Thursday, November 25, 2010

Revisiting Educational Hiring

Earlier this year, in May, the New York Times ran the article "Teachers Facing Weakest Market in Years" (May 19th, 2010). The article claimed that
Even upscale suburban districts are preparing for huge levels of layoffs. School officials and union leaders estimate that more than 150,000 teachers nationwide could lose their jobs next year, far more than any other time, including the last major financial crisis of the 1970s.
Corrections suggested this was nonsense, that the comparison was faulty, and that the Times only offered anecdotal evidence. A commenter, Student1776, suggested that, given there was little evidence for writing an article about how poorly teachers were faring, the Times might have ulterior motives.

Here, Corrections offers a bayesian update on that hypothesis. The summer was not unkind to educational occupations. Below, we offer the residual of HP filtered, deseasonalized educational hiring and separations ("taking out" the long-run business cycle variation and homogeneous monthly effects).  This image indicates to Corrections that teacher separations (including quits) have not suffered much during the recession (click to enlarge).
For those unfamiliar, Hodrick-Prescott Filtering ("HP filtering") is a way of separating business-cycle frequencies from time-series data.  For those interested, we can graph GDP and HP filtered GDP from the first quarter of 1947 to the third quarter of 2010, in billions of chained 2005 dollars (click to enlarge):
For visual purposes, we can zoom in on the first quarter of 1980 to the third quarter of 2010, to help understand the severity of the recent recession (click to enlarge):
 Finally, it is important to note that while deviations from business cycle trends look like quarterly GDP growth, they aren't.   For better understanding, we offer a graph of recessions from 1947:Q1 to 2010:Q3 and deviations from HP filtered GDP in billions of 2005 dollars (click to enlarge).
Hopefully, this brief introduction to the HP filter makes clear that our extraction of business cycle trends are not taking out the phantom educational job loss claimed by the New York Times.  This job loss does not appear to exist on either the hiring or separation margin, unless one's loss function is as biased toward finding teachers in trouble as the one used by the New York Times.

No comments:

Post a Comment