Tuesday, November 6, 2012

Economic inequality, the wrong way [my US elections post]

A research paper about economic inequality in America has been making the rounds: “Building a Better America - One Wealth Quintile at a Time” by Dan Ariely, behavioral economist and best selling author (Predictably Irrational), with economist Michael Norton.

I am going to tell you why, contrary to what Ariely claims, the paper doesn’t say that “Americans Want to Live in a Much More Equal Country (They Just Don't Realize It)”. And hopefully there’s a general point here.

The researchers asked Americans about the distribution of wealth in the US. The respondants grossly underestimated it. They said, for instance, that the two bottom quintiles (40%) possessed 9% of the total wealth, while top quintile possessed 59%. The correct numbers are 0.3% and 84% (!).

Is this shocking? Not really. Surely no one expects people to memorize macro-economic variables. Or do we expect them to “estimate” wealth distribution? based on what?

Incidentally, why did N&A asked about wealth, with income inequality much more on the public mind? Wealth is latent. If I own my apartment and my neighbor rents, I may be much wealthier than her, but no one can see it. Income is more visible - even where salaries are not openly discussed.

It’s possible that people had income in mind when answering about wealth. Maybe they didn’t realize the difference, or they used income as a proxy for wealth. It’s telling to compare the survey answers above to the US income distribution - 11% to the bottom 40%, 51% to the top 20% (Census Bureau, table H-3).

Yes, the survey meticulously explained what wealth was. But do you expect people to know the wealth distribution, if you need to tell them what wealth is?

This point safely nailed, let’s move on to the more juicy part, where people were asked to imagine joining a new nation. When they joined this nation, they were told, they “would be randomly assigned to a place … so you could end up anywhere in this distribution.” They were asked to choose wealth distribution for their new country. The result: among three alternatives – total equality, the current American state of affairs, and that of Sweden – Americans of all stripes wanted Sweden (!).

People also had to construct their ideal wealth distribution for the US. Again – everybody constructed a distribution more egalitarian than the current one. In fact, Ariely writes, “the ideal distribution … was dramatically more equal than exists anywhere in the world.” (!)

I refuse to be shocked, again. We saw people can’t estimate the wealth inequality numbers. I claimed they didn’t even quite grasp the concept. How then is it meaningful to ask them about their ideal distribution? How isn’t it just one big garbage-in-garbage-out?

And when Americans pick a distribution “dramatically more equal” than anywhere on the planet, doesn’t that imply precisely that they have no idea what those numbers mean?

Ariely writes that the study “avoid[s] hot-button terms, misconceptions and the level of wealth people currently have”. But preconception and personal stakes transform the academic to something that pertains on real life. And it’s real life people have opinions about.

That last point is my general one. Questionnaires are bound to take things out of context. Testifies every question you have ever wanted to tick with “it depends”. Questionnaires tend to assume people have opinions about everything, opinions that can fit into the available answers. Personal experience again tells us how wrong this is. People who use questionnaires says this can be fixed with good planning and statistical analysis. Well, if it can, “Building a Better America” isn’t a great example.


There’s more. The paper sometimes simply replaces one bias with another. This doesn’t have much to do with the point I wanted to make, but I can’t resist including it anyway. So as a compromise the type is small.

First, joining the society in a random economic position is supposed to take away incentive (Ariely compares it to blind wine tasting). That’s silly. If you are going to get a random piece of a pie, your rational choice would be an equal distribution, so that you will never end up with a small piece.

And does the size of the pie matter? the economic justification for inequality is that it encourages economic activity and growth. In the survey, all the pies are the same size, an implicit direction to ignore this argument. A similar point was made by a right-wing commentator in the National Review Online. Unlike that commentator, I share Ariely’s social-democratic sentiments. The trick is to be able to think critically no matter what your feelings about the results are.


  1. Very interesting points, thanks for sharing.

    I have a feeling that another issue in play here is that I think humans are probably very bad in estimating heavy tailed distributions. And another crucial difference between wealth and income (besides that of latency/visibility) is that wealth distribution is more heavy tailed than income distribution.

    Lastly, about the "random position" - I agree that this is silly. But given certain utility functions it's still "rational" to want a non-equal distribution, the same way that it could be rational under certain assumptions to buy a lottery ticket.

    1. Good point about the heavy tail distribution, and yet another one a behavioral economist should really be aware of.

      And you are of course right also about the random position (or the Rawl's condition, as it is called in the paper). Preference for a more equal distribution is rational if you assume people are mostly risk averse, but I think it's a reasonable assumption in this case. It's interesting though that when asked to choose between "Sweden" and complete equality, half of the people went of "Sweden". A rational explanation is that people are willing to take small risks, in the same way the a lottery ticket could be rational.

    2. And another thought: it may not be clear from the text, but I don't really expect people would have much clearer idea about an ideal distribution of income.

      At least I personally don't. I have opinions on what stuff everybody should have, and conversely on what kind of wealth is abominable. I can't quantify that in terms of either wealth or income distribution.

    3. Your first point (that maybe people were confused with income) is completely valid, however I disagree with much of the rest.
      Of course questionnaires are messy, but in a garbage-in garbage-out situation you'd expect something random (e.g. around 33% each), not a vast majority selecting the same option. In addition, it looks like on average, people's ideal distribution was more equal than their estimate, and that should cancel out individual biases and misunderstandings (I haven't managed to access the full article so I don't know if this is statistically significant). So, I agree that the findings are exaggerated, but I do thing the study points in the direction the authors say it does.

    4. Thanks for the comment.

      By "garbage-in-garbage-out" I don't mean people chose randomly. But most people just don't have a clear grasp on what these numbers mean about the economy.

      Numbers must be grounded somehow in our experience for us to have opinions about them. Is paying $2 for a cup of coffee expensive? I can answer that because I bought a cup of coffee this morning. Is paying a billion dollars for some infrastructure project expensive? I have no idea. In this case there's deliberately no grounding.