Custom Search

THE TWO PATTERNS To form the four-year Presidential election pattern, I took 80 years of the annual return of the Dow Jones Industrial Average (DJIA) index, starting from 1912 and continuing to 1992. I then organized them into four columns, each representing a type of year relative to a Presidential election:

1 Year before an election
2 Year of the election
3 Year after the election
4 Mid-term year

The 80 years thus form 20 complete trips through the four-year cycle. Over these 80 years, the annual Standard & Poor's 500 index returns have varied from -51.7% to +38.7%, but the average of the four types of years break down as follows:

Pre-election 11.0%
Election7.0
Post-election4.7
Mid-term 2.3
Average6.3

The average return of each type of year is close to what was expected. The pre-election and election years are indeed more attractive on average for investors than the post-election and mid-term years are. In my analysis, the mid-term year performed worse than the post-election year. Other studies, however, have found the reverse.

Constructing the eight-year pattern is similar, but the annual returns are organized into eight distinct columns, not four. The average of each column represents the return for each type of year of the pattern. The results were:

1 First pre-election year: 16.5%
2 First election year: 12.3
3 First post election year: 5.7
4 First mid-term year: 3.3
5 Second pre-election year: 5.5
6 Second election year: 1.7
7 Second post election year: 3.7
8 Second mid-term year: 1.3

Bar charts can help visualize the two patterns. Figure 1 is the traditional four-year pattern shown twice consecutively, while Figure 2 is the eight-year pattern. Each bar represents the average return of the various types of years of the patterns.

On Figure 1, bar 1 is the pre-election year, bar 2 is the election year, bar 3 is the post-election year and bar 4 is the mid-term year. Bars 5 through 8 are repeats of bars 1 through 4. Note how the four-year pattern forms a neat formation where the average return of each successive year is less than the year before. The pre-election year is the strongest, down to the mid-term year as the weakest.

The eight-year pattern in Figure 2 helps illustrate important differences between the two patterns. The first four years follow the same pattern as the corresponding four-year pattern, but the second four years are different; the returns are less. Year six, the second election year, is unusually weak. Had it been stronger, say an average of 4.9, it would have continued the orderly pattern of progressively weaker years. Finally, the eight-year cycle reveals quite a bit more variation in return (between 16.5 and 1.3) than the four-year pattern (between 11.0 and 2.3).

Seeing the Presidential election pattern in this new way suggests a different interpretation of the data. Rather than a single pattern that repeats each four years, perhaps it should be seen as half of a larger and more meaningful eight-year pattern. Knowing which four-year leg of the larger pattern we are in makes an important difference.

What does this eight-year pattern suggest for the remainder of the 1990s? This year, 1994, is represented by bar 4, suggesting that the past three years, 1990 to 1993, were more favorable than what we can expect from 1994 to 1998. Then according to the pattern, 1999 and 2000 should be the next unusually strong years.