Recent Letters to the Editor

February, 2009

I've found oanda.com to be an excellent site for foreign exchange. They regularly update The Economist's "Big Mac" Index based on current exchange rates to help investors gauge currency over and undervaluation. What do you think of this approach?

The short answer is that we regularly monitor it. The Big Mac Index is a practical implementation of the theory of "purchasing power parity" - the idea that, after adjusting for exchange rates, the same basket of goods should cost the same amount in different countries (i.e., it is a version of the law of one price). PPP has some limitations - for example, not all goods and services can be traded across currency zones (say, the labor required to cook that Big Mac), or there may be costs (e.g., transport and tariffs) that distort relative prices. Still, as a rough guide to the balance of pressure for exchange rate changes it is an excellent theory. We combine it with the theory of covered interest rate parity we use in our asset class valuation update, which says that, in order to eliminate profitable arbitrage opportunities, differences in interest rates on equally risky government bonds should be offset by changes in the underlying exchange rate. As we note each month, while all theories that claim to predict exchange rate changes are practically guaranteed to be wrong in the short-run, they provide a good, if rough, medium term guide to the direction in which the complex adaptive system that comprises the economy and financial markets is likely to evolve in the future.

Did the crisis of 2008 undermine your belief in the virtues of diversification?

Thank you for an excellent and timely question. The short answer is that 2008 did not undermine our belief in diversification; rather, it showed us how we - and the rest of the asset management industry - need to improve our approach to it. Let's start with the following table, which shows a range of investments which delivered positive returns in different currencies in 2008:

What failed to work in 2008 was not diversification, but rather many quantitative asset allocation models and qualitative human decision making processes. In essence, in 2008 the world experienced a relatively rare (at least in recent times) regime switch, from a relatively normal regime into one of high uncertainty (with an elevated possibility of deflation). In some ways, this switch was similar to the 2000-2001 popping of the technology bubble in equity markets. Indeed, some of the same signatures are present, including increases in the spread between ten year government bond and AAA corporate bond yields, reflecting elevated liquidity risk. However, the fact that leverage was high and multiple asset classes had become substantially overvalued caused this regime switch to be of a different order of magnitude (alternatively, one could say that while in 2000-2001 we flirted with this regime shift, in 2007-2008 we actually made it).

For many years we have noted two key limitations of many asset allocation methodologies. The first is so-called "model error", which occurs when important risk factors and relationships are not included. Back in 2002, we commented on Kent Obsand's book, Iceberg Risk, which raised this possibility. And in 2008, we very clearly saw that at least two critical factors had been left out of most asset allocation models: systematic liquidity risk, and the tendency of correlations to rise when asset prices are falling (to be sure, there were people like Windham Capital and ourselves who had tried to take the latter risk factor into account). The second limitation of models is "parameter error" which refers to a situation in which a variable is properly included in a model, but given an incorrect value. There is now much discussion in the media about how models failed to completely capture so-called "tail" or "rare event" risk, which is poorly measured using just historical returns. Clearly, the assumption of normally distributed returns for most asset classes has been proven conclusively wrong - tails are actually much fatter, and extreme events more likely than many models have assumed. Perhaps most important of all, in 2008 we also saw a widespread reluctance on the part of too many professionals to disagree with their models, even when their instincts may have been screaming that danger lay ahead. This is both an intellectual problem (why were people so reluctant to openly examine and seek to reconcile this growing conflict?) and an organizational one (undoubtedly many people faced incentives that discouraged them from "making waves").

Going forward, the events of 2008, far more so than those of 1987, 1998, and 2001, seem likely to produce wide ranging changes in the investment management profession. On the modeling side, we believe we will see a shift towards regime-shifting models, and asset allocation approaches focused on hedging exposure to different macroeconomic scenarios. Bridgewater, Windham and GMO already seem to be moving in this direction, as are we here at Index Investor. In fewer cases, we may also see more willingness to adopt disequilibrium approaches like adaptive markets, which enable you to take overvaluation into account without feeling you are betraying your intellectual heritage. Parameter estimation errors will also receive much explicit attention, with greater use of techniques like shrinkage estimators and non-normal distributions. Undoubtedly, the result will be better models, and, we hope, greater willingness to openly recognize and examine apparent conflicts between models' recommendations and investors' common sense. The willingness to do this would also be helped by a shift in the ranking of investors' goals -- with less focus on outperforming index benchmarks and peers, more on risk adjusted outperformance of internal liability driven benchmarks (e.g., are we on track to achieve our retirement income goals), and greater willingness to risk underperformance during uncertain times in order to maximize preservation of capital.

We are not exempt from this process, even though we avoided some of the biggest mistakes that were made in the run-up to the 2008 crisis. For example, as we have noted before, we are reevaluating the role of credit versus government bonds, and of assets like gold (see this month's Product and Strategy Notes) and Swiss Francs that have delivered positive returns during the high uncertainty regime. So, to reiterate our initial answer, we have not lost our faith in diversification; rather our challenge is to learn from experience and adapt our approach to achieving it.

December/January 2008-09

I appreciate the supply and demand of returns methodology you use in your month Asset Class Valuation Update. However, I'm still struggling with how it applies to commodities, and in particular how the equilibration process works. Could you please explain this again?

The supply of returns from a commodity futures-based index comes from four sources. Since futures can be purchased at less than their face value, excess cash is invested (e.g., in real return bonds) to generate returns. There is also a diversification return that comes from investing in a range of commodities (e.g., energy, agricultural and metals) that have low correlations of returns from each other. The third source of return is the so-called "roll yield" that is earned when futures contracts nearing maturity are sold, and the proceeds used to replace them with longer-dated contracts. When futures prices are "backwardated" (i.e., near dated futures prices are higher than longer dated prices), the roll yield is positive; when prices are contangoed (near term prices are lower than longer dated prices), the roll yield is negative. The final source of returns is unexpected changes in the price of the commodity (i.e., price changes that the seller of the contract has not assumed when transacting with the buyer).

Let us now assume a situation like past two years, when the economy was rapidly growing. This caused a tightening of the supply/demand balance for many commodities. In turn, this raised the probability that price surprises would occur (e.g., a positive surprise in the case of an unexpected supply outage, or a negative surprise if demand growth was less than expected). It also raised the premium on having access to physical commodities. When the supply/demand balance is slack, a company that uses a commodity has the choice of either buying, financing, and storing it, or buying sufficient futures contracts that can be exchanged for physical products. Logically, all else being equal, the price of the futures contract should equal the spot price (i.e., the cost of buying the commodity today), plus the cost of financing the purchase and storing the physical product until it is used. However, when supply/demand conditions are tight things change, particularly when there are high costs associated with running short of the commodity in question. Under these conditions, owning the physicals, rather than taking a chance on something going wrong with the delivery of a futures contract, and/or having more flexibility to vary production by using the stored physicals, may seem a more prudent course of action to take. Hence, users of commodities, when supply and demand are tight, may bid the spot price up higher than the futures price, producing a "backwardated futures curve" - and profitable roll yields for investors in commodity futures-based  index products (technically, this is known as the "convenience yield" effect).

So far, so good. Now let's consider the dynamics underway on the supply and demand sides of this market. On the supply side, rising prices will lead to investment in new capacity. And on the demand side, rising prices will lead to searches for more efficient processes and substitute products. Eventually, this will result in a shift in the supply/demand balance, unexpected falls in commodity prices, a shift from a backwardated to a contagoed futures curve, and negative roll yields. But that's not the end of the story.

For all of the Letters to the Editor for this month, please use the following link: December/January 2008-09