Statistical shenanigans or: Sell on the rumour, buy on the news
The first half of 2022 has been dominated by high inflation, central banks hiking rates, and the fear of a new recession. The big advantage for US investors is that they don’t have to rely on some superficial metric like two successive quarters with negative GDP growth to determine when a recession starts and ends. They can rely on the official business cycle dates determined by the NBER. But should investors follow the announcements of the NBER in their portfolios?
The chart below shows the S&P 500 since 1980 together with the recession periods as determined by the NBER. Looking at the chart it is obvious that stock market returns in a recession were overwhelmingly negative. So why not sell stocks when a recession begins and buy them again when it ends?
S&P 500 and recessions since 1980
Source: Bloomberg, NBER
Well, the problem, as with all macroeconomic data, is that it takes time before it is released. The NBER business cycle dating committee examines a broad set of indicators and only once they are sure that the cycle has turned do they make an official announcement. In real life, this announcement often comes a year or more after the cycle has turned. Should investors pay attention to these announcements and sell stocks when the NBER announces a previous cycle trough and sell stocks when it announces a cycle peak?
We have collected the NBER announcements back to 1980 and calculated the average return for the S&P 500 in the 12 months after each announcement. On average, after the NBER announced a cycle peak and the beginning of a bear market, the S&P 500 was up 15%. Only in one out of six cases did the S&P 500 drop in the year after the NBER announced the beginning of a bear market. That was in the long bear market of 2000 to 2003.
When the NBER announced a cycle trough and the beginning of a new expansion, the S&P 500 on average dropped 3.5% with only two cases of positive returns out of the last six cycles (1992 and 2003).
Average return of S&P 500 after NBER cycle announcements
Source: Liberum, Bloomberg, NBER
As a market-timing tool, the NBER announcements are clearly abysmal. The delays on the announcements are so long that it is effectively a countercyclical indicator. This effect is so large that the difference in performance between NBER peak and trough announcements is statistically significant at the 10% level.
But then again, this entire exercise of looking at NBER announcement is an exercise in statistical shenanigans anyway. Who in their right mind would build an investment strategy based on the more or less random announcements of a committee of economists? In fact, who in their right mind would build an investment strategy based on the announcements of economists in general? Just because I can calculate returns and statistical significance values in my Excel spreadsheets doesn’t mean that I should. And it certainly doesn’t mean that I should put my money where my Excel formulas are.
The key to becoming a good investor is to understand the numbers in your spreadsheet, not your ability to calculate them. That is the main challenge for younger investors with less experience and for quantitative investors alike. Quantitative investing can become a case of garbage in, garbage out if you don’t understand deeply what you are doing and why the results are what they are. The case of NBER business cycle announcements is an obvious example, but the graveyard of dead hedge funds and the reefs at the tomb of the unknown Lehman credit analyst should tell you that even highly educated professionals regularly fall prey to this mistake.