There is an article in this morning's Washington Post on the discussions the Fed members had regarding the state of the economy in 2006, just as the housing bubble popped.
In his first meeting as Fed chairman, in March 2006, Ben S. Bernanke noted the slowdown in the housing market. But he said he shared the view that “strong fundamentals support a relatively soft landing in housing,” adding: “I think we are unlikely to see growth being derailed by the housing market.”
The year began with adulation all around for Greenspan. In that January meeting, Roger Ferguson, then Fed vice chairman and now head of the TIAA-CREF financial services group, called Greenspan a “monetary policy Yoda.”
Janet L. Yellen, then president of the Federal Reserve Bank of San Francisco and now the Fed’s vice chair, told Greenspan “that the situation you’re handing off to your successor is a lot like a tennis racket with a gigantic sweet spot.”
Looking back...
In the six years since, Greenspan’s record — seemingly so sterling when he left the central bank after 18 years — has come under substantial criticism from outside economists and analysts. Many say a range of Fed policies under his watch contributed to the financial crisis, including keeping interest rates low for too long, failing to take action to stem the housing bubble and allowing inadequate oversight of financial firms.
Forecasting economic variables is very, very hard. It's hard to forecast for particular microeconomic markets, and it's even moreso with entire economies. What is the theoretical model of the economy? What variables do you include? Why? What variables do you exclude? Why? How do you account for the infinite number of relationships between various economic variables? What about the effect that foreign economies have on the domestic economy? What about the effect that the German economy has on the British economy and, in turn, on the US economy? What about military conflicts and the price of crude oil?
Given the theoretical model, what statistical model do you use? How do you account for things like autocorrelation and moving averages? What data sources do you use? How do you account for differences in how different countries gather and release data?
Then have the forecaster predict two years out? As they say, past results are not necessarily indicative of future performance. So, even if we got the theoretical and statistical models correct for past data, does that mean our models will be correct in two years? Will there be a major war in the period you are forecasting?
The old joke goes "Why did God make economists? To make weather forecasters look good." Go out and find the best tornado scientist in the world and ask that person how many tornadoes will occur in 2014. Asking an economist, say, to tell you what the unemployment rate will be in two years is a lot like asking a tornado scientist how many tornadoes there will be in two years. Even with all their knowledge and skill, Josh Wurman and Reed Timmer can't even do that with any precision worth pay attention to.
We economists may not like to admit it, but we are better historians than forecasters. Our knowledge of what makes the economy tick is very limited and we learn by looking back, not by looking forward. So while it's tempting to jab at the Fed members for failing to see what was going on in the economy, given them a little break.
Even so, this episode is an example of why it's very risky to put too much power into the hands of central planners (yes, setting monetary policy is central planning). Their knowledge is limited, even though those credentials after their names may tell us otherwise.
