Thread by @edenhofer_jacob

In light of the debate about the extent of inequality in the US, I thought it might be helpful to summarise how political scientists think about the demand for and supply of redistribution. The literature is vast -- so this 🧵is intended only a bare-bones introduction. /1

For some empirical background, see this thread by @MortenStostad (though not every graph is uncontested). In general, inequality has risen in most advanced industrialised democracies since roughly the 1980s -- though the extent is debated.
Should we ... /2

https://x.com/MortenStostad/status/1780873004224045442

be surprised by this?
Yes, at least if we follow the logic set out by Meltzer and Richard (1981). The MR model makes two kind of assumptions: economic (drives demand for redistribution) and political (drives supply). Let's examine these. /3
jstor.org/stable/1830813

For the economic assumption, I'll follow @ATabarrok's lecture notes.
Disposable income is given by:
c(x)=(1−t)x⋅n(r)+r, where x denotes an individual’s productivity, n the hours worked, t the tax rate, and r the amount of redistribution received.
Note: n(r) is decreasing in /4

r which places an upper bound on r. There are (non-linear) efficiency costs to taxation [in the paper, these are determined endogenously via labour supply].
All tax revenue is solely used for funding r, and the budget must be balanced. Since everyone is taxed at the same rate /5

and the tax revenue is recycled on an equal per-capita basis, each person receives a lump-sum payment equal to the mean overall income times the tax rate.

➡️Key implication: on net those with above-mean income will see their incomes reduced as taxation rises, with the reverse /6

holding for those with below-mean incomes.
Those with at least mean income will prefer a tax rate of zero. Those with inc. below the mean prefer a positive tax rate, though this is decreasing in income because of the efficiency costs of taxation (see by fig Iversen/Goplerud) /7

Political assumptions:
Voters have single-peaked preferences around their most-preferred tax rate (see fig. by Felix Munoz Garcia). The tax rate is the only dimension of political competition.
➡️ With single-peaked preferences, the median voter theorem holds: The electorate /8

will choose the median voter’s ideal tax rate (and parties will supply the later because they converge to the median voter's position).
What is the MVT's ideal tax rate? The latter depends on the shape of the income distribution (or the distribution of productivity). /9

If the income distribution is symmetric, the mean income equals the median income. Those with at least mean income, here also the median voter, prefer a tax rate of zero.
Result: no redistribution.
If, however, the income distribution is skewed to the right, the mean income /10

is greater than the median. MVT prefers a positive tax rate. Result: some redistribution. As income inequality increases, the income distribution becomes more right-skewed (i.e., more extremely high incomes). ➡️with greater inequality, the difference between the mean and /11

median income increases and by the median voter logic, so does the demand for higher taxes or redistribution. [The amount of redistribution is also inversely related to the efficiency costs of taxation.]

"Redistributive puzzle" [debate how puzzling this is]: The MR model /12

predicts that the MVT demands higher redistribution as inequality rises. Despite the increase in inequality, some/many democracies have not increased redistribution to neutralise pre-fisc inequality. Indeed, the rich seem to be better represented. /13
cambridge.org/core/journals/…

In general, there are two types of explanations for this "puzzle": demand-side ones (having to do with the demand for redistribution) and supply-side ones. I list some (!!!) of them below. Let's work through them. /14

The idea is that, for some reason, ppl's demand for redistribution is lower than their objective, material circumstances (position in the income distribution) would lead us to predict.
Three popular candidates:
1. Misperceptions about one's own position in the inc. dist /15

It's straightforward to see why misp. can (!) undermine the MR model. This happens when the poor are more likely to over-estimate their relative income (to the mean) than the rich. @S_Stantcheva has a great survey in the Deaton review.
/16academic.oup.com/ooec/article/3…

Imo what tends to be overlooked is that one key finding of this lit -- centre-bias, i.e. poorer people tend to overestimate their income, whereas richer people over-estimate them (@dietmarfehr et al) -- is potentially in tension with mis. undermining the MR. That is, centre /17

bias can only co-exist with a pol. system that doesn't respond sufficiently to the demand for redistribution (and sees inequality rising) if mis. are much more widespread among the poor than the rich and/or the poor have actually relatively greater influence on policy. /18

After all, they are the ones who demand "too little" redistribution.
2) Beliefs about inter-generational mobility

“Prospects for Upward Mobility” Hypothesis (POUM): The higher social mobility, the greater the tolerance for higher inequality. If voters /19
academic.oup.com/qje/article-ab…

belwo the median believe that their income will increase in the future and that tax rates are sticky, then they may oppose more redistribution, contra to MR.
Alesina et al. (@S_Stantcheva) show that this might help explain differences in redi. attitud /20
aeaweb.org/articles?id=10…

between Americans and Europeans. The former tend to over-estimate mobility, with the reverse true for the latter.

3) Normative ideals, fairness and altruism
Not all inequality is unjust or unfair and voters might not demand more redistribution if they view increases in /21

inequality as fair. @leo__ahrens also has a new paper out on this:
For a very nuanced take on the role fairness reasoning plays in determining redistributive preferences, see
/22

https://x.com/leo__ahrens/status/1338797134846562305

aeaweb.org/articles?id=10…

@CharlotteCavai1's book. I tried to summarise some of the relevant arguments here: /23

https://x.com/edenhofer_jacob/status/1792976600574898541

cambridge.org/core/books/fai…

There's much more to be said, but let's move on to Shayo's (2009) social identity theory. The idea here is that identity is endogenous: ppl can choose to identify with different groups as circumstances (e.g. inequality) change and they care about the
/24cambridge.org/core/journals/…

relative status of "their groups" (detailed summary below). Shayo's key insight is that higher inequality can increase the poor's incentive to identify with the nation, rather than their class, and, as a result, support less redistribution. The empirical evidence is weak tho. /25

Ok, but the MR model also assumes 100% turnout. But, as e.g. @VerglPolwi and @HannaSchwander argue, as inequality rises, a turnout differential can arise, with the poor abstaining at higher rates than the rich.➡️ The income of the median voter is higher than the income of /26

the median eligible voter, leading even responsive politicians to supply less redistribution than the MR model would predict.
The final demand-side expl. that I want to discuss relates to arguments about the space of political competition having become
/27cambridge.org/core/journals/…

two-dimensional. John Roemer explored the redistributive ramifications of introducing a non-economic second dimension. He (and others) showed that -- even setting aside mis-perceptions, normative beliefs, and identity -- such a second dimension can blunt redistributive press. /28

Now, we need to consider the median position on each dimension and assess their relative salience. If the 2nd dimension is sufficiently salient and parties are vote-max then they appeal to the median on that dimension and if that voter wants less redistribution then the med /29

on the economic dimension, less redistribution results. The culturally conservative poor form an anti-redistribution coalition with the culturally conservative rich. The arguments by Hacker and Pierson, for instance, are an application of that logic. /30
wwnorton.com/books/97816314…

On segregation and the rich's demand for redistribution, see @fetzert's thread.

Let's then turn to the supply side, starting with explanations that highlight the declining power resources of the left. Since the 1980s, unions have declined in strength /31

https://x.com/fetzert/status/1816764704552878334

as have social democratic parties. If we assume that these forces pushed for lower inequality and represented the interests of the poorer, then this explains why redistribution didn't rise in line with inequality, as predicted by MR. /31
academic.oup.com/qje/article/13…

See also this thread.
Others point to pol. institutions. Iversen&Soskice (2006) argued that majoritarian electoral rules favour the right, while PR favours the left. A recent paper provides some evidence for this claim ()
/32tinyurl.com/359ar7fe

https://x.com/Vinceman86/status/1825866988293337252

I'd also encourage you to check this @BJPolS paper by Manow and Doering. The problem with these explanations is that institutions tend to be fairly constant over time and a constant can't explain a change. At a minimum, such institutional theories
/33 cambridge.org/core/journals/…

have to introduce some time-varying element.
Another prominent explanation is that low-income voters are under-represented in legislatures. [I'll say that this strikes me as a better explanation in the contexts with weak parties than in parliamentary systems with strong] /34

party discipline]. I'm also not sure whether this explanation helps us explain the change over time, i.e. have legislatures really become less representative since the 1980s. If anything, I'd guess the reverse is true.
The final supply-side explanation I'd like to mention /35

relates to money in politics and lobbying. See @simonweschle's book for an incredibly nuanced take on this:
/36 cambridge.org/core/books/mon…

This is just a short overview and I think political scientists have yet to fully incorporate many of the insights in the newer economics literature on inequality. Nevertheless, I think the political ramifications of inequality are an important reason /37

https://x.com/edenhofer_jacob/status/1744341648090239364

why so many are interested in what is rather arcane (albeit highly important) debate about measuring inequality (in the US context). I hope this thread helps you to explicate your intuitions about these consequences somewhat better. /END