Leave a comment

Rethinking Volatility: How to Measure Risk in Portfolios

On any given day, our Portfolio Construction Services (PCS) team holds several adviser consultations in which we discuss potential risk and opportunities within their portfolios, based on their clients’ tolerances and objectives. In many of these calls, our discussions tackle similar topics and concerns, one of the most prevalent of which is volatility. In this post, we’ll give you an overview of volatility and discuss why rethinking it—and the way it’s measured—could lead to well-balanced portfolios.

Standard Deviation vs. Downside Risk: The Good, the Bad and the Volatility

Naturally, volatility is of great concern for both financial advisers and end clients. But generally the framework with which we view volatility is negative. It loses us money. And while that can be true, it’s only half the story. The major metric for overall volatility for many is standard deviation. While we believe standard deviation is a good place to start regarding volatility, a deeper dive into the metric is important to understanding a deficiency that otherwise may go unnoticed.

Let’s get nerdy for a quick minute. Data can be distributed, or spread out in different ways. In many cases, data tends to be around a middle value (the mean) with no bias left or right and, if plotted on a graph, takes the shape of a bell (often referred to as a “bell curve”). This is called a normal distribution. Put another way, in a normal distribution, approximately 50 percent of the values fall below the mean, and 50 percent fall above it.

Standard deviation is a measurement of variability that shows how much dispersion there is from that middle value. This is how we measure volatility. On a normal distribution, the calculation for standard deviation takes into account observations both to the right of the mean (positive numbers), and observations to the left of the mean (negative numbers). Those positive observations to the right of the mean represent good volatility. Capturing upside volatility is what generates wealth over time. In fact, it is the reason we invest in the first place. Without it, we’d be better off putting our money under the mattress.

In our PCS reports, which we run using adviser models and walk through during our consultations, we isolate and remove this “good volatility” by including downside risk as a metric. Downside risk eliminates all those positive observations to the right of the mean and focuses solely on those on the left side of the normal distribution—what could be referred to as “bad volatility.” Capturing more of these observations is what erodes wealth over time. Something we are looking to minimize or avoid regardless of our time horizon or risk tolerance. How does the relationship between standard deviation and downside risk manifest itself in models?

In theory, the goal is to have your overall volatility—as measured by standard deviation—as close to your particular benchmark as possible so you capture as much of the upside as possible while keeping your downside risk as far away from the benchmark as possible. This would “skew” your model’s normal distribution so the “bell” is farther to the right than that of the benchmark, which would mean you captured more of the positive volatility and less of the negative volatility in the market. Makes sense right? Sure, but just like all theories, in practical application, it is easier said than done.

On the PCS team, we understand this challenge so we focus on the percentage change, particularly the reduction, of these two measures relative to one another.

For example, a “moderate portfolio” benchmarked against the Morningstar Moderate Target Risk portfolio may have a standard deviation of 7.81 versus the benchmark at 7.91. The model has 1.2 percent less overall volatility than the benchmark. But if one looks at the downside risk, where the model has a 3.96 versus the benchmark with 4.50, the model has 12 percent less “bad volatility” than the benchmark. Clearly, this portfolio’s volatility is in-line with the benchmark, but its bad volatility is quite a bit less, proportionally. Therefore, we would expect a portfolio like this to perform better than the benchmark over the long term.

Rethinking how we view volatility—especially via standard deviation—while acknowledging its relationship to downside risk is an important component of constructing a well‐balanced portfolio.

By the Janus Henderson Portfolio Construction Services Team

 


1 Comment

Do We Rely Too Much on Risk Tolerance Questionnaires?

When you meet a new client you want to know three things: their return objective, their time horizon and their tolerance for risk. You may want to know other things too, but these are the big three.

Calculating a return objective is pretty straightforward. The same is true about time horizon. You gather some facts, make some assumptions and ask the client to think about some things they probably haven’t thought much about. But, in the end, the calculations are pretty simple.

What about risk tolerance? How do you measure it? Can you calculate it at all?

It doesn’t take long to see that there are really two very different varieties of risk tolerance. One is objective and the other is subjective.

The objective variety is easy to understand. If I arrive at my retirement age with more money than I need to live comfortably, my goal should be to take as little risk as possible while maintaining the purchasing power of my assets. Objectively, I have no need to take risk.

On the other hand, if I arrive at my retirement age with significantly less money than I need, objectively I should have a higher risk tolerance. I need to take risk to reach my goal.

This objective type of risk tolerance determination has nothing to do with my internal feelings, attitudes or beliefs about risk. It is all driven by my goals and my time horizon.

Now let’s look at subjective risk tolerance. This is what risk tolerance questionnaires purport to measure by asking questions about our predisposition toward, and comfort level with, risk.

Some questionnaires synthesize our answers into a risk score. Some even use our risk scores to determine our investment strategy. Our comfort with risk drives which portfolio we receive.

These questionnaires assume that our risk tolerance is like an organ in our body that can be “seen” through an X-ray or MRI. The questionnaire is the tool that discerns risk tolerance.

This view is not consistent with the research on risk tolerance. A person’s tolerance for risk is hard to capture and harder to quantify precisely. You may get a general sense of a person’s comfort with, and capacity for, risk-taking, but assigning a meaningful risk score is impossible.

In addition, risk tolerance is fluid not fixed—it changes over time. Today’s bold adventurer is tomorrow’s timid soul. It is also situational. A skydiver may be conservative with her money.

The research also suggests that we are notoriously bad at assessing our own risk tolerance. Getting a sense of our risk tolerance by asking us produces unreliable results.

If we are not good at assessing our own tolerance for risk and that tolerance changes over time, what good is a questionnaire that produces a precise risk score at one moment in time?

A more difficult question is, “why is risk tolerance even relevant?” If a client needs to be aggressive in order to reach his goals, should you put him in a conservative portfolio just because he has a low risk score? The answer seems obvious. Risk, alone, should not drive strategy. Sometimes being a little uncomfortable is a necessary side effect of investing.

A discovery questionnaire can help an adviser solicit important information from a client and stimulate discussion to develop insights about the client’s goals, experiences and attitudes. But substituting a mechanical scoring system, backed by questionable science, for the judgment of an experienced adviser is a step backwards for our industry and ill serves our clients.

scott-mackillop

 

Scott MacKillop
CEO
First Ascent Asset Management
Denver, CO