I'm in My "What a Median Means" Mode

New York Times economics writer David Leonhardt has a nice column today on the money-happiness connection, centered around a new Wolfers-Stevenson paper that provides some evidence against the Easterlin paradox.  But that connection is not what I want to talk about.

At the very end, Leonhardt writes:

"Affluence is a pretty good deal. Judging from that map, the people of the world seem to agree. At a time when the American economy seems to have fallen into recession and most families’ incomes have been stagnant for almost a decade, it’s good to be reminded of why we should care."

Here we go again.  Most families' incomes have NOT been stagnant for almost a decade.  What I suspect he's thinking of here is the data on stagnant median household income.  However, a stagnant median says nothing about whether individual households did better or worse over time.  Median income has been fairly stable (though even that doesn't account for non-wage compensation), but a constant median between time X and time Y is consistent with each household in the population at time X having had their income rise, IF by time Y the right number of new households have entered the population at incomes below the median.  Increased immigration and more people entering the labor market as young workers can cause the median to stay constant even as most households see real income increases through time.

Put differently, a constant median over time where the population at each point in time differs says nothing about how well or poorly any individual households in the original population did over that whole period.  Leonhardt's little throwaway line is simply wrong on the facts.  Not only that, it is overwhelmingly likely the case that the majority, if not a substantial majority, of American households have higher real incomes now than ten years ago, even as the median has remained stagnant.

This is how economic myths get started.

Comments

[David]

Excellent Post, I'd just like to add that even if Median Household Income were stagnant, Household size has also been decreasing. From the Early 1970s it's dropped from 3 to 2.6 now. http://www.msnbc.msn.com/id/14942047/

[DPrychitko]

Steve:

I, too, fully agree with David above that you have made an excellent post. Now, to build upon your post and to make an especially strong argument, why don't you give us clearer statistical details on what has been happening with household income and population growth over time. I can't wait to see your results.

[Steven Horwitz]

Well you can start here Dave:

http://www.treas.gov/offices/tax-policy/library/incomemobilitystudy03-08revise.pdf

I'll hit the relevant highlight (in brackets) for you, which refers to 1996 to 2005, or pretty much Leonhardt's "last decade":

"Median incomes of taxpayers in the sample increased by 24 percent after adjusting for inflation. {{The real incomes of two-thirds of all taxpayers increased over this period.}} Further, the median incomes of those initially in the lowest income groups increased more in percentage terms than the median incomes of those in the higher income groups. The median inflation-adjusted incomes of the taxpayers who were in the very highest income groups in 1996 declined by 2005."

Unlike the median household income data, this is actual household tracking data (like the PSID, which is the other place to go to see this same phenomenon, though for the decade before, as reported in Cox and Alm). That means that it follows the same taxpayers over the ten year period. So 2/3rds of the taxpayers in 1996 saw higher real incomes in 2005. And the percentage gains were larger at the lower end, though Leonhardt says nothing about rich/poor issues.

Feel free to complain about government stats or to criticize the methodology if you want. I can find other data that more or less tell the same story.

And I did make two distinct arguments - one that a constant median doesn't imply constant incomes for individual households and the other was the claim that, in fact, things have gotten better for most. Having provided some supporting data for my latter claim, it's worth noting that Dave has asked ME to provide such data even though Leonhardt's original argument provided no data whatsoever to support his own claim that I am contesting. Why not send him an email Dave and see if he's got data to support his view?

[Frank]

Boffo post Steve!

[DPrychitko]

Thanks Steve. Boffo indeed. Of course, being an old man out here in the hinterland, I admit I had to look up the meaning of boffo. At first I thought it meant something sexual. Then I thought it was related to Buffo (the clown).

At any rate, I understood clearly your first argument. I wanted to know whether the facts and data further support it, and it sure looks that way. Thanks again.

As for me emailing Leonhardt... I don't read the man's columns. But I do read your blog.

[Rafe]

Actually there is a degree of ammbiguity about the term "boffo" or at least about the root "boff".
http://www.thefreedictionary.com/boff

You might be safer to call it a bonzer post.
http://www.worldwidewords.org/weirdwords/ww-bon1.htm

[Hannes]

You have built a good websitee

[Kathy]

This is very interesting site