These Aren't GDP Futures You're Trading

Stocks and the Economy Are Co-Dependent

Equities represent the discounted stream of future dividends. As dividends are generally but not necessarily paid out of earnings -- there are many examples of firms borrowing to pay their dividends -- we should expect stocks to rise if dividends are expected to grow by an amount greater than interest rates will rise. But it doesn't always follow that earnings will rise as the economy strengthens: A strong economy attracts new investment into an industry, which often places downward pressure on operating margins. Nor does it follow that a weak economy is necessarily bad for any and all stocks: Just think back to the 1970s boom in oil stocks or even the recent strength in homebuilders.

In addition, strong economies often are accompanied by rising interest rates. The short and violent bear markets of 1962 and 1987 both were precipitated by rapid increases in short-term interest rates. And to complete the matrix, a weak economy can launch a strong stock market if investors convince themselves that things will soon be much better. This last situation appears to describe the stock market since March reasonably well.

The chart below is striking in how few times the total return on the S&P 500 and the GDP parallel one another. During the two great bull phases of the postwar era, 1949-66 and 1982-2000, the growth rate of total return far exceeded the growth rate of real GDP, and the opposite was true for the 1966-82 trading range and for the post-2000 market.

S&P 500 and GDP Are Different

Source: Bloomberg

In direct answer to the reader's question, we've seen several episodes of falling stocks in a growing economy: 1957, 1962, 1966, early 1984, 1987 and today. All previous episodes were characterized by rising short-term interest rates. Given today's weakly growing economy and the strong but qualified ECRI leading indicator, we might be forgiven for concluding that we won't see a rising economy/falling stocks scenario.

It Is Never That Simple or Linear

Econometric analysis, the art and science of modeling data, is really the art of analyzing residuals, the error term left over after we explain variable Y with some set of variable X. These residuals should be a white noise, or completely random, process. If there's a visible pattern or if they're correlated, that's a sign something is missing in the analysis. That appears to be the case trying to explain the growth rate of deflated S&P 500 total return with the growth rate of deflated GDP.

Something Is Missing

Source: Howard Simons

These residuals are anything but a white noise process. They contain a huge and consistent turn downward during the 1966-82 range and an equally huge and consistent turn higher from 1982-2000. We're in the midst of another too-consistent move right now. If the relationship between stocks and GDP was strong and predictive, none of these features should be visible.

The question then becomes what macroeconomic variables can account for the long periods when stocks either outperform or underperform the economy. I'll try to answer this next week. A few thoughts are in order, though. Several important variables, namely energy prices and currency exchange rates, didn't trade in a free market until the early 1970s, and therefore aren't useful for longer-term analysis. The interest rate ceilings of Federal Reserve Regulation Q didn't disappear until the early 1980s, and financial derivatives didn't reach their current level of acceptance until then, either.

The answer to this question, if it can be found, will help us interpret the financial news a little better. Until then, remember that you're trading stocks and bonds, not GDP futures.

>To order reprints of this article, click here: Reprints