The first graph comes with the following description:
Credit: Professor Amir Sufi. “This maps the unemployment rate in a state in 2010q4 against the percentage drop in construction of new residential units from 2005 to 2010 (i.e., [new units constructed 2010 â€“ new units constructed 2005]/new units constructed 2005).” In other words, you’re looking at points for the % decline in new residential unit construction from 2005 to 2010 by state and for the state unemployment rate at the end of 2010. The GA for Georgia can be found right on the trend line at the top left. States above the trend line have experienced higher unemployment than might have been expected from the decline in new construction, while states below the trend have held up a little better. But the trend is clear. Thanks to Professor Sufi.
The second graph shows both “the correlation and the lag” between housing starts and unemployment. Note that the unemployment rate is inverted. The housing starts are in thousands and use the Seasonally Adjusted Annual Rate (SAAR). Unemployment typically continues to go down for about 18 months after housing starts peak. This might seem counter-intuitive, but housing starts don’t just create employment during the construction phase. The spending of new homeowners and of workers fuels all sorts of other jobs. After the 2001 recession, the correlation slips, with unemployment remaining unusually high even as the building boom took off. That’s likely because so many of those new starts would not ultimately be occupied or even sold. The quandary now is obvious: we don’t need any more housing starts, so what’s the best way to reduce unemployment in the absence of such an obvious spur?
Interestingly, there’s a preponderance of southern states above the trend line in the first graph. It would be interesting to look at the states below trend vs. those above trend to see what specific factors could account for some states holding up better than others after the dramatic decline in housing starts.