If You Want to Know Where Stocks are Headed, Don’t Ask the Economy


Stocks should do well when the economy does well. This seems like a no brainer. After all, a strong economy should mean that more profit and growth is available to corporations, and that should make its way down to shareholders. This, in turn, should reward investors who pursue broad market exposures offered through ETFs like SPDR S&P 500 (SPY), and Vanguard Total Stock Market (VTI) and others.

But does it really work this way, and is GDP growth predictive of future stock returns? To answer this question, we compared both calendar year and quarterly changes in real GDP to changes in the S&P 500 Total Return Index.

First, using calendar year data from 1930-2012, we ran correlation and regression analysis on the S&P 500 Total Return Index lagged one year behind real GDP in the U.S. In other words, we asked the question “Does the change in real GDP in year one predict a change in the S&P 500 in the next year?”

So, is it predictive? In a word, “No.” We found a very weak correlation of 6.3% and an even lower R-squared of 0.004. What about staggering by two years? The results were slightly better, but we still found virtually no predictive power. Correlation improved to only 8.7%, and r-squared remained very low at only 0.007.

What if we use quarterly data? The results were generally the same (i.e., poor). Staggering S&P 500 total returns behind real GDP produced the following:

 CorrelationR-Squared
S&P lagged by 1 quarter0.01050.0001
S&P lagged by 2 quarters-0.08890.0079

Time Frame Matters
While we didn’t expect to find much predictive power in GDP, we were surprised at just how low correlations and r-squared numbers actually were. There was one exception to this, however—longer time frames.

Longer periods of time, like decades, produced better results. This time, because the time period was so long, we did not lag our data. We simply asked, “Does a change in real GDP in a certain decade affect stock returns in that same decade?”

When we look at compound annual growth rates for both GDP and the S&P 500, we see higher correlations and more predictive power in our regression. Of course, as any statistician will tell you, longer periods smooth out your data, so any positive relationship should become amplified (i.e., a positive relationship appears even more positive). The chart below shows the results:

Real GDP and Stock Returns Do Correlate Over Long Time Frames

Source: S&P 500 TR Index from Dimensional Fund Advisors. GDP Data from St. Louis Federal Reserve

Which Came First, the Chicken or the Egg?
Many of the technical analysts reading this article are probably screaming at the tops of their lungs, “It’s the other way around! GDP doesn’t lead the stock market; the stock market leads GDP!”

To test that, we used the same data series, and we lagged both our annual and quarterly real GDP numbers behind the stock market. Both correlations and r-squared improved noticeably.

Lagging GDP by one and two quarters behind the S&P 500 produced correlations of 25% and 34% respectively. The highest correlation was achieved by using annual returns and lagging GDP one calendar year behind the S&P 500. Here, we found a correlation of about 40% and an r-squared of about 0.17. Not great, but better.

While both correlation and regression stats were higher, I’d caution against putting too much faith in these improvements. Our r-squared of 0.17 was an improvement, but I’d hardly call it predictive.

Further, don’t forget the effect that smoothing has on increased correlation and r-squared numbers. As evidence, consider the following: Lagging quarterly real GDP four quarters behind the S&P 500 (the same time period that produced our improved r-squared using calendar year data) produced a very weak r-squared of 0.009. The time frames are the same, but the results are vastly different!

Conclusion
The focus on GDP linked investments has been growing over the past couple of years. PIMCO was granted a patent on its “Global Advantage Bond Index Methodology”—a methodology that weights bond indexes by GDP rather than debt outstanding—in 2012. The bond giant offers its methodology to investors through the PIMCO Global Advantage Strategy Bond Fund (PSAIX).

In 2010, Van Eck filed with the SEC its intention to bring about two GDP linked equity ETFs, the Market Vectors GDP International Equity ETF and a Market Vectors GDP Emerging Markets Equity ETF. Both of these are supported by research from MSCI Barra that suggests GDP-weighting schemes in both international developed and emerging markets can outperform market cap weighted indexes.

While using total GDP to weight broad, global indexes may offer some value over market cap weighting, this is a vastly different from saying that GDP growth in any one country will have predictive power over short to intermediate stock market returns. Our research confirms, if only in part, that GDP growth is probably not a very good predictor of future moves in the stock market. If anything, the stock market is probably a better predictor of GDP than the other way around.


Tags: , , ,