DOES the stock market do better when a Republican is president or when a Democrat is?

The answer is: It's not even close. The stock market does far better under Democrats.

This perhaps surprising finding is examined by two finance professors at the University of California at Los Angeles, Pedro Santa-Clara and Rossen Valkanov, in an article titled "Political Cycles and the Stock Market," published in the October issue of The Journal of Finance.

Professors Santa-Clara and Valkanov look at the excess market return - the difference between a broad index of stock prices (similar to the Standard & Poor's 500-stock index) and the three-month Treasury bill rate - between 1927 and 1998. The excess return measures how attractive stock investments are compared with completely safe investments like short-term T-bills.

Using this measure, they find that during those 72 years the stock market returned about 11 percent more a year under Democratic presidents and 2 percent more under Republicans - a striking difference.

This nine-percentage-point excess can be broken down further into an average 5.3 percent higher real return for the stock market and a 3.7 percent lower return for Treasury bills under Democratic administrations.

This finding raises three other questions. First, is this just some data anomaly resulting from selective choice of sample or quirks of the analysis?

Second, if the effect is real, why don't investors take advantage of the predictable higher returns and buy stocks before elections that Democrats are likely to win, in that way pushing stock prices up and lowering returns? The third, and perhaps the most provocative, question: What is the economic rationale for the difference in returns?

Most Democratic administrations clearly had higher-than-average excess returns, with Franklin D. Roosevelt's second term (1937-41) being the only significant exception. Republicans have been associated with lower-than-average returns, with the only significant exception being Dwight D. Eisenhower's first term (1953-57).

The difference in returns persists even if one looks at subsamples. For example, if you break the sample in two at 1963, the Republicans still come out with lower returns in both periods.

What would cause such a large difference? One possibility is that Wall Street investors expect the Democrats to be bad for the market and sell their stocks before elections that the Democratic candidate is likely to win. Then, when the Democrats do not prove as bad as expected, stock prices rise again.

The authors, though, find that the data do not support this theory - stock prices generally do not tend to decline before elections that Democrats win.

But this finding itself raises another puzzle. If returns are so much higher for Democratic presidents than Republican ones, shouldn't we see investors rushing to the market when a Democratic victory looks likely?

We don't see that happen, either. "In sum,'' the authors write, "the market seems to react very little, if at all, to presidential election news."

Here's another theory. Economic policies under Democratic administrations may tend to be more volatile than those under Republicans - so investors demand higher returns to compensate them for the extra risk. A clever idea, but it is also contradicted by the evidence. If anything, the volatility of stock market returns is slightly higher under Republicans than under Democrats.

One interesting finding is that although both large and small companies do better under Democratic administrations, small companies do especially well, while larger ones do only a little better. The return on the smallest 10 percent of traded companies is 21 percent higher during Democratic administrations, while the return on the largest 10 percent is only 7.7 percent greater. What accounts for this difference?

We don't know.

Of course, there is always the possibility of a spurious correlation. As econometricians say, "If you torture the data hard enough, it will confess to anything." People have been looking for economic predictors of stock market behavior for decades, so it's not surprising that every now and then we find some correlations.

Still, presidents like to think - or at least claim - that they influence economic activity. So a finding that the party occupying the White House has an impact on stock market performance should mean something.

With respect to the questions asked above, Professors Santa-Clara and Valkanov can give a firm answer only to the first: Stock market excess returns have definitely been higher under Democrats than under Republicans.

They also show that some tempting possible answers to the question of why investors don't take advantage of this difference do not work.

They do not try to answer the last question, but they conjecture that the fiscal and regulatory priorities of presidents offer a possible explanation.

Most provocatively, they suggest that the causality might go the other way, with market returns driving presidential elections. Perhaps voters feel wealthier when stock prices are high and then vote Republican; when stock prices are low, they vote for Democrats.

The Santa-Clara/Valkanov finding offers an attractive area of research for both economists and political scientists. But even if scholars eventually come up with a satisfying explanation, we are still left with the financial side of the puzzle: For at least 72 years, the stock market did far better under Democratic presidents than under Republicans. How can it be that investors have failed to take advantage of this seemingly predictable pattern?

Professors Santa-Clara and Valkanov wrote the first draft of their paper in 1999, and they admit that they could have profited handsomely by selling stocks after the 2000 election. Alas, like most investors, they didn't sell at the most opportune time. Even finance professors sometimes misjudge the market.

Hal R. Varian is a professor of business, economics and information management at the University of California at Berkeley.

2