February 9, 2009

Everyone's job prospects are sinking

In discussing new employment data from the Bureau of Labor Statistics when disaggregated by educational attainment, Andy Rotherham argues that "problems are hardly equally distributed and again the returns to education are apparent."

Looking at the table (and at the seasonally-adjusted figures), it's clear that there are differences in January's unemployment rates (12.0% for high school dropouts, 3.8% for college graduates), but there were also differences before the economy fell off the table. How do you tell whether the current differences are changed from prior differences?

My usual tool to use in this situation is the relative odds ratio. Consider one unemployment rate, January's 12.0% for high-school dropouts. The odds that a high school dropout in the labor market was unemployed is the probability of being unemployed divided by the probability of being employed, 12.0%/(1-12%), or about 0.14. The odds that a college graduate was unemployed in January was 0.04. The ratio of the two is 0.29 -- or a college graduate has odds of being unemployed that's a little under one-third of the odds for a high-school dropout. The closer to a relative ratio of 1, the closer the odds are and the less of a difference. A relative odds ratio of 0.29 indicates a HUGE difference.

But the $64,000 question is whether that odds ratio has changed from January 2008 to January 2009. In January 2008, the relative odds ratio was 0.26 for the BA-to-dropout comparison, 0.46 for the some-college-to-dropout comparison, and 0.58 for the high-school-grad-to-dropout comparison. Last month, the relative odds ratios were 0.29, 0.48, and 0.64, respectively. While there is greater movement for the high-school-grad-to-dropout comparison, the story here is that the employment crisis is bad for everyone and slightly compressing the attainment consequences, not expanding them, at least with regard to the official unemployment rate.

There are significant problems with the official unemployment rate, though--it excludes the "discouraged" unemployed and those not in the labor force. Unfortunately, the BLS table on alternative measures (including U-6) is not disaggregated by attainment, so I'll pick on the employment-to-population measure that's in the first table linked above. There, the odds ratios are reversed in meaning (the odds of being employed vs. the odds of being unemployed), and they hardly budge: for the high school grads to dropouts comparison, the relative odds ratio is 2.0 in January 2008 and January 2009; for the some-college to dropouts comparison, the ratio is 3.1 in both months; for the college-grad to dropouts comparison, the ratio is 4.4 in both months. Again, the cross-sectional gaps are huge, but there's no change.

I could go further into the weeds with relative probability ratios, but I'll give you a chance to stay awake and just say that they're going to have the same basic message as relative odds ratios: the job market is getting worse for everyone, and at least at a first glance, being well-educated protects you no more last month than it did a year ago.

Listen to this article
Tags: Andrew Rotherham, human capital, unemployment
Posted in Education policy on February 9, 2009 8:56 AM |