Monday, September 6, 2010

Bonds, Bubbles and Busts



A Bubble in Treasuries?

Is there a bubble in Treasuries? This question has recently been discussed by nearly every financial commentator on the web and in the press, with both sides having amassed convincing but conflicting arguments. Value investor James Montier of GMO had something interesting to say about this question in his blog recently.  You would expect a value type like Montier to warn that Treasuries are overvalued compared to historical norms, and you would be right. The only trouble is that I'm not sure that I agree, and I'm not sure that his argument is even a valid value judgment is today's situation.

In his blog Behavioural Investing Montier published an article on August 31 titled "Bond Bubble - a sterile debate on semantics," which asked : "The issue shouldn’t be whether bond are a bubble or not, but rather are bonds a good investment or not?" Using Ben Graham's definition of an investment as an operation that promises "safety of principal and a satisfactory return," Montier proceeded to estimate the return from Treasuries over the next ten years, so as to see if that return would be in some sense "satisfactory."

Are Bonds a Good Investment or Not?

To do this, Montier estimated the three components of return -- the real yield, expected inflation and an inflation risk premium -- for 10-year Treasuries over the next ten years. He estimated the real yield as 1%, based on the yield of 10-year TIPS. For expected inflation, the inflation swap market implies 2% over the next ten years; alternatively, the nominal bond yield minus the TIPS yield implies 1.5%. For the inflation risk premium, which is a way of accounting for uncertainty in future inflation, he used 0.5%, which is the upper range of current estimates. This implies a return of 4% annually under "normal" inflation conditions. Montier upped the estimate of longer term average 10-year Treasury yields to the 4%-5% range because he questioned whether the current market real yield of 1% is really a "fair price," and because the longer UK experience with inflation linked bonds suggests a somewhat higher number.

Montier's conclusion is "In the ‘Normal’ state of the world bonds sit at close to equilibrium, say 4.5%." The implication is "The current 2.5% yield on the US 10 year bond is clearly a long way short of this." In contrast, a Japanese outcome for the US over the next ten years would have rates around 0%-1%, and an inflationary outcome would have yields around 7.5% (with inflation at 5%).

Montier interpreted these numbers as saying that "In essence, the market is implying a 70% probability that the US turns Japanese."  Maybe 2.5% Treasuries are attractive if you are certain that the US is becoming Japanese, but that is a lot to assume. A rise in yields even to the long-term average level implies some losses.



A Satisfactory Return?

I think that we knew this already, at least in a qualitative way, before Montier published his calculations. We already knew that T-bonds are overpriced according to historical metrics. We understand that an all-out bet on Treasuries at this point is risky, but we also know that all-out bets on other asset classes are also risky. Montier did not address the right question.

The problem is that Montier is arguing on the basis of the range of Treasury bond prices over the limited history of those securities. That history is limited in time extent and in the range of conditions that were sampled. It is not an average over an infinite sequence of all possible financial histories going through all possible states of the world. In other words, it is a sample average, not a known parameter. The values and frequencies seen over the recent past were dependent on the unique and contingent conditions pertaining during exactly that period of time. Why should we assume that the future will be like the past?

Today's financial state is highly peculiar in the US. It is a state of high indebtedness occurring at the end of a period of easy credit and overconsumption. If we look over longer periods of time than Montier examined, we see that periods of excess are regularly followed by periods of sub-par growth and financial crisis. We have examined the work of Reinhart and Rogoff in this blog before. We place much more weight on several centuries of world financial history than we do on the limited history of the past few decades of a peculiar and distinctly American era.

Bubbles are followed by busts, and long term averages can be violated for a considerable period of time. Excessive debt weighs down with an inexorable burden that has to be worked off over time. As a result, the "long term" average may not re-emerge for a long time. I don't think that any Roman descendants are waiting on their British estates for the legions to return from over the channel.

The Only Things That Really Matter in Investing Are Bubbles and Busts

Montier's colleague Jeremy Grantham suggested how to frame problems of this sort in the GMO Quarterly Letter of April 2010. In an article titled "Friends and Romans, I come to tease Graham and Dodd, not to praise them" he wrote on some potential disadvantages of Graham and Dodd-type investing. Grantham attacked the "preposterous belief" that all information is embedded in securities prices and that bubbles and busts can be ignored. Grantham said, in fact, that "I am at the other end of the spectrum: I believe that the only things that really matter in investing are the bubbles and the busts."

In support of this belief, Grantham presented a variety of statistics illustrating that even such "standard" stock valuation criteria as the price-to-book ratio come in and out of fashion, and can over-perform or under-perform over extended periods of time. To extrapolate a little from Grantham's examples, we are already familiar with the work of Reinhart and Rogoff and other historical studies showing how financial crises spawn more crises and are regularly followed by extended periods of sub-par economic growth.




Grantham quotes from Securities Analysis: "Undervaluations caused by neglect or prejudice may persist for an inconveniently long time … and the same applies to inflated prices caused by over enthusiasm or artificial stimulants.” Graham adds: "If ever we were living in a world of artificial stimulus, it is now."

When there is no bubble or bust around, "if you keep your nose clean, you will probably keep your job," by which he probably means that prudent value principles are usually the best investing guide. During bubbles and busts, however, the usual value principles may not be the best way to go. He exhorts: "But when there is a great event, that’s the time to cash in some of your career risk units and be a hero." The end of the age of credit excess certainly qualifies as a great event in my book.

A Margin of Safety

Should we base investment decisions on recent historical distributions or on the dynamics of bubbles and busts? Where does the margin of safety lie? If we take Grantham's advice, "the only things that really matter in investing are the bubbles and the busts." If you are living in a bust, maybe the future consists of something other than a statistical sample of the recent past. Maybe you should think seriously about the life cycles of bubbles and busts.

Treasuries look dangerously over-valued to our eyes, but our eyes have been trained by lifetimes spent in a credit-crazed culture worried about inflation. It is not surprising that, since Montier's article, the yield has fluctuated from 2.50% to 2.70%.  Howewever, if we are in a debt deflation, supported by extraordinary government intervention, maybe there is a better than average chance that Treasury yields will stay low for longer than Montier suggested. It also helps that a contraction in private debt issuance has made room for expanded government issuance without a backup in yields. Or maybe the economy will come back and tank bonds. There is no sure thing here.

Whatever the eventual outcome, there is much more to the Treasury bond situation than a simple comparison to historical average yields. Past returns are no guarantee of future results.