Saturday, June 23, 2018

22/6/18: 'Skeptical' IMF tends to be over-optimistic in its U.S. growth forecasts


In recent weeks, the IMF came under some criticism for posting relatively gloomy forecasts for the U.S. economy, especially considering the White House rosy outlook that stands out in comparison. see for example, WSJ on the subject here: https://www.wsj.com/articles/imf-sees-u-s-potential-growth-at-half-the-pace-of-white-house-estimates-1528995732.

Which begs two questions:

  1. Does IMF have any grounds to stand on its forecasts divergence from the White House? and
  2. Are IMF forecasts for the U.S. economy actually any good?
Firstly, the grounds:



Per above chart, the IMF is not alone in being less than exuberant about forward growth forecasts for the U.S. In fact, it is White House that appears to be an outlier when it comes to 2020-2023 outlook.

Secondly, per the question above, I crunched through IMF's semi-annual forecasts releases from April 2013 on (period prior to 2013 is too volatile in terms of overall fundamentals to take any forecast errors seriously). The chart below summarizes these against the actual outrun:

On the surface, it appears that IMF forecasts in recent years carried massive errors compared to outrun. So I did a little more digging around. I took 1, 2, 3, and 4 years-ahead forecasts, averaged them over different forecast releases, and estimated 90 and 95 percent confidence intervals for these. Here is the resulting chart:
What does the data tell us? It says that IMF forecasts have, on average, overstated actual growth outrun. In other words, IMF forecasts have been over-optimistic, not excessively pessimistic, in the recent past. More that that, IMF's current (April 2018 WEO release) forecast for the U.S. GDP growth is even more optimistic than already historically optimistic tendencies of the Fund imply. In other words, even though the first chart above shows the IMF forecast for the U.S. growth to be pessimistic, compared to that of the White House, in reality, IMF's forecasts tend to be wildly optimistic.

Average error for 1 year ahead forecast for the U.S. in IMF releases has been 0.037 percentage points (very small), rising to 0.476 percentage points for 2 years ahead forecasts (more material error), and 0.867 percent for 3 years ahead forecasts. Augmenting data (to achieve larger number of observations to 2000-2006, 2011-2014 periods, 4 years ahead average forecasts has been 0.867 percentage points above the outrun growth. And so on.

So, to summarize:

  1. IMF is not unique in being less optimistic on the U.S. economy than the White House;
  2. IMF's history of forecast errors suggests that the Fund tends to be overly optimistic in its forecasts and that current official Fund forecasts are more likely to be reflective of significant over-estimation of future growth than under-estimation;
  3. IMF's forecasts more than 1 year out should be treated with some serious caution - something that applies to all forecasters.

No comments: