US unemployment will be 6.6% by June 2009

I recently fell upon a blog that was arguing for the link between new housing starts and unemployment. He showed a graph that looked a lot like this one:



I tend to be very skeptical when people find correlation between two sets of data, but it turns out that housing starts are the real deal. Not only do housing starts correlate well with the unemployment rate, but they are that rarest of animals: a reliable leading indicator. The number of new housing starts can predict 40% of the variation of the unemployment rate 10 months ahead. More specifically, you can predict the unemployment rate 10 months ahead using the following equation:

UE = (2520 / HS) + 3.71

The following graph shows the predicted unemployment rate and the realized rate:



Given the data just released of the number of new housing starts at 854 for August, this predicts an unemployment rate of 6.6% for June 2009. The pain isn't over yet.

Side note: I would have assumed that the absolute number of housing starts would have trended up over time as the population increased, but this hasn't really been the case. I'm guessing this is for demographic distribution reasons: the number of young families buying houses has stayed more constant than the total population.