The Problem with Risk Parity
A critique of RP that is really a critique of a RP Strawman invented by the author. Also makes an empirical case against RP using a backtest, but data since publication proves the opposite point. After reading this, I’m still on the lookout for effective critiques of RP.
Read the original:
J Cooper
Important Points for the RP Investor:
As noted in a previous blog post, I’m very interested in reading critiques of Risk Parity that will challenge my thinking, and help me think more critically of RP as a strategy. So it was with a sense of anticipation of a challenge that I began to read this article on Seeking Alpha by J Cooper (not much of a bio on the author, other than he writes about the cannabis industry). As a long time reader of the site, I’m well aware that it publishes a wide range of contributors and perspectives, each meant to provoke discussion and reader engagement. The provocative title certainly engaged me, so score one for Cooper and SA.
It seemed straight-away that Cooper was arguing with a version of Risk Parity that only partially resembles my understanding of RP. Cooper begins with an assessment of Risk Parity’s popularity at that time (early 2018), noting over $400 Billion invested in RP strategies. On the next page (for ease of reference, I’m referring to pages as it was when I printed it), Cooper then defines RP portfolios as one that are balanced in terms of risk. No argument there, but Cooper does confine this to just two asset classes, stocks and bonds, with no mention of gold, commodities, or other alternatives. True RP for Cooper is 41% stocks (VTI) and 59% bonds (BND).
Seemingly out of nowhere, Cooper then refers to another portfolio, this one at 12% VTI, 60% BND, 11% IAU (gold), 7% VNQ (REITs), and 10% DBC (commodities). That’s more in line with the majority of RP portfolios, keeping in mind that there is no one single RP portfolio, and incidentally calls into question why Cooper would earlier and later refer to a stock/bond only portfolio as Risk Parity. Anyway, Cooper compares that five asset portfolio with a traditional 60/40, noting that the RP portfolio has a lower CAGR (4.3% vs. 7.6%) and lower Sharpe ratio (.65 vs. .8) from 2008 until 2018, but also that it is about 2/3rds as volatile. Referring to the lower Sharpe ratio, Cooper then writes (accurately) that the favored trick of using leverage to bring RP portfolios up to the volatility of traditional portfolios would still have resulted in poorer risk-adjusted returns.
Cooper does note that 2008 to 2018 were bull markets for stocks and bonds, while gold and commodities underperformed, so it really is no surprise that a portfolio with the two good ones and neither of the bad ones would do better than the one with the bad ones included. Cooper also notes that in other backtests, Risk Parity has had higher Sharpe ratios and higher CAGRs, and includes a chart of one such study using the time period 2002-2016.
So far, mostly so good. The back test argument is fine, but the takeaway from there is it all depends on the timeframe for the study. Since Cooper published this in early 2018, I then went to Portfolio Visualizer and repeated the same portfolio tests, this time using data from January 2018 until May 2022. For this time period, Cooper’s five-asset RP portfolio (#1) beat the 60/40 (#2) in terms of Sharpe ratio (.71 vs. .6), on the strength of its much lower volatility (6% compared to 11.5%). Given the higher Sharpe ratio, I then leveraged up the five-asset RP portfolio (#3) so that its volatility would match the 60/40, basically by inserting 88.9% additional leverage in the form of shorted Short-term Treasuries. The result was a clear win for the leveraged RP portfolio in terms of a higher CAGR (8.9% vs. 7.6).
From there, the paper takes a turn to the weird, with four critiques that (to me) just don’t land. His first line of criticism, starting on page five, is that RP ignores asset returns, and is only focused on risk. Cooper writes, “When investing, the primary - only, even - concern for an investor is their returns.” He then seems incredulous that RP investors ignore returns. First of all, I think other things matter to all investors besides returns, and don't really see the "only" part as accurate at all. RP and non-RP investors alike care about returns, risk, and the correlation between assets, just to name a few. Second of all, even just focusing on RP investors, of course returns matter. If I were elected to speak on behalf of all RP investors, I would agree that returns are central, and then you also need to keep an eye on the ups and downs when holding an asset, since if you can’t stomach the volatility, you probably won’t be around to collect on the returns. Additionally, withdrawal rates from a portfolio are super important, and are a function of asset class returns, on the one hand, and the stability of those returns on the other. If returns didn’t matter and all RP investors were concerned about was risk mitigation risk, then wouldn’t all RP investors be 100% in Short-term Treasuries? But they aren’t, right? I’m not sure who Cooper is really arguing with here.
Cooper’s second line of critique is that asset classes do not all offer the same risk-adjusted returns. He then goes into a tangential discussion of how foreign currency fluctuations mean that investors in international equities (or by extension, commodities which are traded on international markets) will always have lower risk-adjusted returns. As someone who invests internationally myself (being an American in Japan), I have found that the foreign currency fluctuations are bi-directional: sometimes they hurt, and sometimes they help. I’m not sure I buy the argument that international investors are destined to underperform, but anyway, I don’t quite see this as an argument against Risk Parity. If it were true, then wouldn’t the savvy RP investor just not invest internationally? Again, not that I speak for all RP investors, but I don’t concern myself too much with the issue of US vs. non-US mostly because they tend to be highly correlated with each other, and there isn’t much diversification benefit.
Third, Cooper writes that RP underperforms when risk-adjusted returns vary because RP allocations completely ignore information about past or future returns. To return to an earlier paragraph, I don’t think RP investors do that, but even assuming they do, Cooper’s example to prove his point is telling, and deserves quoting at length:
Let’s take two assets (so we can ignore correlations) with equal volatilities and construct a portfolio from those. For example, both direct lending and US long corporate bonds have 10% volatilities. Thus, a risk parity portfolio between the two would require no other information, and simply be split 50/50. However, JP Morgan (the publisher of data he relies on earlier) expects direct lending to return 7% per year, while it expects US long corporate bonds to return 3.75% per year. Splitting the two equally is nonsense, if we believe those returns.
Cooper’s right... that would be nonsense, but the argument was DOA with the throwaway line: “so we can ignore correlations”. The point is that no, you can’t ignore correlations here. If direct lending and corporate bonds were perfectly positively correlated, then any RP investor would hold 100% in direct lending. Diversifying here into the lower performing asset would be of no benefit, since you wouldn’t have any risk-balancing between in this case. If they were perfectly negatively correlated, though, then you’d want a 50/50 split and pick up your 6.875% with absolutely no volatility. In the real world, such decisions are not so clear cut, so depending on the actual correlations, the RP investor in this scenario would fall somewhere between 50% and 100% in direct lending. The degree of correlation between the two assets is absolutely critical, and why Cooper thought you could disregard it befuddles me.
His fourth main argument is just plain weird. He essentially writes that asset classes are made-up creations and if we just change the names of things, then we can have Risk Parity, so therefore RP is “trivial.” In his explanation, he invents a new asset class called “bonks” which are 42% stocks and 58% bonds, and then if you have “bonks” in your portfolio then no you can say you have Risk Parity…I’m sorry. I read this section three times, then tried to outline his thoughts, but I really don’t understand it. Basically, he thinks asset classes are just a label, and have no real world distinction, so you can have Risk Parity if you want, just by changing the labels around. The idea of distinct sources of risk, or grouping assets based on how they are similar and dissimilar from other assets, doesn’t seem to have occurred to him.
Cooper closes his argument in the last two pages with a few more head-scratchers. He reiterates the bit about international stocks being a poor investment, but seems to think that this fact damns RP investors who depend on them. Except for the fact that... they don’t. Next, he warns that RP portfolios “will offer worse risk-adjusted performance than other portfolios when the risk-adjusted returns of the underlying asset classes vary,” which is true, I guess. Yes, if the assets in a RP portfolio do badly, then I would expect the RP portfolio to do badly, too. I don’t claim to know, but Cooper writes that knowing what portfolio will do best “does require, of course, that we have some ability to infer future results from either past results or from predictions.” OK, so RP will underperform other asset allocations and we will know for sure if we can predict the future. Mmm, okay, sounds like a plan. I guess he's got a good point - if you can indeed predict the future, you probably don't need Risk Parity (but what if the crystal ball says a RP is best? Mind. Blown.)
Yikes. I guess I’m still looking for a substantive, targeted, and effective critique of Risk Parity. Still, if you’re reading this far, you might want to give the paper a read. If you do read it, feel free to let me know what I’m missing with his “RP is Trivial” point.
Other Resources Related to this Paper:
Here is Cooper’s authors page at Seeking Alpha:
7Author’s Picks
You may also like this article where he goes over his portfolio, as of July 2021. You'll never guess what his own portfolio is similar to! I'll give you a hint - he has around 4% in gold and 12% in real estate in the name of diversification, and is aware of the need to balance out capital accumulation with some assets like cash to help give his portfolio stability when stocks are struggling. It is a growth oriented portfolio, of course, but well in line with the partial adoption of RP principles when one is in the accumulation stage:
J Cooper