Inequality and Redistribution

Photo by Jason Leung on Unsplash

In their seminal article “A Rational Theory of the Size of Government” (1981) Meltzer & Richards argue that democracies over time become more equal economically. This is driven by the median voter, i.e. the voter that decides the election, economical self-interest.

Given that the average income is higher than the median voters income (because some people are much richer), the median voter will vote for distribution until that the income of the median voter is equal the average income.

A couple of things to appreciate behind the model, before moving on to whether it holds empirically and discussing the assumptions. First off, even though the model implies that society becomes more equal, it doesn’t imply that this drift will benefit all members of society. The median voter doesn’t have a motive to secure redistribution for the poor.

Although this claim can seem absurd prima facie, as one would normally assume that redistribution is primarily important for helping the poorest members of society, think of which welfare schemes in society are most strongly protected and best funded. These are primarily middle-class benefits, such as education (subsidies to high-ed are inherently regressive) and short-term unemployment benefits.

A second interesting implication is this gradual converge that happens over time. So, a country is originally a highly unequal society and suddenly becomes a democracy, and then we will expect to see a greater redistribution. And this information is obvious for the dictator and the people, which contributes to some interesting dynamics between inequality, chance of democratization and the returns of oppressing the population as a dictator (see Acemoglu & Robinson 2006).

Do voters vote out of self-interest?
The first argument that can be levied against the argument is whether voters vote out of self-interest at all (1). Given that the costs and benefits of welfare schemes are diffuse, it doesn’t make much difference for the individual voter what party is in party (although there can be great differences on the aggregate level).

With that said, when personal gains or costs are sufficiently high, voters indeed seem to vote out of economical self-interest. This is apparent in zoning-laws, where voters fiercely oppose the construction of affordable housing in their area, as this lowers their real estate value (the NIBMY phenomenon, see Marble & Nall 2017).

The next aspect of the model is to expand with a prospective income aspect, given that voters have a grasp of their life time income and vote accordingly (see permanent income hypothesis by Friedman & Mogliani life-cycle hypothesis). In other words, a student that temporarily low income but expects to have a very high income will not vote to redistribute to her temporary income bracket, but to her prospective income bracket.

Can we find this empirically?
The challenge of the model is that has been hard to validate empirically. From the model, we’d expect to see higher redistribution in democracies than in autocracies, but strangely, this isn’t the case (see Ross 2006).

The great mystery here is the United States, which has relatively limited distribution. One factor that explains why the US has a more limited welfare state vis-a-vis Western Europe is the difference in institutional arrangements, where majoritarian-winner-take-all systems skew towards centre right-governments, whereas proportional systems skew towards centre-left governments and hence redistribute more (Iversen & Soskice(2)).

Another — and perhaps more controversial — factor is the role that racial heterogeneity in the US plays: “Racial animosity in the US makes redistribution to the poor, who are disproportionately black, unappealing to many voters.” (Alesina et al 2001)”. This observation also highlights the importance of thinking in groups rather than individuals.

But on the whole, our baseline assumption that democracies are converging to a redistributed equilibrium has to be modified quite a bit to fit reality.

But experimentally, maybe!
Another approach to the political economy of redistribution tries to attack the question from an experimental side. It may be that voters simply are under-informed and that their preferences actually are in par with what we’d expect from the self-interested voter.

Engelhardt & Wagner (2014) conduct a survey experiment, where Meltzer & Richards (1981) model works quite well. The issue is that voters misperceive 1) the level of inequality in society, 2) their prospects of moving upwards in society. In other words, the demand for redistribution is limited, as voters are too optimistic on how much they are going to earn and hence have limited demands for increased taxation (3, 4).

Hvidberg et al. (2021) also show that voters underestimate the level of inequality in society, and when voters are informed of the actual level of inequality, they perceive inequality as more unfair.

Equality of what?
This informal post has simply run through some arguments related to the Meltzer & Richards (1981) model, without arguing whether inequality is good or bad, or what macro-economical factors are driving inequality, that are not political.

One comment to add here is whether voters care much about the relative levels of wealth, and instead are more preoccupied in absolute gains. So, people are simply of getting more cake, instead of getting a relatively more cake than others. One explanation that voters and the health of society indeed is affected by relative changes is shown by Vinæs et al. (2021), who show voters in Denmark that experienced stagnation or regression of the value of their home voted for the Danish People’s Party (a right wing party that is considered populist).

The implication of the study is that people that are feeling “left behind” are supporting anti-globalist parties and that people may be thinking in relative rather than absolute gains.

1: rationally speaking, voters shouldn’t be voting at all, given that the probability of casting the decisive vote is miniscule and the probability of affecting policy to one’s personal gain is also limited, while the costs of voting are high. See Riker & Ordershooks “The Calculus of Voting” (1968) or my explainer

2: The institutional arrangements are naturally not exogenously given, why it naturally is difficult to demarcate where the desire for redistribution ends and the institution starts. For a review of the effects of very long-run cultural effects in shaping society, see Anne Sofie Beck Knudsen 2021

3: Whether this in fact is irrational from the individual voter is unclear. Consider loss-aversion, where the feeling of losing money matters more than the prospect of earning money. Here, even though you are better off in the society with complete equality, it hurts that you are taxed so much when you had the hypothetical opportunity to earn much more.