Market Returns and the Election Cycle

June 01, 2012 | Doug Oest, CAIA, Managing Partner

With election campaigning in full swing, we have received a number of questions from our clients regarding what will happen to the market if a particular candidate or party wins, or whether certain years of the presidential cycle are better for investors. This week’s chart of the week examines past studies on election years and market returns, as well as other market patterns.

Thus far in 2012, there have been numerous articles focused on finding the relationship between the market cycle and the election cycle. Notable findings of these articles are highlighted below:

• The stock market tends to be positive in an election year. The median return over the 21 election years since 1926 has been 11.1%.

• The stock market has performed better in years where a Democrat has been president (median return of 18.4% vs. 7.7%).

• The first year of a presidential term is typically a poor year for investors (median return of 4.9%), while the third year of a term is typically a good year for investors (median return of 22.7%).

Heading into the 2008 election year, various articles highlighted similar election year market performance, which at that time had a median return of nearly 14%. Of course, 2008 turned out to be one of the worst years for the stock market. This performance was not due to the election year, but rather a massive collapse in the housing market, the credit crisis, and one of the deepest recessions the United States has experienced.

While certain patterns may exist in the return data, the data set is extremely limited; it most likely is a case of identifying random patterns in limited data set. For instance, an investor who only invested in the stock market during odd years would do significantly better than investing in all years, or an investor who only invested in even years. Similar extrapolations can be made based on years ending with a certain digit (3, 5, etc).

In order to be statistically significant, one would need over 2,000 election year data points in order to achieve a 0.05 significance level. Similar levels of observations would be needed for the other data points highlighted in the table. While there may or may not be particular reasons behind these realized returns phenomenon, from a statistical standpoint it would be unwise to base any investment decisions off of them.