Risk Parity Basics
Oct 15, 2012
in Volatility
Risk-parity is a weighting methodology. Given a set of securities in a portfolio, risk-parity overweights lower-than-average volatility securities and underweights higher-than-average volatility securities.
Q. How does it work?
One of the most commonly accepted ways is to start with equal-weight positions and then make adjustments based on the relative volatilities.
Because volatility always embeds a specific time-period assumption, you must specify the time-period (or lookback) you want to use. There is no universal definition on what time-period is correct.
If you choose a shorter lookback period, then your portfolio will adjust more quickly to major changes in volatility. If you choose a longer lookback, then you will have less trading and less whipsaw losses when corrections end quickly and recover.
(for what its worth, we have observed that some index providers use 12-months for the lookback time period and then chooose to rebalance quarterly. But keep in mind that given two different analysts using different assumptions, you will get two different results for the same list of securities within a risk-parity portfolio. As noted, there is no single 'answer').
Q. What is the purpose of all of this?
To us, one of the more interesting problems facing investors are questions having to do with how we weight securities that are in totally different asset classes. Some securities mature at par (bonds) and some are perpetual (stocks) --- some represent paper securities with high yields (preferred stocks) and some can be physical securities that have zero yield (Gold). So long as something is actively traded on an exchange with accurate TOTAL RETURN pricing, risk-parity principles can offer an idea on a weighting methodology that uses the same metric across all of these disparate secruities.
Q. Does risk-parity give an optimal weight?
No. There is no way to calculate an optimal weight for the forward period so risk-parity looks at recent experience and applies that to the future period.
Problem -- if short or intermediate bonds are used, the risk-parity methodology will always very heavily weight bonds (especially lower duration bonds). This is because bonds that have fixed maturity dates will (nearly) always be far less volatile than securities that are perpetual (like stocks). Clearly, low-duration bonds don't have much return potential either so if you were to follow risk-parity, you would end up heavily overweighting bonds relative to something like a 60-40 mix.
Q. Is risk-parity better understood as a concept or as a formula?
While risk-parity sounds quite pleasing, our opinion is that like just about everything in investing, understanding the concept behind it is more important than solving a formula. This goes back to the very basics of investing: given similar return expectations you should choose to more heavily weight the lower-volatility security. This does NOT mean that high-volatility securities are not investable -- it just means that you must have higher return expectations for the high-volatile assets. If your return expectations are indeed higher, then it will make sense to overweight the higher-volatile asset.
Risk-parity will tend to do very well in any period with significant bear markets for an obvious reason, its focus on bonds. Risk-parity will generally (but not always) underperform in up markets for the same reason.
In our view, the point of backtests -- including risk-parity backtesting -- should not be to determine a formulaic 'answer' --- the point is to let backtesting help you digest large amounts of data and be part of the research process that helps you come to a conclusion about what is the right portfolio - in that situation - for that client - at that time.
There are many ways to use volatility to help think about your portfolio exposures and it takes judgment at the end of the day. Given that risk-parity does nothing to adjust for differing return expectations across securities, it should be viewed as simply another tool, not an entire strategy in itself.
-----------------------
Example:
To understand the risk-parity calculation, it is important to realize how a risk-parity portfolio differs from an equal-weight version of a portfolio with the same holdings.
This example will use ETFs from 2 different asset classes: Gold (GLD) and Real Estate Investment Trusts (REITs). Think of it as a way you might want to research how to weight your 'alternatives' allocation. For round numbers, assume alternatives make up 10% of your total portfolio and you want to research how risk-parity handled the past 7 years.
Below would be the weight of GLD using 3-month risk-parity for the period:
Note that in 2006, the risk-parity methodology had Gold as an underweight vs real estate. At the time, real estate was in the tail-end of a major bull market. As 2007 began, the weightings equalized (signalling equal volatility) and then REITs went into a major bear market and Gold came to be about 75% of the mix for about a year. Since the end of 2010, the weightings have been on average near equal. Note also that over that same time, the returns of REITs and Gold have been about the same.
The risk-parity module is embedded within the core-satellite application. Here is a screenshot:
Follow ETFreplay on
