Nearly a year ago, we penned “Return = Cash + Beta + Alpha“: in which we performed “An Inside Look At The World’s Biggest And Most Successful “Beta” Hedge Fund. The fund in question was Bridgewater, and Bridgewater’s performance was immaculate… until the summer when the sudden and dramatic rise in yields as a result of the Bernanke Taper experiment, blew up Bridgewater’s returns for 2013 and at last check, at the end of June, was down 8% for the year. As further explained in “”Yield Speed Limits” And When Will “Risk Parity” Blow Up Again“, an environment in which rates gap suddenly higher (and in the current kneejerk reaction market all moves are purely in the form of gaps as risk reprices from one quantum to another in milliseconds) is the last thing Ray Dalio’s strategy wants. Be that as it may, and successful as Dalio’s fund may have been until now, tonight James Montier of Jeremy Grantham’s GMO takes none other than Bridgewater to task, in a letter in which among other things, he calls risk parity “just old snake oil in new bottles”, and sums up his view about the strategy behind Bridgewater in the following equation:
Risk Parity = Wrong Measure of Risk + Leverage + Price Indifference = Bad Idea
and proceeds to skewer it: ‘At a fundamental level, risk parity is the antithesis of everything that we at GMO hold dear. ” Read on for his full critique.
No Silver Bullets in Investing (just old snake oil in new bottles), by James Montier of GMO
As I have written many times before, leverage is far from costless from an investor’s point of view. Leverage can never turn a bad investment into a good one, but it can turn a good investment into a bad one by transforming the temporary impairment of capital (price volatility) into the permanent impairment of capital by forcing you to sell at just the wrong time. Effectively, the most dangerous feature of leverage is that it introduces path dependency into your portfolio.
Ben Graham used to talk about two different approaches to investing: the way of pricing and the way of timing. “By pricing we mean the endeavour to buy stocks when they are quoted below their fair value and to sell them when they rise above such value… By timing we mean the endeavour to anticipate the action of the stock market…to sell…when the course is downward.”
Of course, when following a long-only approach with a long time horizon you have to worry only about the way of pricing. That is to say, if you buy a cheap asset and it gets cheaper, assuming you have spare capital you can always buy more, and if you don’t have more capital you can simply hold the asset. However, when you start using leverage you have to worry about the way of pricing and the way of timing. You are forced to say something about the path returns will take over time, i.e., can you survive a long/short portfolio that goes against you?
Exhibits 10 and 11 highlight this problem. Exhibit 10 shows the value/growth expected return spread over time. I’ve marked three points with red stars. Let’s imagine you had all of this time series history in the run-up to the late 1990’s tech bubble. Given history and assuming you were patient enough to wait for value to get one standard deviation cheap relative to growth, you would have put this position on in September 1998. If you had been even more patient and waited for a 2 standard deviation event, you would have started the position in January 1999. If you had displayed the patience of Job, and waited for the never before seen 3 standard deviation event, you would have entered the position in November 1999.
Exhibit 11 reveals the drawdowns you would have experienced from each of those points in time. Pretty much any one of these would likely have been career ending. They nicely highlight the need to say something about the “way of timing” when engaged in long/short space.
As usual, Keynes was right when he noted “An investor who proposes to ignore near-term market fluctuations needs greater resources for safety and must not operate on so large a scale, if at all, with borrowed money.”
Risk Parity = Wrong Measure of Risk + Leverage + Price Indifference = Bad Idea
At a fundamental level, risk parity is the antithesis of everything that we at GMO hold dear. We have written about the inherent risks of risk parity before, however I think they can be stated simply as the following:
I. Wrong measure of risk
Many proponents of risk parity use volatility as their measure of risk. As I have argued what seems like countless times over the years, risk is not a number. Putting volatility at the heart of your investment approach seems very odd to me as, for example, it would have had you increasing exposure in 2007 as volatility was low, and decreasing exposure in 2009 as volatility was high – the exact opposite of the valuation-driven approach. As Keynes stated, “It is largely the fluctuations which throw up the bargains and the uncertainty due to fluctuations which prevents other people from taking advantage of them.”
I’ve already discussed leverage in the previous section, enough said I think.
III. Lack of robustness
There are no general results for risk parity. That is to say that adding breadth doesn’t necessarily improve returns. The returns achieved in risk parity backtests are very sensitive to the exact specification of the assets used (i.e., J.P. Morgan Bond Indices vs. Barclays Aggregates). Furthermore, decisions on which assets to include often appear fairly arbitrary (i.e., why include commodities, which, as Ben Inker has argued, may well not have a risk premia associated with them). All in all, the general lack of robustness raises the distinct spectre of data mining, and hence fragility.
IV. Valuation indifference
Proponents of risk parity often say one of the benefits of their approach is to be indifferent to expected returns, as if this was something to be proud of. I’ve heard them argue that “risk parity is what you should do if you know nothing about expected returns.” From our perspective, nothing could be more irresponsible for an investor to say he knows nothing about expected returns. This is akin to meeting a neurosurgeon who confesses he knows nothing about the way the brain works. Actually, I’m wrong. There is something more irresponsible than not paying attention to expected returns, and that is not paying attention to expected returns and using leverage!
As with risk factors (and smart beta), risk parity ultimately comes down to portfolio construction. It is implemented via assets, and can thus be priced. Anna Chetoukhina and I have constructed a model risk parity portfolio using just three simple assets: U.S. equities, U.S. bonds, and U.S. cash. We constructed our risk parity portfolio to have the same volatility as a 60/40 equity/bond portfolio. The relative performance of our risk parity strategy against the 60/40 portfolio is shown in Exhibit 12. As per Anderson et al four distinct periods of performance can be identified. In the early sample (1926-1945) risk parity is an undisputed victor in the performance sweepstakes. However, as they say, payback is a bitch: in the period 1946-1982, the 60/40 took sweet revenge. During the long bull market (in both stocks and bonds) of 1983-2000, the strategies were approximately tied. In the more recent period (2001-2010), risk parity is once more faring better.
Of considerably more interest to me than the performance were the weights held by the risk parity strategy over time (see Exhibit 13). On average the strategy held 44% in stocks, 155% in bonds, and was short 99% cash. The weights, of course, varied considerably over time.
Using these exposures we can apply our expected returns to see what risk parity is priced to deliver (an anathema to the disciples of this strange art, no doubt). The results of this assessment can be found in Exhibit 14. Over long periods of time risk parity doesn’t look very different in return space from a 60/40 portfolio. Currently both the 60/40 portfolio and risk parity display the problems we have previously referred to as the purgatory of low returns.
This is certainly a problem for some of its proponents whom I have heard argue that 60/40 is priced currently to deliver a low return, and thus one should follow a risk parity approach!
Strangely enough, when risk parity is priced to do well relative to the 60/40, it does indeed do well. If you lever up cheap bonds you can get good outcomes (assuming you know something about the path those returns will take, of course). Similarly, if risk parity is priced to do worse than the 60/40, it tends to do so. If you lever up expensive bonds things don’t tend to turn out too well. There is no magic to risk parity (see Exhibit 15).