Monday, May 16, 2011

The 1990 Recession

This post is a very small case study on the 1990 recession associated with the first Persian Gulf war.  This is of interest to me because it occurred in the pre-web era, so hasn't been nearly as chewed over in the blogosphere as more recent recessions.  However, it's recent enough that we have monthly data easily available for almost all variables of interest.  The panel above shows a number of them.

First to summarize the war: on August 2nd, 1990, Iraq invaded Kuwait.  On 17th Jan, 1991, the US began bombing Iraqi positions in Kuwait, and by Feb 28th, President Bush (I) declared a ceasefire.  The effect on oil prices can be seen in the middle panel above, blue curve.  Oil prices were declining in early 1990, and then more than double very rapidly beginning in August 1990.  So the war was very clearly an exogenous event to the US economy, and was clearly causal to the oil shock.

So can we say that the oil shock was the exclusive cause of the recession?  As so often, it's more complicated.  The top panel above shows real GDP and employment (three month changes, annualized).  You can see that the economy was clearly slowing in the second quarter of 1990, before the war.  Although these two variables hadn't started to actually decline until August, the slowdown was more pronounced than at any time in the previous two years.

You can see one possible cause of that in the lower panel - the red curve is housing starts (left scale), and those start declining at the very end of 1989, seem like they might be stabilizing in mid 1990, and then go firmly in the tank when the oil price shock comes.  Consumer confidence (right scale, blue curve) however, is pretty much unaffected until August when it declines dramatically as gas prices hit consumer pocket books.

So what caused the early decline of housing starts?  Well, house prices had peaked in mid 1989, and were declining before the war began:

This may be on account of the Fed raising interest rates though Q2 1989, until they curbed the run-up in housing prices (red curve, middle panel above). Checking the minutes at the time, they weren't thinking about housing prices at all, but that does seem to be something they accomplished notwithstanding.

So was there a recession because of the oil shock, or because of a housing bubble being popped by the Federal reserve?  Probably some of both, I'm guessing.  We don't seem to have a way to know for sure whether, had the war not happened, the economy would still have gone into recession, or, after a brief slowdown in early 1990, would have recovered.  And similarly,  had the economy not already been weakening in Q2 1990, would the oil shock have been enough to trigger a recession?  I believe so, but I can see that the case is not proven beyond a reasonable doubt.


KLR said...

I notice the weekly EIA product supplied series begins in Feb 1991 - maybe in reaction to the war and/or recession? Gasoline supplied was down sharp in Sep 1990, VMT dipped here as I recall. Nice to see these various factors at work in historic examples. Kinda makes you yearn for more recessions...well, maybe not!

This guy has a very different chart of housing starts: Home Sweet American Bubble Investing Pie: Lessons from the Great Depression Part XXIII: The Worst Housing Crash in American History. » Dr. Housing Bubble Blog He shows an absolute peak in 1984 - very sharp, probably an anomaly. A local peak happened in 1989 but just as part of a secular trend down beginning in 1985. So that shouldn't have taken anyone by surprise - or maybe the economy can only handle so few starts, like it can't handle more than 6% of GDP being spent on oil? Other bloggers seem quite taken with housing being critical to the US money machine's smooth operation.

Greg said...

"in the lower panel - the red curve is housing starts (left scale), ... Consumer confidence (right scale, blue curve)"

Mis-labelled in the legend...?

Stuart Staniford said...

Greg - you are correct that the legend in the lower panel is reversed. I'll try to fix later.