This chapter is from the book
This chapter is from the book
This chapter is from the book
Moving Forward
It would seem that knowledge of the limitations and the empirical facts of the last decade would have forced change by now. But it hasn’t. An industry survey published in 2011 says that despite the renewed focus on risk management, a wide gap still exists between mean-variance and quantitative strategies.
- Investment managers at financial institutions know, in principle, that basic mean-variance portfolio theory has it limits, but our findings clearly show that, in practice, mean-variance analysis is still the industry workhorse. Possibly to blame for this state of affairs is an absence of consensus on the most appropriate model.9
If we cannot rely on current practice and there is no consensus on how to move forward, what is the next step? How do you frame the possibilities? In the end, maybe it is a matter of taking a step back and asking the fundamental questions. The most basic question is: as investors what do we want and what tradeoffs are we willing to make? One of the answers that I think frames the issue as well as any I have seen is from the Ennis article mentioned above.
- Investors want three things. They want some downside protection. They want to capture the equity risk premium to the maximum extent consistent with their preference for downside protection. And most would also like to garner excess return (alpha), although we know that, by definition, only about half do so over any particular span of time.10
I think he is exactly right. Downside protection will always be in demand. Equity risk premiums have historically been 2% to 3% over bond returns. Over long periods of time, this risk premium has been responsible for incredible wealth creation. And with research and other techniques, investors will always look for investments that will outperform market averages. Of course, different investors will put more or less weight on each objective. For example, institutional strategists may play more heavily in risk premiums. Aggressive traders will emphasize alpha and quantitative risk control, but the basic elements are there to describe a wide range of investor goals.
Taken together, the three objectives seem very reasonable. But in practice, it is hard to get them—at least, with any sizeable exposure to equities (and bonds too at this point).
Why is this? For one, there is a natural tradeoff between the goals of providing downside protection and capturing risk premiums. When I first started looking at this issue, I didn’t understand why it is so difficult to add a risk budget or drawdown limit to a diversification framework. At some point, the incompatibility began to dawn on me.
If you try to impose a drawdown limit, it interferes with equilibrium. If you rely on equilibrium, it is never obvious how much downside there is. A gap seems to exist between modern portfolio theory and related mean-variance portfolios—which are great at capturing risk premiums over the long term but lack a risk discipline—and quantitative strategies that have great risk disciplines but are not so good at capturing risk premiums.
The question is whether it is possible to bridge the gap and at what cost? And if you try to find a middle ground between premium capture and risk control, how do you do it?
Imagine you are a trustee of an endowment, and the fund is down 10% for the year. You were hoping for a return of 8%, so now you’re off almost 20% from where you expected to be. You may have to start looking at spending cuts. You know that if the fund drops another 10%, it will threaten core functions. If the fund drops another 20%, it is difficult to think about what will happen. What do you do? Do you sell assets now to slow the rate of decline? Or do you hold on and hope for a rebound?
Institutions normally have a policy statement to guide trustees through this decision. The policy statement is a strategic plan written in anticipation of market ups and downs. Most encourage riding out the rough times. As part of maintaining the strategic allocations between asset classes, most recommend adding to underperforming assets during a downturn. The plan realizes that rebalancing involves doing the opposite of what most people will feel like doing. For instance, if the equity market is declining, instead of selling equities into market weakness, the plan tells you to maintain the proportion of equities to fixed income. That means buying more equities. However, buying more equities actually accelerates the losses if the market continues to go down.
According to equilibrium models, this makes sense because it is the best way to capture risk premiums. When the market recovers, or restores equilibrium between asset class valuations, you make more by having bought the cheaper asset. But it is not the best way to provide downside protection.
Objective 1. Some Downside Protection
The Harvard experience during the financial crisis is particularly important, as described in this press release:
- Harvard Endowment Hires New Chief Investment Officer, January 14, 2010
- Boston – Harvard University named a new CIO after the school’s endowment dropped $26 billion last year. Long admired for its investment savvy, Harvard was forced into heavy cost cuts and interrupted its high-profile campus expansion.
I think what happened at Harvard happened to a lot of institutions and people. At some point, losses get too heavy and there is nothing you can do other than start hoping for a turnaround. Once you are down 20%, it seems too late to start managing risk. Instead, you start reminding yourself of deeply held beliefs such as “don’t time the market,” “buy on the dips,” and “think long-term.”
Harvard has been at the center of academic theory and practical implementation. It has taken modern portfolio theory to its limits, and most of the time it has paid off. However, sometimes the only way to avoid a 30% loss is to start doing something about it when you are only down 5% or 10%.
Objective 2. Capture Risk Premiums in Line with Risk Tolerance
In trying to explain why many portfolios lost more than the worst-case outcomes predicted by asset allocation models, one researcher looked at how much risk is really in a mean-variance portfolio. He modeled portfolios under stress using a typical correlation matrix. Then he compared the predicted performance to the actual performance of these portfolios in market crashes. The two weren’t even close. So he tried it again, this time using a correlation matrix built from information about asset behavior during prior market crashes. This time, the results matched almost perfectly. The problem was the way the correlation matrix was estimated, using average rather than stress relationships.
Why doesn’t everybody use a correctly constructed matrix? Because it can mean cutting equity allocations by as much as 75%, and few funds are willing to do this. Especially now. Giving up the opportunity for equity risk premiums at a time when bonds are so highly priced might be more risky than doing nothing. If equity allocations are reduced, current low yields on fixed income will not support the promises of pension plans and other institutional sponsors that have assumed annual returns of 7% to 9% or the retirement income needs of many individuals.
Objective 3. Some Alpha Opportunities
Going after alpha opportunities is almost irresistible. The history of Wall Street is the history of story telling—whether it is an undervalued stock, a reversal in a trend, or a chart pattern—and nothing has changed. I love a good story too. It is part of being an investor and an optimist.
There are two interesting issues related to alpha. One, the Efficient Market Hypothesis (EMH), has been debated for decades. The other, idiosyncratic risk, seems to be fairly well accepted. EMH addresses the effectiveness of active management such as stock picking, compared to broad asset class exposure. In other words alpha versus beta. Probably more research has been done and material written on this topic than any other in investing. Tests of the EMH going back over 30 years have consistently shown that beating the market with either technical or fundamental analysis is tough. And if current hiring trends are any indication, then EMH is winning. Stock pickers are out; asset allocators are in. As Ennis says, it only works about half the time for most of us.
Idiosyncratic risk is non-diversified risk. The issue is whether or not you can expect to be compensated for taking this kind of risk. It is generally thought that the market only provides an extra return for taking an extra risk if that particular risk cannot be diversified away. If you want a credit risk premium, the market should reward you if you buy a diversified portfolio of bonds. However, it is not obligated to reward you if you buy one bond that turns out to be bad, such as Enron, Worldcom, or Greece. If you want an equity risk premium, the market should reward you if you have broad exposure to equities, not if you buy an individual company stock. In other words, theoretically compensated risk is diversified risk or beta risk, not alpha. Traders and quantitative investors understand this and therefore don’t rely on equilibrium or mean-reversion to protect them from losses. Because the nature of the risk is different, it makes sense to manage it differently as well.
The first step in moving forward for any investor is to find the right balance between seeking downside protection, capturing risk premiums and finding alpha opportunities. After finding a balance, strategy implementation is really an engineering problem. That is, the decisions about the kinds of investments most likely to meet the objectives and the trading rules to manage them. And to realize that in practice, the objectives often compete with each other.
For instance, if you want downside protection, you could interfere with the capture of risk premiums. If you want alpha, you shouldn’t expect to capture risk premiums or find any protection from equilibrium. If you want to capture risk premiums, how much downside protection can you really expect?
In terms of existing portfolio construction, I am not suggesting diversification models don’t add value—just to recognize what they can and cannot do. The most important decision is when and how to begin managing losses or mitigating volatility. If you don’t want to accept the possibility of large losses, then the strategy needs to manage risk actively so that losses are addressed earlier rather than later.
There are two ways of doing this. The first is to stay within the MPT/MVO framework by adding risk management features other than diversification (such as hedging and insurance) and to find securities that add real diversification when you need it most—during market crashes. The second is to go outside the diversification framework to add more dynamic quantitative elements.