Econometric Forecasting is Worthless
Econometric forecasting, or the use of mathematical or statistical models to predict the future of some sector of the economy, is pretty much a bunch of garbage, pure voodoo. Anyone who predicted the timing of current downturn got incredibly lucky. And the so-called experts mostly missed the boat. Among them Raymond Torto of CBRE’s Torto Wheaton Research who said back in 2006, “”We’re not predicting a crash, not predicting a disaster” and went to predict ‘investment in real estate remaining healthy, with a solid amount of liquidity in the market, and prices for assets remaining high.’
Even in October 2007, in the midst of the residential subprime implosion and when the CMBS market was already closed down for the foreseeable future, Torto Wheaton wrote, “the volatility in the markets has created buying opportunities, especially in sectors that are no longer priced for perfection yet continue to have a favorable outlook.” The truth is that most who bought in late 2007 and into early 2008 are regretting it.
Torto and Wheaton are widely respected and extremely bright; this all goes to show that even the best and the brightest can’t make successful predictions about when and how dramatically the market will turn for better or for worse.
All that’s safe to say is that the economy is cyclical, and clearly we’re in the “down” portion of the cycle, and that means it’s time to buy, when no one else is willing buy. In 2006 and 2007 I was pushing my clients to sell sell sell! Admittedly not with purely altruistic notions, but it was clear pricing then was not sustainable. Now that we’re in an unrivaled buyers markets, I’m finding most buyers are sitting on their hands, waiting for what they perceive as the “bottom.” Unfortunately, you can’t time the bottom. You just have to dive in and start buying.
Like this post? Subscribe at the right and get more like it in your email every day!
Tags: econometric forecasting
February 16th, 2009 at 2:37 pm
Absolutely right, econometric forecasting is worthless in most situations, but it’s worth knowing when and why. If (and it’s a big if) the system under study is slow moving, and we know what all the factors affecting outcomes are (it is a closed system) and how all the factor inputs influence each other, then econometric predictions will work. Most real-world future situations are too open, too complex, too fast moving, too multifaceted to be predicted in this way.
There’s a great essay “Why Can’t We Predict?” Neil Duncan; New Scientist, Vol. 136, Issue 1841 (03 October 1992) that absolutely nails this issue in a few pages.
I offer an analysis of the limits of quantitative/econometric foresight in my new book, Future Savvy (American Management Association, NY, 2009) The book has a summary forecast-testing battery, showing forecast consumers how to assess and discern quality in the future thinking they read and hear, and very high on this list is knowing when quantitative methods will work and when they won’t.