Friday, August 15, 2008
Macroeconomic data is notorious for difficulty in measurement and heavy revisions (for up to 5 years in some cases). The labor market data, compiled by the Bureau of Labor Statistics (BLS) is quite possible the worst of them all; any monthly report is likely to be revised up or down by up to 152,000 jobs.
This chart was published in the Wall Street Journal that accompanies an article about Economists’ predictions over the possibility of a Fannie and Freddie bailout (59% probability that the US Treasury will bailout the two mortgage giants). Not pertinent to the article, but what I found terribly interesting is the distributions of forecasts for GDP, inflation, and jobs (payroll).
Each block represents one forecaster’s prediction for GDP, inflation, and jobs over the next year. The GDP forecast distribution is tight, ranging from a little over 0% growth to around 2.5%. The economy will expand, but everyone agrees that the expansion is below the economic potential. The inflation forecast distribution is also rather tight, where most predictions range from 3%-5% annual inflation. Inflation will subside, but pressures are likely to remain.
On the very other hand lies the jobs forecast distribution, which is clearly wide and totally meaningless. According to the poll, the next 12 months will see an average change of jobs ranging from -100,000 to 120,000 jobs.
Nobody knows what the job market report is and nobody ever will. Further, and to a forecaster’s dismay, our previously solid predictor of the jobs report, the weekly unemployment claims report, has become tainted by needless government aid extensions.
The financial markets move big on labor reports because it is often a good indicator of the underlying macroeconomy. If the labor market is falling (growing) sharply, then consumer spending and sometimes the rest of the economy is also deteriorating. The employment situation is one of the biggest variables that the National Bureau of Economic Research uses in dating a recession. The 2001 recession saw 1.6 million jobs slashed, which dragged down economic activity enough to call a recession even without the famed 2 consecutive quarters of negative GDP growth.
So, to avoid any unnecessary confusion to financial markets (where volatility is already hot), I propose that the monthly labor report, which is always released on a Friday, be released at 5pm, rather than at 8:30am, because the report is inherently unpredictable and subject to heavy revisions.
Wouldn’t it be great if the BLS released the report outside of working hours? Portfolio managers can sit back and drink their martinis as the job losses are reported, rather than needlessly reallocating assets based on a unreliable report. The managers could take the whole weekend to realize that the job number may be revised up or down by 152,000 jobs, relieveing some of the current volatility in the S&P 500, the Nasdaq, corporate yields, or whatever your favorite asset class.
Please leave comments. Best, Nontruths