Monday, 27 April 2009

Do crises cluster? If so it could be a problem for Larry Summers' solution to the debt crisis

In a recent article Simon Johnson provides a very interesting summary of what he takes to be the cornerstones in the thinking of Larry Summers about the current financial mess. There are five key components to the Director of the White House National Economic Council's perspectives, as spelled out in a speech that Summers gave last Friday, and Johnson, who has analyzed the speech, outlines them below including, in parenthesis, some comments of his own.

1. All crises must end. The “self-equilibrating” nature of the economy will ultimately prevail, although that may take massive one-off government actions. Such a crisis happens only ”three or four times” per century, so taking on huge amounts of government debt is fine; implicitly, we will grow out of that debt burden.

2. We will get out of the crisis by encouraging exactly the kind of behaviors that “previously we wanted to discourage” two years ago. It is “this insight, this view” particularly with regard to leverage (overborrowing, to you and me) that “undergirds the policy program in the United States.”

3. There is a critical need to support financial intermediation and to ensure it is adequately capitalized, with a view to the risks inherent in the current situation. He then said, with a straight face, that the current bank stress tests are designed with this in mind.

4. Growth in the 1990s and more recently was based too much on finance (this appears to be a relatively new thought for Summers). The high and rising share of finance in corporate profits “should have been a warning”. The next expansion should be based less on asset bubbles and more on investment in key public services.

5. The financial regulatory system “in fundamental respects has been a failure”. There have been too many serious crises in the past 20 years (yes, this statement was somewhat at odds with the low frequency of major crises statement in point 1).
The one that bothers me the most is the first. There is always a danger when considering crises - not only financial but also natural disasters - of thinking in terms of periodicity i.e. that is the reasoning behind the view of Summers that they only come along four times in a century. From this it seems to follow that since we've had the big one we won't get the next one for a generation or so. There are several problems with this view.

Firstly the sample size is too small for any meaningful application of quasi-statistical techniques to determine for example the average period between crises or the dispersion of them. We have so little to go on with regard to how they are distributed and it would be foolish to have the presumption that they are periodic and normally distributed. Just like earthquakes there is no safety in assuming that there could not be two major quakes on a fault line in quick succession even if there has not been a large seismic event for much longer than the alleged time to wait period.

How does one discretize or individuate a financial crisis? What of the 1930's - was that a single crisis event or a series of inter-related critical events?

From several academic studies of major disruptions in time series data, there is a strong case for believing that critical episodes of illiquidity and volatility have a tendency to cluster rather than to be evenly distributed across the time spectrum.

In my opinion it would be very unwise to assume that we've had our crisis and now - as Larry Summers appears to be suggesting - have a breathing space to re-build our economy and repay our debts. There might be another leg to this crisis - indeed there could be several more.

The fact that, even if turns out that we do have a breathing space, we might not take the opportunity to re-build our economy and pay back our debts - for lack of political will for example - is of course a different topic.

No comments:

Post a Comment