A little over a month ago, Scott Vincent took aim at much of academic finance by publishing a paper entitled Is Portfolio Theory Harming Your Portfolio? In it, he makes a strong argument for active investment management.
Vincent, founder and managing partner at Green River Asset Management, LLC, a hedge fund based in Baltimore, agreed to answer Five Good Questions.
1. In your recent paper, you argue that modern statistical finance isn’t all that it’s cracked up to be, and that active management can and does work. On what do you base those conclusions?
The fundamental underpinnings of quantitative finance have been shown to be flawed and impractical as applied in practice. Most of quantitative finance depends on the basic concepts and statistical math introduced in CAPM and Portfolio Selection (two of the key theories behind Modern Portfolio Theory or (MPT). We tend to assume that because these theories appear sound, and the math behind them is flawless, they will generate accurate results when applied.
The problem is that mathematics and statistics are rigid disciplines, and they need certain rules and relationships to hold true if they are to generate reliable results. For those rules to be upheld when we apply these theories in financial markets, a set of assumptions is required. Those assumptions are simply unrealistic with respect to financial markets, where relationships are highly fluid. This means that quantitative finance generates unreliable results.
The most important example of this is shown in MPT’s treatment of risk.
When Harry Markowitz wrote "Portfolio Selection" in 1952, he was the first to formally recognize and try to quantify the tradeoff between risk and return. It was incredibly important work — this tension between risk and reward became the standard to describe how just about all securities are priced.
But to make his model work, to fit it to financial markets, he needed a quantitative measure of risk, and he chose variance of returns. Why?
There was no compelling empirical evidence that variance and returns were strongly correlated. It seems instead that variance was chosen because we needed a quantitative measure for risk, and it made some sense that riskier assets would be more volatile. To be fair to Markowitz, he left room for "the judgment of practical men" to alter variance as needed. But in practice, when large computations are done very rapidly, there's little room for judgment.
Just because an asset varies little in price, should it be thought of as safe? What about a highly volatile stock that trades below the level of net cash on the company's books and that generates cash?
I see that as an almost guaranteed return, while MPT would deem it risky. The theories rely on a host of other unrealistic and unfounded assumptions including that returns fit stable, normal distributions and that all investors are rational and have the same views on expected risk and return.
Another important problem with quantitative finance is that it relies on historical measures for model inputs. Practitioners look at the volatility or correlations of the past to predict the risks of the future.
In a dynamic environment, however, circumstances change and so do volatilities and correlations, so historic measures give inaccurate predictions of future volatilities. There's also the practical question of which time period to use in measuring historical volatility. Is it one year, one month, one day, or one hour? You'll often get dramatically different answers for each.
MPT's assumptions don’t square with reality, so we shouldn’t be surprised that its key concepts don’t hold up when put to the test in financial markets. Numerous empirical studies have shown that taking on more risk (as represented by volatility) doesn’t reliably deliver additional reward (see paper for references).
Studies also demonstrate that returns are neither stable nor normal, and that investor behavior is far from rational. If you're looking for further proof, note that some of the founding fathers of MPT and its key early proponents (Harry Markowitz, Eugene Fama, and Paul Samuelson, to name a few) have admitted to the theory’s serious shortcomings.
Some people would argue, “So what? These quantitative techniques give us an answer that’s close enough most of the time, so there’s no real harm in continuing to employ them.” However, these folks probably don’t fully understand where this line of thinking leads us.
When an equity portfolio manager adds another position to his 150-stock portfolio in order to increase his beta exposure, he’s usually not analyzing the fundamentals of the company to detect whether the stock is priced inefficiently. He’s buying it based primarily on one number, and this can create inefficiencies.
When a fixed income manager buys a triple-A rated CDO based primarily on a default correlation number rather than an understanding of the fundamental risks inherent in the underlying credits, we get inefficiencies (which can lead to calamity, as we’ve recently witnessed).
The misapplication of financial theories gives us a precise measure of risk, which imbues us with a false sense of security and allows us to build very large portfolios. Yet when the predictably unpredictable happens, and correlations break down, we realize — too late — the mistakes we’ve made.
With regard to active managers, I argue that, on average, they aren’t really that active at all. The average actively managed fund holds approximately 140 positions and turns over around 100% per year, meaning that the average fund probably holds much more than 140 positions during the course of a year.
Holding so many positions means that these funds tend to behave like very expensive index funds. The human judgment component of management is reduced by the desire on the part of these managers to closely track their benchmark index. So, when you see studies that compare average “active” managers to index funds, you’re not really getting a fair representation of “active” management results.
Stripping away the influence of portfolio theory (which encourages broad diversification) involves isolating and evaluating the relatively small group of equity managers who rely heavily on judgment to build concentrated equity portfolios. Empirical data from multiple studies (see paper) show that, in fact, these concentrated managers persistently outperform indexes.
I generally take empirical work with a grain of salt, as we all know that data can be tortured to advance an agenda. It's worth noting, however, that if we compare the studies that focus on teasing apart the influence of more active, concentrated management, to the broad, all-inclusive active management studies, there’s a large change in the signal received.
It's understandable that investors would abandon active managers for the better returns of index funds. What's less understandable is that we would abandon the idea of active management and the use of judgment to make investment decisions.
The quantitative alternative that promises an efficient frontier doesn't make sense, nor does it work when applied in practice. Human judgment, on the other hand, has never been proven ineffective at uncovering inefficiencies. The real issue is that we need more products that employ truly active, concentrated strategies.
We don't have the space to address the issue adequately right here, but it's interesting to ask, Why do our capital markets continue to embrace quantitative finance so enthusiastically, despite its widely acknowledged shortcomings? And what are the ramifications?
I think the short answer to the first question is that we have a lot vested in our financial system that depends on quantitative finance.