In our latest blog, RoZetta Institute PhD Candidate – Luca Parlamento – provides a high-level overview of yield enhancement strategies focusing on the pre-packaged spread of capped-uncapped variance swaps that has attracted recent media attention.

 

Luca Parlamento (pictured) is a PhD candidate at Macquarie University in Sydney undertaking RoZetta Institute’s unique Industrial PhD Program

In terms of financial innovations, the late 2010s will be remembered as the years of “yield enhancement” (also referred of carry, risk recycling, and risk-sharing) strategies.

In bull markets, many buy-side [1] players tend to increase risk exposure in the effort to outperform competitors and benchmarks. Few basis point differentials in performance seem essential to be able to market the fund to existing and prospective clients. [2]

Behavioural biases, short-term incentive structures, and industry turnover tend to conceal the memories of the last crisis, making fund managers and allocators less prone to “waste” money on insurance-like strategies. On the contrary, they have incentives to add risks to generate “extra” returns.

Given the last years’ performance of US Equity Markets, it does not seem a stretch that some longstanding bulls fell into complacency; they may have added some of these strategies as the cherry on top of their risk-on exposure.

To also influence the supply and demands of these strategies, it is also the overall demand by hedging strategies [3]. Managers that have been defensive in the first stage of a bull market, perhaps counter intuitively, tend to have even low willingness to further under-perform in a continuation of the bull market by hedging.

According to academic theory, hedging must have a negative risk premium, as it reduces the non-diversifiable risks of a portfolio. In “The VIX Premium” (2018) Ing-Haw is actually “puzzled” for theories of why investors hedge volatility.

In the zero-sum games of derivatives markets, a negative premium for a market participant needs to be a positive premium for another. Financial media often quote the gross size of the derivatives market as a clickbait keyword, which to a large extent, is a meaningless figure.

A more critical question is who ultimately hold the risk and if this entity can be forced to react given a change in mark-to-market. A huge notional cash-secured put book (which involves writing a put option and simultaneously setting aside enough cash to buy the stock) does not require any forced adjustment. In comparison, other lower notional products can lead to painful unwinds if sitting in portfolios where their leverage effect is not well understood/modelled.

Leverage, which is not necessarily a nasty feature, makes an investment vulnerable to risk and margin constraints. That is, there is an additional risk for the investor to consider; she can be forced to post extra cash to maintain the position and to avoid liquidation at a loss. There are almost daily stories of naive retails accounts hit by margins calls [4]; however, most of them do not have any edge or any risk management strategies, making their risk of ruin [5] highly likely since the start trading. Retails traders also tend to panic at the worst possible time.

Institutional portfolio managers, where portfolios are overseen by independent risk teams, don’t usually liquidate positions because they are in a state of panic. They more likely need to reduce exposure because they must meet predefined risk metrics (e.g., Value-at-Risk or more merely a fixed drawdown levels [6]) or outflow. Only very few buy-side investors, like Warren Buffet [7]  have embedded in the business model permanent capital to be able to absorb significant losses without being forced to de-risk their portfolios. As Levine (2020) points out, “A lot of investors are less patient, disciplined and value-oriented than Buffett, but a lot of other investors are just as patient and disciplined and value-oriented but have to answer to impatient limited partners.” [8]

That is, the necessity to “outperform” competitors in a low rate environment in combination with the excellent back-tested performance [9] made “yield-enhancement” strategies hard to pass for some investment committees. It’s also likely that some less sophisticated investors failed to see that ALL yield enhancement products are a form of short volatility.

The usual “entry-level” analogy to explain short volatility strategies is to compare it to a storm damage insurance. Returns’ profile from the side of the insurance company or the option seller is similar. Both exhibit steady small gains followed by rare but higher in magnitude losses; besides, both endeavours expect a positive mean return, given they should require a premium to be able to bear violent, but rare shortfalls.

Nonetheless, the analogy breaks given in nature of storms are exogenous events (the existence of insurance policy does not make the storm any more or less likely). In contrast, the price of insurance of financial instruments can become an endogenous event. That is, the cost of the derivative can influence the outcome of the underlying, especially as the derivative can be more significant for pricing than the underlying market (This phenomenon is well documented with CDOs during the great financial crisis (GFC), or in the Volmageddon event of February 2018).

The framework – of “stability lead to instability” introduced by Minsky (1985) – seems an acceptable approximation in a context where multiple investors, especially pension funds, have fixed target returns. For instance, for the broad class of option selling strategies, a lower price for a class of options would sell a higher number of options to generate the same dollar amount. Thus, there are clear implications to supply demand of options, volatility, and perceived risk.[11]

In the mid-2000, this “insurance-like” function was predominantly offered by Investment Banks’ balance sheets. After the Financial Crisis of 2007/2008 regulations (e.g., Dodd-Frank, CCAR) and banks, internal governance (E.g., Capital Consideration, Balance Sheet Costs) lead to the adjustment of banks’ risk appetite to continue this function [12].

Banks have established dedicated, sales and traders to be able to intermediate these insurance premiums in new wrappers. As a new asset class, these yield enhancement products are not malicious. They can potentially offer an excellent orthogonal stream of return to well-constructed portfolios if they are traded at the “right” price (disaster insurance has been overpriced between 2009 and 2012 on the wake of GFC [13]).

What made them explosive in the covid-19 led crisis is that they were traded by almost price inelastic yield-starved mandates. As Orr (2020) [14] reported, “Many traders believed that they were effectively getting free money. Banks were paying to get risk off their books in order to pass stress tests regulators imposed after the last financial crisis. But the thinking went – the insurance would never actually payout – because such a dramatic crash had never happened.”

Pre-March 2020, when market volatility was low, these premiums appeared to be slightly but consistently overpriced. Given the speed of the crash in response to Covid-19 has never been experienced before it was tempting to conclude that the small, but very steady returns, would have continued.

The not well-known “trick” of short volatility strategies is that if some parameters are set accurately, they can perform magnificently in back-tests. In the age of “AI” these products can be calibrated with few lines of code never to lose money under any historical market shock. It was possible to come up with strategies that tested well through 1987, 2001 dot.com, and 2008/2009 crisis, giving the underwriter a false sense of security. It is now possible to overfit the March crash given the market rebounded in early May 2020.

One example for all: Uncapped variance vs. Capped

There is one trade that recently received media attention [15] : the pre-packaged spread of capped-uncapped variance swaps.

“Variance swaps are instruments that offer investors straightforward and direct exposure to the volatility of an underlying asset such as a stock or index. They are swap contracts where the parties agree to exchange a pre-agreed variance level for the actual amount of variance realized over a period. Variance swaps offer investors a means of achieving direct exposure to realized variance without the path-dependency issues associated with delta-hedged options.” Allen (2006) [16]

Without going into a detailed technical analysis here it is worth noting that most institutions historically want to capture monthly variance risk premia without bearing unlimited losses. The solution is to sell a variance swap with caps to banks. These caps are often set at 2.5 times the strike of the swap capping realized volatility above this level. Thus, their counterparties (banks) are usually left to be long variance capped 2.5X at whatever level the swap has been fixed. The goal to sell-side traders is mainly to capture the bid and ask spread, and under normal circumstances, they will look to offset the risk with other counterparties.

However, for capped variance swaps, there is a complication given there are not fixed default Cap strikes (any new swap will have a different strike). For those reasons in the interdealer market, variance swaps are usually traded uncapped. In mid-2000 banks they often had these kinds of “dirty hedges” [17] in their trading books, given they could collect extra returns.

The latest CCAR stress test, discussed in detail in “Dodd–Frank caught Covid-19: a study on the performance of OTC volatility-based trading strategies during the March 2020 crisis” Parlamento (2020) made it very difficult for banks to hold these risk since 2015/2016. So they started to pitch this optically appealing spread to the “yield-starved” buy-side.

An example

Buy-side 1 -> wants to capture monthly risk premia, and she asks multiple banks to price Monthly Variance Swap. She sells the capped monthly variance, at best bank BID level.
Banks 1 -> he buys the Capped Variance from buy-side 1. Immediately he is looking to offset the risk to collect the premium ( MID price – BID price).
Banks 1 -> he find counterparties into the interdealer market to sell uncapped variance while locking a positive spread.
Banks 1 -> the sell-side trader has left long capped and short uncapped variance swap. This trade very expensive to hold after changes in regulations.
Banks 1 -> he can repackage the risk within his “yield-enhancing” offering to and try to collect another spread (effectively intermediate the risk without violating regulations).
Buy-side 2 -> inherited the uncapped / capped variance, locking a favourable premium.
Buy-side 2 -> looks like a genius until March 2020.

In March 2020, such a strategy would have lost in the magnitude between 200x/800x of the monthly premium (according to the premium received and timing) [18]. For example, for a locked premium of USD1M, the loss could have reached between USD 200M and 800M in March alone. So even for the most optimistic scenario it would take about 17 years of premium to break-even, also making all existing trades and traders involved deep underwater.

Example of spread of capped-uncapped variance swaps P&L behaviour 

Inputs Vega Not   1,000,000.00 Strike 13.17% USD Cap Level (2.5X) 34,573,151.59
Date SPX Price Return (%)  Day Vol (%) Uncapped P/L (USD) Capped P/L (USD) Spread
(USD)
Returns ( X of premium)
02/18/2020 3370.29                       500,000  
02/19/2020 3386.15 0.5 7.453                      194,642 –             194,642                       500,000 + Infinity
02/20/2020 3373.23 -0.4 6.069                      420,176 –             420,176                       500,000 + Infinity
02/21/2020 3337.75 -1.1 16.785                      241,449 –             241,449                       500,000 + Infinity
02/24/2020 3225.89 -3.4 54.113 –                 4,305,458             4,305,458                       500,000 + Infinity
02/25/2020 3128.21 -3 48.811 –                 7,951,601             7,951,601                       500,000 + Infinity
02/26/2020 3116.39 -0.4 6.01 –                 7,724,891             7,724,891                       500,000 + Infinity
02/27/2020 2978.76 -4.4 71.702 –               15,924,475           15,924,475                       500,000 + Infinity
02/28/2020 2954.22 -0.8 13.132 –               15,922,799           15,922,799                       500,000 MTM losses
03/02/2020 3090.23 4.6 71.453 –               24,063,377           24,063,377                       500,000 MTM losses
03/03/2020 3003.37 -2.8 45.259 –               27,158,045           27,158,045                       500,000 MTM losses
03/04/2020 3130.12 4.2 65.619 –               33,978,905           33,978,905                       500,000 MTM losses
03/05/2020 3023.94 -3.4 54.784 –               38,646,417           34,573,152 –                 3,573,265 -7
03/06/2020 2972.37 -1.7 27.306 –               39,590,764           34,573,152 –                 4,517,612 -9
03/09/2020 2746.56 -7.6 125.425 –               65,270,232           34,573,152 –               30,197,080 -60
03/10/2020 2882.23 4.9 76.539 –               74,653,296           34,573,152 –               39,580,145 -79
03/11/2020 2741.38 -4.9 79.536 –               84,808,359           34,573,152 –               49,735,208 -99
03/12/2020 2480.64 -9.5 158.658 –             126,070,401           34,573,152 –               90,997,249 -182
03/13/2020 2711.02 9.3 140.979 –             158,589,131           34,573,152 –             123,515,980 -247
03/16/2020 2386.13 -12 202.641 –             226,080,890           34,573,152 –             191,007,738 -382
03/17/2020 2529.19 6 92.431 –             239,896,262           34,573,152 –             204,823,111 -410
03/18/2020 2398.1 -5.2 84.488 –             251,391,994           34,573,152 –             216,318,842 -433
03/19/2020 2409.39 0.5 7.456 –             251,197,432           34,573,152 –             216,124,280 -432
03/20/2020 2304.92 -4.3 70.368 –             259,084,124           34,573,152 –             224,010,973 -448

Source: Bloomberg

 

Looking at Bloomberg data and using the VIX as reference Variance was trading was about 14.2v on 18th February (the last day of VIX trading before settlement, and one of the typical days for rolling these strategies). Counting the bid-ask spread is not unreasonable that some buy-side players sold variance at 13.17v Capped at 33v on a 1M USD Vega Notional. We do not have exact data for the spread of the uncapped package purchased that day. However, the range quoted by the sell-side to sell uncapped and buy capped variance has been between 0.25v and 1v historically.

In the table above, we assume that the buy-side would have received 0.5v to enter the spread trade, so 500,000 USD “premium” ( the payout is not premium, but close enough to keep the analogy with insurance). It is uncanny the exponential progressions of accumulated realized losses, which totalled -224M (-448X) in just 12 days since the Cap have been breached. On top of the P&L hit, the holder of the risk would have started to receive margins calls since early March.

The most exciting mystery is maybe, which exact day many Portfolio Managers who have employed this strategy (and looked like geniuses for years) would receive their termination letter.

Conclusions

We do not believe that it is possible to disentangle the optimism of corporate America before the crisis [19], the emotional component on the traders in response to a health crisis[20], and the leverage embedded in financial engineering products [21] to analyse the fastest market correction [22] in history.

However, it seems safe to say that many investors who chased yields over the last five years have suffered exponentially higher losses than the one they anticipated in a crisis scenario. To minimize fees, some institutions may have internalized the dealing of these products without having the right capabilities, and they may have rewarded unknowingly funds who have were selling disaster insurance on their behalf.

On the bright side, it should be clear to allocators the value of the portfolio manager who delivered during the meltdown.

To quote Warren Buffet, “Only when the tide goes out to do you discover who’s been swimming naked.”

Notes
[1] The financial institutions who broadly invest in securities for money-management purpose
[2] Krishnan, Hari P. going into details on these phenomena on “The Second Leg Down: Strategies for Profiting After a Market Sell-off. John Wiley & Sons, 2017.”
[3] It is not totally correct thinking about yield enhancing strategies as the exact opposite of hedging, but on a general framework the expected stream of returns tends to be reverse-correlated.
[4] https://www.bloomberg.com/news/articles/2020-05-08/oil-crash-busted-a-broker-s-computers-and-inflicted-huge-losses
[5] Ralph Vince’s book “Portfolio Management Formulas” (1990) discuss in details the topic
[6] This is typically for hedge fund platforms
https://www.ft.com/content/8605e046-696b-11ea-800d-da70cff6e4d3
[7] Arguably the structure of Berkshire Hathaway have been fundamental to make capital “sticky” even after many years of underperformance during the dot.com.  He also have been able to be opportunistic in 2008. Warren Buffet structure allow him to sit on cash and wait until companies are cheap before buying them. Given he has permanent capital, so he can never be forced to sell, even if initial timing is not perfect.
[8] https://www.bloomberg.com/opinion/articles/2020-05-08/oil-prices-were-a-beautiful-mystery
[9] Post great financial crisis (GFC) there was a run to buy insurance which latest until 2012 which arguably inflated the performance of short volatility strategy.
[10] Parlamento, L. (2019). “February 2018 Volatility Black Monday: Evidence from Quant Factors gone Wild”. Working paper
[11] A lot can be written on the argument, but is out of scope of this article.
[12] Parlamento, L. (2020). “Dodd–Frank caught Covid-19: a study on the performance of OTC volatility-based trading strategies during the March 2020 crisis”. Working paper
[13] Supply and Demand dynamics were the big seller of financial insurance (for example AIG) went out of business in face on increasing demand to have tail-hedge strategies.
[14] https://www.institutionalinvestor.com/article/b1lffwvwdh7xtq/Amateurish-Trades-Blew-Up-AIMCo-s-Volatility-Program-Experts-Say
[15] https://www.theglobeandmail.com/business/article-alberta-pension-manager-loses-4-billion-on-investment-bet-gone-wrong/
https://www.theglobeandmail.com/business/commentary/article-response-to-aimco-debacle-shakes-trust-in-albertas-public/
[16] See, Allen, Peter, Stephen Einchcomb, and Nicolas Granger. “Variance swaps.” London: JP Morgan (November) (2006).
[17] Not an exact hedge, usually involve keeping some form exposure to correlation breaking / tail exposure. Risk that otherwise they will be too expensive to hedge
[18] Details discussed on “Dodd–Frank caught Covid-19: a study on the performance of OTC volatility-based trading strategies during the March 2020 crisis”
[19] Corporates levered up exponentially issuing debt to buy back stocks
[20] https://www.cnbc.com/2020/04/15/op-ed-coronavirus-market-swings-can-lead-to-an-amygdala-hijack.html
[21] Not just yield enhancement, but also CTA, Risk Parity have contributed to the speed of the drawdown according multiple commentators.
[22] https://www.cnbc.com/2020/02/27/this-is-the-fastest-stock-market-correction-in-history.html

 

The views and opinions expressed in this article are those of the author only, and do not represent the views or opinions of the author’s employers, the employers’ affiliates and employees, or any of the individuals acknowledged below. The author makes no representations or warranties, either expressed or implied, as to the accuracy or completeness of the information contained in this article, nor does the author recommend that this article serve as the basis for any investment decision. This article is for information purposes only.