CEPR - Center for Economic and Policy Research


En Español

Em Português

Other Languages

Home Publications Blogs Beat the Press NYT Falls for Bad Numbers on Household Income

NYT Falls for Bad Numbers on Household Income

Thursday, 22 August 2013 04:23

It's good to see the NYT taking a serious interest in the wages and income of typical families. Unfortunately, they have picked a bad measure to highlight in relying on the reports of Sentier Research.

The Sentier measure is moved to a large extent by erratic patterns in income reported by respondents to the Current Population Survey fielded by the Census Bureau. Because the sample size in this survey is relatively small (the overall survey has 60,000 households, with only one quarter answering the income question each month), there will frequently be large movements which almost certainly reflect sampling error rather than actual changes in the economy. 

For example, the Sentier index showed a sharp drop in before tax income in the early months of this year which has since been reversed. It showed an even sharper drop at the start of 2012, which was reversed over the course of the year. There were no obvious economic developments that could explain the drops in either year or their subsequent reversal. These movements were simply random fluctuations in the data, which is common in this series. That is why economists generally do not pay much attention to its short-term movements.

A much better analysis of trends in income can be found at the Economic Policy Institute's (EPI) website. It has been providing solid analysis of wage and income trends for more than two decades. Unlike Sentier Research, EPI does not charge for its research findings.

Comments (8)Add Comment
written by kharris, August 22, 2013 7:16
I once made reference to EPI analysis and a client called to "inform" me that EPI is a partisan organization and not to be trusted. That's my only data point, but I would not be surprised to learn that reporters get similar "help" when they highlight work from EPI.

Heritage and AEI are OK, apparently.
partisan and free
written by Jennifer, August 22, 2013 9:00
@kharris I am guessing that line gets used often. Also, I might guess that the fact that EPI does not charge for its research findings is a factor in being taken less seriously. If you have to pay it *must* be of higher value is a common mindset.
written by kharris, August 22, 2013 11:41

I guess we judge economic analysis the same way we pick kittens.
Sample Error in the CPS?
written by LSTB, August 22, 2013 12:01
Dean writes:

Because the sample size in [the CPS] is relatively small (the overall survey has 60,000 households, with only one quarter answering the income question each month), there will frequently be large movements which almost certainly reflect sampling error rather than actual changes in the economy.

I'm surprised by this because 15,000 people is still a very healthy sample size. It could be biased if the 3/4ths of non-reports are low-income people, but in my experience with CPS data the income numbers only start getting wacky when looking at smaller subsets like 25-34-year-old professional degree holders.
written by ltr, August 22, 2013 6:52
I think Sentier has proven quite reliable over time. Also, I like EPI a lot but from my experience I think the criticism of Sentier is too harsh.
written by watermelonpunch, August 22, 2013 8:20
I don't have an opinion on how reliable I think EPI is in relation to Sentier.

But I'm going to take this opportunity to mention my bafflement at the inclusion in articles about incomes, of stating the obvious things like that the summer of 2011 was a low point and that people over 65 haven't had it quite so bad as the rest, and superfluous mentions of the recession ending in 2009.

But then, after all, I'm currently in the market for a FREE kitten.

Look at chart showing income by Sentier's measure
written by Dean, August 22, 2013 8:30
The chart jumps around in a way that it is implausible for income to actually change. This is why economists generally have not used this data for monthly changes. Taking an annual average gives you a reasonable number. A 3 or 4 month average does not.
looks like Christmas hiring to me....and re: biased stuff
written by pete, August 23, 2013 3:20
Its not a drop in 2012 early, its a spike late 2011...looks like spikes are annual. Simple.

Regarding biased "research", one would guess that the backers generally are concerned with the output. As I understand it, the EPI is mostly union funded, with some Soros money, etc. I would think these folks expect supportive research, much like the Kochs fund generally libertarian ventures, and big business funds the Heritage and other conservative/statist spots. That said, I think Dean and his crew are quite careful in their analysis. It is just that the topics they wish to pursue are problably biased. I guess that is sins of omission rather than commission, and to be understood. Overall there is decent competition for ideas, and I think it best to avoid the narrow view of research.

In the academic world, research is undertaken, responded to, elaborated on, sometimes findings are reversed, new data, better techniques, missing variables, etc. Same is true in the popular kind of research such as unvetted papers at places like EPI. Apparently these researchers do not attend academic-type conferences, so there is little peer review, simply shooting from the hip. But there is clearly sufficient give and take.

Write comment

(Only one link allowed per comment)

This content has been locked. You can no longer post any comments.


Support this blog, donate
Combined Federal Campaign #79613

About Beat the Press

Dean Baker is co-director of the Center for Economic and Policy Research in Washington, D.C. He is the author of several books, his latest being The End of Loser Liberalism: Making Markets Progressive. Read more about Dean.