Universa Investments LP published a paper to challenge the rationale behind the effectiveness of a leveraged diversified portfolio, in particular the infamous 60/40.
Universa Investments has always stressed the importance of mitigating the largest drawdownsas the most effective way to improve compound returns over time. However, risk mitigationneeds to be cost-effective. We have long argued that relying solely on diversifiers like bondswith the primary objective of protecting against equity drawdowns is not efficient. The abilityof bonds to hedge equity risk is statistical, rather than mechanical in nature. The hoped-fornegative conditional correlation between equities and bonds when equities experience steepdeclines has not been consistent over time. Notably, bonds provided no hedging benefit at allin 2022.
As we saw in this previous post, this is not 100% true: Treasuries provided a good, maybe not excellent, tail risk hedge. 2022 was a different story but…not even traditional tail hedging strategies worked in that environment. Indeed, Treasuries move linearly while options offer convex protection: we need to allocate less to options to achieve the same level of hedge. But this is not a free lunch, cause Treasuries are EV>0 while plain vanilla put strategies are EV<0.
“Interestingly, one of Asness’s objections to the pitch for 100% equity by Anarkulova et al. is what he considers as the flawed reasoning of looking at historical results and telling investors to “buy the thing that has gone up the most over the long term.” It seems ironic that Asness is relying on the same historical data, now including bond returns, to support his own assertion.”
I understand Cliff’s point (in a century, will we advocate for 100% Bitcoin portfolio?) but also what happens if you take it to the extreme like it is done here. The stock-bond combination is (should be?) an acknowledgement of the benefits of diversification, similar to the takeaway expressed here when facing an uncertain future.
Universa updated the AQR study, starting from January 1926 until February 2024, arriving at the following results (portfolios are rebalanced monthly, gross of transaction costs):
Based on this, Universa concludes that the argument for a leveraged 60/40 is not that strong; the difference in returns is too small to be statistically significant.
The Cost of Leverage at 1m T-Bill + 0.5%
Let’s first analyse this point, because high costs directly impact the strategy performance. The cost of funding is closer to 3m LIBOR (R.I.P.) and 3m T-Bills rates so more like 1m + 10/15bps than 50bps. If you do not believe the theory, here is a real-world example:
NTSX includes ALL the costs: fees to rebalance, funding and a higher management fee compared to both SPY and IEF. If you are a pension fund and run the strategy by yourself, you can even save part of the management fee. For retail investors, we can say that NTSX equals a 1m T-Bill + 30/35bps all-in cost (financing, management, trading).
[Asness original study assumed a cost of leverage equal to 1m T-Bill and the 55%-levered portfolio had 0.8% overperformance. The period under study ended in 1993.]
Here is a drill-down of the unattractive aspects of the strategy, according to Universa, point by point:
1) The performance of the leveraged 60/40 portfolio has been inconsistent, underperforming 100% equities over extended periods in the past.
Over the entire 30-year period from 1950 through 1979, the leveraged 60/40 portfolio underperformed the all-equity portfolio by 90 bps per annum.
A 30-year underperformance would break even the most convinced believer in a strategy. But let’s not forget that Universa employs an over-penalising cost of leverage. Shifting the line up by 20bps in the above graph would shorten the lemon period.
What makes this argument tenuous is the opposite point of view: the 100%-stock portfolio underperformed for 61 years in total (vs 37) and for a 30-year-long stretch as well (again, using the conservative Universa’s data). I understand that the scared money manager would prefer the known (100% stocks) for the unknown (something something leverage) but this is more a career risk than anything else (better to fail conventionally than succeed unconventionally).
What would you choose for your investments? The tracking error issue is a known beast for the readers of this blog. Remember, without tracking error, there would not be over performance (yes, there’s no guarantee for it).
2) There has been little or no benefit in major equity drawdowns
The investor must leverage more than 60% of their capital to match the volatility of an all-equity portfolio and, potentially, achieve only a slightly improved return over an extended period. It seems absurd that an investor would undertake such an endeavor while being exposed to the same significant drawdown risks.
Universa compares the drawdowns for the strategies during the three worst periods for equities of the past century: the 1929 Stock Market Crash/Great Depression, the 1970s Oil Crisis and the 00s Lost Decade (Dotcom Crash and Great Financial Crisis). In all these occasions, the lows of the two portfolios are almost identical.
Here is Cliff’s take on the same topic in his original paper:
I do not know you but I usually expect to earn a higher return only if I take a higher risk. The fact that the leveraged portfolio only matches the 100%-stock in its shittiest moment while providing higher returns overall is a huge victory…for me? Sure, the leveraged portfolio might as well come to your place and iron your shirts but we are not that picky, are we?
It is Eric Crittenden’s “wine tasting test” all over again: imagine if I present you the two strategies without telling you how they invest, would you pick the one with equal max-drawdowns but lower returns?
It seems absurd that an investor would maintain such a cumbersome yet simplistic strategy and endure all of the drawdown risk of 100% equities in the hope of eking out a slightly better compound return over time – running the risk of long periods of underperformance.
I guess at Universa they do not know the existence of the “stock picking” concept? 😉
3) Virtually all of the long-term outperformance of the leveraged 60/40 portfolio came during the 40-year secular decline in interest rates.
Ever heard about log charts? 😉
The 60/40 with 62.5% leverage is in essence just a long position of close to 100% in equities plus a huge leveraged bet on the spread between long duration bonds and short duration bills.
Yes, plus the correlation between stocks and the interest rate curve being in contango/backwardation? Sort of non-diversified $RSSY? 😉
I had the same issue in the past, as I wrote in one of the Model Portfolio’s updates: where to ‘attach’ the leverage component. At the end of the day, any investment is a long position vs cash…no? We earn the Equity Risk Premium because that premium is not constantly above the cash return. Japanese stocks can be a relatable example: the excesses of the 80s brought the Nikkei to ‘front-load’ all the returns of the next 30 years. Shall we conclude that the ERP does not exist in Japan?
High inflation in the 70s is responsible for the strategy’s relative losses in that decade as much as for the gains later; ditto for the extra gains due to ZIRP that led to the losses in 2022 and 2023. You cannot expect a uniform distribution of overperformance. Pull and push.
The focal point here is that there is a valid reason why the Interest Rate curve should have a positive slope. Should I expect it to be always so? Definitely not. Should I take advantage of it? Probably yes. That said, we should not divert our focus: if we like the 60/40 and want to achieve higher returns, the way is a levered 60/40. Easy Peasy.
4) Leverage involves risks like margin calls that the academic backtests miss.
The collapse of LTCM in 1998 serves as a powerful reminder of the dangers of leverage, even when applied to seemingly low-risk instruments like bonds, albeit at a much larger scale in that case. Recall that a leveraged arbitrage strategy involving on-the-run and off-the-run Treasury bonds was a major contributor to LTCMs losses.
LOL, LTCM was levered 30:1, here we talk 1.6:1
Another rather important difference is that $NTSX and other similar products get leverage by using futures. And futures on the most liquid asset in the world, US Treasuries. True, when there is a high level of stress in the market, arbitrage actions (the one that keeps futures’ embedded cost of leverage in check) are harder to perform. But contrary to LTCM, here we are long the flight-to-quality asset. If the value of Treasuries goes up during a crisis (as it should), I do not have to post more margin, I have to post LESS.
If Treasuries would stop providing Crisis-Alpha then yes, leverage in this case might be an issue. Wake me up when it happens though. [if you are one of those smart-asses thinking “and if the US loses the role of The World King? What about then?” let me tell you this: it is not a change that will arrive overnight. Everything would shift towards the new risk-free asset, this strategy included].
We must also anticipate scenarios where the usual market dynamics deviate during a crisis. For example, if equities were to drop by 30% or more, it’s conceivable that the bonds may not increase as expected. We had a glimpse of such an anomaly in 2022.
WUT? 2022 wasn’t an anomaly. There is a relationship between what the stock market is doing and what the FED does. For leverage to be a problem, we need a material downturn and we need it to happen fast. 2022 wasn’t either.
Universa drew a cute table that never happened in the real world. It is again about Crisis-Alpha and what I described at the end of this post. If in 2022 the market would start to dive meaningfully, I am pretty sure (but this is just my opinion, who knows), the FED would have at least stopped raising rates. S&P500 at -30% -> recession -> mission accomplished (even if inflation would have stayed high). The levered 60/40 would have had an awful performance, worse than 100%-stocks but I am not arguing that here; I am simply stating that even in that scenario this use of leverage would not have been an issue.
This paper is marketing material for Universa. I understand their push towards tail-hedging strategies, I use them both in the Model Portfolio and my actual portfolio. But I find the arguments….a stretch at best, considering their main point is that institution investors would have a hard time implementing a strategy that today, thank you $NTSX, is available to retail at a 20bps all-in cost.
What I am reading now:
Follow me on Twitter @nprotasoni