Tuesday, November 2, 2010

Causality: Economics and Politics Edition

Kevin Hassett, via The Oregonian, makes a fundamental error of the sort that you all know will set me off: mistaking correlation for causation.  This one seems so classically wrong that I offer it as a teaching moment on this election day.  Here is the thesis:

Chances are good that the power of the federal government, squarely in the hands of Democrats the past two years, is about to become divided between the two parties. Modern American history suggests that this is the best of all worlds. With due respect to Abraham Lincoln, a house divided against itself often prospers.

Since 1970, the levers of federal government -- the White House, Senate and House of Representatives -- have been in the hands of one party, whether Republican or Democrat, 30 percent of the time. By most any measure, the U.S. economy has been healthier the other 70 percent of the time.

His evidence to support this thesis:

From 1970, median GDP has grown 3.3 percent in years of divided government (1970-1976, 1981-1992, 1995-2002, 2007- 2008), compared with 3 percent when government was unified (1977-1980, 1993-1994, 2003-2006, 2009-2010).

The effect has been more pronounced in recent years. Since 1981, when Ronald Reagan took office, median GDP has grown 3.3 percent in years in which the government was divided and 2.8 percent when government was unified. Since 1993, when Bill Clinton took office, the economy has grown 3.6 percent per year when power is split and 2.8 percent in the other times.

How about unemployment? Divided government might be exactly what today's jobless should be wishing for.

Since 1970, median unemployment has been 6.1 percent under one-party rule, 5.7 percent when both parties have some control. The spread narrows (6 percent vs. 5.8 percent) since 1981 and widens (6 percent vs. 4.9 percent) since 1993.

Equity markets have practically jumped for joy at political division. Since 1970, the Standard & Poor's 500 Index has increased at a median rate of 13.5 percent per year in divided times and 9 percent per year under one-party rule. That spread grew (14.6 percent vs. 9 percent) since 1981 and even more so since 1993 (19.5 percent vs. 9 percent).

Open and shut. Uh, except for a few inconvenient facts. First, economic policy of the type made in the Capital Building in Washington, DC has very significant lags: the effects of a new piece of legislation are generally not felt for years. Second, divided governments are often the result of an unhappy electorate, and what makes electorates the most unhappy?  Poor economic performance.  Finally, the business cycle has always been a part of the economy (even when we thought we had achieved the 'great moderation') which means that poor performance is generally followed by good performance.

Thus it is just as, or even much more, likely that the causality works the other way: poor economic performance, especially around midterm elections cause divided governments and natural business cycles causes economic improvements after such elections.  So it is the economy that causes divided governments not divided governments that cause the economy.

This time will be no different: it is virtually certain the the economy will improve in the next two years (there is no where to go but up) and it is virtually certain that the Republican party will control at least the House.  Will the fact that the Republicans control the house have any real effect on the economy in the next two years?  Not really.  The roots of the current economic downturn run from the deregulation of Reagan to the financial reforms of Clinton to the loose monetary policy of Bush and has basically nothing to do with a specific party of what branches of government they control.

Hey, it is election day: VOTE!

No comments: