As a certified numbers geek, I have always tinkered with quantitative models to explain and/or forecast the economy and financial markets. Specifically, I developed five different models designed to estimate future returns on the S&P 500 Index. Attempting to predict short term stock returns is futile so they forecast annualized returns over the next ten years. Each model is based on several variables that are generally believed to influence stock prices; I tried to avoid spurious data analysis in which there may be correlation without causation. Among the variables are volatility, economic growth, interest rates, inflation, corporate profit margins, P/E ratios, the level of margin debt, and the equity weighting of the average investor.
Over the long term, these models have an excellent track record with 75% to greater than 90% accuracy. However, things have gone awry over recent ten-year periods with the spread between predicted and actual returns ranging from 2% to as much as 7% per annum. Those gaps are large enough to concede that the models were just plain wrong. Most important, actual returns exceeded the forecasts in every single case!
So, what are some possible explanations for my models going off of the rails? First, I could be a victim of the well-known problem that models built on what is known as “in sample” data sometimes fail when they are applied to real world, “out of sample” data. A second possibility is that the financial world has fundamentally changed in some way. When something seems out of whack, it may simply represent a temporary aberration that will eventually return to normal. Alternatively, it could be what financial types term a paradigm shift. It is very difficult to tell! Most of the time, we can expect regression to the mean, but there are occasions when “it is different this time.” Finally, the fact that the models have significantly underestimated actual returns could mean that we are in a stock bubble in which long-held financial relationships are temporarily suspended in favor of euphoria and “new era” thinking.
For whatever it is worth, my models are predicting annualized returns of 1% to 5% over the next ten years which certainly seems pretty pedestrian in light of the raging bull market currently underway. Anyway, what’s a poor quant to do?