Do People Have Weird, Abstract, Pareto-Damaging Preferences?

by Will Wilkinson on April 18, 2008

Lane Kenworthy shows some evidence that when asked to choose one of five pictures that best represents their preference for their country’s income distribution, people tended to pick one of two options — options D and E:

D and E are identical in their population shares at the bottom. The difference between them is that D has a larger share in the middle, whereas E has a larger share at the top. Average income is higher in E. Inequality is lower in D.

[...}

I wouldn’t go so far as to conclude from this that people tend to value low inequality over high incomes. Other ways of posing the question might yield different results. But it does suggest that inequality matters to people.

I find these pictures a bit hard to interpret myself, and I think the idea that people have some kind of standing preference over the shape of the national income distribution is plain bizarre. The question embodies and encourages a nationalist orientation to economic patterns, as if this is the natural level at which to look at economic patterns, as if this is the natural level at which people will have preferences about such patterns. But why think people actually have prior preferences about such things? The national income distribution is not experienced. The local income distribution isn't experienced. Differences in local visible consumption may be experienced, and it seems plausible that people would have preferences about that. But that's not what the question was about. Anyway, this seems to me a bit like asking about my preference over the proportion of luxury to compact cars in the nearest parking garage. Why would I have one?

Anyhoo, if people really have these preferences then many of them are malicious. E is a world in which many people are better off than in D but in which no one is worse off. It's Pareto Awesome! If so many people really do like D better, does that tell us that there is latent support for egalitarian political institutions, that such support is based on a deep moral error, or both?

How about tackling the question a bit more rigorously? Will Ambrosini points us to Matthew Rabin and Gary Charness’s "Understanding Social Preferences with Simple Tests" [pdf], which tests plausible local preferences for equality, efiiciency, and reciprocity in a lab setting:

Our findings suggest that the role of inequality-reduction in motivating subjects has been exaggerated. Few subjects sacrifice money to reduce inequality by lowering another subjects’ payoff, and only a minority do so even when this is free. Indeed, we observed Pareto-damaging behavior more often when it increased inequality than when it decreased inequality. While this comparison is itself confounded by other explanations, our data strongly suggest that inequality reduction is not a good explanation of Pareto-damaging behavior.

My faith in humanity is restored. Sort of.

Viewing 5 Comments

    • ^
    • v
    Why should people have preferences about the distribution of wealth in a society? I suspect that when they are asked to declare a preference among these different graphs, they may imagine themselves making a choice about which society they will find themselves randomly placed into -- from behind a Rawlsean veil of ignorance.

    By that standard, E is also the clear winner, but before making my choice, I'd like to know more about the absolute wealth of each society, and also how big an increment we have from one bracket to the next, and whether that increment was standard. I could easily imagine favoring distribution A if for example the lowest bracket was $1,000,000 per year. I might likewise find A untroubling if the differences between the brackets were trivial and if the overall standard of living were still high.

    But then... the original question still stands: Why ARE we arguing about the justice of patterned distributions? What are the processes that get us to each of them?
    • ^
    • v
    These diagrams are hard to interpret suggesting a lot of measurement error (if interested analysts have a hard time, what do we expect of survey takers that are just doing the pollster a favor). I got the interpretation wrong in my post!

    My reinterpretation: I interpret each picture as a way to slice up the same pie (i.e. the total GDP is the same in each) and each row a size of a slice of pie. In my experience, people often believe the size of the pie is fixed and they believe wealth distribution is a zero-sum game. The X's are the number of people with each size of slice. Under this interpretation, the middle row doesn't represent the same amount of income in each picture.

    I also interpret this as a "behind the veil" experiment, so there's no discounting of others' incomes. Finally, in this sort of experiment people are considering the distribution of others' income, not necessarily their own, but they have preferences over others' incomes. This kinda changes the meaning of Pareto improving; its more of an individual's criterion. (If people didn't have social preferences, then clearly none of these distributions would be PI over any of the others... some people would be hurt by the change.)

    Given survey takers had this interpretation of the diagrams, E isn't necessarily a Pareto improvement over D. In fact, D may be a Pareto improvement over E. It depends on how much diminishing returns to others' income there are for the person behind the veil.

    This is why its not clear what sort of social preferences are being suggested by this survey data. It would have made the "people have preferences for equality" case more clearcut if the survey had option E as Will describes, i.e. everyone better off. As it is, it could be that people have a preference for efficiency, they see D as a Pareto improvement over E and that would explain a preference of D over E.
    • ^
    • v
    I agree with pushmedia1. The idea is that there's the same amount of total wealth in each society, now how should it be divided? So E is not pareto superior (if that's a correct phrase) to D, and Will's worry is misplaced.

    What was interesting to me was that none of the models had an idealist/communist option, which would look like this:

    **********************

    And it's just obvious to me that people would have a preference about these. If your options are:

    (a)
    *******
    *******
    *******

    versus

    (b)
    *
    *
    *
    *
    *
    *
    *
    *
    ***********************

    then option (b) seems likely to have a very large poverty rate, and a few very very rich people. I've been to places like that, and they suck. A lot.

    So, Will, what's the mystery?
    • ^
    • v
    Well, in the recent debate, Obama indicated that he wants to raise the capital gains tax from 15% to 28% even if it reduces the government's revenue.

    So he thinks, and/or he thinks his target demographic thinks, that it would be better to make wealthy people worse off even if it makes nobody better off (in fact, it would probably make everyone worse off).

    He said the reason was "fairness".

    http://www.cato-at-liberty.org/2008/04/17/obama...
    • ^
    • v
    I found a study by Philip Michelbach et al to be interesting. See:http://digitalcommons.unl.edu/psychfacpub/186/ )
    Survey participants were asked to choose between different income distributions as "outside observers giving their advice". It was intended to be an experimental test of Rawls theory of justice. So far so good. But the results suggested that females had a strong bias toward an egalitarian income distribution while males tended to favour efficiency maximization. I'm not sure to what extent this might just reflect the survey sample, but that is not the point I want to make.

    What I am wondering is what would happen if the survey respondents were actually to choose where to live behind a veil of ignorance of everything except the level and distribution of income of different societies. I imagine that there might be a subsequent migration away from both the egalitarian societies and those with greatest focus on efficiency.
 

Trackbacks

(Trackback URL)

close Reblog this comment
Powered by Disqus · Learn more
blog comments powered by Disqus

View My Stats