Last month, when economists were still banging their heads against the wall trying to determine if the turmoil unfolding in China’s equity markets would play a role in the Fed’s decision making come September 16/17, we took a look at how forecasters have fared when it comes to accurately predicting the timing of “liftoff.”
By way of introduction, we said the following:
Economics (like sociology and political science and astrology) isn’t a real science. It’s a pseudo-science. And as is the case with other pseudo-sciences, it’s flat out impossible to discover laws and immutable truths, no matter what anyone told you in your undergrad economics course. The economist’s job description looks something like this: make predictions that are almost never right and then make up any reason you want to explain away the fact that you were wrong. These explanations run the gamut from intentional obfuscation via opaque statistical tinkering (“residual seasonality”) to comically absurd attempts to turn common sense into an excuse for poor outcomes (“snow in the winter”).
Economists’ collective inability to accurately forecast economic outcomes is exacerbated when they try to predict what other economists think about said economic outcomes or, as we put it previously, “when ‘forecasters’ are surveyed on the timing of a Fed hike (or cut) what you get is one group of economists trying to guess at what another group of economists mistakenly thinks about the direction of the economy. We might call this ‘stupidity squared.’”
Now that the Fed has officially admitted that “data dependency” is at least to some extent a myth, meaning the market is, as Deutsche Bank amusingly put it over the weekend, “now observing itself from another angle as an observer of the observer of the observers,” it’s worth taking another look at the following graph from WSJ:
Note that in the August survey (so just one month out) around 82% of economists were sure that "liftoff" would come at the September meeting.
We close with the following chart from The New York Times which largely speaks for itself: