Gail,

I would like to draw you attention to two errors in Paul Krugman's column of Friday, March 12, "No More Excuses on Jobs."

These two errors have nothing whatsoever to do with opinion, ideology or interpretation. They are both simple factual errors that deserve correction.

First, Krugman writes:

"But wait — hasn't the unemployment rate fallen since last summer? Yes, but that's entirely the result of people dropping out of the labor force."

Krugman makes it clear that he is referring to employment statistics derived from the Department of Labors "household survey." These statistics are available to the public on the DOL's website at http://data.bls.gov/cgi-bin/surveymost?ln. If you go to that page and pull up the numbers for the first two series -- "Employment level: civilian labor force" and "Civilian labor force level" -- you will have the numbers you need to calculate the unemployment rate, which is itself available as the fifth series.

You will see that no matter which summer month last year is used as the starting point, the statistics show BOTH an increase in employment AND a decrease in the labor force. Therefore it is an error to say that the falling unemployment rate is the result "entirely" of a shrinking labor force. The unemployment rate would have fallen even if the size of the labor force had been held constant in the calculation.

Yes, Krugman's column concerns how unreliable the "household survey" is, in his opinion. Nevertheless, he himself uses it to make this point. And his point is in error, regardless of what he thinks of the reliability of the statistics he has chosen to use.

Considering that this particular column largely concerned how to interpret labor statistics, it is incumbent upon The Times to confess Krugman's error and correct it.

Second, Krugman writes,

"My own piece, 'My Economic Plan,' was fairly typical: I called for extended unemployment benefits, temporary aid to state and local governments, and rebates for low- and middle-income workers.

"Maybe this more or less textbook response to a depressed economy wouldn't have worked. But we'll never know, because the administration rejected all such proposals."

Perhaps the administration favored such proposals less than tax-cuts, but the fact is that all three were in fact enacted with the president's signature.

Here is a link to a description of the extension of unemployment benefits (made subsequent to the October 4, 2002 Times column to which Krugman refers). http://www.whitehouse.gov/news/releases/2003/01/images/20030107-5_unemployment-010720-515h.html

Here is a link to a Congressional Budget Office description of last year's tax bill, including $20 billion in aid to states. http://www.cbo.gov/showdoc.cfm?index=4249&sequence=0

Here is a link to a White House document detailing the rebates to low- and middle-income households in last year's tax bill.
http://www.cbo.gov/showdoc.cfm?index=4249&sequence=0

These actions may have not been taken in the exact form that Krugman would have liked, but he is in error to say that "all such proposals" were "rejected" by the administration. Further, and more important, he is in error to say that "we'll never know" whether such measures would have helped the economy in virtue of the fact that they have not been tried -- they have, indeed, been tried.

Considering that this particular column concludes with an indictment of Bush's economic policies based on their results, it is incumbent upon The Times to confess Krugman's error and correct it.

Please feel free to contact me directly if you need further information.

Sincerely,
Donald L. Luskin

Here's the reply I got from Collins:

Dear Mr. Luskin,

I've discussed your objections with Paul Krugman and am convinced that everything he wrote was well within the realm of acceptable opinion writing.

On the issue of the unemployment rate, Paul notes that the decline since June is due to a decline in the labor force participation ratio, and that the employment-population ratio is actually down. That seems to me to be ample support for his argument that the unemployment rate did not reflect an improved job situation.

On the second issue, the clear sense of the column was that the administration did not employ a strategy for improving the economy that centered on the ideas Paul likes -- such as extended unemployment benefits and aid to local governments as opposed to high-end tax cuts. I don't think he meant to convey the thought that none of those elements occurred at all. Your argument would be a good point to make in a debate, but it is not the kind of thing that triggers a correction.

Finally, I'm going to have to call a halt to our correspondence on these matters. I always try to respond to readers who have complaints or questions but as a matter of self-preservation, I've never allowed myself to get involved in ongoing dialogues. You've been using me as a channel for your arguments with Paul, and while I must admit I've found them interesting, I simply can't keep the conversation going indefinitely.

Best wishes,
Gail Collins

Last things first. This "correspondence" wants to "call a halt to" consists of maybe three emails from me to her over the last two years (of which this is one), and maybe two responses to me from her (of which this is one). How this is "using me as a channel for your arguments with Paul" or why she thinks I expect to "keep the conversation going indefinitely" is beyond me. Perhaps she has me confused with someone else.

Now, on to the matter at hand.

On the issue of the unemployment rate, she is basically just repeating the error. Here's how you can prove to yourself that Krugman is wrong about this. Go to the Department of Labor website and download three data series: the "Unemployment Level - Civilian Labor Force" (series LNS13000000 -- this is the number of people in the civilian labor force who are unemployed); the "Civilian Labor Force Level" (series LNS11000000 -- this is the number of people in the civilian labor force); and the "Unemployment Rate - Civilian Labor Force (LNS14000001 -- this is the fraction of the civilian labor force that is unemployed). Let's put them into an Excel spreadsheet, staring with June 2003 and taking it up to the most recent figures in February 2004. Figures are in thousands, by the way.

What does this spreadsheet tell us? First, as Krugman says, the unemployment rate (column D) has fallen since last June. Second, also as Krugman says, the civilian labor force (column C) has shrunk since June. Third, as I pointed out, it is also the case that the number of unemployed people (column B) has declined since June. Note that the unemployment rate (column D) can be calculated from scratch by dividing column B by column C, as I have shown by doing so in column E.

If Krugman were correct that the fall in the unemployment rate were "entirely the result of people dropping out of the labor force," then you would expect to see any fall in the unemployment level (column B) at least matched by a fall in the civilian labor force (column C). But no. As you can see in the spreadsheet below, the fall in the unemployment level (column B) is far greater than the fall in the civilian labor force (column C), no matter which summer month you take as the starting point.

To turn this into a final proof-positive that Krugman is wrong, look at one more spreadsheet. Here I increase the unemployment level (column B) by any losses since June in the civilian labor force (column C). The effect of this adjustment is to build back into the number of unemployed people and reduction that could be attributable to shrinkage of the labor force. That done, I recalculate the unemployment rate (column E). Even penalizing the drop in the number of unemployed people by the number of people who have left the labor force, today's unemployment rate -- a hypothetical 5.9% -- would still be lower than it was in June. If Krugman were right -- that the drop in the unemployment rate were due "entirely" to a reduction in the labor force, then the hypothetical unemployment rate would now be at 6.3%, where it was last June.

Now let's go back and compare this to Krugman's self-defense, as transmitted through Collins: "Paul notes that the decline since June is due to a decline in the labor force participation ratio, and that the employment-population ratio is actually down." Well, fine. The "labor force participation ratio" and the "employment-population ratio" are indeed down. But so what? Those truths have nothing to do with the error Krugman wrote in his column. He made a very specific statement about the "unemployment rate," which (as I have shown above, and as explained by the Department of Labor) is not based on either of the data series mentioned in Collins's email. His defense is in irrelevant distraction -- not the "ample support for his argument" that Collins dutifully swallows. There is no support whatsoever -- "ample" or anything else -- for Krugman's use of the word "entirely" in his column.

Now, briefly, on to the second matter -- Krugman's statement that "the administration rejected all such proposals" for "extended unemployment benefits, temporary aid to state and local governments, and rebates for low- and middle-income workers." Collins' response is, in essence to restate Krugman's position in terms that make it true: "the administration did not employ a strategy for improving the economy that centered on the ideas Paul likes."

Can Collins not see that there is a world of factual difference between the sweeping absolutism of the original statement and the opinion-centered nuance of the restatement that she wrote? Is her restatement not, indeed, a correction?

We're discovering something important, here, about the Times columnist corrections policy. Maybe we've been wrong all along to think that the solution is to have someone other than the columnist himself responsible for corrections. Maybe we've been wrong to think that such a thing would assure accountability. Because for that to work, we'd have to believe that there were people at the Times who had the integrity and the courage to admit that a mistake had been made. It would be no improvement at all to have Gail Collins or anyone reporting to Gail Collins manage columnist corrections.

Correction: As originally posted, the dates Jan 2003 and Jan 2004 were entered in the spreadsheets instead of Jan 2004 and Feb 2004, respectively. It has been corrected above.

Posted by Donald L. Luskin at 12:55 AM | link    

Tuesday, March 16, 2004

Posted by Donald L. Luskin at 12:01 PM | link    

Posted by Donald L. Luskin at 11:58 AM | link    

Monday, March 15, 2004