A major shift in U.S. macroeconomic policy has occurred since 1960: The economy has been managed to expand prior to an election and contract thereafter. Over the 1961-78 period, stocks returned an average 21.7 per cent in the year beginning two years prior to a presidential election, 15 per cent in the year before the election, 3.6 per cent in the year immediately following and 15.2 per cent in the second year following. The odds strongly favor stock prices rising relative to trend over the two years prior to a presidential election.

The authors tested a trading strategy that involved purchasing stock on the last trading day of October two years prior to the presidential election and selling at the close on the last trading day of October preceding the election. Whereas, over the 1960-78 period, a buy and hold strategy returned two per cent per year, the basic trading strategy returned 5.4 per cent per year.

The finance literature asserts that price swings in the stock market are random, hence unpredictable and unexploitable. But many past tests of the efficient market hypothesis failed to test it against a powerful alternative. The four-year election cycle of stock prices provides just such an alternative. Except over the short periods (day, week or month) studied by academics, changes in stock prices are not random.