In January 1991, the S&P 500 had a Price/Earnings ratio in the mid-teens. The yield differential on corporate bonds, compared to long-term Ts, was about average, and had been rising to that point.

Meanwhile, the yield curve (long-term T’s to 3moTs) was pretty steep, and had been steepening.

Finally, CPI was rising at an uncomfortably high rate, year over year.

The NBER eventually called that period a “recession” beginning in July of 1990, but they didn’t “call it” until April of 1991.

Sounds a lot like now. Hmm.

Using data available from AAII, one could do a regression analysis with the S&P 500 index’s PE, the corporate bond spread and its year over year change, the long-term-to-short-term T spread and its year over year change, and the year over year change in headline CPI as independent variables.

I suggest the subsequent year’s change in the S&P 500 index be the dependent variable.

I’ve already done that work. I may (or may not) post it later. Any nerd could do it. Some of y’all probably should do it, as well.

The adjusted R-square for the regression, along with the t-Stats and P-values for each coefficient, could be provided as proof you did your “homework.” Alternately, submissions detailing the variable whose coefficient had the highest (least significant) P-value might be accepted as proof of work.

Discussion points could include macro reasons why each variable type’s (valuation of the index, yield curve, corp spreads, and CPI change) coefficient has the sign that it does, or illumination of the differences between P-value of a coefficient, additional predictive value of an added coefficient, and relative formula weight of said coefficients in a regression.