Last month (https://convex-strategies.com/2020/11/19/risk-update-october-2020/) we referenced a speech from Federal Reserve Vice Chair of Supervision, Randall Quarles, and noted that we should expect to see further formal policy papers coming out on the topic of ongoing support of levered Non Bank Financial Institutions (NBFIs). Well, courtesy of the Financial Stability Board (FSB), here it is and she’s a beauty!
https://www.fsb.org/wp-content/uploads/P171120-2.pdf
The 60-page note has some useful timelines on the events preceding, during and following the pandemic related market shocks of Q1 this year. It is effectively a 60-page justification of moral hazard inducing behaviour by central banks, and even self-identifies as such a couple of times (pages 40 and 43) in the note (three times if you count the mention in the summary). Most interesting, in our opinion, is Section 3 where they discuss the “financial backdrop” leading into the shock, what we might call fragility. They, of course, go to great lengths to note that “the shock originated outside the financial system” while, nevertheless, acknowledging that “developments in the lead-up to the pandemic may have contributed to the severity of the reaction in financial markets”.
Figure 1: FSB: Financial Conditions and Debt Accumulation
Source: FSB Holistic Review of the March Market Turmoil, 17 November 2020
They could hardly state it any more gingerly:
“Certain pre-existing financial vulnerabilities may have amplified financial market reactions to the shock. Relatively easy financial conditions, stretched valuations in some asset classes, compressed risk premia and (more recently) a large amount of sovereign debt with negative yields further encouraged a search for yield.” Thus “it may also have increased the popularity of investment strategies reliant on low market volatility, short-term funding and high leverage”
We cannot help but wonder just how much time was spent on the final edits to wordsmith in such softeners as “(more recently)” and “it may also have”?
As we have noted several times over the last year or so, the point of all this work dates back to the “repo shock” of September 2019 and the justification for increasingly backstopping levered NBFIs. Section 3 of the paper provides some background on the evolution of NBFIs and the additional sub-categories of Insurance Corporations & Pension Funds (ICPFs) and Other Financial Intermediaries (OFIs).
Figure 2: FSB: Growth and Composition of NBFIs
Source: FSB Holistic Review of the March Market Turmoil, 17 November 2020
We are not sure precisely what to read into it, but we do find it interesting that they break down the details into a broad array of sub-categories and are still left with the large green slice of the pie that is a completely unexplained “Others”.
At the centre of the overall discussion is the period around the peak of the market stress in mid-March, and what the FSB folks have termed as the “dash for cash”. Section 4 goes into reasonable detail as regards the multiple sources of liquidity strains on markets. They do a good job of dancing around what we would consider the key issue at task, that being the mass of leverage utilized by these market participants. Or, how we like to phrase it, the lack of loss absorbing capital or “uncapitalized tails”.
What was driving the sudden “dash for cash” was not some unforeseeable phenomenon of foregone liquidity in underlying securities markets but, rather, an inevitable rush for the exits of the unsupported risk due to flawed risk methodologies and questionable incentive structures. Returning to last month’s example of the leverage food chain with the mortgage security REIT as the example, we can use our simple Value at Risk (VaR) calculator tool to show this.
Again, using 18-month daily return series, with one up till February 28, 2020 and the other through March 19, 2020, we can construct simple normal mean-variance distributions, generating VaR and cVar numbers. First, we highlight the Bloomberg Barclays Non-Agency CMBS Total Return Index, ie. “the asset”, then the listed share price of one of the larger mortgage security REITs, Annaly Capital Management, ie “the equity”.
Figure 3: Bloomberg Barclays Non-Agency CMBS 18mth daily returns data to 28 Feb 2020
Source: Convex Strategies, Bloomberg
Figure 4: Bloomberg Barclays Non-Agency CMBS 18mth daily returns data to 19 Mar 2020
Source: Convex Strategies, Bloomberg
Figure 5: Annaly Capital Management 18mth daily returns data to 28 Feb 2020
Source: Convex Strategies, Bloomberg
Figure 6: Annaly Capital Management 18mth daily returns data to 28 Feb 2020
Source: Convex Strategies, Bloomberg
The proxy asset has what looks like only a moderate increase in negative skew and kurtosis, but on a proportional basis it is quite significant, with a worst day p/l event that was 3x your 95percentile VaR projection. The problem here, on what our FSB friends would consider a liquid and efficient security, is the lack of capital supporting those losses. According to Bloomberg, Annaly has 9.07x financial leverage (defined as Average Total Assets/Average Total Equity). At that level of leverage, the flaws in risk measures at the asset level become noticeably amplified at the equity level. Clearly, the mis-measured risk of the asset, using a normal Gaussian distribution for something that is actually negatively skewed and fat tailed, results in an inevitable (near) wipe-out event at the equity level (saved by the policy response).
The asset, peak to trough, declined 11.52%. The equity, peak to trough, puked 66.57% before the big save came in. “Dash for cash”, margin calls, liquidations, etc. are all just the inevitable rush for the exits where capital is insufficient to support the heretofore mis-measured risk.
Figure 7: Bloomberg Barclays Non-Agency CMBS Index
Source: Bloomberg
Figure 8: Annaly Capital Management
Source: Bloomberg
It is all this uncapitalized tail risk that will continue to necessitate extraordinary support to keep the system from going into yet another systemic meltdown. As the FSB themselves put it, “the measures taken by central banks were aimed at restoring market functioning, and not at addressing the underlying vulnerabilities that caused markets to amplify the stress.” The below, from a Morgan Stanley piece, gives a simple visual of the lengths that central bankers are having to go to keep the underlying vulnerabilities at bay, and goes some way in explaining the ongoing asset inflation that seems overly detached from fundamental economic circumstances.
Figure 9: G10 Average Monthly Market Intervention/QE
Source: Morgan Stanley
Can this scale of central bank intervention go on indefinitely? History seems to indicate that it is unlikely but, as tends to be the case, there is no shortage of folks that will enthusiastically argue that “this time is different”. However, one possible source of future constraint on never ending central bank largesse could come in various forms of concerns about the wealth segregating effects of their asset inflation side-effects (or objectives, depending on your perspective). The second order effect of pricing large swaths of populations out of their own housing markets has come to the attention of a few other prominent voices of late. We took great interest in the letter from the NZ Finance Minister to the RBNZ, asking that their policies factor in the implications on housing prices. A mere couple of days later, similar sentiments were raised by the former head of the PBOC.
As many of you will be aware, the RBNZ is the original “inflation targeting” central bank, starting a trend that all would eventually follow. Thus, it is not all that crazy to think that they could be a leader in factoring in the implications of housing prices on monetary policy. Even in the most Keynesian circles of central bankers, you can generally get them to concede, at least in private, that housing prices were probably more germane to recent financial and economic cycles than deviations in various tracked consumer pricing indices. Fortunately, we didn’t have to wait long to hear the RBNZ’s response. As has been the case across any venturing down this path for some time, they made it very clear that house price measures should reside only in the domain of their financial stability mandates and, as such, dealt with through the means of various macro-prudential measures. They clearly state their resistance, if not so much their reasoning, against embedding house prices as somehow a part of their monetary policy remit.
You can save yourself some time and just summarize the response with the ‘Doublethink’ inspired simplicity of paragraph 46: “Finally, the addition of a housing consideration to the monetary policy remit could also make the goal of monetary policy confusing and reduce financial market efficiency”. End of discussion.
So where does the above leave us? It leaves us in a uniquely fragile environment, we would argue, where unprecedented levering of mis-measured risks necessitates extraordinary and endless policy support the world over. The second order effect of the policy support is the gap between asset valuations and underlying fundamentals, and the societal destabilizing impact of the resulting wealth segregation. It is the classic Catch-22. As we have said so often, central banks have done such a great job of preventing and extinguishing fires, thus allowing/inciting an unprecedented amount of fire risk to build up, that they see no choice but to throw everything at preventing and extinguishing fires.
If that thinking is correct, then what is to be done? We would argue much as always; focus on performance in the tails through building convexity in your portfolio. As promised last month, we can use our little VaR calculator to evaluate (wrongly measured) risk on a multi-asset portfolio and get some sense of the impact on kurtosis, not just from skew but also from correlation. We will use the same sort of analysis as above, 18-month daily returns over two periods, one ending February 28 2020, the other March 19 2020, but this time on our traditional 60/40 portfolio of S&P beta and US Treasuries, versus an adapted version of our Always Good Weather (AGW) portfolio of 30% S&P, 30% Nasdaq 100, and 40% VIXM the VIX mid-term ETF.
[We have used VIXM instead of usual CBOE Eurekaheadge Long Vol index for the daily data calculations for VAR]
Figure 10: 60/40 Balanced Portfolio 18mth daily returns data to 28 Feb 2020
Source: Convex Strategies, Bloomberg
Figure 11: 60/40 Balanced Portfolio 18mth daily returns data to 19 March 2020
Source: Convex Strategies, Bloomberg
Just as one would expect, the numbers at the end of February significantly under project potential risk. The revised distributions, post the 3 week pandemic shock, show the painful increase in realized loss and negative skew, and the subsequent convex increase in kurtosis. Obviously, the 40% allocation to fixed income did not act as an effective portfolio risk mitigant. You can probably guess what happens if, hypothetically, we take that 40% allocation and put in something explicitly risk mitigating and allocate half of our growth allocation to higher octane Nasdaq.
Figure 12: Adapted Always Good Weather Portfolio 18mth daily returns data to 28 Feb 2020
Source: Convex Strategies, Bloomberg
Figure 13: Adapted Always Good Weather Portfolio 18mth daily returns data to 28 Feb 2020
Source: Convex Strategies, Bloomberg
As if by magic, we now get a theoretical jump to positive skew through the 3 week shock period, and a large increase in right side (ie. profitable!) kurtosis. Lower initial VaR, despite the increased allocation to the more volatile Nasdaq 100 Index, and much less divergence on realised downside performance. Far too often we get the bemoaning pushback that the long volatility hedge costs money. This criticism holds true only if one makes the mistake of looking at it on a stand-alone basis. Poor risk mitigating strategies cost money because they impair your ability to hold more assets that will truly participate in the upside, eg fixed income and supposedly diversifying absolute return hedge fund strategies. Good risk mitigating strategies allow you to take on more upside-participating risks and more than pay for themselves through time.
Figure 14: Balanced vs Adapted Always Good Weather Sept 2018-Feb 2020
Source: Convex Strategies, Bloomberg
Figure 15: Balanced vs Adapted Always Good Weather Sept 2018-March 19th 2020
Source: Convex Strategies, Bloomberg
Let’s expand those out through the end of November. As various monthly realizations plot in the tails, the superior convexity in the Adapted AGW portfolio continues its tendency to pull away in the compounding race.
Figure 16: Balanced vs Adapted Always Good Weather Sept 2018-Nov 2020
Source: Convex Strategies, Bloomberg
And look at the longer period VaR distributions. Yes, we know it is a bad way to look at risk, but it is how the bulk of the financial world, courtesy of regulatory dictate, actually looks at risk. At least, with the long vol cutting off the fat left tail, we have eliminated the biggest flaw of the methodology.
Figure 17: 60/40 Balanced Portfolio daily returns data Sept 2018 to Nov 2020
Source: Convex Strategies, Bloomberg
Figure 18: Adapted Always Good Weather Portfolio Sept 2018 to Nov 2020
Source: Convex Strategies, Bloomberg
The positively convex portfolio has 1) lower VaR (thus needs less capital to carry), 2) significantly lower cVar, 3) positive skew (versus negative skew for the Balanced Portfolio), 4) positive upside kurtosis, and 5) superior compounded returns. The hedge does not cost money! We humbly suggest not to evaluate your goalkeeper on the same metric that you use to measure your strikers. Association Football matches, much like compounded returns, spend ~95% of their time in the middle of the pitch, and only 2.5% in both of the respective penalty boxes. All that matters, all that is relevant to the outcomes of the match, we believe, is what happens in the penalty boxes.
Read our Disclaimer by clicking here