BofA "Explains" Why Optimistic Economist Forecasts Have Been So Wrong In The Past 5 Years

A few days ago, when looking at the latest quarterly compendium of humor, aka the IMF's world economic outlook release, we showed the reason why economists are such horrible forecasters: all they do is extrapolate trends.


Of course, this is not even remotely close to economic analysis as it assumes a perfect world devoid of any realitym and would hardly even pass as an Art 101 project.

This, however, appears to have ruffled some feathers among Wall Street's economic community, especially among those who extrapolate trends for a living. Such as BofA's Michelle Meyer, who overnight released an amusing note titled "The Random Act of Forecasting" in which she tries to explain how 7 years after the advent of central planning, first in the US and soon everywhere else, everyone has been so massively wrong when it comes to forecasting the "imminent" recovery.

Amusing, because it contains the following attempt to scapegoat a "series of mini shocks":

we can look back to the start of the recovery and identify the shock or vulnerability each year which impaired the recovery.

BofA may call them mini shocks. Others may call them reality. And this is how Bank of America justifies the fact that year, after year, after year, reality has failed to match up with Bank of America's (or the Fed's) model. From BofA:

  • 2010: The first full year of the recovery was a growth recession with a collapse in inventories (after the restocking was complete), and continued private sector deleveraging.
  • 2011: There were a series of events, including the Japanese tsunami, spike in oil prices and US debt downgrade by S&P.
  • 2012: The crisis in the Eurozone intensified with concerns over a Greek exit and a breakup of the Eurozone. The policy response abroad was lackluster and there were concerns of another financial crisis.
  • 2013: The combination of the sequester, debt ceiling fight and government shutdown created an environment of heightened uncertainty and fiscal restraint.
  • 2014: The polar vortex delayed economic activity and led to a permanent loss of growth.
  • 2015: Rapid appreciation of the dollar and heightened uncertainty about the winners and losers from plunging oil prices has hurt growth. A small part of the weakness may be related to the weather and the dock strike.

Again, as we said: "reality."

Of course, when one models "reality" while excluding the impact of reality, things end up being forecast incorrectly.

So just in case this wasn't enough of an excuse for why economists have now become the most useless and overpaid profession on The Weather Channel, here are some more excuses from economists why, for the past 5 years, they has been dead wrong.

The choppy data create difficulties in the forecasting process. When the data weaken, even if there is a tendency to explain it away due to special factors, expectations are naturally set lower. This is shown in the data surprise index — inflections in the index represent the periods when the data are shifting relative to expectations. A low reading in the data surprise index is not just indicative of weaker data flow, but also of expectations that were set too high.


The mood swings in this cycle have been exceptional. We have oscillated between calls for a V-shaped recovery to a double-dip recession to cries of secular stagnation. It is as though forecasters are behaving as hormonal teenagers. The reality is that this has been the recovery of fits-and-starts and the hysteria in forecasting is due to the following factors:


Return to normalcy: there is a tendency to forecast a return to the “steady state”, which is a function of prior business cycles. However, in this recovery, looking at past cycles has been misleading. Recoveries from major banking and real estate crises are both weaker than the normal cyclical rebound and more sensitive to shocks. As we reminded readers last week, our core story for the last six years has been that the US was in a slow, fragile recovery, and growth would only begin to pick up once balance sheets fully healed.


There has been a slow capitulation that the concept of normality has changed after this recession. Forecasts for potential growth have been slashed along with expectations for the equilibrium Fed funds rate. The consensus, based on the Blue Chip survey, has taken down its forecast for potential GDP growth from 2.8% pre-recession to 2.6% in 2010 and 2.3% today.1 The FOMC has similarly revision down its long-run GDP forecast (Chart 2). Forecasts for the equilibrium Fed funds rate have followed suit (Chart 3). Part of this revision would have happened regardless of the business cycle, given the aging of the population. However, it was probably a faster realization and perhaps more dramatic given the other overhangs on the economy. As the long-run estimates are adjusted, forecasters rethink estimates for growth based on a new trajectory of potential.


Trend extrapolation: It is exceptionally difficult to identify turning points in the economy. Although there were warnings of excesses in the economy and the usual signals were clear, including an inverted yield curve, forecasters continued to pencil in a recovery. In the October 2007 SEP, the FOMC was expecting growth of 1.8 - 2.5% in 2008 and 2.3 - 2.7% in 2009. Even in January 2008, as the economy was slipping into recession, the FOMC was still expecting growth of 1.3 - 2.0% for the year. Indeed, they didn’t pencil in a decline until the October 2008 SEP — they marked down 2009 growth to -0.2 to 1.1% from 2.0 to 2.8% in June. 2 The challenge is that it is difficult to fight the current trend. Many models include a trend variable, which could be as simple as a moving average of the dependent variable.


Point estimates: Economists, the private sector and the FOMC all give point estimates. It is understandable as we have to be held accountable for a forecast and need to differentiate ourselves from others. However, point estimates give a false sense of precision. In reality, we should be forecasting with confidence bands around our forecasts, so that we can illustrate the distribution of risks. For example, our forecast this year is for 2.9% growth, but we wouldn’t be surprised to see growth anywhere from the low 2’s to the mid 3’s.

Moreover, with point estimates, it is difficult to capture the vulnerability of the economy and possibility of shocks, in either direction. This is a critical challenge for the Fed’s SEP and we would therefore not be surprised if they decide to implement confidence bands, particularly around the Fed funds forecasts.

And our all time favorite:

Confirmation bias: which comes first the forecast or the data analysis? With so much data to sort through, forecasters must guard against focusing just on data that supports their case. Consider the recent debate about incipient wage inflation. A couple years ago, the short-term unemployment rate was supposedly a signal of imminent inflation. This was confirmed by a particular measure of wages —average hourly earnings for production and non-supervisory workers (AHEPN). With inflation dipping recently and particular weakness in the AHEPN series, those indicators seem to have been forgotten. When the facts change, find new facts!

It's our favorite, because as Upton Sinclair once put it, this is the "It is difficult to get a man to understand something, when his salary depends on his not understanding it" excuse. Everything else follows from there.