Why do we overweight NFP ?

Should we put so much weight on NFP?  Statisticians suggest "probably not."

The WSJ lays the blame for the latest obsession (in part) at the feet of Fed Chair Bernanke: 

"Yet from a statistician’s point of view, the market’s reaction to the jobs data is hard to fathom. Over the past six years, the economists’ consensus has missed the reported number, on average, by about 82,000 jobs, according to Bianco Research. That might look like a big difference. But as a percentage of total payroll employment — 135 million — it’s actually very small, less than one tenth of 1%. Because some miss is always expected, it would take a much bigger divergence from Wall Street’s expectation — say a number of more than 200,000 above or below consensus — to be meaningful, says David Juran, a statistics professor at Columbia Business School.

For a while, it seemed that the bond market had come to its senses. At the end of last year, according to Bank of America, the sensitivity of bond prices to the jobs report had fallen sharply. Now, though, the Federal Reserve’s new chairman, Ben Bernanke, has stated that future monetary-policy moves will depend on changes in economic data. Hence, like it or not, traders — and Wall Street economists — will be glued to their screens this morning. "If I’m a trader I have to trade off of it, because everyone else does," says Paul Kasriel, chief economist at the Northern Trust Co. in Chicago."

Interesting stuff, though I have two criticisms:

First, The Street  has been fairly obsessed with NFP for years, so its not really fair to lay all the blame on Bernanke. I read his "Data Dependant" schtick as a ready excuse for doing what he wants to do anywway. This is especially the case, given what we know about how subject to revision this data is, and how long it takes for the Fed’s action to be felt in the economy.

Second, the statistical take given above is glib and misleading — Yes, its true that the average miss of 82,000 jobs as a percentage of total 135
million payroll employment is small, less than one tenth of 1%.

But we track NFP to ascertain how many new jobs are being created within the economy. That provides insight into how robust the cycle is, how much money consumers will have to spend, etc.

A miss of 82k per month — when the trailing 12 month average of new jobs created is about 175k per month — is off by a huge percentage: ~47%.

Payroll Day
WSJ, March 10, 2006; Page C1

What's been said:

Discussions found on the web:
  1. Curmudgeon commented on Mar 10

    memo to the geniuses at Columbia Business School: a big average forecasting miss means the number has… drum roll… high informational content.

    if everybody nailed it every time, it would be a snoozefest, like say leading indicators.

    the employment report is the first info on the previous month, gives you some ideas on hours worked in different sectors, industrial production, construction, personal income…sets the tone for the whole month.

  2. fred c. dobbs commented on Mar 10

    A miss of 82K on the forecast is nothing. The statisticians at the BLS (your fav villians) say the payroll number they report is statistically significant within 99K or so. Once again, we must learn to look at the trends, not the monthly number.
    Of course bond traders pay too much attention to essentially meaningless monthly data, but what else can they do? They aren’t qualified for any productive work.

  3. kharris commented on Mar 10

    “A miss of 82k per month — when the trailing 12 month average of new jobs created is about 175k per month — is off by a huge percentage: ~47%.”

    Let’s remember what it is that is being forecast. We aren’t talking about the underlying number, but the first difference. I can come within 0.5% of the monthly jobs tally in my sleep. Can you get within 0.5% of IBM’s month-end stock price with only the prior month-end price as the last level you know? The lower limit for both stock prices and payroll employment is zero, right?

    Try getting within 42% of the monthly change in a stock’s value before casting asparagus toward forecasters of jobs, or retail sales, or whatever. Or try publishing a jobs forecast yourself, month after month.

  4. Barry Ritholtz commented on Mar 10

    kharris —

    The “asparagus” I was casting wasn’t on the jobs forecastors — it was on the statisticians who claimmed the 82,000 per month average miss was “as a percentage of total payroll employment — 135 million — it’s actually very small, less than one tenth of 1%.”

    Thats sleight of hand — its the new job creation that we track — not the total.

    Back to decaf for you

Read this next.

Posted Under