How Investors Can Achieve True Diversification — And Better Returns
Institutional investors often fail to build diversiﬁed portfolios because the asset classes in which they invest actually have largely the same risk. To achieve true diversiﬁcation — and better returns — institutions should instead construct portfolios that contain a broad selection of risks.
I FIRST PRESENTED THE IDEAS CONTAINED IN THIS article at an Institutional Investor conference in January. Before taking the podium I struggled to come up with something that would convey the importance of what was to follow. I decided to inform this audience of professional pension fund and endowment managers that their luck had run out: They were going to learn about investing in risk, either from me or from someone else, but — not unlike having dinner with the in-laws — it was going to happen, so they would be well served to get it over with. The doors were locked, the audience lost hope, and we pushed forward.
For most of the audience, 2008 had been a shock, characterized by a serious loss of self-esteem and devastatingly poor performance that had been exacerbated by a fundamental lack of risk diversification. But such an outcome was not inevitable. In what follows I describe and substantiate how, through the use of a novel, rational allocation process, investors can achieve significant improvements in portfolio diversification, thereby potentially increasing the return and decreasing the risk of their portfolios.
Post-2008 most investment portfolios have remained one-trick ponies, effectively possessing a single source of variance, a typical proxy for risk. In the case of private investors, with a “standard” 60-40 stock-bond portfolio, roughly 97 percent of the portfolio’s variance can be explained by the equity allocation. Perhaps surprisingly, the typical U.S. pension fund doesn’t fare much better, despite a more broadly diversified asset allocation, full-time investment professionals and, in many cases, externally hired experts — variously referred to as “guys from out of town” or “guys with shiny shoes.” Accordingly, the typical U.S. pension fund has in excess of 90 percent of its variance explained by a single risk factor.
There are three basic reasons for this. First, when the bulk of an investment allocation is directed toward its most volatile component, then the emergence of a single dominant risk factor is a virtual certainty. Second, many investors mistakenly assume that nominal, or apparent, diversification is the same as real diversification. This is, of course, not true. To understand why nominal is not real diversification, we must pause for a moment to consider the three basic allocations in a typical diversified institutional portfolio: equities, fixed income and alternatives. With respect to the equity allocation, when you own every stock on the planet, you have in effect eliminated any sources of idiosyncratic risk and now own a single, rarefied risk factor: systematic equity market risk.
With respect to fixed income, when a typically underfunded U.S. pension plan decides to target a 6 percent rate of return on its bond portfolio in a world where ten-year U.S. Treasuries are yielding an anemic 1.8 percent, it is forced to acquire a potpourri of below-investment-grade bonds. In so doing, it has effectively converted the portfolio of daddy-needs-a-new-pair-of-shoes bonds and its associated duration risk (that is, the sensitivity to changes in interest rates) back into equity risk. Conceptually, this happens because default probabilities on corporate debt are typically calculated by the market using a model developed by Nobel Prize–winning economist Robert Merton. The Merton model structures the problem as a call option on the assets of the company issuing the debt; a decrease in the portfolio’s credit quality, therefore, effectively represents a transformation of fixed-income duration risk back into equity risk.
Then there’s the curious case of so-called alternatives. Spoiler alert: These investments are also highly correlated with equities. As an example, the HFRI Global Hedge Fund Index has been correlated versus the S&P 500 index at greater than 0.9 for approximately the past two years. For those of you now defensively pointing out that you have both real estate and private equity, I say, “Whatever helps you sleep at night.” Were it not for the fact that these assets possess neither frequent nor accurate pricing, they would both correlate above 0.9. The bottom line is that the typical U.S. pension fund is diversified in name only, with the broad equity market representing most of the risk.
The third explanation for concentrated risk factor exposure is the inertial effects of outmoded thinking. Investment professionals have historically been trained to think of portfolio construction as an optimal, sign-constrained (that is, long-only) allocation to a collection of assets; the effect of this is to concentrate portfolio risk within a single factor. The failure of both individual and institutional investors to achieve adequate diversification is a direct consequence of the failure to recognize that, contrary to conventional wisdom, investing should not be a process of selecting assets; rather, it should be a process of selecting and accepting risks.
To understand this distinction, consider the return you would require to hold an asset that is completely devoid of any form of risk — an idealized asset immune to any market or implied volatility, credit exposure or inflation, possessing virtually infinite liquidity. The return required to hold such an asset would be very low indeed, if not zero. Now consider how the required return would vary if the asset were a stock, a bond or a commodity. Such a definitional distinction is irrelevant. For example, if the asset were a commodity, one could simply form a corporation to purchase it, have the company issue equity or debt, and — presto — the commodity would be nominally reclassified without modifying any of the actual underlying economic exposure.
Last, consider how the required return would change as a result of an increase in an identified risk factor. If the idealized asset suddenly became dramatically more volatile, with a significant left-tail skew — that is, if large losses became much more likely than large gains — or more sensitive to movements in the U.S. dollar or yield-curve shifts, how would the expected return change? Clearly, it would have to increase, because investors typically require compensation in direct response to the magnitude of the increase in an identified risk. The bottom line is that you are compensated as a result of the risks you assume and not in any direct or obvious way as a product of the types of assets or classes of securities in your portfolio. Furthermore, if you wish to introduce a productive measure of control into your performance outcomes, you need to identify the risks you are assuming and make your investment decisions accordingly.
Investing in risk is a nebulous and difficult concept for many investors. “How do I buy risk?” “What risks should I be looking to buy?” “What sort of compensation should I expect?” “The last time I looked at the menu of options available for my 401(k), no such choices existed.”
The reality is that investors currently do invest in risk. However, in most cases the risks are not well understood and, as noted above, typically boil down to a single exposure: systematic equity risk.
It should be noted that there is nothing inherently wrong with assuming systematic equity risk. Even under the strictest assumptions of the Efficient Market Hypothesis, one can expect some degree of compensation for investing in a broadly diversified basket of equities; the only theoretical constraint is that predicting the return in any period with any useful degree of precision will be impossible. Those of you facile in the dark arts of stochastic calculus may wish to refer to Nobel laureate Paul Samuelson’s excellent paper, “Proof That Properly Anticipated Prices Fluctuate Randomly.” Nonetheless, there is pretty solid evidence that, on average, you get compensated for assuming equity market risk; there’s so much evidence that it even has its own name: the equity risk premium.
THE EQUITY RISK PREMIUM, WHICH IS typically defined as the expected excess return of equities over the risk-free rate (while remaining coyly silent on the precise meaning of the term “risk-free rate”), has been well documented as a relatively persistent and positive feature of the equity markets. It has also been established that it is by no means a certainty that one will obtain a positive outcome, and that’s why it’s called a risk premium and not merely a premium. A recent study produced by Robert Arnott, founder of Newport Beach, California–based asset management firm Research Affiliates, revealed that there has historically been an approximately 15 percent chance that the risk-free rate will actually exceed the return on equities even over periods of time as long as 20 years.
There are many competing theories that have been developed to explain why this risk premium exists. They run the gamut from explanations rooted in the Efficient Market Hypothesis — essentially, when you have diversified away all that can be diversified away, you are confronted with a degree of inescapable market volatility that deserves compensation — to far more entertaining explanations rooted in the frailties of human behavior. Most notable among the latter is a combination of myopia and loss aversion. In brief (and, wow, am I paraphrasing), when looking too frequently at investment return data, our “inner punk” (my term, not theirs) has an asymmetric sensitivity to losses versus gains and tends to cause mere mortals to abandon their investments before achieving an appropriate long-run return. An interesting study, reported on a few years ago by the Wall Street Journal, linked significant improvements in trading to the ability to engage in just the sort of risk-neutral decision making common to those with organic brain disorders resulting from blunt-force trauma or chronic alcoholism. It seems that the path to a higher Sharpe ratio may be shorter than you ever imagined.
As luck would have it, and in sharp contradistinction to the typical equity-risk-heavy portfolios foisted upon the public by legions of experts — recall the out-of-town guys with shiny shoes — there are other well-documented risk premiums one can harvest in the financial markets. Such risk premiums, if harvested correctly, can provide relatively robust returns over time while maintaining a reasonably low degree of statistical dependence versus other assumed risks.
One common risk premium harvested by many of the world’s more sophisticated investors — here defined as art collectors or users of private jets — is forward-rate bias. For the uninitiated, back in 1923 British economist John Maynard Keynes wrote a seminal paper in which he presented the notion of “uncovered interest rate parity.” The basic idea was that when you have two currencies with varying domestic interest rates, the one with the lower interest rate will have an unbiased tendency to appreciate versus the one with the higher interest rate, so that there is no persistent reward associated with borrowing in the low-interest-rate regime, converting into the other currency and lending in the higher-interest-rate regime — a sensible appeal to the gods of market efficiency.
This wonderful paper, richly deserving of the accolades it received, turns out to be empirically incorrect. Since the time of its authorship, the paper’s core conclusion has been refuted in countless studies by many of the leading lights of econometric research, many of whom share a vested interest in the Efficient Market Hypothesis. Such research has demonstrated again and again that, even on a risk-adjusted basis, you tend to get rewarded for lending in high-interest-rate currencies while funding such loans by borrowing in low-interest-rate currencies. The question is, how is this possible? The foreign exchange markets are open around the clock. They are the deepest, most liquid markets in the world, and everyone knows about this game.
There are a number of competing explanations. One of the more compelling ones relates to the distributional profile of the returns associated with what is frequently referred to as forward-rate-bias investing, or simply “carry” investing. The distribution of returns is, to use a ten-dollar word, highly leptokurtic: essentially, possessing a high probability of a very large loss. As the argument goes, humans tend to demand a higher rate of return when they are subject to a potentially larger than expected loss. Another explanation posits the existence of supply-demand imbalances for investment capital across national economies. Though these explanations have merit, they are not correct. As I pride myself in having all the right prejudices, I will share the correct explanation with you.
Several years ago I became obsessed with understanding why the Japanese yen was failing to appreciate versus the dollar as per the prescription of uncovered interest rate parity. A review of the publicly available data on the Bank of Japan site on Bloomberg revealed the answer. In an act that served to reveal in part that I had too much time on my hands, I calculated that the BoJ had for the preceding three months been printing yen, converting them into dollars and buying U.S. Treasury securities at a rate of $137 million an hour, including weekends.
So there’s your answer: A dominant player in the forex market was pursuing a nontraditional economic objective. The BoJ was clearly not attempting to maximize the risk-adjusted return on a currency portfolio; rather, it was pursuing the mercantilist objective of benefiting Japan’s local producers by making the prices of domestically produced goods more internationally competitive.
It would be wrong, of course, to single out the BoJ for pursuing this action. It is in fact quite common for the world’s central banks to engage in this activity, and as long as they do — to wit, as long as there is more than one currency — investors can participate in the market and effectively tax this behavior. It is worth noting that a veritable industry has cropped up around this empirical feature of the forex markets. It is, for example, not uncommon for even very simple expressions of carry strategies — for example, being long the three highest-yielding Group of Ten currencies and being short the three lowest-yielding (with a pinch of not a darn bit of insight, rolled once a month) — to explain the preponderance of the returns of even the most sophisticated global macro investors. Recall the definition of “sophisticated.” In the interests of public safety, there is one required caveat: To avoid any career-defining return outcomes when employing this risk premium, it is important to remember to implement an investment approach that effectively carves off a portion of the left tail through the use of a rigorous money management process — in layman’s terms, if it starts to move against you in a scary way, have a plan. A little financial engineering can go a long way to improving the outcome.
Another highly useful risk premium is variously known as commodity momentum, commodity trend trading or, more to the point, capturing positive first-order serial correlation — that is, it went up (down) last period, so chances are it will go up (down) this period, so I’m going to get long things going up and short things going down.
To understand the role this risk premium can usefully serve in a portfolio, it is worth considering some of the core stochastic properties — technical gibberish for how prices jiggle around over time — of commodity price movement. The two most important features of commodity price movement are significant right-tail skew (that is, large, sudden price increases) and mean reversion. This last feature — that the price tends to end up back where it started — was undoubtedly responsible for precipitating the Wall Street truism that bulls make money and bears make money, but pigs get slaughtered.
Anyone who has lived with a commodity investment can tell you that the preponderance of its returns will typically be dominated for very long periods of time by very sudden, short-lived events. The basic intuition is that for many commodities there is a short-term highly inelastic demand. If the supply is suddenly disrupted, large upward price movements will be required to clear the market. In the short term, in the absence of a relevant substitute, you and your neighbors will continue to insist on eating, heating your homes and going to work. This feature of commodity price movement provides for a very useful risk factor exposure. By engaging in an investment strategy that replicates a straddle, being long both a put and a call option on the traded asset so that one profits from either a large up or down price movement, it is possible to capture a highly valuable characteristic of commodity price movement: namely, that the peak periods of performance for this risk premium tend to be precipitated by significant exogenous shocks such as wars, frosts, droughts, hurricanes or pod borers munching through cocoa pods and coincide quite usefully with the periods when the rest of your portfolio is being ravaged by the very same exogenous events.
The other hallmark feature of commodity price movement — mean reversion — is a critical feature that is more often than not overlooked by commodity investors. Basically, when the drought ends or peace breaks out, commodity prices tend to drop back to lower long-run equilibrium levels. Therefore a momentum investment strategy rather than a simple long-only approach tends to yield much more satisfying results over time.
Risk premiums, when properly conceived and harvested, represent a core, elemental source of both risk and return. As a result, they generally maintain a relatively orthogonal (a good cocktail party word that means statistically independent) profile that provides a superb building block when constructing a portfolio. Why is statistical independence so important? It is important because it is a powerful ally in developing well-risk-managed, profitable portfolios. As an example, all else being equal, a portfolio consisting of three uncorrelated assets will have significantly lower volatility — on the order of 20 to 25 percent less — than a portfolio containing an infinite number of 0.5-correlated assets. Additionally, this three-asset portfolio will make more money over time. This increase in the compounded rate of return is directly attributable to a reduction in volatility — a point that is analytically provable. Essentially, the geometric mean return, or compounded rate of return, is equal to the average return minus a volatility correction. Having access to a collection of relatively independent building blocks provides a real opportunity for improving risk control and enhancing return.
BEFORE MOVING ON, IT IS WORTH issuing a warning: There is no agreed-upon list of what constitutes a risk premium. Importantly, there are investment programs masquerading as risk premiums that are in fact conflations of risk premiums. Perhaps the most notable example of this is convertible bond arbitrage. In my experience, this investment strategy is generally not well understood from a risk-factor-exposure standpoint by most of those executing and investing in this strategy.
Although there are a number of potential convert-arb strategies, the usual approach is to purchase a convertible bond, typically a corporate bond with an associated warrant (a call option on the stock) and then periodically delta-hedge the warrant — that is, sell the amount of stock necessary to effectively neutralize the equity market exposure of the warrant at the current price level. By doing so the investor effectively creates a straddlelike payoff on movements of the equity component of the convertible security. If the market moves up significantly, the warrant appreciates more than the loss associated with the short position in the stock (the delta hedge). If the stock moves down significantly, the profit associated with the short stock position will be greater than the loss attributable to the warrant. A sharp move in the price of the stock, either up or down, will generate a profit.
This payoff feature is what causes many investors to mistakenly refer to convertible arbitrage as a long-volatility strategy. The fundamental error that most investors make when evaluating this strategy is to ignore the inherent short-volatility characteristics of the fixed-income component of the convertible security. Convertible securities are usually lower-credit instruments that tend to suffer mightily during periods of systemic risk and therefore possess a payoff profile that can be very effectively modeled as a short exposure to fixed-income volatility. Empirically, this feature tends to dominate the return outcomes of a convert-arb portfolio during periods of excessive market volatility.
At the portfolio level convertible arbitrage is a combination of equity beta, fixed-income beta and various market volatility exposures — a number of risk premiums. During a systemic risk event— technically speaking, when the market is going to hell in a handbasket — a convert-arb portfolio will have the precise look and feel associated with being short a call option on the U.S. ten-year note; you end up losing money at an increasing rate as U.S. Treasuries move higher in response to the market risk event. Make no mistake, this effect will dominate. The bottom line is that referring to convertible arbitrage as a risk premium is wrong and somewhat harmful, as it obscures the true underlying risk factor exposures and may serve to severely complicate the portfolio construction process.
Without becoming overly concerned with the math, it is an established fact that by increasing the number of statistically independent building blocks (N), risk can be significantly reduced while increasing expected return. As a rule of thumb, this effect is particularly pronounced up to about ten independent blocks but diminishes rapidly thereafter.
Getting a large-valued N is very important from a tactical standpoint because applying even sound asset allocation techniques, such as risk parity, to a small number of risk factor exposures can result in very poor outcomes. It is worth noting that risk parity is an optimization technique and not an asset class, as many experts and investors mistakenly believe. Risk parity simply targets, or minimizes, volatility subject to the constraint that the contribution to portfolio risk is equal across investments. Take some of the risk parity products that are now available. Having reverse engineered a number of them, I have found that such programs were marketed during a period characterized by two key empirical market features: a multi-
decade secular bull run in the U.S. fixed-income markets and a reasonably reliable anticorrelation in stocks versus bonds. The presence of these two key empirical features has provided many investors with a misleading sense of optimism with respect to what performance will look like going forward. This is a classic quandary, well documented in the academic finance literature, known as the peso problem — essentially, a situation in which in-sample, or historical, data is not representative of the full distribution of possible outcomes out of sample (that is, in the future). Unfortunately, life is out of sample, at least for those of us lacking a DeLorean with an aftermarket flux capacitor.
To understand just how easily something could upset these narrowly defined, long-only, small-N portfolios, all that is required is plausible, existential proof of their frailty. To that end, I now present for your amusement the sci-fi epic “A Tale of N.” So grab your blanket and juice box; it’s story time . . .
Stardate 2014: Politicians grown weary of the Federation Reserve’s war against deflationary forces and the moribund pace of economic growth decide to launch a rearguard action. On the advice of Borg representative PK, they mint a trillion-dollar platinum coin that they call the Krugermand, which actually contains only 17 Fed dollars of platinum. Placing the new coin on reserve with the Federation, the government then uses the excess funds to dig and fill holes in the desert until the economy revives. This final reckless act of fiat money production inflames inflationary expectations, disrupting economic activity, simultaneously crushing both the bond and stock markets.
You probably remember that the trillion-dollar coin has already been floated as a possibility by a blogging attorney operating under the pseudonym Beowulf; for the name of the coin, however, I alone deserve scorn. In Beowulf’s blog he rightly points out that the Treasury has already been granted the necessary seigniorage power to mint such a coin as a means of evading debt ceiling restrictions.
If this story doesn’t meet your minimum credulity threshold, perhaps something in the horror genre would be more to your taste: An employee of the Federation has a grand mal seizure while sitting at his computer terminal, convulsing forward with his snout coming to rest on the terminal’s print key. He goes undiscovered for more than two years. In the intervening period he prints $84 billion a month, which is used to purchase an amalgam of mortgage-backed and Treasury securities — essentially, the same ending as the first story.
THE PATH TO DIVERSIFICATION IS not without challenges. Identifying and gaining access to a broader set of complementary risk factors has proved challenging, perhaps nowhere more so than through the use of hedge funds.
Many of the challenges of hedge fund investing have been well documented. They run the gamut from such obvious issues as a lack of adequate risk-position transparency and infrequent liquidity to more-headline-grabbing issues, such as the arbitrary gating of investor capital, the issuing of money-losing side pockets and, of course, the ever-popular, bestseller-inspiring frauds, Ponzi schemes and use of material nonpublic information. The desire to include additional independent risk factors unfortunately introduces a range of unintended and potentially harmful exposures.
Despite such obvious risks, the hedge fund industry has experienced massive growth as a result of the perception that fortune favors the bold and financial rewards await competent, diligent investors. The evidence for this is far from conclusive, however. In his book, The Hedge Fund Mirage, Simon Lack argues that rich, asymmetric fee structures have resulted in the less than desirable outcome that 85 percent of all investment profits earned by the hedge fund industry have accrued to the managers, while on an aggregate dollar-weighted basis, the industry has underperformed the three-month U.S. T-bill. Though many have quibbled with Lack’s methods for deriving these results, there has been far less debate concerning the veracity of those results.
Setting aside the performance debate and recognizing the fact that many investors have had positive experiences dealing with certain managers, there remains the important yet frequently overlooked issue of the impact of rich performance-based compensation on the portfolio construction process. Aside from the obvious economic drag, performance fees introduce a significant distortion to the process of portfolio optimization. In brief, what constitutes optimal in the presence of performance fees is quite different from optimal construction in the absence of such fees. This difference is a result of the direct analytical linkage between diversification, a good thing, and netting risk, a bad thing.
To better understand this problem, consider the following simple example. Mrs. Smith manages the Happy Lucky Golden Super Diversified Fund of Funds. In a given year, because of her extraordinary efforts to produce a highly diversified portfolio, she finds herself in the unpleasant position of having half of her managers up 100 percent and the other half down 100 percent. On a gross basis she is able to break even for her clients during a difficult year. However, with the 20 percent performance fee structure — an industry-standard fee — the 100 percent return achieved by the good managers becomes a net return of 80 percent. Subsequently, the net return at the portfolio level is now –10 percent; the 10 percent difference between gross and net return represents the cost of having to net out the incentive fees. Mrs. Smith’s revealed talent for producing a diversified portfolio has directly contributed to her clients’ losing money. From a technical standpoint performance fees introduce a new term into the objective function; we must now account for netting expenses that are directly and positively related to improvements in portfolio diversification. Incentives matter, and performance fees serve to increase overall portfolio concentration and risk.
Oddly enough, over the long term successful investing requires honesty. We are required to acknowledge what is possible and what is not. The acceptance of obvious truths, however, especially when they are unpleasant, is difficult for most of us. When confronted with unpleasant information such as “It’s not the sweater that makes you look fat,” “You didn’t make the rhythmic gymnastics team” or “You can’t forecast financial markets,” there is a fairly typical reaction cycle of denial, anger, bargaining and then, one hopes, acceptance. It is high time to get started on this process for investment portfolios.
First, forecasting returns with any useful degree of accuracy is very difficult. Though legions of experts proffer such forecasts, frequently referred to as capital markets assumptions, the reality is that such forecasts are not the least bit accurate and arguably have done more harm than good. By systematically determining investment allocations through an unholy alliance of actuarial assumptions and woefully unreliable return assumptions, many defined benefit plans have found themselves critically underfunded. This condition creates a harmful feedback loop in which, in response to being underfunded, asset-class return assumptions become increasingly unrealistic while the portfolio becomes increasingly concentrated and fragile.
Let’s review the evidence concerning return forecasts. A 2001 survey published by the thoughtful team of P. Brett Hammond, Martin Leibowitz and Laurence Siegel reviewed the estimates of the equity risk premium made by approximately 20 of the leading lights of academia and the world’s investment community. The reported forecasts were, of course, point estimates, without the requirement that they also report a confidence bound. Nonetheless, we gain an initial insight into the degree of uncertainty faced by forecasters and users of forecasts; the bid-offer spread on the point forecasts was 700 basis points. Things become a great deal more daunting, however, when one includes the prediction error associated with those point estimates: A 95 percent confidence bound around such forecasts is typically on the order of plus or minus 20 percent. From a statistical standpoint you can pretty much pull your forecast from a hat without any real loss of accuracy. Basically, a hat is as good as an expert and, generally speaking, costs less (unless, of course, you went shopping with my daughters — then all bets are off).
Oddly enough, the reaction cycle to this truth is still in the denial stage despite the awarding of a Nobel Prize in 2011 to New York University economics professor Thomas Sargent for a commanding body of research that substantiated the difficulties associated with forecasting returns. In a brilliant television advertisement aired recently by Ally Bank, Sargent is glowingly introduced to an auditorium full of formally attired guests and asked by the moderator if he can tell the audience what CD rates will be two years from now. His answer is a terse no, followed by an awkward silence, a plug for an adjustable-rate CD and closing credits.
In the interest of beating this issue into submission and promptly moving past the anger phase directly to bargaining, I offer the paraphrased results of a nonpublished study conducted by my former colleagues at Commodities Corp., the entity that eventually became the alternative-investment unit at Goldman Sachs Group. One of the objectives of the study was to uncover the proportion of the time directional traders simply got the direction right. Forget about a point forecast with confidence bounds. How about “You bought it and it went up” or “You shorted it and it went down”? Note, I refer to them as directional traders to distinguish them from investors engaging in providing insurance to the markets — that is, doing something that is the spiritual equivalent of selling a deep out-of-the-money option. It is a well-known feature of the option markets that one can make an investment with an arbitrarily high probability of success on any trial; the equilibrating effect is that the investor is then exposed to an associated arbitrarily large loss when it occurs.
At that time, Commodities Corp. had access to a vast treasure trove of individual trading records for what turned out to be many of the most successful directional trading managers in the history of the hedge fund industry. After examining the trading activity of approximately 100 managers, my former colleagues discovered that the best manager, judged purely on this criterion, got the direction right about 60 percent of the time. However, given the large number of managers analyzed — and the statistical reality that one would expect some degree of variability from such a large random sample — this 60 percent-right outcome was basically consistent with the null hypothesis that none of them was any better than 50-50.
So there you have it. We must now move from bargaining to acceptance. When you hear pundits decisively forecasting returns one year into the future, out to two significant digits, bow your head and recognize it for what it is: a marketing opportunity for their respective employers. For those of you required to engage in this activity to put bread on your table, remember that the trick to forecasting well is to forecast often; over the long run you’re probably going to get the direction right only half the time.
In the “what’s possible” category, we have a little good news. As a society we have advanced to the point that we can do a better than random job of forecasting volatility and therefore correlation — a mathematically related endeavor — over short periods of time. Notice how low my expectations are. To put this in context, a three-month forecast of equity market volatility, using advanced but relatively standard methods, has a prediction error typically in the range of plus or minus 1.5 percent.
I hope we have now moved to the acceptance phase, with a core set of investment principles. First, the right portfolio building blocks are risk premiums — well-documented, robust, intuitive, liquid, investable risks — rather than asset classes. Second, there is a right way and a murky way to gain access to risk premium exposures. Third, selected risk premiums should be well-defined sources of return that retain a high degree of statistical independence rather than conflations of risk factors. Fourth, we are well served to have up to ten legitimately different building blocks. Fifth, though we can’t forecast returns with a persistent, useful degree of accuracy, we can get a reasonable handle on volatility and correlation over a short time frame.
The right way to proceed is to construct a portfolio that contains a broad selection of risk premiums, weighted to ensure that the contribution to risk across them is equalized, with due consideration given to individual volatilities, the portfolio’s dependence structure and the overall portfolio volatility.
When proceeding in this fashion, the portfolio is likely to be measurably improved. Statistical analysis will reveal a significant increase in the number of sources of variance. In my opinion, this objective improvement in diversification will serve to maximize the likelihood of both a measurable decrease in portfolio risk and an increase in the growth rate of wealth over time.
Andrew Weisman is chief investment officer and portfolio manager for the liquid alternatives division at Denver-based Janus Capital Group. He has more than 25 years of experience in portfolio construction and risk management, particularly in alternative asset strategies. The opinions are those of the author as of April 2013 and subject to change at any time as a result of changes in market or economic conditions. The comments should not be construed as a recommendation of individual holdings or market sectors, but as an illustration of broader themes.