Product and Strategy Notes: Technology and Active Management; Hedge Funds Update; and Update on H5N1

On the Frontier of Technology and Active Management

As we have frequently noted, successful active investment management ultimately comes down to successful forecasting, of either fundamental asset values and/or the future actions of other investors. In turn, successful forecasting requires either access to superior information, and/or to a superior model to derive actionable investing insights from it. While almost every investor will, at some point, generate a superior insight, the data show that it is extremely difficult to do this consistently year after year. Superior information sources dry up or are copied, while superior models are replicated by competitors or have their key assumptions invalidated by changes in the structure of the real economy. This is why, as the investment horizon lengthens, the percentage of active managers who outperform index funds shrinks so dramatically.

However, the exploding number of hedge funds and still large number of actively managed mutual funds provide ample evidence that there are still many managers and investors who believe it is possible to outperform a well-diversified portfolio of index products over the long-term. In point of fact, the success of those index funds requires this belief in active management. If at least some investors did not believe they could "beat the market", nobody would search for new information or build new models, and asset prices would soon become inaccurate, thereby invalidating indexing’s basic premise.

What most people fail to appreciate is just how difficult "beating the market" has become. Essentially, there is a "technological arms race" underway among the world's most active (and highly-compensated) managers, which generally include investment banks' proprietary trading desks and hedge funds. This arms race has four main components. The first is developing new means to fuse multiple types of information (e.g., quantitative and qualitative data) in order to make sense of complicated and fast changing markets. The second is developing new models that simultaneously capture the multiple processes underway over multiple time frames within and across the markets for different financial assets. The third seeks to build game theory insights into analytical and trading models. Game theory studies the behavior of players in competitive situations, such as the contest between investment managers to earn the highest returns. This reflects the fact that, professional traders are, in a sense, like poker players. To succeed, both require not just an understanding of the odds (e.g., the distribution of past price changes for a given asset), but also insight into other players' likely behavior (to put this more technically, trading and poker both involve a mix of decision and game theory). The fourth component of the technological arms race focuses on the minimization of so-called "trading friction", or the costs involved in translating insights into portfolio positions. These costs include not only explicit commissions and bid/ask spreads, but also the impact of a trade on market prices. One name that has been given to this arms race is "intelligent finance", the title of a recent overview by Pan, Sornette, and Kortanek.

This month, we'll highlight one example of this trend. In the years since the September 11, 2001 terrorist attacks, the U.S. and other intelligence agencies have made substantial R+D investments in the field known as "NIMD", for "Novel Intelligence from Massive Data Sets." This is a technological response to the failure of the intelligence agencies to "connect the dots" in the period before 911, when, in hindsight, clues were available that, if they had been correctly combined, would have provided warning of the impending attack. We should note here that we are referring not to "strategic warning" (i.e., what and why?), which existed, but rather to so-called "operational warning" (how?) and "tactical warning" (when, where and who?), which were lacking. Crucially, many NIMD investments are focused on mining textual and other non-numeric data to automatically generate hypotheses, marshal and evaluate evidence for and against them, and generate conclusions for human review. This new technology goes well beyond simple "search", and focuses instead on "sensemaking." Its implications are enormous, and its application to active investment management is obvious.

In fact, they are beginning to occur. One example of this is a fascinating working paper by Feng Li of the University of Michigan. In "Do Stock Market Investors Understand the Risk Sentiment of Corporate Annual Reports?", he analyzes 10-K filings (which are available electronically) using a simple algorithm that counts the frequency of what Li calls "words related to risk and uncertainty." He finds that companies with a higher number of these words in their 10-K have a much higher probability of negative earnings and returns during the subsequent year, and shows how a trading strategy based on this system would, in hindsight, generated substantial alpha (returns above the relevant market index).

Another implementation of this new technology is known as "news-based algorithmic trading." This approach correlates historical news events and asset price changes to build models that can identify (and perhaps automatically trade on) similar "high price impact" news events in the future. See, for example, the paper "Which News Moves the Euro Area Bond Market?" by Anderson, Hansen and Sebestyeu. Another example is the so-called "heat index" produced by Relegence Corporation (www.relegence.com) for professional traders. It provides real time tracking of more than 10,000 news feeds, and highlights companies and other key phrases mentioned with the highest frequency. Today, this is being used to warn of impending increases in an asset's volatility (which is correlated with an increasing number of news mentions).

In sum, there is a reason that hedge funds and investment bank proprietary trading desks are now filled with more physics, math, and psychology PhD's than many university departments - consistently successful active management (i.e., the ability to consistently generate alpha) is becoming an ever-more capital and technology intensive, and extremely difficult game to play successfully over the long-term.

Hedge Funds Update

Hedge funds have once again been in the news, with Federal Reserve Bank of Atlanta holding a conference dedicated to the issues they raise, and the European Central Bank publishing a commentary on them in its latest Financial Stability Review. One of the most controversial subjects in the world of hedge funds is the extent to which existing indexes provide an accurate picture of the risks and returns from these investment strategies. The essential problem is that hedge funds self select when they will start reporting their results, and to which index they will contribute them. This creates a number of potential biases including selection (only successful funds will choose to report), survivorship (the indexes don't reflect the results of funds that stop reporting) and backfill (when firms start reporting, the index is "backfilled" with the fund's results up to that point, which often aren't matched by subsequent performance). Many academic papers have asserted that the net impact of these biases is a substantial overstatement of hedge fund returns, and an understatement of risk. In particular, a paper by Malkiel and Saha ("Hedge Funds: Risk and Return") that found a 4.5% over statement of average annual return has been a particular target of criticism by hedge fund supporters. Their counterargument is that survivorship bias is actually quite low, because it isn't just failing hedge funds that stop reporting, but also successful funds that are closed to new investors (the apparent logic being, "why report to an index if you aren't seeking new funds from investors?"). At the FRB Atlanta forum, Malkiel and Saha responded with another paper ("Why Do Hedge Funds Stop Reporting Their Performance") that found funds that ceased to report are more likely to be failing than delivering high returns.

Another issue that came up at the conference was the lower returns on so-called "investable" hedge fund index products, compared to those on the broader hedge fund indexes that they seek to replicate. Two explanations were offered, and both make sense. The first is that, because they allow more frequent capital withdrawals than the typical hedge fund, the investable index products should logically have lower returns because they carry less liquidity risk. The second point is that these investable products are based on larger hedge funds, and hedge funds' performance (like mutual funds') tends to decline with size (e.g., because great $25 million dollar investment ideas are easier to find and profitably execute than great $1 billion investment ideas).

Another set of papers looked at just how successful hedge funds have been at delivering alpha - that is, returns above some type of investable asset class benchmark index. In "How Smart Are The Smart Guys?", Griffin and Xu make a number of interesting observations: the median equity hedge funds trades twice as much as the median mutual fund, its portfolio has a lower correlation to the market index than its median mutual fund peer, and it has a more pronounced tendency to prefer small value stocks. However, the authors cannot prove that "hedge funds in general are any better at long stock picking or timing sectors than mutual funds," leading them to "question the ability of long-equity hedge funds to add value." They acknowledge, however, that "hedge fund firms seem to have more differential ability in stock picking than mutual funds," lending support to the widely held belief that the most talented active managers have moved from mutual to hedge funds because of the superior potential compensation they provide for active managers who are either very lucky or very skilled. In a related paper ("Sources of Hedge Fund Returns: Alphas, Betas, and Costs"), Ibbotson and Chen find that, in aggregate (and this varies across styles), underlying asset class returns (beta) account for the majority of hedge fund net returns after manager fees (5.4%, versus average alpha of 3.7%).

As always, the problem lies in identifying hedge fund managers who are truly skilled at generating alpha. In this regard, two other papers provide a sobering perspective. In "A Portrait of Hedge Fund Investors", Baquero and Verbeek show that while investors are quite quick to pull their money from underperforming funds (unlike the average investor in actively managed mutual funds), they show no reliable ability to pick winning managers in advance. In "Hedge Funds: Performance, Risk and Capital Formation", Fung, Hsieh, Naik and Ramadorai find diminishing returns to scale as hedge funds grow in size, as well as declining risk adjusted returns as they have become more popular and attracted more managers and capital. Specifically, they find that Funds of Hedge Funds (their proxy for an investable index) have seen declining returns, and on average deliver zero alpha (though about one in five did manage to deliver statistically significant alpha). Another paper ("Do Funds of Funds Deserve Their Fees on Fees?" by Ang, Kropf, and Zhuo) argues that because of the services they provide investors, funds of hedge funds deserve the fees they charge. However, other commentators have argued that the so-called "multistrategy" fund (in which a single fund company utilizes multiple strategies) is a more efficient approach than the fund-of-funds structure with its higher level of fees. Time will tell who is right.

This brings us to the European Central Bank paper. Its primary concern is the potential impact of the growth of hedge funds on the stability of the financial system. It begins by noting both the rising correlation of returns between funds within different hedge fund categories (with equity market neutral and global macro notable exceptions to this trend) and the rising correlations between the categories themselves. For example, our own research shows that, in the two years ended April 30, 2006, the average monthly return correlation between hedge fund categories in the value weighted Tremont Hedge Index has risen by .27 compared to the five years ended in April, 2004. In the eyes of the ECB, "the increasingly similar positioning of individual hedge funds within broad hedge fund investment strategies is another major risk for financial stability which warrants close monitoring despite the essential lack of any possible remedies. This risk is further magnified by evidence that broad hedge fund investment strategies have also become increasingly correlated, thereby further increasing the potential adverse effects of disorderly exits from crowded trades. It is difficult to gauge what could cause correlated sell-offs and how damaging these could be, but one possible trigger could be an abrupt end of the recent global search for yield possibly induced by the tightening of global liquidity conditions. A further slowdown of inflows into hedge funds or even widespread redemptions could also exert pressures on individual hedge funds to liquidate increasingly less liquid holdings, as more hedge funds seem to be venturing into less liquid markets in order to earn [additional returns from] the associated liquidity premium."

What is an investor to make of commentaries like these? The first lesson is that, as is usually the case, when a system becomes homogenous - that is, when it ceases to contain sufficient diversity - it becomes very vulnerable to sudden and substantial changes. In the opinion of the European Central Bank, this is also true for financial systems, and the increasingly homogenous alpha generation strategies employed by many of the world's 8,000 plus hedge funds. The second lesson is that, despite the intense competition among hedge fund managers, there are still those who can generate alpha, whether that is due to superior insight into fundamental value or the psychology of other investors. As always, the problem is how to identify these skilled managers in advance. The latest research shows that people who invest in hedge funds are no better at this than mutual fund investors, though the former are quicker to cut their losses.

In our opinion, this leads to a third lesson - while uncorrelated alpha is a very welcome addition to any portfolio, it is so hard to consistently generate that an average investors long-term asset allocations to these strategies should be relatively small. As we have noted in other writing in this journal, the equity market neutral strategy seems to most consistently deliver alpha that has a very low correlation with returns on broad asset class indexes. That is why it is our preferred vehicle for a long-term policy allocation. In addition, if one wanted to "outsource" the tactical shifting of one's allocations between asset classes, it would be logical to add a second allocation to a global macro type strategy. Unfortunately, there are still relatively few investment vehicles available to individual investors that make it easy for them to implement these allocations. Most mutual funds that claim "market neutrality" fail to achieve it, and are, in fact, long/short funds with a long bias. They retain more exposure to overall equity market risk than we would like to see. As for global macro funds, there are still too few in existence, and those that are available often use too few asset classes. In the United States, the Pimco All Asset Fund (PASAX), with its long-term five percent target real return of remains our favorite by far.

H5N1 Update

As we have repeatedly noted in our economic updates, the possibility of an H5N1 influenza pandemic is the principal "wildcard" global economic scenario we have been monitoring. The most recent developments have not been encouraging. Particularly in Indonesia, there is growing evidence of more efficient human-to-human transmission. The fact that the U.S. government has sent a portion of its strategic Tamiflu stockpile to "an unnamed Asian country" is evidence that we are not the only ones who have reached this conclusion. However, because of poor reporting by local health authorities, the mortality rate from the more transmissible virus is unclear at this time. For cases who have been hospitalized, it has been quite high, but nobody seems to have a good idea of the total number of people who have been infected. As always, we recommend the World Health Organization website (www.who.int/csr/disease/avian_influenza/en/), the U.S. Center for Disease Control site (www.pandemicflu.gov) and Recombinomics Inc. (www.recombinomics.com) for the most up-to-date information on this evolving situation.

| 2006-2007 Benchmark Portfolios | This Month's Issue: Key Points | Retail Financial Services Trends and Opportunities | Product and Strategy Notes: Technology and Active Management; Hedge Funds Update; and Update on H5N1 | Global Asset Class Returns | Asset Class Valuation Update | What Happened to the Financial Markets in May? | This Month's Letters to the Editor: Convergent/Divergent Investment Strategy; EMN Funds vs Long/Short Equity Funds; Equity Market Neutral in Portfolios; YTD PCRDX Performance; Japan Overvalued?; Forecasting Volatility; Portfolios During Economic Downturn; Timber and Residential Property; Concentration of Swiss Equities in Swiss Portfolios; and Changing Asset Allocations |