Fundhouse Insights Latest Articles

The statement “Forward PEs look attractive” is often offered as an astute observation. In fact, it’s almost a truism. Why? Because the perpetual, almost childlike optimism of analysts is such that the market almost always looks good based on next year’s earnings forecast.

Recently, I reviewed a comparison of Wall Street analyst consensus S&P500 EPS growth forecasts versus actual results from 1995 to 2012. The findings were extraordinary, to say the least.

Over the 17 years of data, the analyst average forecast was for EPS growth of 14.3% per annum, which compared to subsequent actual growth of 7.0% per annum! If the forecasts had been on the button, even on average, the S&P500 would be trading at over 5000 today. What is even more remarkable is that over the 40 years prior to 1995, average real EPS growth for the S&P500 had been a modest 1.7% per annum. With inflation at 2.5% per annum over the 17 year forecast period, based on historical real EPS growth, the analysts should have been expecting growth closer to 4% per annum than 14% per annum.

Throughout those 17 years, the analyst consensus NEVER ONCE forecast lower earnings in the coming year. That is to say, they never once saw a downturn coming and, even when a downturn had started, they could never see it continuing into a second year. As it turned out, five of the 17 years had negative EPS growth, including the back-to-back negatives of 2007, 2008 and 2009.

The most pessimistic consensus one-year forecast was for 7.7% EPS growth – higher even than the average actual 7.0% per annum EPS growth over the period. You read that right – at their most gloomy, the concensus never once forecast below average growth. (The analysts must all come from Garrison Keilor’s Lake Woebegone where all the children are above average.)

Looking for some redeeming grace, we wondered if perhaps the analysts were just perpetually and consistently optimistic. That is, if we shaved, say, 7% per annum off the consensus forecast each year, could we get some indication of whether the years ahead might bring good or bad EPS growth?

The answer was a resounding no. In fact, using a standard statistical test, we came up with an R-squared of 4.5%. (An R-squared of 100% indicates that the predictions entirely explain what subsequently occurred, while an R-squared of 0% indicates that the predictions have no predictive power whatsoever. And, the R-squared test is an exceedingly generous one – even a forecast that was always 100% wrong would get a perfect R-squared score because, after the fact, we could just use the opposite of the forecast and get the right answer.)  But even under this soft test, the analyst consensus scored just 4.5 out of 100!  That is, it had pretty much no predictive power at all.

farrelly’s knows from experience that forecasting is a tough game. But using forward PEs to assess market valuations? It’s nuts, and you can clearly see it’s nuts.

Tim Farrelly is principal of specialist asset allocation research house, Farrelly’s Investment Strategy, available exclusively through PortfolioConstruction Forum. Tim is a member of PortfolioConstruction Forum’s core faculty of leading investment professionals.