Occurrence and Duration of Recessions in Relation to the Political Party in Power

As the eight-year administration of President George W. Bush comes to an end, we are almost certainly in a recession. We won’t know for sure until we get fourth-quarter GDP numbers early next year, but most believe that we are well into a financial crisis that is the worst since the Great Depression. The current presumed recession will be the second in the eight years of President’s Bush’s two terms. This follows the eight-year administration of President Bill Clinton when the country did not go into recession. This prompted me to ask these related questions. 1) Is the country more likely to go into a recession when a Republican or a Democrat is elected President? 2) Do recessions last longer when they occur during a Republican or a Democratic Presidency?

The longest data set I could find on the timing and duration of recessions in the United States came from the National Bureau of Economic Research (NBER) and can be obtained at www.nber.org/cycles. It encompasses the period from the beginning of 1857 to the present. This interval contains 32 recessions. The Political Party in the White House can be found from any number of sources and is certainly not in dispute. The data set with which I started was a month-by-month record of the Party occupying the Presidency and whether or not the country was in recession. Because we don’t yet know whether and when a recession has begun in 2008, I used a data set that ended in 2007.

The first question I posed was this: given that a Republican is in the White House, what is the probability that the country is in recession? I then posed the same question about a Democrat. The answers were that 37% of the time that a Republican was in the Oval Office, the country was in recession. This was almost one and a half time more likely than when a Democrat was in office: the country was in recession 26% of the months when a Democrat was in office.

So, this seems to answer the first question: the country is more likely to be in recession when a Republican is President than when a Democrat holds the post. However, I’m sure that a number of people would immediately point out that if I am really interested in assessing responsibility for recessions, perhaps it’s best to look at who was President at some point prior to a recession, rather than at the same time. If the country is in recession when a new President takes office (as will almost certainly be the case for President-elect Obama), does it make sense to attribute the economic downturn to him or her? I don’t think so.

What is the appropriate time lag to use when trying to investigate cause and effect for a recession? I doubt there is much agreement about this. National Budgets are for one year. The time between Congressional elections is two years – one could argue that pertinent economic policy decisions might follow this interval in some way. I suppose one could also argue that the full term of a President might be needed before the full impact of policy would be manifested in the national economy. I doubt many would argue for a longer time lag, but who knows?

I decided to investigate two- and four-year time lags. Therefore, I reconstructed my data set, but this time I shifted the Political Party in the White House ahead two and four years in two separate analyses. In effect, this allows one to ask if the Party in power at a point in time has an effect on whether or not the United States is in recession two (and four) years later. These two analyses reduced the longevity of the data set I could use: I could not use 2006 and 2007 for the two-year time lag assessment, and had to further exclude 2004 and 2005 from the four-year analysis.

The results are almost identical to those with no time lag. With a two-year lag the probabilities become 37% and 25% for Republicans and Democrats, respectively. For a four-year lag, the numbers become 38% and 24%. With no time lag or with a time lag of two or four years, the probability of the country being in recession is one and half times more likely when a Republican is President. This difference is statistically significant.(1)

This still leaves a question unanswered: is this increased likelihood of recession with a Republican President due to the greater chance of a recession starting with a member of the GOP in office, or is it because recessions associated with Republicans last longer?

Excluding the current term, which I will not include for reasons stated above, my data set consists of 37 Presidential terms. Of these, Republicans were in office 22 times and Democrats 15. As I mentioned earlier, this timeframe encompasses 32 recessions. Republicans were in office when 20 of these began, while Democrats occupied the White House when 12 recessions started. This may seem like a substantial difference, but recall the prevalence of GOP Presidents during this time (22R, 15D). The ratio of 20 to 12 is not significantly different from that of 22 to 15. The ratios of “Republican” to “Democratic” recessions vary little when a two-year (16R to 15D) or four-year (17R to 13D) time lag is incorporated into the analysis. Therefore, we can conclude that recessions are not more likely to begin when one or the other Party is in power.(2)

This suggests that the enhanced proportion of years in recession associated with the Republican Party is not due to increased frequency of Republican-associated recessions. However, when I examined the length of recessions, I found that Republican-associated recessions were substantially longer: the difference was similar if no time lag is introduced (R: 21.0 months, D: 14.3), a two-year lag is used (R: 23.9, D: 12.7) or a four-year time lag is examined (R: 22.9, D: 13.5).

The recession that in all likelihood began earlier this year is clearly associated with the Republican administration of President George W. Bush. If history is a guide, it will be one that is far longer than average. Let’s hope that is not true.

Statistical stuff:
(1)When analyzing this data, there is a problem with the lack of independence of data. Clearly, each month does not represent an independent assessment of the relationship. At what interval of time are they independent? I’m not sure there is an “answer”. I decided that yearly intervals were sufficiently independent, so I restricted my statistical analysis to January of each year. I used a 2x2 contingency table analysis evaluated by chi-square. The results for zero time lag as well as two- and four-year time lags were all significant (p<0.05). The chi-square values were 4.6, 4.2, and 4.4 for these three analyses, respectively.
(2) A 2x3 contingency table analysis with “Political Party” and “Number of Recessions” as factors did not even approach statistical significance (0.8 < p < 0.7). Clearly, these numbers are not independent and therefore violate the assumptions of the test. I include it only to underscore that there is not even a hint of a difference here.

Welcome!

Welcome to my new blog! For several years, I have used publicly available data in the realm of macroeconomics and politics to gain understanding about what is really going on in our country and world, and to convey this information to others in a simple and straightforward manner. I have been a columnist, primarily in the Carroll Star News, and have spoken at a number of meetings in West Central Georgia. I have decided to create this blog to continue this process. I hope you will find my entries useful. I welcome the opportunity to answer questions that arise in the minds of readers – feel free to contact me if there is something you’d like me to investigate.