National Economic Growth and the Political Party of the President

Economic factors seem to play a major role in elections. It’s hard to argue against the notion that poor economic times during election years tend to lead to a change in the Political Party that holds the White House. However, there is a larger question. How many times have you heard from pundits that this Party or that Party is better for the economy? What is the truth about this issue?

There are a number of different measures of the health of the economy at any point in time. Perhaps the most widely used is the Gross Domestic Product (GDP). This is a measure of the market value of all good and services produced in the country. There are valid arguments to be made about other measures being more valid, such as those that reflect the condition of a majority of Americans, for instance. Here, I’ll examine GDP.

I obtained the historical values of GDP from the Bureau of Economic Analysis (BEA) website. The BEA is part of the Department of Commerce, and therefore part of the Executive Branch of our government. The available data set begins with 1929 and consists of annual values. I chose data that were adjusted for inflation, so the annual changes I computed would reflect change in “real GDP”. I then examined these annual changes in relation to the Party of the President. The data set included information through the GDP of 2007, so annual change values were available for the years 1929 through 2006. This period was divided almost evenly between the Parties: Democrats held the White House for 40 of these years, while Republicans held it for 38.

The average annual change in real GDP was greater when Democrats held the Presidency (+5.2%) was much greater than when a Republican was in the White House (+1.8%). The difference was nearly a factor of three and is statistically significant (1).

There are at least two problems one might have with this analysis. First, perhaps it’s inappropriate to attribute GDP growth for a year to the incumbent of the White House during that year, particularly during the first year of a term. In fact, one could argue that if one is going to attribute GDP to a President, perhaps it’s best to give credit (or blame) to the President in the previous year, when policy decisions were being made that ultimately affected the GDP growth. This suggests “shifting” the data by one year before performing the analysis, or perhaps even two years.

With a shift of one year, the difference between the two Parties is still quite large: when Democrats were in the White House, the real GDP change (+5.1%) was still 2.7x greater one year later than the same value when the GOP held the Presidency (+1.8%). This difference was still quite pronounced when a two-year shift was used (Dems +4.7%, GOP +2.2%).

There is also a purely statistical problem (1). However, I think one can avoid both the time shift problem and the statistical one by extracting only one number from each Presidential term and performing the analysis. What number should be used? Let’s avoid any “time lag” problems by focusing on only the last two years of a Presidential term. Further, let’s use and average of the last two years of each term. So, the data set used is the average change in real GDP for years three and four of each of the Presidential terms since 1929. The result is that real GDP grew by nearly twice as much in Democratic (+4.9%) than in Republican (+2.6%) administrations. This difference was highly significant statistically (2).

One has to conclude that in the entire data set available from the Bureau of Economic Analysis (1929 on), the real growth of the United States’ Gross Domestic Product was much greater in Democratic administrations than in those of Republicans.
___________________________________
(1) An analysis of variance (ANOVA) resulted in a value of F = 9.47 with 1,76 df, p<0.003. Yes, I know that there are questions about the independence of the data.
(2) ANOVA resulted in F = 16.3 with 1,17 df, p<0.001. I don’t see any problem with data independence here.