As we can see from Table 1, full cycles occur approximately every four years. During this period, bull markets averaged about three years, while bear markets averaged less than a year. A more detailed investigation that includes presidential election cycles for the period from 1941 through 2000 reveals some interesting findings. Stock market lows have occurred surprisingly close to mid-year congressional elections, or approximately two years before presidential elections. (See Table 2.)

Table 2
Presidential Elections and Market Troughs

Table 2 clearly shows a clustering of bear market lows around the congressional election period, or about two years into the presidential term. As can be seen, three of the 16 bear market lows occurred in year one of the presidential term, 12 in year two, one in year three, and none in year four (the election year).

Potential Investment Strategies Based on the Political Cycle

Price data for the S&P 500 Index was compiled and averaged on a weekly basis over the period from 1942 to 2003. Analyses of these data suggest that a potentially lucrative investment strategy would have included buying on October 1 of the second year of the presidential election term and selling out on December 31 of year four. This simple strategy would have sidestepped practically all down markets for the last 60 years. For the most part, bear markets have historically occurred during the first or second years of presidential terms. (A bear market is defined here as the S & P 500 Index’s decline approximately 15 percent or more over a period of one to three years, while a bull market is an environment of consistently rising prices.) As shown in the above table, no bear markets of this magnitude have occurred during an election year for the time covered.

It is also apparent that markets are subject to change from time to time because of unforeseen macro events. As a result, some cycles have been shorter and some longer than the norm. For example, 1946 to 1949 was a shorter cycle, while 1982 to 1987 and 1994 to 1998 were longer than average cycles. One cannot point to any one factor that has directly caused bull market runs of the last two decades to be longer in duration than those during previous decades. To speculate on the above is beyond the scope of this paper, although fundamental economic conditions and the role of the Federal Reserve are factors.

The Test

Based on discussions above and the notion that the S&P 500 Index seems to bottom approximately two years into presidential terms (Table 2), we can construct a hypothetical test for two investors that calculates the dollar return for two simple alternative investment strategies. In neither case is allowance given for possible dividends or interest earned. In addition, commissions and taxes are ignored for the purpose of simplicity. Prior to the 1952 presidential election, the U.S. government generally did not attempt to influence the economy in any significant way, so the period before 1952 (except for the World War II period) is not useful for testing this strategy. Therefore, the test begins with the 1952 election period.

Imagine that the first investor had consistently purchased the S&P 500 Index 27 months before presidential elections and had sold near election time on December 31 of the election year. Because a 27-month period seems to provide better returns than other studied periods before the election, a 27-month period was selected for this test. This strategy kept Investor 1 out of the market from January 1 of the inaugural year through September 30 of the second year during the test period. On the other hand, imagine further that Investor 2 bought the S&P 500 on the first trading day of the inaugural year of each presidential election during the test period and liquidated the portfolio on September 30 of the second year of the presidential term. Would either or both of these simple procedures have consistently made money for the investors? Table 3 below reveals the results on both a percentage change basis and dollar return.

Table 3
Percentage and Dollar Returns of Two Investment Strategies
(Dollars Amounts are Cumulative)
Starting Value for Investors 1 and 2 = $1,000

The findings in Table 3 are extremely interesting. The first investor put up money 13 times and did not lose money in any period. Gains ranged from a high of 70 percent prior to the 1976 election to a low of 16 percent before the 1960 election. Investor 1 saw the original investment of $1,000 grow to $72,701. This is a percentage gain of 7,170 percent. Investor 2 was not so fortunate. This individual also anted up 13 times, but lost six times. The largest loss of -36 percent was seen after the presidential election of 2000. Investor 2 saw the original $1,000 shrink to only $643, or a loss of -36 percent, in nearly five decades. That percentage is based on nominal dollars and does not include the impact of inflation.

Graphing the percentage gain and loss makes the difference quite obvious.

However, when you look at a graph of the cumulative dollar difference between the two strategies, the difference is even more dramatic!

Final Thoughts

It is not the intent of this paper to forecast the stock market. Even where patterns exist, there is enough variability that it is risky to try to anticipate specific turns in the market. Yet we have identified a potentially profitable four-year stock market cycle that has worked well over the better part of the last century. Investing for the 27 months before a U.S. presidential election certainly seems to be more profitable than investing during the 21 months after the elections.

However, just when you think that you have figured it all out, you find another pattern that can suggest different possibilities. For instance, another analysis shows a highly intriguing re-occurrence in the stock market index. During the entire twentieth century, every mid-decade year that ended in a “5″ (1905, 1915, 1925, etc.) was profitable! This is not to say that all of these years had uninterrupted ascending trends, but by year’s end there had been impressive gains. Whether that pattern was a fluke or will continue in the 21st century is anyone’s guess. And, 2005 is also an inaugural year.

However, trying to figure out such patterns can certainly make life interesting.