Confirmation Bias, Perhaps?

by Don Boudreaux on November 10, 2011

in Data, Inequality

Here’s a letter to the Washington Post:

Fareed Zakaria writes that “The most comprehensive comparative study, done last year by the Organization for Economic Cooperation and Development, found that ‘upward mobility from the bottom’ … was significantly lower in the United States than in most major European countries, including Germany, Sweden, the Netherlands and Denmark” (“The downward path of upward mobility,” Nov. 10).

Not so.

Here are this OECD-study’s three bullet points summarizing findings on economic mobility directly (rather than findings on the connection between family background and educational achievement).  Does the U.S. stand out from “major European countries”?

“* Parental or socio-economic background influences descendants’ educational, earnings and wage outcomes in practically all countries for which evidence is available.

“* Mobility in earnings across pairs of fathers and sons is particularly low in France, Italy, the United Kingdom and the United States, while mobility is higher in the Nordic countries, Australia and Canada.

“* Across European OECD countries, there is a substantial wage premium associated with growing up in a better-educated family, and a corresponding penalty with growing up in a less-educated family. The premium and penalty are particularly large in southern European countries, as well as in the United Kingdom. The penalty is also high in Luxembourg and Ireland. In these countries the wage premium is more than 20%, while the penalty is some 16% or more (relative to wages earned by individuals raised in a family with average education).”

As for this study’s other measures of social mobility (which examine family-background’s influence on students’ educational achievements), on these, too, U.S. mobility simply does not stand out as being significantly or consistently lower than in other – including major European – countries.

Donald J. Boudreaux

Be Sociable, Share!



17 comments    Share Share    Print    Email


Euro-JoeDoe November 10, 2011 at 4:44 pm

Is it easier to move upward in social ladders here in Europe? The answer is yes! But it’s not because we have some how better system. It’s only because gaps between social classes are much smaller here than in US.

Don Boudreaux November 10, 2011 at 4:48 pm

Yes. This point is important – and, coincidentally, one that I made earlier today at a talk I delivered at Duke’s law school.

Methinks1776 November 10, 2011 at 7:08 pm

That is always my first thought when I see these statistics. In Europe, everyone is poorer.

EG November 10, 2011 at 4:57 pm

Seen this study before, and its not very convincing. The statement Zakaria makes is simply wrong. He thinks this data represents mobility from the…bottom. Thats not the same thing as upward mobility. Second, income brackets as measured in the US are different than from Germany, or Sweden or whatever. We’re not using equivalent units here.

Third, there’s no reason to suppose that inter-generational mobility should be greater in the US, or that a greater mobility in this field would be a desirable or meaningful thing. You may have very little upward mobility compared to your parents, but if your parents were already well off, vs if they were homeless, makes a big difference. The mobility gains in the US may have been realized in the last generation at a greater level than in some European countries. If some countries in Europe went through decades of economic stagnation, and jump-started in the 90s, whereas the US saw steady growth for decades, we’d expect these results.

Fourth, immigration. If 10-15% of the population of some Scandinavian countries is composed of relatively recent immigrants from third world countries, their inter-generational mobility would be huge. America’s gains from immigration were small over the last few decades.

Jon Murphy November 10, 2011 at 4:58 pm

“Second, income brackets as measured in the US are different than from Germany, or Sweden or whatever. We’re not using equivalent units here.”

Yes we are. This is OECD data. Their own defined brackets.

EG November 10, 2011 at 5:21 pm

Oops! I was thinking of another study, not this OECD one. Now that I read this one, its even less related to what Zakaria is talking about. The OECD study is looking at something else.

Its the PEW study that uses income brackets as a comparison without adjusting for the spread of those brackets.

Bill November 10, 2011 at 5:30 pm

Leave to Don to look at and cite from the OECD study and screw up the impression Zakaria wanted to convey.

muirgeo November 10, 2011 at 6:14 pm

Well the study does confirm what Zakaria asserts not to mention the other TWO studies Zakaria references also confirming the same. Don is completely incapable of objectivity. He’s very particular when the data doesn’t go his way and lets in the loosest of data when it supports his position. See some of his references to Mark Perry’s crap that reaccepts without a second thought. Confirmation bias indeed.

Invisible Backhand November 10, 2011 at 8:21 pm

He recc’ed a Marc Perry link to a climate change denier speech a five days ago:

It’s also the one where RegardsKen slipped up.

muirgeo November 11, 2011 at 1:24 am

Yeah it’s all so boring I couldn’t care less except that the bankers have taken over the world on propagandizing this philosophy… and the ignorant people willingly submit or support the subversion.

I used to laugh at my paranoid father-in-laws concerns about the Bilderberg Group taking over the world… now the new Italian Prime Minister is a board member….

vikingvista November 11, 2011 at 2:56 pm

I suppose they are interested in the study of showing to what extent there is an aristocracy or caste system, which is pretty silly. If they want to remove the effects of individual income mobility from the total population income variance, they simply need to follow each individual’s income over time in the study.

Given the trouble they went to, you have to wonder why they didn’t attempt to answer the question directly.

Charlie November 11, 2011 at 9:55 am

“found that ‘upward mobility from the bottom’ … was significantly lower in the United States than in most major European countries, including Germany, Sweden, the Netherlands and Denmark”

“Mobility in earnings across pairs of fathers and sons is particularly low in France, Italy, the United Kingdom and the United States, while mobility is higher in the Nordic countries, Australia and Canada.”

I don’t understand your objection. Are you saying the problem is the word “most” meaning more than half or “major” or “most major”? Or are you saying we are lower but not “significantly lower”?

Here are the results from the paper:

The U.S. is more mobile than:

Great Britain

The U.S. is less mobile than:

Canada (not in Europe)

I can’t find anything wrong with Fareed’s statement that relates to your criticism. Though, in fairness, I don’t really understand what criticism you are trying to make, so it’s not that surprising.

Maybe a valid criticism could focus on “from the bottom.” I didn’t read the whole paper, but the graph people seem to be using doesn’t seem to take where you start into account. The same correlations could reflect sons born into middle class and staying there, poor sons showing a lot of social mobility, and rich sons not being able to match the high earnings of their fathers. I’d have to read the whole paper to know if that held up, and I really don’t have time right now.

Dom November 11, 2011 at 11:03 am

Charlie: The text to the graph you are pointing to — Fig 5-1 — says (my emphasis in CAPS):

“…OECD evidence on intergenerational persistence in wages is obtained by
estimating the percentage INCREASE OR DECREASE in individuals”.

Does this imply that the countries with high intergenerational mobility are countries in which sons may make LESS than their fathers? If so, then the countries at the top are countries in which fathers provide a floor — not a bad thing.

(I might be wrong in this interpretation. I’m not an economist. I’d appreciate anyone who can straighten me out.)

Charlie November 11, 2011 at 5:49 pm


Here is a way to think about what they did. If I draw a random sample of people in a country, I will find that on average a person earns the average wage in the country.

Now I will examine what happens when I divide the sample into two groups. The first group has fathers who are “high earners” and the second group has fathers who are “low earners.” Now I want to see what information this provides. If in separating the two groups, both groups still earn the average wage (ie. knowing the father’s wage provides no knew information), that would be a high mobility country. If the wage was exactly the same as the fathers, that would be a low mobility country. A high mobility country has a correlation of 0. The extreme low mobility has a correlation of 1. (this is not literally what they did, just a simpler way to think about regression).

To address your question directly, first, I’m pretty confident that on average sons earn more than their fathers. There was median growth in these countries over the sample. So it’s not that everyone made less than their fathers. Is it possible that high earning fathers produced lower than average earning sons? In that case, the numbers on the Y access would be negative, which they aren’t.

So basically, what that chart is showing, is that if you are an American, Italy or the UK, if I know who your father is, I know much more about your likely income than if you are born in Denmark, Australia, and Norway. And further, if you have a high earning father in America you are much more likely to be high earning yourself than if you were born in Denmark (and vice versa, a low earning father is more likely to have a low earning son).

Ryan Vann November 11, 2011 at 2:31 pm

Zakaria, that is the dude that constantly harps on how the entire US Constitution should be rubbish binned, right?

vikingvista November 11, 2011 at 3:23 pm

That’s like harping on how the earth should be round.

steve November 11, 2011 at 9:04 pm

Previous post:

Next post: