Senaste poster
Google Scholar
« Privata vinster, socialiserade förluster. | Main | Tillit i Sverige över tid »

Some problems with the “Building a Better America” paper by Norton & Ariely

[På engelska för att det kan hända att någon som inte kan svenska är intresserad av just denna post…]

The paper “Building a Better America—One Wealth Quintile at a Time” by Michael Norton and Dan Ariely has been cited a lot – here’s an example:

Wealth in America is heavily – extremely – concentrated among people in the top quintile. It’s not that wealthy people have a bit more than the middle class and a lot more than the lowest quintile. No. Wealthy Americans own almost 85% of assets in America. That should be surprising to you because when Ariely and Norton surveyed people to find out how much wealth folks *think* the top quintile owns, they estimate about 58%. Even that inequality is too much, the respondents think.

Here is the full paper, published in Perspectives on Psychological Science. It contains the picture shown below:


The paper is typically said to show the following: When asked, americans would like US inequality to be lower than they think it is. And when asked what they think it is, it is lower than the actual level of inequality. The level of inequality where that americans actually prefer is more like the level in Sweden.

But what distribution are we talking about, income or wealth? The claim below the pie charts says: “Pie charts depict the percentage of wealth possessed by each quintile”. This was confusing to me: I thought wealth inequality in Sweden was rather high. So I looked ut up.

Actually, the distribution of total assets in Sweden (in 2004, table 4 in this report from Statistics Sweden) looks like this (assuming that i converted from deciles to quintiles correctly):


The net wealth distribution looks like this:


It is hard to say if A&N use net wealth (assets minus debt) or assets for US, but it is obvious that they are not using wealth data for Sweden at all. The Swedish circle looks more like the Swedish income distribution to me.

At, both distributions are actually said to be income distributions:

When given the choice between the income distributions of Sweden and the United States, 92% of respondents preferred that of Sweden

So... This post was originally intended to be about this paper by Kimmo Eriksson & Brent Simpson, titled

“What do Americans know about inequality? It depends on how you ask them”

It shows that the way A&N ask questions about distribution is biased towards underestimating inequality, because people are not very good at doing math involving percentages (as I read their paper, at least). They show that people make the same mistake for the distribution of web page popularity and school teacher salaries.

Furthermore, E&S note that the math-problem seems to drive the results in N&A:

when made aware of the logical connection between the Percent measure and the Average measure, many people do not stand by the low inequality responses elicited by the Percent measure.

I guess the argument in Eriksson & Simpson hold for both income distributions and wealth distributions, but it it still worth noting that the often cited stylized fact that americans prefer to live in Sweden inequality wise, comes from comparing the US wealth distribution with the Swedish income distribution…

PrintView Printer Friendly Version

EmailEmail Article to Friend

Reader Comments (3)

Ska det stå "as I read their paper, at least"?
22 mar | Unregistered CommenterAndreas
Självklart. Tack!
23 mar | Unregistered Commenterbergh
Så här skriver de i not 2: "We used Sweden’s income rather than wealth distribution because it provided a clearer contrast to the other two wealth distribution examples; although more equal than the United States’ wealth distribution, Sweden’s wealth distribution is still extremely top heavy."

Mycket märkligt resonemang.

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
All HTML will be escaped. Hyperlinks will be created for URLs automatically.