Risk Basics: It’s Simple Math
Coverage of the 2010 Risk Management Conference.
BY Robert A. Garvy and Jason Stefanelli | October 28, 2010
Risk matters. It can definitely ruin your day and affect the long-term returns of plan sponsors and the investment managers who get hired to achieve them. Peter Bernstein once said that the revolutionary idea that defines the boundary between modern times and the past is the mastery of risk: the notion that the future is more than just what has been determined by some oracle or soothsayer, but instead, showing how to understand risk, measure it and weigh its consequences allows us to achieve the advancements and benefits we see and have experienced in today’s world. The study of risk lends itself readily to quantification; this language is the language of numbers such as information ratios and t- statistics, which can be very powerful tools for investors to help them gauge risk.
Many plan sponsors are still trying to recover from the recent financial crisis. We are hearing a lot about Black Swan events and how they seem to be appearing a lot more frequently than expected. Accordingly, tight risk controls are important within investment strategies as portfolios become more vulnerable to these events and are more susceptible to losses. Trying to avoid these losses is often overlooked by investors – particularly relative losses. In the equity arena, managers are often compared to indices, to benchmarks, passive portfolios or alternatives. The relative losses to those indices have a significant impact on long-term performance. In our view, relative downside protection is equally (if not more) important than upside potential. Even though it sounds simplistic, a portfolio declining 50% requires going up 100% just to get back to even. In today’s environment, investors need alpha more than ever. The S&P 500 Index is relatively flat over the last 10 years, funding ratios of pension funds are down, and retirement savings plan balances are under stress. Needless to say, investors are under stress and in the current low nominal return environment, alpha becomes harder to find and more important than ever.
So what kind of tools can help plan sponsors in this current environment? First and foremost must be an analysis of the investment process. Subsets of this include information ratios, t-statistics, and the normal due diligence process and qualitative factors on manager selection. The key to long-term success in our view is compounding in positive space. For example, a strategy that can compound just 1% to 2% excess return annually will have a meaningful impact over the long term. Investors should look for an “a priori” expectation or reason for thinking that an investment process should work and provide a future positive average relative return. “A priori” means “known without reference to experience.” An “a priori” expectation is a powerful starting point and can be achieved through mathematics. Mathematics involves absolutes.
Secondly, investors should look for effective risk management. Risk reduces the long-term return of a portfolio and can come from a number of sources. One key risk investors should consider is model risk, which looks at correlations and historical parameters and relationships from the past. These relationships may have little or no bearing on what happens in the future. All models are subjective to these risks and need to be adjusted with practical oversight. For example, large active equity bets in a portfolio should be taken with a large diversification of securities. Don’t be fooled by faulty assumptions either. Statistics that look at past behaviour and use this data over past periods to predict the future are a dangerous game if only looked at in isolation.
Lastly, efficient implementation is a key tool to help make plan sponsors compound in positive space. Even if you have a manager producing positive alpha, if dissipated away due to transaction costs or implementation shortfall, this can become a big problem longer term. Investors should look for a statistically significant high information ratio with low tracking error to reduce the deviations around the alpha source. Large tracking error can affect expected return. To summarize this point, a good “a priori” reason for thinking an investment process should work, combined with low one-tail significance, should provide a high degree of confidence to investors. The longer an investment process has been in place producing a positive alpha, the higher the t-statistic. A high t-statistic means the greater the probability that the investment process has resulted from something other than luck – likely investment skill. Plan sponsors should note that a manager with a high information ratio, long performance track record and statistically significant t-statistic can still disappoint you if you fail to consider a few caveats.
First of all, investment returns are not typically normally distributed. Returns are often normally distributed over very long periods of time, but in short time horizons are not. Secondly, it still could be luck! For example, even a one in ten thousand chance (or 0.01% one-tail significance) that a manager’s track record is a result of luck is still possibly a lucky outcome (although less likely over long periods of time). However, these types of statistics combined with an “a priori” reason to believe a process should work are what will give the plan sponsor more confidence about what they are trying to achieve with their investment objectives.
An unfortunate outcome of today’s current environment is the unrealistic expectations that managers, consultants and plan sponsors have placed on assessing whether or not an investment manager that is underperforming has lost their touch. More specifically, what is the probability that a series of negative returns is normal for an investment process? Periodically, even high information ratio strategies that are working normally can experience negative relative returns of extended time periods.
Investors should use tools available to them to gauge risk. Information ratios and t-statistics can be helpful but only when combined with an “a priori” expectation. Investors should be realistic in their expectations and consider making sure that investment policies reflect appropriate freedom to allow for mathematical truths. By defining a rational process of risk-taking, plan sponsors can better understand the nature of risk and the art of science and decision making for the future while adapting to our modern market environment. In this spirit, risk should be a choice rather than a fate.
Robert A. Garvy is chairman & CEO of INTECH; Jason Stefanelli is managing director, Janus Capital Group Institutional