-modern technological and political differences could have influenced which societies acquired the capacity to project military power and establish colonies. The development of the capitalist world economy could then have amplified those earlier differences. Or it could be neither. Hickel and Sullivan move too quickly when they seize only on the world-systems side of the ledger to explain non-convergence.

The spatial fragility, and robustness, of Deep Roots

But there’s one potential problem for the Deep Roots explanation, which is important and interesting in itself. Conley and Kelly (2025) have examined a large group of Deep Roots/persistence studies and argue that conventional cross-country regressions, which are typically employed in this literature, often mistake spatial patterns for independent observations, which is very problematic for the validity of their statistical inference.

The potential problem is this: historical technologies (or geography, institutions, and so on) and present income are all spatially autocorrelated. Countries close to one another tend to resemble one another, so standard errors that treat them as largely independent can be much too small. In other words, statistically significant results (coefficients with a small p value) could very well be artificially inflated and non-significant in actuality.

This is exactly what seems to be the case. Across 30 influential, even famous Deep Roots/persistence studies they reanalyze, only a few results remain conventionally significant under their more demanding procedures for addressing spatial autocorrelation. And you guessed it, Comin et al. (2010) is one of the main victims.

In the Comin specification reported by Conley and Kelly, the original nominal p-value is 0.002, which is very nice. But after adding a simple spatial trend, which is one way to test robustness, it becomes 0.829. See figure below, though note that the row is mislabeled. The result concerns Comin, 1500 AD (not Comin, 1000 BC). The replication code makes it clear that they’ve just mislabeled the row.

In their spatial-basis regression, the coefficient remains marginally significant with an ordinary heteroskedasticity-robust standard error, but the associated placebo exercises indicate that this procedure over-rejects, so it’s not really informative. With their preferred large-cluster inference, the p-value is 0.23. Again, not significant.

However, and bear with me here just a bit more, the Comin et al. result scrutinized by Conley and Kelly corresponds to the place-based specification in Comin and colleagues’ Table 8A, not the stronger migration-adjusted specification in Table 8B. In other words, Conley and Kelly simply didn’t test the most relevant Comin et al. result! But I did.

When I tested the key Comin et al. result to see if it’s robust to spatial dependence (and yes, I freely admit I did so in hopes of seeing it crumble), I found that it is. I threw virtually everything Conley and Kelly suggest at it, but the migration-adjusted 1500 AD technology result remains significant across the vast majority of tests. When spatial controls and large geographic-cluster inference are introduced, the coefficient was positive every time, and significant at the standard 5% significance threshold in 29 of the 30 different specifications I ran.

I also carried out the recommended simulation diagnostics. These ask whether (1) we would get similar results if we replaced the historical technology variable with fake spatially patterned noise, and (2) if the modern income pattern could itself be generated by spatial trend plus spatial noise. The first test is easily passed, while the second is more uneven: at the 5% threshold, only about half of the targeted specifications pass all diagnostics (though at the 10% threshold they all do). That’s pretty okay.

All this to say we should be careful how we interpret persistent developmental differences between countries (outside the precious and special temporary periods where these shrink notably). You can’t just look at non-convergence between core and periphery and attribute it to “the core siphoning off surplus from the periphery.” There are multiple competing explanations for the persistent rich-poor gap. I also have to admit it’s a bit funny to me to imagine that the “new core” countries such as, say, Malta, Taiwan, Slovenia, Estonia, Cyprus, or Singapore are somehow ripping off African economies, as if that’s how they became rich.

Markets are, in general, an invaluable tool for promoting human well-being. It’s very easy to forget that whenever two people enter into an economic exchange, it’s because they both expect to be better off as a result of this exchange than they would have been without it.

But it is false to claim, on this basis, that the more our society relies upon competitive markets to organize the production and distribution of goods and services, the better off we will be. Sometimes moving closer to the ideal of perfect competition will make us better off, but sometimes it won’t.

Not only does the Invisible Hand Theorem offer no guarantees in this regard, it doesn’t even generate a presumption in favor of the market pattern of organization [emphasis mine]. The argument for market solutions must be made on a case-by-case basis; it cannot be derived from an abstract model of how an ideal economy would function.

First, take a step back.

With the “Invisible Hand Theorem,” Heath means the mathematical statement that if you have a perfectly competitive market, you’ll achieve Pareto efficiency. In simpler terms, if the economy and people are set up in a certain way (like no market power and full knowledge of prices, etc.), then we automatically get to a desirable state where nobody’s welfare can be further improved without harming the welfare of at least one other person. All positive-sum interactions have been exhausted and nothing has been left on the table. We can rest easy knowing that resource allocation is efficient.1 The fun part is that we get that even if there’s no overseer consciously planning and guiding the individual moving parts of the economy (hence, the invisible hand).

Okay, most educated people know this already, and they’re also well aware of the obvious drawback of this way of defending, say, capitalist markets. The drawback is that the theorem (also known as the First Fundamental Theorem of Welfare Economics) doesn’t really work in practice. In the real world, we have some market power, only partial information, externalities, and so on. The many assumptions that generate the conclusion do not obtain in reality; they’re too demanding.

But, and I promise I’ll get back to Heath in a moment, I think many defenders of free markets usually shrug off this inconvenience. They point out that though real markets might not be perfectly competitive, they’re at least close to being so, which means that real markets are also close to being Pareto efficient. In other words, though we can’t say the invisible hand works in its totality, it does work approximately. And so even if we can’t totally defend real world markets on the theorem’s basis, we can still use it for an approximate defense of markets.

Sounds intuitive, right? If perfect competition is unattainable, the economy should at least be made as competitive as possible. More concretely, if one price cannot be corrected, the other prices should still be allowed to reflect marginal social costs. And if one market remains distorted by subsidies and overregulation, the other markets should be liberalized.

Heath says this is simply false. You can’t say that because perfectly competitive markets are Pareto efficient, almost perfectly competitive markets are almost Pareto efficient. Relatedly, we shouldn’t try to maximize competitiveness so as to move closer to the desired efficient state of the economy. The theorem just doesn’t do anything for you as a free market defender once you enter the real world.

More competitive markets can be less efficient, and vice versa

Why? Because of the theory (or theorem) of the second best. In 1956, Lancaster and Lipsey published a paper in which they showed that once even one of the conditions required for the first-best optimum cannot be satisfied, satisfying the remaining conditions one by one is no longer guaranteed to improve welfare/efficiency. Because the conditions interact, the best feasible arrangement may even require additional departures from the first-best rules. Correcting one distortion while leaving another in place may improve welfare, but it may also reduce it. There is no universal mathematical guarantee that “as free markets as possible” are most efficient. The intuitive defense from approximation just doesn’t work, unfortunately.

Heath illustrates the approximation problem with a journey to Hawaii. Travelling 98 per cent of the distance to Hawaii is not necessarily 98 per cent as good as arriving there. It may leave the traveler in the Pacific Ocean! So, if Hawaii is unavailable, travelling in a completely different direction to Las Vegas may be preferable. The point is that proximity in one dimension does not necessarily imply proximity in the dimension that matters, meaning that a market can resemble the conditions of perfect competition more closely without producing an outcome closer to Pareto efficiency.

To move on from Heath’s toy example, consider a standard simplified case that combines monopoly with pollution. A monopolist normally restricts production and charges a price above the competitive level. Considered by itself, this is inefficient because some mutually beneficial exchanges do not occur; namely, consumers who value additional units more than their production cost do not receive them. Suppose, however, that producing the good also creates pollution that is not included in its market price (a classic case of externality). Now, a competitive market would produce too much of the good from the standpoint of social welfare, on account of the externalized pollution. So, the existing non-competitive arrangement (where a monopolist restricts output) might be inefficient when considered alone, but it might also partly offset the external cost of pollution, making the net outcome perhaps more efficient!

Put simply, if the monopoly is removed while the pollution remains unpriced, output will increase. Competition has been strengthened, and one first-best condition has been restored, but the additional pollution may be more harmful than the benefit created by the additional production. The economy has moved closer to perfect competition in an institutional sense but may have moved farther from the welfare optimum.

So:

1. you can’t say that the Invisible Hand Theorem shows real world, imperfectly competitive markets are approximately Pareto efficient,

2. you can’t use the theorem to argue for removing distortions in the real world and moving markets closer to perfect competition,

3. and you can’t use it to argue against government-induced distortions or characterize them as inevitably worsening efficiency when they move the economy further away from the perfectly competitive benchmark.

I get why Heath says that Lipsey and Lancaster have “handed the left one of the most powerful arguments against laissez-faire capitalism ever developed.”

Here’s how he puts it in the book at more length:

[T]he model was not entirely science fiction; it did represent some kind of an approximation of the real world. A debate immediately broke out among economists (and other interested parties) about whether the approximation was close enough to justify drawing any sort of “real world” conclusions from its results.

But in pursuing the debate on these terms, they overlooked an even more fundamental problem. Everyone assumed that if the assumptions of the model mapped onto the real world in some approximate way, then the results of the model could also be mapped back in much the same way.

This was the error diagnosed by Lipsey and Lancaster. As it turns out, the extent to which the model approximates the real world with respect to competition says nothing at all about the extent to which it maps onto the world with respect to efficiency. So even if it were incredibly realistic, it would still be completely useless when it comes to addressing public policy questions.

Theory of the third best

Okay, so the theory of the second best destroys one naive, simplistic defense of the free market. But does it really remove, as Heath puts it, the “presumption in favor of the market pattern of organization?” I don’t think so. Heath doesn’t discuss this in his book (though he indirectly gestures at some of it), but there also exists the theory of the third best.

Yew-Kwang Ng came up with this pushback. I was randomly reminded of him and thought of this issue a couple days ago as I was rereading David Friedman’s Hidden Order. Friedman doesn’t discuss Ng’s work in the book, but he mentions him on the dedications page among all the other, as he calls them, “co-conspirators living and dead, the colleagues from whom I have learned.” I thought that was pretty cool. Anyway, Ng provides an epistemic or knowledge-based pushback to what flows or doesn’t flow from the theory of the second best.

The theory begins with three different levels of policymaking competence. In a first-best world, the economist knows how the economy works and can satisfy all the conditions needed for the optimum. There are no immovable distortions, or they can all be corrected. The task would then simply be to identify the optimum and implement it.

In a second-best world, at least one distortion cannot be removed. Perhaps, as intimated before, monopoly is unavoidable or an externality cannot be measured. Lipsey and Lancaster indeed showed that the remaining first-best rules no longer have to be desirable one by one, so the best attainable policy may include additional distortions away from the free market, such that they compensate for the unavoidable one.

But, Ng points out, standard second-best analysis still assumes something very demanding, which is that the policymaker understands the economic system well enough to calculate the right compensating distortions. The policymaker knows which imperfections are present and how they interact, or how people will respond. But do they know all of that? Probably not, so Ng’s more realistic third-best world removes the assumption. Now, the policymaker not only faces an imperfect economy but also has an imperfect understanding of it. Of course, the theoretically correct second-best solution may exist, but it’s not likely anybody can identify it with confidence (there might also be implementation issues).

Consider the polluting monopolist again. Second-best theory says that the monopoly may reduce pollution by restricting production, which is obviously possible. But preserving monopoly as an environmental policy would require enormous confidence that the environmental benefit of lower production exceeds the costs created by market power. If the government has good evidence about all the relevant factors here (pollution damage, how much is produced, demand, etc.), then yes, second-best analysis could be useful. But if the government has very little evidence, it should not treat the monopoly’s possible environmental benefit as established. It might be better to encourage competition and also develop a direct way of controlling pollution, which sounds much more like the first-best “free market” situation.

In short, the third-best theory’s upshot is that governments should not confidently preserve nor introduce indirect distortions whenever they might in principle offset another imperfection. Under uncertainty, which is pervasive in the real world, simple and targeted corrections, and policies that are easy to revise, are probably better than any fine-grained Big Government attempts at engineering a second-best optimum.

Where does this leave us? Heath is clearly right that the Invisible Hand Theorem cannot settle real policy disputes in the abstract, as at least many right-wingers (among those who still find it in themselves to defend capitalism and the free market) falsely believe. But Ng gives us an important reason to think that the absence of such a theorem does not leave markets and complex intervention on an equal footing. I’d say we have a defeasible presumption in favor of market-oriented arrangements that economize on policymakers’ knowledge, combined, where intervention is necessary, with measures that address identifiable problems as directly as possible.

I imagine most people would nominally and even confidently advocate for the position while at the same time thinking that, if you have the common cold and take a pill, and then stop having the cold the next day, this indicates the pill probably “worked” and produced a causal effect.

But that’s kind of like confusing correlation for causation! The fact that taking the pill came before recovering does not tell us what would have happened had the same person not taken it. Even if a lot of people (instead of just one) take a pill when they have a cold, and then stop having it, this succession of events does not by itself establish causality.

It’s completely possible that people who tend to take pills when having a cold are also people who rest, drink fluids, and take other precautions for ending the sickness, or are just generally more health conscious and thus in better physical shape. So who’s to say it was the pill specifically that helped? It could be so many other things. That’s confounding and it shows up in regressions as omitted-variable bias. Moreover, it’s completely possible that people take a pill for the common cold only after the sickness drags on for several days and starts becoming annoying, or when their symptoms are at their worst. But if so, the apparent effect of the pill may simply reflect natural recovery or regression to the mean. After all, the common cold goes away on its own in roughly a week.

In more fancy terms, we have an endogeneity problem. And in the social world, where people are constantly optimizing and selecting, endogeneity is pervasive.

The best way to distinguish these possibilities and truly arrive at a causal estimate of what the pill is or isn’t doing would be a randomized experiment. We would take a sufficiently large group of people with comparable colds and randomly assign some of them to receive the pill and others to receive an indistinguishable placebo. Because assignment is random, the two groups should, on average, be similar not only in the characteristics we can observe – such as age, physical condition, and the initial severity of the cold – but also in relevant characteristics we cannot observe. They should also be equally likely to contain people whose colds were about to end naturally. If the pill group then recovers more quickly than the placebo group, we have much stronger grounds for attributing the difference to the pill itself. The placebo group gives us the crucial and otherwise missing counterfactual. That’s an estimate of what would have happened to the treated group had it not received the treatment.

So, instead of just mindlessly uttering “correlation is not causation,” which yeah, we all sort of learned in high school or at the university, we’d be better off understanding what it actually means and applying it to particular cases where we are prone to falsely identifying a correlation as a causal relationship.

This works in the other direction as well, mind you. People robotically trot out the slogan all the time – or, well, when it ideologically suits them. But it’s clear most don’t understand it. A relatively recent example of that, which stuck with me, was the cool Funke et al. study on how populists wreck societies.

If we want to find out whether electing a populist leader causes deteriorations in economic growth and just overall damages democracy and society, that’s going to be very hard. Clearly, just looking at whether societies with populist leaders are doing worse means mistaking correlation for causation. It could be that poorly performing societies are more likely to elect a populist leader (due to reasonable discontent with the existing mainstream status quo). In that case, we have reverse causality. It’s not necessarily that populists wreck societies, it’s that wrecked societies are more likely to elect populists.

Or, to take omitted-variable bias, it could very well be that when populists come to power, societies really start going downhill, but for reasons that have nothing to do with populists coming to power. Don’t mistake correlation for causality. For instance, societies with populist leaders may also be experiencing a completely unrelated commodity-price collapse. That external shock could both increase public frustration – making voters more likely to support a populist candidate – and independently reduce economic growth or increase opportunities for corruption.

Now, the paper I mention is very well aware of all this. Hence, it tries to account for reverse causality and omitted-variable bias as much as is possible when working with observational data (not true randomized experiments) by explicitly constructing a plausible counterfactual or an estimate of what would have happened to each country had it not elected a populist leader.

But we still get the inevitable knee-jerk reaction:

So let me explain.

For every populist episode, the authors construct a “synthetic” version of the country from a weighted combination of countries that did not elect populists but had followed a similar economic path beforehand. (Well, a statistical algorithm constructs the synthetic control, so there’s no individual choosing and picking by biased authors.)1 No single country may resemble pre-populist Venezuela particularly closely, for example, but some combination of several non-populist countries might reproduce its earlier growth trajectory remarkably well. That synthetic Venezuela then provides a benchmark against which the actual Venezuela after the populist’s election can be compared.

This is meant to, first, tone down concerns related to reverse causality. Perhaps the country elected a populist precisely because its economy was already deteriorating. But if the synthetic country was constructed to mirror that deterioration before the election, it should – roughly speaking – capture where the country was already heading. If the real and synthetic countries move closely together before the populist comes to power and then clearly diverge afterward, it becomes harder to argue that the post-election decline merely reflects a bad trajectory that was already underway. Yes, in an important sense, this is just another correlation, but it’s a very different kind of correlation than just naively observing a time series.

The same strategy also helps with omitted-variable bias. Perhaps some third factor - like falling commodity prices or political instability – both makes the election of a populist more likely and independently damages the economy. A long and close pretreatment match can make it more plausible that the synthetic country captures the treated country’s exposure to important common forces. (Of course, it doesn’t guarantee equal exposure to new shocks after treatment.) The remaining gap between the populist country and its synthetic counterpart can be interpreted more plausibly as the causal effect of populist rule itself than the effect of some obvious confounder.

Another methodologically sexy thing this setup offers us is that we can run all sorts of diagnostic/falsification tests that help us decide how convincing the synthetic counterfactual is.

In one such test, the authors pretend that every populist entered office five years earlier than he or she actually did and reconstruct the synthetic comparison using this fictitious starting date. The obvious idea here is that if the method produced a large economic decline whenever the researchers arbitrarily declared that “treatment” had begun, there would be little reason to interpret the decline following the real starting date causally. The uncovered “finding” would just be a meaningless artifact of the statistical method.

But that’s not what happens. The actual and synthetic economies continue along very similar paths for the five years following the fake treatment, when no populist has yet taken office. They begin to diverge only after the populist actually enters government.

This is no definitive proof of causality, but it’s reassuring in several ways.

1. It counts against reverse causality. If the apparent economic damage merely reflected a deterioration that had already caused voters to turn toward the populist – and that would have continued regardless of the election – we’d expect the treated country to begin falling behind its synthetic comparator before the populist took office.

2. The test also provides some evidence against strong anticipation effects. There’s no obvious decline during the years in which voters or businesses might already have begun responding to the populists’ impending rise.

3. It’s also at the same time a test of whether the synthetic counterfactual can actually predict the treated country’s (or countries’) path instead of merely reproducing the data on which it was constructed. The algorithm builds the comparison using only data from before the fake treatment date. It’s then asked, in effect, to predict what happens over the next five years. Since those five years were not used to construct the match, the close fit is not automatic. But in this case, the synthetic country really does continue to track the actual country, at least until the populist takes office. That makes the counterfactual look much more convincing.

A method that finds no effect when treatment did not occur, continues to predict the treated economies during an untreated period that was not used for matching, and finds a divergence only after the actual treatment is pretty neat stuff. You can’t just say “but correlation doesn’t equal causation” to discount it.

Of course, I’m also not pretending all of this makes “the fundamental problem of causal inference” disappear. These are observational data after all, not a true randomized experiment! The synthetic comparison is only as credible as the assumption that it approximates what would have happened without the populist. An unobserved shock that affects only the populist country after the election could still be mistaken for an effect of populism. Nor can matching on the past guarantee that two countries would otherwise have continued along the same path at all times.

But these are objections to the credibility of the particular counterfactual, not reasons to dismiss the study by mindlessly repeating the slogan. The authors know that correlation is not causation, thank you very much. The whole purpose of the paper and their research design is to ask what additional comparisons and assumptions would allow the observed relationship to be interpreted causally.

They have a dataset which spans 1995-2020 and includes 23 erosion spells in 22 countries. Here are the countries and erosion events:

Their key finding is that higher income inequality is strongly associated with a higher probability of democratic erosion. In their basic model, predicted erosion risk rises from low single digits in relatively equal democracies to above 30 percent in the most unequal democracies. They illustrate this with Sweden, the United States, and South Africa: Sweden’s relatively low inequality corresponds to a low predicted risk (4%); the United States’ higher inequality corresponds to a higher predicted risk (8%); South Africa’s very high inequality corresponds to much higher predicted risk (31%).

And, of course, they don’t leave things at associational, correlational language. They’re clear they think this is causal (emphases mine):

What are the causes of this rise in democratic erosion? Cross-national statistical analysis points toward one robust finding: The more unequal income distribution is in a democracy, the more at risk it is of electing a power-aggrandizing and norm-shredding head of government. … The inequality effect is robust across a number of statistical approaches.

They do many robustness checks. The inequality result holds when they change the erosion coding, use alternative datasets, use wealth inequality instead of income inequality, switch from country-year to election-year analysis, use rare-events logit, add country fixed effects, control for regional differences, and control for polarization, GDP, democracy age, and state capacity. They describe the inequality-erosion relationship as robust across more than 100 model specifications.

That’s all good. But I don’t think they’ve actually causally identified their result.

The usual suspects of reverse causality and omitted-variable bias are only weakly addressed, if at all. Somewhat bizarrely, they mention the first problem once in the entire paper (in the abstract, at that!) and never speak of it again, let alone methodologically address it. They include a few controls to address observed confounding, but they again leave it at that.

In any case, their study is just one the several that we have on the topic of how inequality relates to democracy. Some of the existing correlation research shows, like Rau and Stokes, that inequality increases are associated with democracy decreases. But just as many papers (of similar or higher quality) show there’s no relationship between the two variables. See my discussion here. Unfortunately, no one seems to have undertaken a more plausibly causal approach to the issue so far.

Probing their result

To try and get around that problem as much as possible, I’ve set up a more demanding, quasi-experimental econometric design and tested whether relatively sudden and large inequality increases reduce democratic quality over time compared to a plausible counterfactual. Across many, many, many different analyses, I find virtually nothing but a null result. This is a paper I have under review currently, so I can’t talk about it too much. It also just came back from the first round of reviews, with the reviewers liking it but asking me to do additional checks and tests, so we’ll have to wait and see what they think of my revision. I’ll have a post up once it hopefully passes the process.

In the meantime, however, there is still some stuff we can do and talk about.

I downloaded the Rau and Stokes dataset because I wanted to just see, visually and perhaps only for descriptive purposes, what’s going on, over time, with cases of democratic erosion compared to non-eroding countries. Are there any increases in inequality right before the erosion, or perhaps in the years leading up to it? Is it just that countries that are on a higher inequality level are (regardless of the trend) more likely to erode compared to those with low levels of inequality?

Most importantly: if we try and create matching sets of countries, such that we are looking at eroders and non-eroders with otherwise similar traits (development, polarization, state capacity, etc.), what’s going on? On the Rau/Stokes hypothesis, I’m expecting to see non-eroding matches to be much less unequal compared to eroders, even after matching on all the other relevant characteristics.

The idea is that if high inequality really is the key difference separating democracies that end up eroding compared to democracies that don’t, then we should probably see this inequality gap between the two sets of countries (pre-erosion). Of course, we shouldn’t just naively compare eroders to non-eroders, because there are many other differences between them besides inequality. Non-eroders will naturally be less unequal.

However, if we compare eroders specifically to those non-eroders which are otherwise similar to eroders on many traits but for inequality, then we’d have something to work with. If it turns out there’s no big inequality difference, that’s a bit iffy for the thesis. But if a big inequality difference does show, such that comparable non-eroders are much less unequal than eroders, then that would go some way in corroborating the idea; namely, that it’s high inequality in particular (not just poverty or low state capacity) that matters for erosion.

Here’s what I get. Note the scale (Gini coefficient, y axis).

What’s going on?

1. Eroders (in red) are much more unequal in levels than never-eroding democracies (dashed grey line). That’s as expected and tells us nothing interesting.

2. Eroders have a flat inequality trajectory in the decade before erosion. That’s a bit weird, but okay, perhaps inequality is just a background risk condition for erosion such that high inequality level raises the risk, while low inequality level reduces it. It could be that levels, not changes, are important, which is fair enough.

3. Crucially, eroders are just a bit more unequal (in level terms) than matched non-eroders (blue line). Well, that might be a problem for the thesis. In plainer terms: democracies which end up eroding have roughly the same inequality level and trend as matched democracies that don’t end up eroding.

And here’s an analysis which includes both never-eroders as well as not-yet-eroders as donors for matching. This is in some ways a weaker test, but it also has some upsides. But the difference is again very small (even smaller).

Now, does this matching exercise really tell us anything? What if the matching-based procedure simply isn’t appropriate? More practically speaking, what if we carried out a falsification test and used the matching-based analysis on some variable for which we’re much more confident should matter in democratic erosion. If possible, it should be something that was also indicated as important and significant by the Rau/Stokes paper itself. One candidate could be state capacity. Clearly, democratic erosion should be easier to carry out in democracies with lower state capacity, though it actually only shows a mixed signal in the paper. An even better candidate is political polarization. It’s again theoretically very likely, but even more importantly, the Rau/Stokes paper finds it to statistically rival inequality in terms of effect size and significance.

Now, if we do this and find small gaps again, then perhaps we’re just generating false nulls. But if the expected gaps show up, then maybe the inequality “null” is onto something.

Here it is, with inequality in panel A for comparison.

Even just eyeballing it, the difference is very stark. (To make vertical distances comparable across panels, each panel spans the same amount of variation, i.e., two full-sample country-year standard deviations for that variable.) Eroders and matched non-eroders are almost indistinguishable in terms of inequality, but they clearly moderately differ with respect to state capacity (especially in the last few years before erosion happens), and they strongly differ as far as polarization is concerned, again especially in the few years before erosion starts.1

Here are the more precise numbers so we’re not only eyeballing. The standard-deviation differences are 0.12, 0.28, and 0.49, respectively. See below. If we’re restricting matching solely to never-eroders (which is arguably more appropriate), the SDs are 0.20 (non-sig), 0.37 (non-sig), and 0.49, respectively.

Now, I’m of course not saying this quick-and-dirty analysis of mine overturns the Rau/Stokes results. Additionally, it doesn’t provide causal evidence that inequality doesn’t matter for erosion. And it doesn’t provide causal evidence that state capacity and polarization do matter. Unobserved confounding is still a problem (though no more a problem than in the original paper). Moreover, although I was able to significantly reduce standardized differences between countries after matching, they’re nothing to write home about, so the “counterfactuals” here are really so-so.

But I would say this is some descriptive/associational evidence against the descriptive/associational evidence provided by Rau/Stokes. I’m simply not sure what to think of their results, even as associational evidence goes; let’s leave it at that.

I want to use the book as a launching pad for my own take on populism, which aligns closely to how Hanania thinks about the phenomenon. Before doing that, however, I can’t but at least briefly touch on his ideological past first. If you’re a center-left liberal like me, you must think Hanania’s radical-right past is insane and abhorrent. I certainly do. And that’s not for the silly reasons some of my fellow ideological travelers might adduce, like him merely affirming the overall reality of behavioral genetics, IQ, or evolutionary psychology. If you’re not wholly beholden to activism and are in touch with empirical evidence, you don’t deny these in a general sense.

If anything, genuine and level-headed leftist sensitivity to inequality and a concern with egalitarianism and social justice are wholly consistent with accepting the real constraints and push factors associated with biology or intelligence. See The Genetic Lottery by the progressive Kathryn Harden, The Cult of Smart by the communist Fredrik de Boer, The Darwinian Left by the liberal Peter Singer, or Behave and Determined by Robert Sapolsky.

No. What rightly freaks you out is that Hanania’s past is simply a slew of racist and misogynist cringe, outright genetic (group) determinism, and other extremely misguided things that might seem edgy and cool to male adolescents but are actually some combination of wrong, immoral, and outright dangerous. And it’s not just you or me. Perhaps surprisingly, he himself states much the same in the book, noting that he, quote, “hates” and is “ashamed” of his past.

I must also admit from the get-go that though I like how critical he’s become of low-information illiberals of all stripes (but especially MAGA), I also still can’t at all understand his incredibly misguided decision to support Trump over Harris in 2024. If you’re interested, he tries explaining it here.

I think I first came across his name in 2020, when I read his WaPo piece on the likelihood of civil war in America. That was before his terrible past was revealed. In the piece, he said that despite what we might feel in our gut, a civil war in the modern US is virtually unthinkable if we go by the available quantitative evidence. The article was basically a breezy academic literature review. No shock-jock posturing, no caustic rhetoric, and certainly no racism or misogyny. To my mind, it was a delight to read, with a surprisingly high information density for a random column. I wholeheartedly agreed with his evidence-based take and am happy to report that, 6 years later, there’s been no civil war in the US, despite everyone being convinced at the time that the end was nigh. “If only more columnists wrote like that,” I thought to myself at the time.

There are other things I could say, but I’m more interested in whether the influential things he says today about populism are as measured and true as his 2020 WaPo piece was.

Populism writ large

Usually, we define political populism as basing itself on the Manichean division of society into “pure people” and “corrupt elites.” Populists are also illiberal in the sense that they might embrace the electoral part of democracy (this is, after all, how the “pure people” can express their will in the first place) but they absolutely reject and despise the liberal part of democracy. That is, they don’t like the rule of law (allegedly wielded by unaccountable “corrupt elites”) and the separation of powers, which only complicate and stifle a ruler’s immediate will and thus prevent him – it virtually always is a him – from benevolently channeling the will of the “pure people” whom he allegedly represents.

Hanania broadens this somewhat. He wants to go beyond a purely political definition of populism. As he puts it,

Yes, there are populist and nonpopulist political leaders. But there are also populist and nonpopulist media sources, talk-show hosts, sellers of medical and nutritional products, and forms of entertainment.

For him, populism is a more general anti-elitist sentiment, where the “out-of-touch” elites are understood as those (usually higher-educated) people who are embedded in established societal institutions, such as universities, professional boards, mainstream media, and perhaps legacy non-government organizations. Importantly, elites need not be rich or politically powerful on this view. What’s key is that they get their status from the mentioned long-standing institutions and attendant peer recognition. Populists, on the other hand, strive to get their status by rejecting and denigrating mainstream institutions and by constructing a more direct relationship with a (usually not as educated) mass audience.

This is why it makes sense to call not only Trump or Orban populist, but to use the appellation as well for non-politicians like Joe Rogan and RFK, or Tucker Carlson and Gary Stevenson.

Incorrect explanations for the populists’ rise

Pundits, including some scholars, incorrectly believe that Trump, Orban, and other populists rose to power (or at least in prominence) over the past 2 decades primarily because voters correctly perceived the failure of the political establishment to deliver on the goods.

The story goes something like this. Our legacy political elites, who are/were supportive of liberal democracy, neoliberalism, globalization, and less restrictive immigration policies, have thereby enriched themselves immensely while bringing us devastation in the form of collapsing working-class communities, exploding and unstopping income and wealth inequality, decades of stagnating living standards, and the infiltration of untold numbers of immigrants, who are now responsible for a serious increase in crime across societies. Many voters correctly perceived this with their own eyes and responded by punishing the ruling elites. They withdrew their support and turned instead to (mostly right-wing) populists who were telling them the exact same thing: our communities are collapsing, incomes are stagnating, immigrants are corrupting our societies, and economists and other experts know nothing.

The story continues by admitting that perhaps these populists alternatives don’t necessarily have all the solutions and could themselves be terrible (like Trump), but at least they are in lock-step with the people and their preferences, are not lying to the people (like the old political establishment did), and are not supportive of disastrous liberal or neoliberal policies.

I think that’s virtually completely false at every turn. That is:

1. It isn’t true that legacy political elites have presided over societal failure. It isn’t true that neoliberalism and globalization have wrecked Western societies (or countries across the globe).

2. It isn’t true that, within the past 2 decades, income and wealth inequalities have just kept going up and up and up everywhere.

3. Nor have real incomes stagnated for the vast majority of people anywhere in the world.

4. It’s also not true that voters (especially populist voters) are correctly perceiving the broader social reality they are a part of.

5. Nor is it at all clear that those voters who have objectively seen the least progress or witnessed the highest inequality or had been hurt (by losing their jobs, say) or live in countries/regions where the immigrant influx has been the strongest are disproportionately more likely to vote populist.

Hanania more or less agrees with all that. Drawing on a 2025 Ipsos poll, he even says that subjective (as opposed to objective) grievances are not likely behind the rise of populism. I’m not completely sure about that. I think we actually do have some evidence that subjective economic grievances play a role in populism, one Ipsos poll notwithstanding. One key instance about which I’ll write a longer post is inequality. Research shows, and I have myself found, that though objective inequality measures don’t predict populist voting, subjective perceptions of inequality (which are often wrong, by the way, and don’t correlate well with objective measures) are fairly okay predictors of populism.

In any case, here’s a just simple throw-away scatterplot with the “system is broken” variable from the Ipsos report Hanania leans on. This doesn’t really tell us anything, but it is a surface-level corroboration of his idea that even subjective grievance isn’t necessarily linked with populism. Like I said, more, better, and more fine-grained data complicates this picture and suggests subjective grievance plays a role.

At the same time, he notes that in terms of stated values, mass publics and parliamentarians (at least in Europe) actually do diverge when it comes to immigration. So, he intimates, a possible mechanism for defusing some of the attractiveness of populism might be hiding in there. I’m actually not sure how much parliamentarians could do to defuse it simply by toning down their values regarding immigration. That’s because mass publics (and especially populist voters) are ignorant of many relevant facts about immigration or are actively misinformed. So it can’t all just be a discrepancy in values; it’s that voters, especially populist ones, tend to live in an alternate reality as far as immigration is concerned.

A better explanation

Hanania’s preferred explanation is basically communications technology. That might sound too simple, but I think it’s actually right. In fact, I’m convinced that any explanation of populism that doesn’t put changes in the information ecosystem front and center is not really trying to explain the phenomenon.

The old, gate-kept media had one enormous advantage, which is that it imposed a tax on crankery. If you wanted to reach a mass audience, you generally had to pass through institutions that, however flawed, cared at least somewhat about whether you were obviously insane and whether you could be embarrassed by evidence. That’s no longer true. Just think of the Joe Rogan Experience, Bret Weinstein’s The Dark Horse podcast, or Tucker Carlson’s show, or Candace Owens, or any of the other dozens and dozens of incredibly popular but ludicrous political podcasters out there.

Unfortunately, the internet, podcasts, YouTube, X, TikTok, and the rest of the decentralized media ecosystem have made it possible for anyone to reach a mass audience directly. In many ways, this can be cool. For one thing, I wouldn’t have this Substack before the internet. But the democratization of speech came with a serious downside, which is that it allowed people to build status not by being right, or careful, or respected by competent peers, but just by being compelling to a mass audience. And the mass audience does not, on average, reward boring truth. It rewards gossip, moral panicking, and conspiracy theories.

So people are just irrational? No.

Thing is, people in real life don’t typically fall for this bullshit. Though many perhaps indulge in conspiracy theories because they seem fun to them, or whatever, I maintain most people don’t really take them seriously. To understand this, consider Hugo Mercier and Sacha Altay’s distinction between reflective and intuitive beliefs.

Some beliefs are linked to action. If I think the pan is burning hot, I keep my hand away. Perhaps more importantly, if an ordinary person thinks a real child is trapped nearby, they call someone to intervene or even try to help themselves. These are what Mercier and Altay call intuitive beliefs. But other beliefs are more like accepted propositions and stories one is willing to endorse. People say “yes, I believe that” without letting the claim guide their actual perception and costly action. These are reflective beliefs. They are not necessarily consciously dishonest, but they are in important part insulated from the rest of practical cognition.

Here’s a good example:

Consider the belief in the so-called Pizzagate conspiracy theory, according to which high-level Democrats were sexually abusing children in the basement of a restaurant from the suburbs of Washington D.C. Although, according to polls, millions of Americans endorsed this conspiracy, for the overwhelming majority of them holding this horrendous misbelief appeared to have little or no consequence.

The exceptions are notable. One individual--who might have held the belief intuitively--stormed the restaurant and demanded at gunpoint that the children be freed. By contrast, all other behavioral responses consisted in expressive behaviors, such as posting negative reviews of the restaurant. Such behavior would be completely incongruous if the belief were held intuitively (imagine having seen pictures of children being abused in a restaurant, and reacting by posting a one star review of the place).

We have thus, for everyone but maybe the lone individual who tried to free the supposed victims, either no consequences from the belief, or consequences that would not follow if the belief were held intuitively. This suggests that the vast majority of those who endorsed Pizzagate did so reflectively. Note that doing so shielded these individuals from nearly all costly consequences from holding that belief, by contrast with the single individual who might have held it intuitively, and who served a multi-year jail sentence.

Like I said, in an important sense, most people don’t really believe the bullshit they spew or enjoy listening to on a podcast – they “believe” in it like they believe in WWE (wrestling). It’s fun make-believe. Unfortunately, when it comes to voting, where you’re shielded from the costly consequences of your beliefs (on account of your own vote not counting for much), conspiracy theories and misbeliefs can flourish. And when lots of voters act on them in this “safe” environment, crazy people get elected despite the fact that from your own individual perspective, voting on the basis of such belief was not dangerous and costly. And then you get insanity the likes of Brexit or Trump.

So, the general theory I’m imagining goes something like the following.

1. Though most people are not detached from reality in their ordinary lives, and can’t easily be duped where and when it counts, there are always a few in any population who are either grifters or just a bit crazy-leaning.

2. Nowadays (but not in the past), the voices of some of these outliers are able to get out to the public via podcasting and social media, namely, the fall of the gate-keepers.

3. Many ordinary people hear these salacious, inane ideas and conspiracies, and quite a few adopt them reflectively (though not intuitively). They tend to be entertaining, after all. As a result, these listeners go on with their lives normally and don’t reap the costly consequences of holding the aforementioned crazy beliefs. (They’re too rational for that. It’s mostly just bar-room talk and entertainment. Just think of the following examples: despite so much misinformation, most Americans absolutely did get the Covid vaccine, and they happily use 5G technology every day!)

4. However, when it comes time to vote, many act on these very same misbeliefs, which they might discount in everyday life, where they’d carry costly consequences with them (like, again, not being able to use your 5G phone). From an individual voter’s perspective, nothing bad will happen when you are “owning the libs” at the voting booth - which is true, individually speaking!

5. But because millions of people act this way, something bad does happen in the end; something bad that couldn’t have been prevented by an individual voter who valiantly resisted his podcaster-bro opinions. So why not indulge them? It’s a classic social dilemma. As long as people have the ability to listen to crazies online, and there’s a healthy supply of politicians willing to capitalize on this, weird things will happen at the voting booth.

The above sketch is how I square my acceptance of a broad version of rational choice theory with the irrationality of so many (populist) voters, while also working in the fact that the internet, and especially social media and the decline of legacy media, are key in unleashing populism.

One skeptical reply might be that, though theoretically elegant, this “rational irrationality” explanation for populist voters doesn’t work in the real world. That is, even if it’s true that an individual voter’s vote has an infinitesimal chance of affecting anything, the majority of voters probably don’t know it’s true but instead wrongly (but causally efficaciously) think that their vote matters immensely. If so, the whole explanatory edifice collapses.

I take the objection seriously in principle, but I don’t think I wholly buy it, primarily because we have studies which indicate that, actually, only a small minority of voters wrongly believe their vote is likely to be decisive, while half or more than half correctly know it’s very unlikely they’re going to count for much. Here are the results from one:

Respondents were asked what the chances were that the outcome would be decided by a single vote, and four response categories were offered: ‘very high’, ‘somewhat high’, ‘somewhat low’ or ‘very low’.

The modal response in both provinces was ‘very low’, this was chosen by 48% of respondents in Quebec and 59% in BC. However, another 22% of Quebeckers and 15% of British Columbians said the chances were ‘very high’ or ‘somewhat high’.

The scourge of populism

There’s other stuff in the book, like whether populism might work better in the developing world, where Hanania brings in El Salvador’s Bukele and Argentina’s Milei. Here, I wholeheartedly agree with what Scott Sumner laid out in his review. I have nothing to add.

I think we really do need an expansive definition of populism that goes beyond the standard poli-sci term. In that vein, if you’re interested in how populism fails when it comes to its critique of mainstream economics, check out this post of mine. And click the one below if you want a takedown of populist ideas as applied to anti-capitalism:

I mean this first in the sense that I grew up left-wing and hung out virtually exclusively with left-wing people my entire life, even becoming a Marxist activist in my twenties, being affiliated to a formal group of young Marxist intellectuals, and being part of a fledgling Slovenian socialist party. (The party has since entered parliament and eventually helped form a coalition government).

But moreover, like Dan, I became somewhat less stereotypically left-wing over time for cognitive reasons similar to his own: taking Darwinian cynicism seriously, learning textbook economics, finding out about public choice theory and behavioral genetics, encountering “progress studies” or “the new optimists” like Steven Pinker and Hans Rosling, and just generally reading outside Marxism and being interested in empirical evidence.

The analytical Marxist head start

I’ve mentioned this before, but it bears repeating that the flavor of Marxism I had been most attracted to was, surprisingly, among the first important stepping stones on my way toward becoming more pro-capitalist and eventually shedding the label. Analytical Marxism and especially the Marxist historical sociology associated with Robert Brenner made me realize early on (at least latently) that capitalism:

1. is an immensely productive and dynamic system, which makes even ordinary people better off than in the past (directly through the market as well as indirectly through all the redistribution that it allows for),

2. has an internal economic motor all its own, due to which this enrichment is not at all dependent on outside plunder, colonialism, or slavery,

3. works this way primarily through the rational logic of material incentives, not ideology and mass manipulation,

4. makes possible the various downstream social benefits (on account of massive economic development), such as reducing the likelihood of wars of territorial conquest and political unfreedom/authoritarianism.

I won’t be unpacking these in any detail here (but parts of this post can help). I’m also sure some might paradoxically object either that all of that is already in Marx and so it shouldn’t threaten socialist ideology in the least bit or that what I’ve said is not really Marxism but instead a bastardized, bourgeois, de-radicalized version of Marxism, so I was never really a Marxist and hence my “deconversion” is not surprising. I think neither objection really works, but that’s an argument for some other post.

Here’s a Brennerian gem of analytical Marxist insight into why and how (agrarian) capitalism works, from his famous essay “Property and Progress: Where Adam Smith Went Wrong,” so you get the hang of where I was coming from:

As you can probably see, this was no ordinary, unsophisticated, dogmatic, run-of-the-mill Marxism. Here’s another example of what kind of a weird Marxist I aspired to be, courtesy of Erik Olin Wright:

There are four specific commitments that I think characterize Analytical Marxism and justify considering it a distinct "school" of contemporary Marxist thought:

1. A commitment to conventional scientific norms in the elaboration of theory and the conduct of research.

2. An emphasis on the importance of systematic conceptualization, particularly of concepts that are at the core of Marxist theory. This involves both careful attention to definitions of concepts and to the logical coherence of repertoires of interconnected concepts.

3. A concern with a relatively fine-grained specification of the steps in the theoretical arguments linking concepts, whether the arguments be about causal processes in the construction of explanatory theories or about logical connections in the construction of normative theories.

This commitment to elaborating the details of arguments is reflected in one of the hallmarks of Analytical Marxism: the use of explicit, systematic models of the processes being studied. The nature of these models may vary quite a bit, from formal mathematical models to less formal causal models. But in each case there is a belief that the possibility of theoretical advance is enhanced when we are able to generate systematic explicit models of the processes under study.

4. The importance accorded to the intentional action of individuals within both explanatory and normative theories.

However, what started pushing me more over the edge, intellectually speaking, were two additional and important occurrences.

Facts about progress and trouble with democracy

First, in February 2018, I read Steven Pinker’s Enlightenment Now. I felt somewhat uncomfortable hearing his name, but I read the book nevertheless. I did so partly because I felt obliged – we all liked to reference how Marx had to read the greatest bourgeois economists of his day before coming up with his theories. But partly also because I saw him, i.e., Pinker, as a theory-free empirical scribbler whose data I could use within my own framework. And lastly, I felt at least some of his points would be easy to uncover as wholly ideological and wrong.

Now, maybe I was naive and my haters will have a field day with this, but as I was turning the pages, I almost immediately felt the force of most of the statistical facts and figures he presented. It was like magic; it felt undeniable. Some of them I kind of knew already, and with many others I just felt it’d be tendentious (and pretentious) of me to try to wave them away or discredit them in some contrived way. I browsed through Rosling et al.’s Factfulness sometime after and felt similarly … perturbed? (Then again, even a devout socialist like Nathan Robinson is willing to admit he agrees with “perhaps 80 percent or more” of the statistical facts in Pinker’s book. And he says this in a review titled “The World’s Most Annoying Man.” Maybe I was not just being naive.)

I started asking myself more and more: if capitalist modernity is so clearly associated with radical reductions in poverty and hunger (and violence to an extent, though that’s more complicated), while it immensely boosts median income, health, and political freedom and human rights, what exactly is the big, revolution-demanding gripe I have with capitalism? In today’s capitalist societies, even inequality can’t be said to have exploded compared with pre-capitalist times.

Second, in July 2018, I almost randomly discovered Jason Brennan’s Against Democracy and, as I was reading it, again found it hard to argue against the facts about political ignorance he presented. What struck me even more was the rational-choice and incentives-based framing of it all (rational ignorance/rational irrationality), which I’d previously found among the most convincing aspects of analytical Marxism and Brenner’s account of capitalism. I couldn’t controvert the facts Brennan was citing in Against Democracy, and I had already been naturally attracted to the very same theoretical framework he was using (albeit as a non-Marxist in his case). What’s a man to do?

I also realized immediately that if democratic decision-making is in principle this problematic, it might spell disaster for any (socialist) economic system that is fundamentally based on the same logic.

Of course, if my Marxist friends are reading this, some of them are rolling their eyes and screaming “market socialism, dude, market socialism!” Right, but I was never a market-socialist kind of guy. Marx convinced me that the dreaded “law of value” prevails in any system based on market competition and so virtually all the issues he has with capitalism will reemerge in a “socialist” system based on market competition – regardless of who the owners are. Competition is at the root of most problems in capitalism for Marx, so I never could get behind market socialism, where worker collectives compete against each other. For me, this was just another form of capitalism. So, as far as alternatives to capitalism go, I was much more on board with non-market democratic decision-making as far as what to produce and how much, and how to allocate it all.

A crazy thing I remember liking was Albert and Hahnel’s deliberative “parecon,” or participatory economics, model. As I was rethinking things, I thought (to paraphrase Rick and Morty): isn’t parecon actually just the capitalist market with extra steps? The extra steps being endless cycles of “Iteration Facilitation Boards” meetings (look it up) and workers/consumers’ councils fruitlessly going back and forth in deliberation over what to produce and where people should be employed. Holy hell.

Anyway, I realized that democracy is not a panacea alternative to markets and it even has its own set of structural problems that are hard or impossible to overcome.

Institutional economics and public choice: “Marxism” by any other name?

After discovering Brennan, I almost immediately (August 2018) found out about Daron Acemoglu and James Robinson’s Why Nations Fail. I had been scouring the internet for (critical) discussions on Brennan’s book and found this NDPR discussion forum over at the Pea Soup blog. You can still see, even today, Brennan mentioning Why Nations Fail in the comment section, which I devoured (the comment section and the book, that is). This was how, after years of pronouncing on the economy and economics, I finally made some effort – for the first time – to delve deeply into mainstream economics, including actual textbooks such as Cowen and Tabarrok’s Modern Principles, along with intros by Krugman, Weil, Mankiw, and others.

As I did that, I realized that social dilemmas or collective action problems are pervasive in human affairs and have nothing specific to do with capitalism. In principle, they are just as likely to reappear in socialism. The power and importance of incentives and institutions (as opposed to hopes of people just being more solidaristic and other-regarding once capitalism is replaced) was reaffirmed and boosted in my mind. I also finally grasped Hayekian insights about prices and information.

I was totally hooked by Why Nations Fail. That’s partly because Acemoglu and Robinson cite Brenner (the analytical Marxist – don’t confuse Brenner and Brennan) at a key point in the book.1 So I didn’t feel that much dissonance. But also because they just generally use the rational choice approach I loved in analytical Marxism, and broadly say “capitalism” works in the same way I had already believed it works (the points 1–4 at the beginning of this post). So when they added other tidbits about how “extractive” and “inclusive” institutions work – including the dynamics of social power and inequality more generally – everything sort of clicked, even if I had to revise away some of my Marxist beliefs.

As a result of reading Brennan’s Against Democracy, I naturally found out about Bryan Caplan’s The Myth of the Rational Voter and Ilya Somin’s Democracy and Political Ignorance. Relatedly, partly as a result of reading these books, partly due to trying to get my hands on as many mainstream neoclassical economics textbooks as possible, I happened upon public choice theory at that time. Here again, the rational-choice, incentives-based theorizing made a whole lot of sense to me, and it also strangely linked up with some of my preexisting Marxist ideas. The big one is how capitalists and politicians can end up in bed with each other, screwing over the public.

Rent-seeking, the principle of concentrated benefits and dispersed costs, regulatory capture, the behavioral symmetry thesis: all of these made a lot of sense, had practical, real-world evidence behind them, and were not that alien to a Marxist who knew his “state theory” well. Of course, capitalists want to earn as much money as possible, public be damned, and politicians want to gain as much power and status (but also money) as possible, public be damned. It’s only (well, primarily) the institutional frameworks and the logic of incentives that may or may not channel this default, presumed selfishness into good social outcomes, such as quality products at low prices or quality public schooling and no (or few) political affairs and scandals. Doesn’t matter if we’re talking about the economy or politics. Doesn’t matter if we’re talking about capitalism or socialism.

But if so, I thought, this is going to make any kind of a possible alternative socialist future quite a bit more complicated. If we’re going to tear down the existing system and put something else – and ostensibly completely different – in place, we’d better make sure the incentives are going to be sensible. We can’t rely on the goodwill of anonymous strangers holding office (or wielding economic power) for society to actually be taken care of by them.

Discovering Bruce Bueno de Mesquita et al.’s The Logic of Political Survival and its breezier cousin The Dictator’s Handbook was also a huge public choice eye-opener. Structurally speaking, politics and ruling is at least typically constrained by certain tendencies which appear everywhere, regardless of how good a person is or whether we’re talking about capitalism or socialism or anything else.

As I put it in my 2021 book Rational Choice and Democratic Government, distilling Bueno de Mesquita and others:

The first rule of politics is: get power and hold on to it for as long as possible. Even politicians who are primarily moved by the self-sacrificing motive to do good for others cannot – must not – go against this first rule. One has to possess power before one can pursue the goal of promoting the public interest. Now, this by itself is not necessarily problematic. But it does start to bring up issues when we consider the second rule of politics, which flows directly from the first: get key supporters on your side. Even an exceptionally civic-natured politician cannot achieve power or, if she has already got it, hold on to it for very long without strong support from other people.

Here, two issues immediately open up.

First, even if a candidate for the leader is herself exceptional, the majority of her key supporters are probably ordinary, self-interested people. And what can a potential leader do to secure the loyalty of such people? What can she offer them in exchange for their support? Regulatory capture, unreasonable subsidies, packing state institutions with one’s own (but not necessarily meritorious) supporters, enacting policies that do not make sense in the medium- to long-term, etc. Everything comes in handy.

Second, the brutish rules of the political game tend to eliminate, or filter out, the exceptional, civic-minded, altruistic political candidates, and instead structurally attract those who are willing to do what it takes to come to the top and secure power for themselves. Two specific mechanisms are operative here. Either most of the exceptional people are turned away from politics from the start or, barring that, they tend to get replaced in time with other, less exceptional people by their own disgruntled supporters. Structurally, at least mild government failure should be pervasive.

There’s so much more I’d like to say about this, including answering why liberal democracies nevertheless work pretty well, but it’ll have to wait for some other post.

The empirical record of “economic freedom”

There are also just lots of other things to mention that happened to me in the period between 2019 and 2021, but let me focus only on one last big one. In the second half of 2019, I really got interested in empirical measures of “economic freedom,” which can at least partly be used to statistically assess the correlates of being more (or less) capitalist as a society. The most widely used measure in the academic literature is Fraser Institute’s Economic Freedom of the World index. It’s composed of 5 general components, which are in turn composed of many subcomponents. Whether the index is “really” measuring capitalism (or neoliberalism, or market society), to what extent it’s doing that, and what all the good and bad correlates are is a topic for another post.

But I can say that at least some of the index’s components (and subcomponents) are clearly very relevant for discussions about different economic systems and institutions. Size of government, freedom of trade, and regulation are probably the key ones. Under size of government, you can look at transfers and subsidies, or top marginal taxation, or government ownership of assets. Freedom of trade is self-evidently neoliberal. Regulation includes, respectively, subcomponents such as the regulation of business, labor markets, and credit. Those are themselves made up of interesting categories, such as collective bargaining, ease of hiring and firing, and so on.

So, when I found out, over the years, that we have literally a thousand+ empirical papers (most of which are correlational, but some of which can get us plausible causality), and that a lot of them, though definitely not all, show economic freedom to be associated with nice social outcomes, like growth, income, and peace, this further nudged me in the pro-capitalist direction. (The jury is still sort of out on whether economic freedom boosts inequality, by the way! And there’s lots of nuance to unpack with economic freedom in general if we’re not being ideological, which we shouldn’t be.)

But at this point, am I even left-wing still?

Here are just some assorted reasons why I think I’m very recognizably left-wing even as I’ve become more pro-capitalist over the years. Of course, there’s also the simple fact that I believe myself to be left-wing, and political typology tests and things of that sort consistently place me left of center. That’s at least some evidence.

1. At the highest level of abstraction, and speaking in metaphysical terms, I don’t believe in free will, or in social hierarchies being somehow “natural.” Moreover, I just personally find power hierarchies distasteful. I detest deference to power and status titles. I literally get somewhat uncomfortable when I’m forced to attend academic ceremonies and there are all these titles and procedures and formal fancy attire with golden chains and medals that signal where everyone is in the status hierarchy. Bleh. I couldn’t care less. It’s just not my thing.

   Now, I get that capitalism has hierarchies: so am I hypocritical? I don’t think so. I see them at best as a necessary evil that should be begrudgingly tolerated so we get all the other, actually good things out of capitalism. Moreover, the alternative society-wide systems that have been tried also have hierarchies, which are often even bigger and uglier, and there are little to no benefits for the common person from them.

   I know I should say more about my disbelief in free will, but that gets messy quick and so should wait for another post. Suffice it to say that I find proper, libertarian free will verging on the unintelligible. My thoughts and actions are either caused, in which case they’re determined by preceding factors, like other desires I have or my character and the environment, which I didn’t freely choose. Or they’re uncaused, which seems impossible or in any case precludes the possibility of “me,” whatever that is, choosing them.

   As for compatibilism, I think it tends to be a cop-out, because it tends to define free will as simply “will” or “intentional action” or “action originating from within the person” or “incentives-responsiveness.” To my mind, the compatibilist notions have nothing to do with free will as we usually think about it (“could have done otherwise under the exact same conditions”) with all its moral baggage surrounding just deserts and true sovereign control. Listening to Christian List and browsing through his book was interesting though.

   In any case, on my view people are not ultimately self-made in any deep sense, so I find it hard to take strong claims about moral desert, punishment, or entitlement at face value, in contrast to how right-wingers tend to reason. I’m much more inclined to think in terms of luck, circumstance, genetic determinants, social conditions, and so on than in terms of who “deserves” to be above or below whom.2

   (Before you come after me, especially philosophers in my crowd, believe me when I say I know it’s much, much more complicated than this. Also, even if I’m wrong, I’m just demonstrating to you that I don’t believe in free will, not that I’m correct in doing so.)

2. I think governments can really help where markets fail in certain important cases. Sometimes, public choice theory simply proves too much, given the empirical record we have of how governments in liberal democracies work in practice, i.e., quite okay. Aside from all the textbook cases about externalities and so on, I’m also referring to things shown in books like Lane Kenworthy’s Progress for the Poor. Some people simply cannot participate in the market due to illness, or old age, or whatever; others are generally unlucky through no fault of their own (again, in the grand scheme of things, everyone is just more or less lucky or unlucky on my no-free-will view). You absolutely need the welfare state (see this post of mine). Perhaps a “free-market welfare state” like Denmark or Ireland?

   

3. Relatedly, and despite being overall quite pro-capitalist, I don’t think moderate government ownership is bad. The evidence just isn’t there. The same is true of corporate taxation. Corporate tax cuts don’t boost economic growth. And we don’t need higher (or rising) inequality for the sake of getting more economic growth. The theoretical case for the idea is plausible, but the evidence goes against it. (Not that inequality is nearly as bad as people make it out to be.)

4. I think it’s silly to say, as right-wingers are wont to do, that if a worker accepted a contract willingly, without being physically coerced, there’s simply no exploitation that could be going on in any reasonable sense of the word; case closed, nothing to see here. In fact, there’s just so much else that can be going on. There absolutely is, in certain labor markets and to a degree, market power and concentration which makes some workers vulnerable vis-a-vis capital owners and unfairly drives a wedge between how much they’re paid and what they’re contributing to the firm. You don’t have to buy Marx’s labor theory of value to recognize some workers are exploited to an extent, even if they’re consenting.

5. Notwithstanding the recent takedown/non-replication of Chenoweth and Stephan’s 2011 book Why Civil Resistance Works, I think non-violent civil resistance is overall good and good for social progress. Student movements, protests, sit-ins, anti-war rallies, labor movements and unions might be overrated by some on the left, but they were historically important, in part. Does that mean I think the amount of working time got basically slashed in half over the past century and a half solely because of unions, like many leftists argue? No, economic growth and labor-saving innovation was probably more important if you go by things like economists and/or historians’ consensus; but yes, unions and all the rest of it also played a role, especially in Europe. Also, when workers and students mobilize, more political freedom typically follows – though the effect size is not massive.

6. I’m naturally cosmopolitan in outlook and can’t emotionally connect with nationalism and ethnic-based categorizing. Like, not at all. (Let alone connect cognitively, given its bloody history and tendency to be exclusionary and brutal and just anti-human.) Obviously, I despise racism like any normal person, but even some studies of and discussions on “cultural persistence” make me a bit queasy (see my two empirical pushbacks). I also have relatively strong utilitarian intuitions, so I can’t get how people overridingly use disgust and aesthetic preferences to form their moral opinions on, e.g., homosexuality, people with different skin color, etc. It just seems so obvious to me that people should be left alone in how they behave, what they look like, and what they believe in, as long as they are not physically or generally unfairly impacting on anyone else who’s not consenting.

7. I was not brought up religiously. I’m also, I think, by disposition virtually incapable of entertaining supernatural/religious claims, and I emotionally and aesthetically strongly dislike organized religion. Of course, I’m aware of its likely sociological upsides (1, 2, 3), but I can’t personally relate – on literally any level. I think it’s partly the stale hierarchy of it all, the (to me) weird rituals and tacky aesthetics, the bullshit substantive beliefs, non-evidentiary reasoning, weird sex dynamics (both in terms of gender and religion’s strictures on sexual relations), and many other things. But I should stop because I don’t want to offend, though I’ve probably already done that. Even though I can’t “get” religion, I’m wholly tolerant of its existence and even more tolerant, of course, of its practitioners.

   What can I say, I’m a non-US, European atheist, secular person. You know what we are like.

There are more reasons I could list, but that can wait for another post. Reading the points above, I’d be shocked if people (still) sincerely questioned my center-left, social-democrat leanings. You can absolutely be pro-market/pro-capitalist and left-wing at the same time.

Gentle cooperation is sometimes sad to be the perfect alternative to brutal market competition. Instead of selfishly exploiting and predating on each other – trying to beat each other – why don’t we try to just cooperate? Everyone could work together for mutual, collective benefit. Capitalism with its dog-eat-dog market competition is a nasty, distrustful world, while some sort of cooperative alternative would be nice and pro-social. That’s what the analytical Marxist G. A. Cohen tried to argue for in his little book Why Not Socialism?

In reality, the funny thing is that market competition actually tends not to stand in the way of cooperation. In fact, market competition is precisely a form of cooperation, and something that we use to generate positive-sum, pro-social outcomes. Not every single person is necessarily a winner, of course. When, as a consequence of competition, wasteful, unproductive companies go bust, they are losers, while the rest of society (other owners, workers, and consumers) wins. But that’s exactly the pro-social, “cooperative” result we want as a community.

Here’s how Joseph Heath puts it in Cooperation and Social Justice, with reference to a central theme in Cohen’s book:

[I]f one begins with the moral principles elicited by Cohen’s camping trip thought-experiment [i.e., reciprocity, scarcity-responsiveness, efficient allocation, need-guided production, etc.], but then takes seriously the scaling-up problem [i.e., we are not 4 friends on a camping trip, but an anonymous society of millions of strangers], focusing on institutional arrangements that have an actual track record of success at promoting large-scale cooperation, and in particular, at instituting a complex division of labour, it is not difficult to make the case that a market economy constitutes a realization of those principles.

It is not a direct realization – and therefore will be experienced by participants as morally counterintuitive – only because of the kluge-like way in which these institutions bring about the desired cooperative outcome. … The competitive market arrangement is … morally counter-intuitive, because it achieves the cooperative outcome only indirectly, as a by-product of a system of adversarial relations.

Market competition is an adversarial system of social cooperation, and it is a way of generating what could be termed “cooperative outcomes.” Okay, that much should have already been clear if you’re a thoughtful and educated individual who had learned to dismiss emotional knee-jerk reactions and hunter-gatherer intuitions, which are simply misleading in our modern world.

But now, if that’s true, wouldn’t it be neat if we could measure whether, under the influence of market competition, people themselves also really become more pro-social and cooperative? Well, there’s a cool 2018 paper which contains multiple studies with different methodologies from different countries that show precisely that.

Patrick Francois and colleagues have shown that:

Increased competition across firms exposes subjects to increased group beneficial behavior on the part of their co­workers and increases their own such behavior. In competitive markets, firms unable to elicit this cooperative behavior are likely to be outcompeted by firms that are more successful in doing so, leading to the proliferation of firms exhibiting cooperation.

Workers in these settings experience, and themselves internalize, more cooperative norms. They then report more positive answers to the generalized trust question, which explains the cross-­sectional and panel correlations we have reported here.

Let me take you through their studies. The main but not the only outcome is generalized trust, measured with the familiar survey-style question about whether most people can be trusted or whether one should be careful in dealing with others. This is obviously just an imperfect proxy for how people think about the likely moral behavior of others, and perhaps also of how they themselves would act in a risky social situation. But if trust (and cooperation/contributions in economic games) goes up, this might be some indication that society is more cooperative and that pro-sociality is spreading.

Cross-sectional US survey data: more competitive industries, more trusting workers

The first empirical study uses US survey data from the 2004 General Social Survey workplace module. The authors connect each employed respondent to a measure of how competitive that respondent’s industry is. Industries are treated as more competitive when sales are less concentrated among the largest firms. They then ask whether workers in more competitive industries are more likely to give a trusting answer to the generalized trust question. The answer is yes! After adding many controls for personal traits and workplace conditions, workers in more competitive industries still report higher trust. The figure below shows this. The authors are of course cautious here, because the evidence is merely cross-sectional. It shows an association, but it cannot prove that competition caused trust to rise.

After banks got deregulated, competition and trust went up

The second study is designed to get closer to causality by using US banking deregulation. Beginning in the late 20th century, states changed rules that had limited interstate banking. These reforms increased credit availability, which prior research connects to more firm entry and greater market competition. Because the timing varied across states, the authors are able to compare trust before and after deregulation within states while controlling for many confounds (state differences, time, individual characteristics, and so on).

The event-study graph (below) shows two things: (1) that firm entry and exit rise after deregulation (so more competition), and (2) that trust begins rising a few years after the reform. Importantly, there are no clear preexisting trends in trust before deregulation. Their preferred estimates imply a yearly rise of about 1.4 percentage points in the share of people reporting trust after reform.

(The authors also consider whether something else might be going on. For example, deregulation could have affected income growth, migration, or other state-level economic changes, and those could have affected trust independently. Supplementary analyses examine these alternatives and are in line with the competition idea. So the argument that deregulation made local economies more contestable, which increased the importance and prevalence of cooperation within firms, and so residents then started reported higher generalized trust seems plausible.)

Tracking trust of the same German workers over time as they shift industries

The third study relies German panel data, focusing on survey waves from 2003, 2008, and 2013. This one tracks individuals over time and uses (the same) workers who move across industries. The authors match each person’s industry to a competition measure based on the Herfindahl-Hirschman index, where more dispersed revenue across firms means higher competition. The question is whether a person who moves into a more competitive industry becomes more trusting, relative to their own earlier trust level. And the answer is again yes! The figure below shows that increases in industry competition are associated with increases in reported trust among movers.

This is very cool because, remember, it follows the same individuals instead of only comparing different people at one point in time. So we should be less concerned about the possibility that naturally trusting people simply choose competitive industries. Instead, we see that as a person moves to a more competitive industry, they grow more trusting. If they move to a similarly competitive industry, they remain as trusting as before, and if they move to a less competitive field, they become less trusting! Also, the authors again report checks suggesting that workers who move to more competitive industries were not already on a path of rising trust before the move. They also say the result does not appear to be explained by higher income growth after moving.

But why? Turning to experiments

The paper then explains the theory linking these findings. The authors distinguish between two ways competition could be important. The first is a relatively immediate incentive effect. When firms face stronger outside pressure, poor cooperation inside the firm is more costly, because lower effort or coordination failures can translate more directly into loss of business. Under such conditions, cooperative norms may become easier to sustain.

The second mechanism is a longer-run selection effect, where we’d say that firms which manage to create effective cooperative norms might do better and expand (and perhaps be copied), or attract workers, while firms that fail to do so may shrink or disappear. This is the phenomenon of cultural selection (as in the most alluring mechanism of cultural evolution). In this account, competition changes which norms are more viable and which organizations are more likely to persist.

The laboratory experiment the authors ran is presented as a way to examine the mechanism that’s actually at play here. Subjects in France played repeated rounds of a two-person public goods game, with new anonymous partners each round. In the noncompetitive version, each person chose how much of an endowment to contribute to a shared account. Contributing helped the pair overall but was individually costly, so a purely self-interested player would contribute nothing. In the competitive version, however, the pair’s payoff from the shared account depended on whether their pair contributed at least as much as another randomly assigned pair. This created a setting where group cooperation could be more attractive because the group’s success depended on matching or beating another group. After the game, subjects answered the generalized trust question.

And we see the same result again. Contributions were higher in the competitive condition than in the standard public goods game, and they did not decline in the same way over repeated rounds. In the control condition, contributions generally fell over time. In the competitive condition, contributions started higher and remained higher. More pro-sociality and cooperation.

The authors also find that people exposed to more cooperative partners and competitors in earlier rounds tended to contribute more in later rounds, even though they were not paired with the same people again. It’s wasn’t all just about direct payoff calculation; instead, there was also a kind of cultural-evolution like learning from observed group behavior. Subjects seemed to adjust their own later conduct after seeing that others, including rival groups, had contributed more.

Finally, the experiment finds that subjects in the competitive condition were more likely to report generalized trust afterward. In the noncompetitive condition, a little over half gave a trusting response, while the competitive condition raised the probability of a trusting response by about 15 percentage points. The authors argue that the trust increase was most connected to those participants whose contributions rose over the experiment after encountering high-contributing partners or competitors. They interpret this as evidence that competition exposed some subjects to more cooperative behavior, led them to behave more cooperatively themselves, and then made them more willing to say that people in general can be trusted. So, competition may contribute to prosocial norms not just through material incentives (which is obvious and cool on its own), but by making cooperative group behavior more useful and more likely to spread. There might be some cultural selection going on.

So, yes, certain forms of economic competition are not the obverse of cooperation but are forms of cooperation that, in turn, lead to cooperative outcomes.

Yes, there is a vast literature that has accumulated on this topic since the 2000s. But I think Matt’s reading of it might be biased. Of course, if he’s only saying, minimally, that ordinary people (that is, poor or middle classes) are somewhat more likely to perceive their governments as being unresponsive in comparison to richer people, and that there is some reality to this perception, I have no issue with him. But if the idea is that liberal democracies, such as the US and other advanced societies, are virtually plutocratic, hollowed-out non-democracies where ordinary people have close to no say compared to rich folks, are locked out of the system and just generally not represented, this is just emphatically not true.

Here’s a relatively brief list of the key points that I think should be made in any non-biased review of the political science literature Matt is referring to. See for yourself if any of the points surprises you, given your own priors and the image that popped into your head after reading Matt’s X post.

(By the way, he read the post and wasn’t wholly pleased with my framing of him. Here’s his response for context. I don’t want people to get the wrong impression: Matt is decidedly not saying that currently existing liberal democracy is a plutocratic sham.)

1. The responsiveness gap exists, though there are dissenting studies as well. Everyone knows about the infamous 2014 Gilens and Page study. But that’s nothing, really. There are so many more studies than just “Gilens and Page.” And many studies in this area (most, actually, not just many) have indeed found that governments are more responsive to the preferences of wealthier citizens than to those of poorer ones. This holds both for the US and elsewhere; mostly Europe, but globally as well in some larger samples that we have.

However, a non-trivial number of papers also show that either there’s no such gap in policy responsiveness, or find the bias running in the opposite direction – that is, toward lower-income groups. The literature is not uniformly damning, so let’s not present it as such.

That said, I think it’s fairly clear that the overall weight of evidence does suggest a gap probably exists. This should just be accepted. Here’s how a recent state-of-the-art review puts it:

[M]ost studies find a positive income gradient in political responsiveness, indicating that political outcomes respond more strongly to the preferences of higher-income groups. …

Some studies find relatively equal levels of political responsiveness [1, 2, 3], some find that political outcomes best reflect the preferences of the poor [4, 5], and still others find that political outcomes respond most strongly to the preferences of the middle class [6].

This divergence in results is often downplayed in scholarly and public discussions of the literature, yet it stands out clearly here.

Now, the obvious and important next question is what its magnitude is and what is actually driving it.

2. The gap is quite modest. On the matter of magnitude, the evidence gives reason for considerable restraint in how we characterize the problem. Persson and Sundell, working with a sample of 30 European countries, recently found that the policy preferences of low-income citizens were congruent with actual policy outcomes 53.66% of the time. For middle-income citizens, the figure rose modestly to 55.12%, and for high-income citizens to 57.06%. The gap between the bottom and top of the income distribution, in other words, amounts to roughly 3 percentage points – which is why the authors titled their study “The Rich Have a Slight Edge” and not anything more dramatic. If taken at face value, this is no obvious demonstration of robust, undeniable political oligarchy.

Schakel and colleagues, examining a larger sample of 43 countries, arrived at similar numbers. The least advantaged social group, defined by both low income and low education, had preference-policy congruence of 52.7%, while the most advantaged – high income and high education – had congruence of 57.6%.

3. But perhaps a “better measured” gap would be bigger? There is a structural reason to suspect that these estimates may actually understate the true degree of representational inequality. That’s because the preferences of people at different points in the income distribution tend to be substantially correlated. Rich and poor citizens frequently want the same things, or at least do not disagree as sharply as one might expect. This is itself an important empirical fact that tends to be underemphasized in discussions about political inequality, but it also has a methodological implication. Namely, when the groups being compared hold similar preferences much of the time, there is a natural ceiling on how unequal representation can appear to be.

As one formulation of this point in the literature goes, even if the rich are the ones effectively driving public policy, the preferences of lower-income groups will still be registered as “represented” much of the time – simply because their preferences happen to coincide with those of the more influential group. Congruence figures, in other words, can look equal even if the underlying process of influence is quite unequal. (A related statistical problem is that if the preferences of different income groups are highly collinear, as they are, multivariate analyses designed to isolate the independent effect of each group’s preferences become unstable and potentially uninformative. But let’s leave that for another day.)

4. Okay, run a more decisive test then. Branham and colleagues devised an approach intended to sidestep this problem by restricting the analysis to cases of disagreement between income groups. By focusing only on instances where the preferences of high- and low-income citizens diverge, we have at what amounts to a cleaner test of differential responsiveness. But it turns out that when examining only these contentious, high-stakes cases of real disagreement, the rich in the US do not dominate overwhelmingly at all. The probability of the rich getting the policy outcome they prefer, as against the middle class, comes out at roughly 53% versus 47%. This is something, but again no oligarchy worth its salt would be this weakly tilted in favor of the elites.

5. Lower income, less information? A further complication for any simple account of plutocratic capture has to do with the role of political information and how it interacts with measured preferences. Elkjær and Iversen have pointed out, through simulation, that if wealthier citizens are better informed about politics and policy – which the evidence suggests they are – then standard statistical models studying changes in policy over time can produce the appearance of rich-only influence even if that is not true. What this means, methodologically, is that studies of short-term policy changes may be telling us more about differential information than about differences in political power. Elkjær and Iversen’s proposed remedy is to study long-run levels of policy instead of short-term ones, since the former is less susceptible to this form of bias.

When they use this approach on a sample of 21 advanced democracies, the results are again complicated. Short-term changes in spending do appear to track the preferences of the rich, but the long-run levels of policy appear to be set by the preferences of the middle class. As they themselves conclude:

middle-class power has remained remarkably strong over time, even as inequality has risen. The authors conclude that the rich have little influence on redistributive policies, and that the democratic state is apparently not increasingly constrained by global capital.

Elkjær’s subsequent work using Danish data shows much the same. Denmark displays the same pattern of differential responsiveness seen elsewhere, but the result appears to be driven primarily by the fact that wealthier Danes hold preferences that are more closely aligned with standard macroeconomic policy stances. As he puts it:

unequal responsiveness reflects asymmetries in information about fiscal policy across income classes, as opposed to being a structural bias inherent in capitalist democracy.

6. It might be more about education and culture than income and economic issues. Much of the literature on political inequality has been organized around income as the primary dimension of stratification, but a (slowly) growing body of work suggests this focus may be missing something important. There are only a few studies here, so we should be careful. But, as Schakel and colleagues have found, education appears to have a stronger effect on policy representation biases than income does, when both are examined. Moreover, the largest and most consistent gaps in both congruence and responsiveness appear not on economic policy questions but on socio-cultural ones. And on the economic dimension, responsiveness to lower-income citizens is actually relatively okay. It’s just on the socio-cultural dimension – questions of immigration, identity, social liberalism, and the like – where lower-educated and lower-income citizens find their preferences least reflected in policy.

If this is true, unequal political representation is not necessarily so much about redistribution or economic policy, as one would naturally think, but much more about the cultural and social agenda, where the preferences of the less educated diverge most from those of the educated professional class.

7. And it’s actually about Europe more so than the US! One would think that if any capitalist liberal democracy is being warped and skewed toward oligarchy, it’d be the US, perhaps on account of its weaker labor union tradition, higher inequality, corporate power, lesser regulation, and so on. But if anything, the reverse seems to be the case, which is very weird. Here’s how the systematic review mentioned before reports this puzzling result:

published research suggests that differentials in responsiveness are weaker in the United States compared to other developed democracies. … Moreover, studies of the United States are more than four times as likely as comparative studies to observe a pro-poor bias …

Considering that the countries studied in the comparative literature (Canada, Denmark, Germany, Netherlands, Switzerland, and developed democracies more generally) have lower levels of economic inequality, stronger left parties and labor unions, higher turnout rates, and less money in politics than the United States, these results are puzzling. They go against conjectures that the United States might be a unique case regarding political inequality. They also challenge long-standing theories in comparative political economy, which explain the more egalitarian political outcomes in Europe partly as a function of political and economic institutions, a stronger working class, and greater political influence of low-income groups. …

Why social Europe appears to experience more unequal representation than liberal America is puzzling and warrants further scrutiny.

Next time you or your friends do the whole “Gilens and Page (2014)” spiel, remember these points!