Pension surplus risk at all time lows

The Funded Status Volatility index (PSRX) from NISA is a useful tool for understanding pension risk exposures. It is an aggregate of the risks for pension based on the combination of assets and liabilities associated average allocations for the 100 largest pension funds as measured through public information and 10-k filings. The index, which represents $1 trillion in AUM, will move up and down with the volatility of all asset classes. Hence, falling equity and bond volatility as well as changes in the discount rates will translate into a falling PSRX index. 

Still, the starkness of the volatility decline in the index is telling. There is less risk with the funding status of pensions but this also means there is less upside potential. The risk of seeing a large change in your poor funding status is about 50% lower than what it was 10 years ago. For the underfunded, pension problems will not go away. Those in good financial shape can sleep well at night. Yet, if we move back to the good old days of 10 years ago, there will be a lot of pension risk. 

This low volatility environment has generated a search for holy grail investments - good returns with the ability to provide diversification. If volatility is headed higher, pension will want something that can dampen risk. If volatility stays low, pension will want some absolute returns to help get out of their underfunded status. Find me unique managers who can make money! 

Where are these managers? They have to be iconoclasts who will take risks and not be like everyone else. Unfortunately finding these managers is not easy. To offer diversification and return, these managers may have higher stand alone volatility, larger active shares versus a benchmark or peers, and a willingness to be nimble in the face of changes in risks and economic cycles. A safe manager which has appeal to the masses is not needed. Yet, with fewer hedge fund start-ups, growing consolidation, and a focus on asset gathering by many successful managers, investment risk-taking is less likely. 

Volatility and managed futures - Is there are relationship with the VIX?

Many believe that managed futures is a long volatility strategy because the strategy is like being long a look-back straddle. We believe there is a more nuanced story associated with long gamma exposure, but let's use the prevailing wisdom of long volatility as a starting point for a discussion. 

The question related to the current environment is simple. Should you buy managed futures because volatility is low? Since volatility is low, this could be a good time to buy a "long straddle" strategy. The relationship between low volatility and managed futures can also be viewed through thinking in terms of valuation. Does low volatility means that managed futures funds are cheap? 2017 inflows suggest that even though recent performance has been negative and many large trend-followers are in drawdowns, investors are putting more money into the space based on a expectation of higher volatility. 

So is there evidence on the the connection between volatility and managed futures? We have looked at this issue in earlier posts, Managed futures and VIX - beware high daily volatility - embrace high monthly volatility and Managed futures and volatility - The type of vol matters. Our conclusion is that there is no simple answer and there is such a thing as good and bad volatility. Higher long-term volatility which increases the spread of prices is a good for managed futures, but spikes in short-term volatility is detrimental. At tight or range-bound spreads as measured by vol, there may be limited opportunities. If volatility increases and thus increases the range in prices, trend-following should be more profitable. Nevertheless less, volatility spikes may cause positions to be unnecessarily stopped and thus less profitable. 

There has also been some extensive research in this area which suggests that there is a connection between volatility moves and managed futures, but it is also complex and suggests that the trading time frame of the manager matters. This useful analysis was done by Christian Lundstrom and Jarkko Peltomaki in their paper, "Beyond Trends: The Reconcilability of Short-term CTA Strategies with Risk Shocks" in the winter 2015 Journal of Alternative Investments. At a simple level, the researchers found that the VIX level leads to different performance across managed futures styles. They also found that managed futures performance will be impacted by whether volatility changes were expected or unanticipated. 

Exhibit 3 shows that managed futures returns are sensitive to the level of VIX. Short-term traders will do better when the volatility is high and trend-followers actually do better when volatility is low. There is a return difference in performance based on the timeframe of the styles used. Exhibit 4 shows that short-term traders do better when there is a volatility shock. Short-term traders will exploit short-term changes in volatility. 

A deeper analysis suggests that short-term CTA's will generate positive returns for volatility shocks and the impact is non-linear. The linear response to shocks is not present for trend-following CTA's although they also have positive non-linear response to shocks. The return response to expected volatility is slightly negative.

The response to volatility shocks differs with the level of volatility. There will be more positive returns to shocks when overall volatility is higher. There is either an insignificant or a negative response to volatility shocks when volatility is low.

My reaction to this work is that the behavior of CTA's to expected and unexpected volatility is not completely consistent with descriptions of managed futures strategies and requires more research. The idea that managed futures is a long volatility strategy is not completely validated with respect to the VIX index. Of course, managed futures trade many markets and a single volatility index will not capture the behavior of these other asset sectors. Clearly, the trading time frame has to be matched with the time frame for volatility to get a better representation of style behavior to volatility. Nevertheless, if an investor wants to gain some protection or diversify against equity volatility shocks, he should hold an allocation to short-term CTA managers.

How many managers should you review before you pick one?

Choices, choices, choices. There are just so many managers to choose for a portfolio. Look at the major database, and you will find hundreds of managers of all sizes, styles and skill. Using some simple criteria, the list can be reduced significantly. Minimum size, minimum track record, max drawdown, and max volatility could be just a few ways of reducing the pool of managers, yet there could still be a sizable number of managers. So how many managers do you have to look at before you find the right one? 

This a classic decision science problem given a specific name, "The Marriage Problem" and more recently "The Secretary Problem". The lay-out of the problem is simple. How many candidates for marriage or a job should you review before you make a decision? This is assuming that once you make a decision the game stops and you cannot go back. If you look at all candidates and have not made a decision, you will have to take the last one. You offer or go on with the process. Martin Gardner solved the problem in a 1960 Scientific American article. For an in-depth review on the math see, "Who Solved the Secretary Problem?" and the rejoinder.

A simple algorithm based on optimal stopping times has been developed which states that 36.8% of the total sample of choices should be reviewed before a decision is made. So if you take a large sample and reduce the number based on some simple criteria, you may still be left with 10 candidates. The question is how many should you interview before you make a final decision. The answer is 36.8% (4 out of 10) of the sample or 1/e should be reviewed. Then choose the first candidate that is better than the set reviewed earlier. Of course, if the best manager was in the initial 36.8% of the sample, you will be stuck with a second best solution. Nevertheless, you will end with a good choice if you follow this algorithm. 

Now this may seem far-fetched as an explicit way to choose managers within the hedge fund space, but like many fun math problems there is a kernel of useful information that can help with other more serious problems. So what can be done for hedge fund managers selection and due diligence? 

1. Use a filter mechanism to cut the number of managers to a smaller size based on set minimum standards. A simple filter could be size, length of track, and worst drawdown.

2. Set criteria for what you are looking for in the due diligence.

3. Set the number that can be initially interviewed.

4. Start reviewing managers to get a "feel for the sample" formed.

5. After reviewing the initial sample, pick the first manager that is better than the set reviewed. 

This is not perfect and can be subject to criticism, but it forms a simple algorithm that can start the process and lead a good outcome. I am open to other ideas, so let me know what you think. 

Quant research and managed futures - Key areas of focus

Managed futures research is hard. This is especially the case in the quantitative area. There always are new models being tested by almost all managers, but finding a truly new model or process that adds value is truly difficult. Data mining is an issue.

Managers gain new data everyday but most of the test data has existed for years. This past data are not changing and has been mined extensively. Many of the techniques used to extract signals are well known. The objective functions or goals for most programs and investors are well known. Most managers want to achieve a return to risk above one net of fees. The risks that have to be controlled are also well known. Most managers think that volatility should be set at a target and drawdowns have to be minimized and should be below a well-defined threshold. So it is not easy to find something new that is not subject to data-mine issues and meet these constraints. Consequently, it is important to determine where research could or should be conducted and handicap the chance for something new.

Quantitative research can be separated into four major areas:

1. Signal extraction

2. Portfolio construction

3. Risk management 

4. Model switching 

While we see there are significant areas for further research, each areas of research may produce different levels of benefit. To start a discussion, here are some simple thoughts on where research will be fruitful. 

• Signal extraction - The process of finding improved signals has strong potential using some new techniques like machine learning and AI; however, there are significant barriers to entry based on the knowledge of how to program and developed these models. Unfortunately, we have seen the use of AI models in the past only to be disappointed with the results. New price-based models using older techniques may be limited. There can be some strong gains from the use of fundamental models; however, these models may be subject to issues of coefficient stability.

• Portfolio construction - Employment of new portfolio construction techniques may be limited, but there can be many new ways to form weighting schemes based on the preferences of managers. This can be a fruitful area of research but is closely related to the utility and risk averse of the manager and not the type of model used. 

• Risk management - After the strong movement to VaR and risk-weighting schemes over the last few years, there may be limited room for improvements except for the adjustments based on a deeper focus on differences in the distribution (skew and kurtosis).

• Model switching - This may be a fruitful area of research based on determining model performance or forecasting failure and how to dynamically adjust allocations across models based on factor analysis

All of these areas require a fair amount of theory and data research work in order to generate significant results. In fact, our point is that anyone should be wary of a new algorithm or a new approach. Most of what passes for new may be the preference choice of the manager and not a new technique. There is nothing wrong with making models, risk management, or portfolio construction preference choices which may lead to better returns, but that is not the same as a new method of finding value. Perhaps a key research area is focusing on how decisions are made and systematized.