Reality is more complicated than economic models

10 Aug 2011 by Jim Fickett.

Nicholas Bloom, economist at Stanford, predicts a recession this year, based on the idea of an “uncertainty shock”. He points out that, on average, output and employment drop following large jumps in the volatility index, VIX, and one such jump has just occurred. Unfortunately, a look at the historical data do not bear out the predictive value of jumps in the VIX.

Today Economist's View pointed to an interesting post on VoxEU entitled, The uncertainty shock from the debt disaster will cause a double-dip recession.

The main idea is that when the VIX index rises substantially, this indicates great uncertainty on the part of investors, businesses, and consumers, and when the populace is highly uncertain, they pause spending, causing a recession.

The potentially explosive combination of Eurozone debt contagion, vulnerable banking systems, and European and American political paralysis has pushed stock-market volatility to levels nearly as bad as the days following the 11 September 2001 terrorist attacks. Nobody knows what happens next. This column reviews research on 16 previous shocks and concludes that today’s uncertainty shock will create a short, sharp contraction in late 2011 of about 1% with a rebound coming in spring 2012. …

When people are uncertain about the future, they wait and do nothing.

  • Firms do not to hire new employees, or invest in new equipment if they are uncertain about future demand.
  • Consumers do not buy a new car, a new TV, or refurnish their house if they are uncertain about their next paycheck.

The economy grinds to a halt while everyone waits.

A graph of the VIX shows that the jump in the VIX over the last few days is comparable to that in past momentous events:

The short VoxEU post is based on a longer research paper. In the paper, Nicholas Bloom of Stanford selects for study all events where the VIX makes a sufficiently large move or, more technically,

These 17 events are chosen as those with stock-market volatility more than 1.65 standard deviations above the Hodrick–Prescott detrended (λ = 129,600) mean of the stock-market volatility series …

(Using a measure of standard deviations above a detrended series means that one is looking for large changes in the VIX, rather than high levels of the VIX.)

The events so selected, along with the date of maximum volatility, are as follows:

Event Max volatility
Cuban missile crisis Oct 1962
Assassination of JFK Nov 1963
Vietnam buildup Aug 1966
Cambodia and Kent State May 1970
OPEC I, Arab–Israeli War Dec 1973
Franklin National Oct 1974
OPEC II Nov 1978
Afghanistan, Iran hostages Mar 1980
Monetary cycle turning point Oct 1982
Black Monday Nov 1987
Gulf War I Oct 1990
Asian Crisis Nov 1997
Russian, LTCM default Sep 1998
9/11 terrorist attack Sep 2001
Worldcom and Enron Sep 2002
Gulf War II Feb 2003
Credit crunch Oct 2008

The main content of the research paper is to construct a model of what happens following an uncertainty shock, and then to study that model. The paper does not give much detail on what actually happened historically. Here is a graph showing the actual course of industrial production (one of the most prominent variables studied in the paper) following each of the 17 events.

(Here the aftermath of economic shocks are shown with solid lines, oil shocks with dashed lines, and violence-related shocks with dotted lines. The lowest line is the 2008 credit crisis aftermath, which was not studied in the paper because data were not yet available.)

The model derived by Bloom predicts a drop in industrial production that hits bottom at 4 months. It is not surprising that a model trying to capture average behavior would show show such a drop, because there was a large drop in a few cases. Nevertheless, more of the shocks show an increase than a decrease in production at 4 months. So the average case does not provide any basis for predicting a recession.

This is a very good illustration of why I am generally skeptical of economic models.

I've argued against various predictions of recession lately. This does not mean I think a recession is unlikely – quite the contrary. But I would like to see evidence and, so far, none of the arguments I've seen really holds water.