Tuesday, July 5, 2011

Hussman Weekly Market Comment: Chutes and Ladders

A side note - we make an important distinction between "risk" and "uncertainty." As I've noted before, risk describes the "spread" of a known probability distribution - for example, the chance that you'll roll something other than a 7 given that you know you're throwing two six-sided dice. Uncertainty emerges when you don't even know whether the dice have six sides, so you are forced to entertain the possibility that your entire model of the world is incorrect. In the past two years, our research efforts have been singularly directed at measuring uncertainty and dealing with it in a very disciplined way.

The simple fact is that we are all playing a game of Chutes and Ladders where it is not at all clear which game-board is applicable. To believe strongly in a certain investment outcome is to imagine that there is only one correct model of the world, and that the correct model is in hand. Investors appear very eager to apply post-war norms to the economy, and to apply the elevated valuation norms of the past two decades to the stock market. I doubt that these models represent the correct view of the world, but our approach is to allow for these possibilities and dozens of alternate ones.

As I've noted in several recent weekly comments, our eventual resolution of the "two-data sets problem," as I described it in 2009, is to discard the belief that one model, or any small number of models, is the correct way to view the world, and to instead proliferate dozens of models each estimated over different subsets of historical data and different subsets of indicators. These still emphasize measures of valuation, market action, credit spreads, interest rates, sentiment and other factors with strong theoretical and historical support, but there is no single "master" among them. The performance of the ensemble is typically much stronger (and generalizes better to new "out-of-sample" data) than any of the components.


Turning to the subject of monetary policy, according to the Fed's consolidated balance sheet, the Fed now holds assets $2.87 trillion, versus $52.87 billion in capital, putting it just north of 54-to-1 leverage. Now, given that the average duration of the Fed's holdings works out to something in the neighborhood of 6 years, an increase in interest rates of roughly 30 basis points is now enough to completely wipe out the Fed's capital. Needless to say, given the recent spike in Treasury yields, we would have observed a huge hit to the Fed's reported capital in recent weeks if the Fed was to allow such losses to actually flow through to the capital line of the balance sheet. Of course, that's why the Fed changed its accounting a few months ago to allow such losses to show up as a "negative liability" - basically consuming interest that the Fed receives on the Treasury debt it holds, which would normally be remitted back to the Treasury for public benefit (yet another example of altering accounting rules to effectively allow insolvency).