Diversification Causes Correlations

The Bond insurers diversified out of their niche of municpal bonds into real estate backed securities and suddenly these two markets that previously seemed to have low correlation were highly correlated as the sub prime crisis brought down the Bond Insurers and their problems rippled into the Muni market.

(I say seemed uncorrelated, but of course they are highly dependent since a high fraction of municipal incomes comes from taxes relating to real estate values.  That is a major problem with the statistical idea of correlation – statistical approaches must never be used uncritically.)

But the point of the first paragraph above is that interdependencies do not have to come from the fundamentals of two markets – that is to come from common drivers of risk.  Interdependencies especially of market prices can and often do come from common ownership of securities from different markets.  The practice of holding risks from seemingly unrelated risks or markets is generally thought to create better risk adjusted results because of diversification.

But the perverse truth is that like many things in real economics (not book economics) the more people use this rule, the less likely it is that it will work.

There are several reasons for this:

  • When a particularly large organization diversifies, their positions in every market will be large.  For anyone to get the most benefit from diversification, they need to have positions in each diversifying risk that are similar in size.  Since even the largest firms had to have started somewhere, they will have a primary business that is very large and so will seek to take very large positions in the diversifying markets to get that diversifying benefit.  So there ends up being some very significant specific risk of a sudden change in correlation if that large firm runs into trouble.  These events only ever happen once to a firm so there is never, ever any historical correlations to be found.  But if you want to avoid this diversification pitfall, it pays to pay attention to where the largest firms operate and be cautious in assuming diversification benefits where THEY are the correlating factor.
  • When large numbers of firms use the same correlation factors (think Solvency II), then they will tend to all try to get into the same diversifying lines of business where they can get the best diversification benefits.  This results in both the specific risk factor mentioned above and to a pricing pressure on those markets.  Those risks with “good” diversification will tend to price down to their marginal cost, which will be net of the diversification benefit.  The customers will end up getting the advantage of diversification.
  • Diversification is commonly believed to eliminate risk.  THis is decidedly NOT TRUE.  No risk is destroyed via diversification.  All of the losses that were going to happen do happen, unaffected by diversification.  What diversification hopes to accomplish is to make this losses relatively less important and more affordable because some risk taking activity is likely to be showing gains while others is showing losses.  So people who thought that because they were diversified, that they had less risk, were willing to go out and take more risk.  This effect causes more of the stampede for the exits behaviors when times get tough and the losses that were NOT destroyed by diversification occur.
  • The theory of a free lunch with diversification encourages firms who are inexperienced with managing a risk to take on that risk because their diversification analysis says that it is “free”.  These firms will often help to drive down prices for everyon, sometimes to the point that they do not make money from their “diversification play” even in good years.  Guess what?  All that fancy correlation math does not work as advertised if the expected earnings from a “diversifying risk” is negative.  These is no diversification from a losing operation because it has no gains to offset the losses of other risks.
Explore posts in the same categories: Assumptions, Correlation, Diversification, Enterprise Risk Management, ERM


You can comment below, or link to this permanent URL from your own site.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: