Equities

Special Report

Highlights Volatility strategies are a useful tool for asset allocators. They can be used for both alpha generation and risk mitigation, but they have to be managed properly within a fund’s total risk management framework. Dedicated tail-risk hedging can reduce volatility, but can be very costly depending on the holding period. Short volatility strategies can generate alpha, but can also incur large losses when volatility spikes. Long volatility and also relative-value volatility strategies are much better alpha generators. A simple and easy-to-implement rule-based dynamic hedging strategy using short-term VIX futures reduces equity portfolio risk significantly without sacrificing return. The Sensational Headlines The COVID-19 pandemic-induced financial market volatility has put two major pension funds in the proverbial spotlight. First, CalPERS was questioned about its October 2019 decision to unwind its tail-risk hedging program that would have generated a payoff of more than US$1 billion during the March equity market selloff.1 Then, AIMCo was said to have lost over C$3 billion in its short volatility program, and was also forced to shut the program down.2 With such high-profile stories making the rounds, it is not surprising that we have received questions about tail-risk hedging and volatility strategies from many clients: Should long-term investors hedge tail risk? Is short volatility not a suitable strategy for pension funds? What are the efficient ways to manage large drawdowns? Chart 1The High Profile Failures: Not Uncommon Before we attempt to answer these questions, we want to first point out that tail-risk hedging and short-volatility strategies are negatively correlated, as shown in Chart 1, panel 1. It is normal for short-volatility strategies to suffer large drawdowns when tail-hedging strategies make handsome gains in periods of extreme financial market stress. This is largely due to the nature of volatility. As shown in panel 2 in Chart 1, VIX futures curves are normally in contango (the far-month contract is higher than the near-month contract), so a plain-vanilla short position in VIX futures benefits from positive rolling yields, while a plain-vanilla long position suffers from negative rolling yields. When VIX spikes, however, the futures curve turns into large backwardation (the far-month contract is lower than the near-month contract) in a fast and furious fashion, hence the large insurance-like payoff. The short-volatility and tail-hedge indexes in Chart 1 are from CBOE Eurekahedge, which has a suite of volatility indexes. As shown in Table 1, these indexes track the average performance of hedge funds that employ various volatility strategies, including tail-risk volatility, long volatility, short volatility and relative-value volatility. Table 1CBOE Eurekahedge Volatility Hedge Fund Indexes*  The performance statistics of these indexes are shown in Table 2. It is clear that not all volatility strategies are created equal. Below, we explore in more detail how these strategies should be used. Table 2CBOE Eurekahedge Volatility Index Performance Statistics Tail-Risk Hedging Is Not Free Tail-risk hedging has been in the news of late, given the unprecedently sharp drop in equities in February and March and also the untimely decision by CalPERS to unwind its tail-risk hedging program last October. So, what is tail-risk-hedging exactly? How does it work? Tail-risk hedging strategies aim to profit from large drawdowns in risky assets. Unlike the traditional approach of diversification that reduces the weighting of risky assets (for example, a 60-40 equity-bond portfolio is less risky than a 100% equity portfolio), tail-risk hedging attempts to allocate a small percentage of capital, say 3-5%, to a specially designed insurance-like payoff, while maintaining exposure to the risky asset. As such, tail-risk hedging is like buying an insurance policy against a catastrophic event. The premiums paid may or may not be recouped, depending on how likely it is that a catastrophic event may occur and how long one has held the insurance policy. The Universa Tail Fund is one of the two tail-risk funds that CalPERS made the untimely decision to redeem. The fund returned 3,600% in March alone, and 4,440% in the first quarter of 2020. As well, according to reports, a portfolio with 96.7% in the S&P 500 and 3.3% in Universa’s tail-risk fund would effectively have mitigated the S&P 500’s large loss in March, and would have also produced a compounded return of 11.5% since March 2008 versus 7.9% for the S&P 500.3  The performance of the Universa Tail Fund seems to be very different from the average hedge fund in this category, as shown in Table 2 and Chart 1. The CBOE Eurekahedge Tail Risk Hedge Fund index is an average of eight hedge funds that employ tail-risk strategies to achieve capital appreciation during periods of market stress. Since December 2007, when the index started, it has had two outsized monthly gains: 37.5% in March 2020 and 27.5% in August 2011, when MSCI US equities lost 12.7 and 5.5%. However, such benefit is very costly from a long-term perspective because the index has generated an annualized loss of 2.5%, even through April 2020. Its arithmetic average during the period is about -1.6%. To better understand why Universa has been doing so much better than the “average” tail risk hedge fund, we replicate a stylized exercise by Universa published in October 2017.4 The only difference is that we use the MSCI US equity total return index instead of the S&P 500 index. The payoff structure of 9 to 1 means that when the MSCI US calendar year return is less than -15%, the hedge would generate a return of 900%. In other years, insurance premium is not recouped at all, i.e. there is a loss of 100%. The original exercise by Universa designed such a payoff structure because it aimed to have an average payoff of zero in the period from 1996 to 2016. As shown in Chart 2, the biggest advantage of the tail-hedged portfolio (97% MSCI US + 3% Insurance) is its much smoother return stream, with a standard deviation of 12.9% compared to 17.7% for the unhedged MSCI US equity portfolio based on calendar year returns from 1970 to 2020 (as of March for 2020). Also, the skew is improved to -0.1 from -0.7. In terms of return, however, it is highly variable depending on the period chosen. The hedged portfolio outperformed the MSCI US total return index by about 70 basis points annualized from 1996 to 2016, consistent with the result from the original exercise by Universa.5 Outside this period, however, the average return of the payoff stream really depends on how often US equities fall below -15% yearly. In the 50-year period from December 1969 to December 2019, the average return of the insurance payoff was -20%, and the tail-hedged program underperformed MSCI US by 26 basis points annualized. Chart 2Universa Exercise Replica* For 12/1969 - 3/2020 This simple stylized exercise shows that both the starting point to initiate the tail-risk hedge and the length of time to hold the hedge are very important for a tail-risk hedge to work, not to mention generate spectacular results. Like a catastrophic insurance policy, a tail hedge should not be considered as a stand-alone strategy but as a hedge to the underlying portfolio. It is critical to design the right payoff structure, which in turn requires a view on how often a large drawdown will likely happen in the forecast period. It also takes special skill to find the right instruments to implement such a payoff structure and manage it accordingly. As we will show in the section on page 9, a dynamic approach is needed to ensure the hedge is on only when it’s needed to reduce cost. In fact, Universa did mention about using extreme valuation as one indicator to identify periods with high likelihood of downside risks.6 It also locked in a massive gain in March 2020,7 another indication of the “dynamic nature” of tail-hedging management. Bottom Line: From a long-term perspective, tail-risk hedge does not significantly improve compound returns, but it does reduce volatility significantly. Unless an investor has the skill to dynamically manage a hedge program, passively holding a tail-risk hedge can be costly in terms of return, even though it does improve risk-adjusted returns. Is A Short-Volatility Strategy Suitable For Pension Funds? The CBOE Eurekhedge Short Volatility index lost 20.8% in the first four months of 2020, in which March was the worst month in its history since December 2004, with a loss of 15.8%, while April was the best month with a gain of 9.3%. The annualized return since December 2004, however, has been 5.4%, and 73% of monthly returns have been in positive territory (Table 2). On the other hand, AIMCo had to shut down its volatility trading program in March because of its large $3 billion loss, or about 2.5% of its $119 billion of AUM. It is not known why a small volatility program was allowed to lose more than the fund’s total full-year value-add target. Chart 3Volatility Measures: Implied Vs. Realized There are different ways to short volatility. One is to sell options on the underlying assets. This approach, however, is also impacted by the price level of the underlying assets. VIX futures, as shown in Chart 1, panel 2, are a way to bet on the change in implied volatility. Another way to short volatility is via variance swaps, which bet on the change between realized variance at the expiry of the swap and the strike variance, which is set according to both historical variance and implied variance.8 Because variance is the square of volatility, the payoff of a variance swap is convex, i.e. when volatility spikes up, a short seller loses more money than when volatility decreases. As shown in Chart 3, VIX, the implied volatility, peaked on March 16, and realized volatility peaked on March 27. However, the difference between realized and implied volatility did not peak until April 6, and remained positive through the end of April. As such, a short volatility program via variance swaps would have experienced severe mark-to-market losses daily from mid-March to early April, even though equities bottomed on March 23.   However, such a spike happened in 2008 as well. Any back-test would have included such an occurrence in 2008. Granted, the magnitude of the current spike is larger than that in 2008, but it reversed quickly down to the 2008 level. We may never know why AIMCo’s short volatility program suffered such outsized losses. The only guess is that it may have used variance swaps, and the embedded leverage made the size of the program not appropriate for the total fund. Bottom Line: Short volatility can be a useful tool for alpha generation. The key, however, is risk management. It should be properly sized within the overall risk management framework of the total fund. Volatility As An Asset Class? Tail-risk hedging using volatility is too costly in general, while shorting volatility outright can be disastrous. Some argue that investors should not have anything to do with volatility strategies. On the other hand, other investors treat volatility as an asset class for both alpha generation and risk mitigation. Chart 4 shows the CBOE Eurekahedge Relative-Value Volatility index and the Long-Volatility index together with the MSCI US equity index, and Bloomberg Barclays US aggregate bond index and US Treasury index. The relative-value volatility index can be long, short, or neutral on volatility (Table 1). As shown in Table 2, it has achieved an annualized return of 7.6%, only 60 basis points less than MSCI US equity return of 8.2%, but much higher than the 4.3% and 4.5% respective return from Bloomberg Barclays US Treasury index and aggregate bond index in the period from December 2004 to April 2020. Its standard deviation of 3.9% is much lower than the MSCI US (14.7%) and very close to Treasurys (4.1%) and aggregate bonds (3.2%). For this specific period, in fact, this index even has a much better risk-return profile than a typical 60/40 US equity/aggregate-bond portfolio, which scores a 7.1% annualized return with 8.9% standard deviation. With almost zero correlation to both stocks and bonds, this index serves as an ideal addition to a balanced equity-bond portfolio (Chart 5). Chart 4Volatility As An Asset Class Chart 5Relative-Value Vol Strategy Improves The Performance Of A 60/40 Equity/Bond Portfolio The challenge, however, is that this index is an average of 35 hedge funds that employ relative-value or opportunistic-volatility strategies that can be long, short, or neutral on implied volatility.9 Because of this, capacity constraints for investors to get into those funds may exist, which could produce diverging performances. Even the long-volatility strategy (Chart 4, panel 2), which in theory suffers negative rolling yields when the VIX is in a normal range, has generated a 5% annualized return. It has a negative correlation of 0.46 with MSCI US equities, comparable to the negative correlation of 0.5 between the Tail-Risk index and MSCI US. Given the much better statistics of this index compared to the Tail-Risk index, it should be a less costly alternative to the Tail-Risk Hedge index (Table 2). To illustrate how these two strategies work to mitigate downside risk in the MSCI US equities, we compare a series of portfolios that allocate from 0-100% of capital to MSCI US and 100-0% to the two volatility strategies, respectively. As shown in Chart 6, the long-volatility strategy is a much better risk mitigator to the MSCI US equities index than the tail-hedge strategy at all levels of allocations for the period from January 2008 to April 2020.  Chart 6Risk Mitigation Using Long Vol Vs. Tail-Risk Hedge Dynamic Hedging Using VIX Futures The CBOE Eurekahedge volatility indexes are based on average returns of the funds in each index. They are not investable. Also, hedge funds in these indexes may have capacity issues to accommodate large investors. In this section we run a simple rule-based hedging strategy using VIX futures to illustrate how investors can use volatility strategies in-house as an alternative tool to mitigate risk. We use the S&P VIX short-term futures index for this exercise, because it can be easily replicated in-house. This index is constructed based on rolling daily 5% of the front-month contract to the second-month contract. This means the index always has one month to expiry. It also means that daily rolling averages out the rolling yield for any given month.  The rule is simple: invest in the short-term volatility futures only when the VIX is outside its normal range. Since its inception in 1990, the VIX average is about 20. To test how different thresholds and rebalancing frequencies work, we test four different VIX thresholds: 25, 30, 35 and 40 with both weekly and monthly rebalances. The rebalance rule is: if the VIX is greater than a threshold at the end of one period, then in the next period, 5% of the fund is allocated to the S&P short-term VIX futures index and 95% is allocated to MSCI US. Otherwise 100% goes to MSCI US equities. For comparison, we also run a static hedge that has 5% in VIX futures and 95% in the MSCI US index.  The monthly rebalanced results are quite interesting, as shown in Table 3 and Chart 7: Table 3Dynamic Hedging Using VIX Futures Chart 7Dynamic Hedging Works Despite a terrible risk-return profile on its own, VIX futures can be a good risk mitigator when the hedge is put on only when the VIX is above a certain threshold.  Even though the 60-40 wins in terms of risk-adjusted return, dynamically hedged portfolios have better returns than both the 60-40 and US equities. The results are also robust when we do a weekly rebalance. Three conclusions can be drawn from Charts 8A and 8B, and Chart 9: Chart 8ADynamic Hedging – Monthly Rebalance Chart 8BDynamic Hedging – Weekly Rebalance     Chart 9Simple But Robust Dynamic Hedging   Hedging reduces volatility significantly. The lower the VIX threshold is, the larger the volatility reduction in the hedged portfolio compared to the unhedged. Hedging also improves average returns, albeit at a smaller scale compared to the reductions in volatility. Depending on the rebalancing frequency, the return improvement differs. For the monthly rebalance, the best VIX threshold lies between 30-35; for the weekly rebalance, the best is when the VIX threshold is at 30. Hedging is not needed all the time because volatility is within a normal range most of the time. Even when it spikes, it does not stay high for an extended period of time. Bottom Line: A simple rule-based dynamic hedging approach using VIX futures can substantially improve an equity portfolio’s risk-return profile by decreasing volatility significantly without sacrificing return. In a low interest rate environment, dynamic hedging using VIX futures can be a good alternative to a 60-40 equity-bond mix.   Xiaoli Tang Associate Vice President xiaoliT@bcaresearch.com   Footnotes 1  https://www.institutionalinvestor.com/article/b1l65mvpw5xpts/The-Inside-Story-of-CalPERS-Untimely-Tail-Hedge-Unwind 2 https://www.institutionalinvestor.com/article/b1l9c8n9lgdj1r/AIMCo-s-3-Billion-Volatility-Trading-Blunder 3 https://www.bloomberg.com/news/articles/2020-04-08/taleb-advised-universa-tail-risk-fund-returned-3-600-in-march 4 https://www.universa.net/UniversaResearch_SafeHavenPart1_RiskMitigation.pdf 5  https://www.universa.net/UniversaResearch_SafeHavenPart1_RiskMitigation.pdf 6 https://www.universa.net/UniversaResearch_SafeHavenPart2_NotAllRisk.pdf 7 https://www.bloomberg.com/news/articles/2020-04-08/taleb-advised-universa-tail-risk-fund-returned-3-600-in-march 8  https://en.wikipedia.org/wiki/Variance_swap 9 https://www.eurekahedge.com/Indices/CBOE-Eurekahedge-Volatility-Indexes-Methodology  

The SPX catapulted to fresh recovery highs, on the back of optimism surrounding the successful reopening of the economy along with the ongoing support of easy fiscal and monetary policies. Sentiment is not as extended as in February or during previous SPX tops in the past few years, as we highlighted in recent research.1 Equity market internals signal that there is likely a bit more gas left in the tank, despite the roughly 1000 point rise since the March 23 lows. The S&P deep cyclicals/defensives share price ratio, has led the broad equity market bottom and continues to herald additional gains for the SPX (not shown). Deep cyclicals include tech stocks, but even if IT were excluded, the cyclicals ex-tech/defensives ratio still troughed prior to the SPX and is gaining steam. Importantly, the turn in our Global Trade Activity Indicator corroborates the message that the cyclicals/defensives ratio is emitting (see chart). Further, the recent breakout in the JPM EM currency index along with budding evidence of China’s economic recovery and likelihood of a stimulus package (not as large as the GFC, but bigger than the early-2016 manufacturing recession one) suggest that global growth is slated to recover in the back half of the year. Bottom Line: We remain constructive on the broad market’s prospects  over the coming 9-12 month time horizon. For more details, please refer to this Monday’s Weekly Report.   Footnotes 1  Please see BCA US Equity Strategy Weekly Report, “There’s No Limit” dated May 26, 2020, available at uses.bcaresearch.com.

Dear client, Along with an abbreviated report this week we are sending you this Geopolitical Strategy service report written by my colleague Matt Gertken, BCA’s Geopolitical Strategist. Matt argues that US social unrest is structural and therefore can still cause volatility, while the market’s recognition that Trump is an underdog is also a risk. I hope you will find this report both interesting and informative. Kind Regards, Anastasios   Portfolio Strategy While we remain constructive on the prospects in the broad equity market over the coming 9-12 month time horizon, a flare up in geopolitical risks and uncertainty around the upcoming election could serve as catalysts for a much needed breather in equities. Recent Changes Last week our rolling stop was triggered and we downgraded the S&P biotech index to neutral and booked gains of 5% since inception.1 Table 1 The SPX catapulted to fresh recovery highs last week, on the back of optimism surrounding the successful reopening of the economy along with the ongoing support of easy fiscal and monetary policies. Sentiment is not as extended as in February or during previous SPX tops in the past few years, as we highlighted in recent research.2 However, greed is slowly showing up on our radar screens as investors that have missed out on the rally are chasing performance. Additionally, the market action has an element of a short squeeze. Equity market internals signal that there is likely a bit more gas left in the tank, despite the roughly 1000 point rise since the March 23 lows. While the S&P transports index has neither made new all-time highs nor outperformed the SPX year-to-date, one economically hypersensitive sub-group, trucking, has been revving its engines. The S&P 1500 trucking index has stealthily joined the “new all-time highs” club. The highly fragmented trucking industry has an excellent track record in leading the S&P 500 and the current message is that the path of least resistance remains higher for the SPX (Chart 1). As large parts of the economy are reopening, this index seems to have priced in a full recovery and a return to normal in the back half of the year. The jury is still out on the economic recovery’s shape and the risk of a second viral wave is significant, but stocks continue to climb the proverbial "wall of worry". Chart 1Trucking As A Leading Indicator Importantly, another extremely pro-cyclical equity market indicator, the S&P deep cyclicals/defensives share price ratio, has also led the broad equity market bottom and continues to herald additional gains for the SPX (Chart 2). Deep cyclicals include tech stocks, but even if IT were excluded, the cyclicals ex-tech/defensives ratio still troughed prior to the SPX and is gaining steam. Chart 3 shows the GICS1 sector returns since the March lows and technology is similar to the overall market’s return. The deep cyclical trio (energy, industrials and materials) have outperformed the tech sector, and bested defensives by a wide margin. Chart 2Cyclicals Are Besting Defensives Chart 3GICS1 Sector (%) Returns Since The March Lows Our Global Trade Activity Indicator corroborates the message that the cyclicals/defensives ratio is emitting (Chart 4). The recent breakout in the JPM EM currency index along with budding evidence of China’s economic recovery and likelihood of a stimulus package (not as large as the GFC, but bigger than the early-2016 manufacturing recession one) suggest that global growth is slated to recover in the back half of the year. Chart 4Looming Global Growth Recovery Nevertheless, it is quite unnerving that the SPX has broken out to fresh recovery highs despite bleak economic fundamentals and rising political and geopolitical risks. One potential negative catalyst that could cause a healthy reset is the rise in the polls of Democratic presidential candidate Joe Biden ahead of the November elections. Chart 5 shows that over the past year, the S&P 500 has moved in lockstep with the relative odds of a Republican versus a Democrat getting elected President. But recently, a wide gap has opened warning that the SPX is vulnerable to a pullback. In truth, the online gambling community has been slow to react to the erosion of President Trump’s platform due to pandemic and recession – so his odds could fall further in the near term. At the margin, a Biden win should be negative for the stock market because his party is perceived as more hostile to businesses and the specter of higher taxes could trip up the SPX. Our Geopolitical Strategy service has highlighted this risk in recent reports, including on May 15.3 Tack on the persistently high reading in the Baker, Bloom and Davis Policy Uncertainty Index and the risk/reward tradeoff for the overall market tilts further to the downside at the current juncture (Chart 6). Chart 5Do Not Neglect (Geo)Political Risks Chart 6High Policy Uncertainty Is A Red Flag Bottom Line: While we remain constructive on the SPX over the coming 9-12 month time horizon, a flare up in geopolitical risks and uncertainty around the upcoming election could serve as catalysts for a much needed breather in equities.   Anastasios Avgeriou US Equity Strategist anastasios@bcaresearch.com   Geopolitical Strategy Social Unrest Can Still Cause Volatility Highlights Social unrest in the US is driven by structural and cyclical factors as well as election-year opportunism. It can still cause volatility. Unrest will weigh on consumer and business confidence – adding to already ugly fundamentals. The market has come around to our view that Trump is an underdog in the election. This is a risk to equities since a Democratic victory will bring full control of government. President Trump has low legal or political constraints to deploying the military if violence gets worse in the streets. This increases tail risks of a civilian death that amplifies the unrest. A “silent majority” of voters could give Trump a polling boost as a “law and order” candidate later this year. This could require us to upgrade his odds of reelection. The US dollar faces long-term headwinds but we are unlikely to reinitiate our long EUR-USD trade until the US election cycle is complete. Feature Chart 1Markets Skyrocket On Stimulus & Reopening Economic reopening and stimulus are winning the day as investors continue to look forward to a time when growth and corporate earnings recover yet inflation and risk-free rates remain suppressed. Judging by the breakout of cyclical versus defensive stocks and risk-on versus risk-off currencies, the rally could continue and the gap between stock markets and macro fundamentals could widen further for some time (Chart 1). The market is looking through the most widespread social unrest since 1968 in the United States, which emerged due to the death in police custody of a black man, George Floyd, in Minneapolis. History suggests that over a one-year horizon, social unrest can be ignored – but in the near term it could yet provoke volatility. This risk is underrated because the market already believes that the unrest is a known quantity without material impact, yet this report shows otherwise. We see four new risks, the first three negative for the market. Chart 2US Consumer Sentiment Is Vulnerable Consumer confidence and activity could worsen in the face of historic national unrest. The slight uptick in improving consumer expectations could reverse (Chart 2). President Trump’s odds of reelection could fall permanently, triggering a downgrading of long-run earnings expectations. A mistake could cause unrest to reach an unknown critical threshold that strikes fear into investors about US stability. The US debate has moved on from racism to “fascism” as Trump’s opponents criticize him for his authoritarian rhetoric and deployment of military forces to secure parts of Washington, DC. Structural factors are driving the riots which means they may smolder and additional incidents could cause them to flare up throughout summer and fall. The deployment of troops to quell civil unrest – as in any country at any time – could easily lead to bloody mistakes. The upside risk is that Republican senators will capitulate even sooner on fiscal spending measures, seeing that their corporate power base is likely to feel more concerned about the collapse of society. The House Democrats and President Trump already share an interest in larding up the spending, so it was only a matter of time till the senate caved in anyway. If the next $2 trillion arrives without the June-July hiccup that we expect, then the market could power higher (Chart 3). Chart 3Global Fiscal Stimulus Continues To Grow In this report we show why US social unrest is structural and how it can still bring equity volatility. Also, the online betting market has caught up to our view that Trump is the underdog in the election. The prospect of full Democratic Party control could start to weigh on US equities. The upside risk to this view would be markets cheer Biden – which is unlikely for long – or if the violent protests create a “silent majority” that helps Trump win the swing states. If his polling improves in the wake of the riots – and the stock rally continues unabated – then we may upgrade his reelection odds from 35% to 50% or higher. Bottom Line: A pullback would be a buying opportunity, but a 10% correction could easily transpire given that a falling market reduces Trump’s odds greatly and could kill the market’s faith in Trump reflation policy from 2021-24. How Social Unrest Came To The United States The US was ripe for a major bout of unrest, as we have highlighted in past reports such as “Populism Blues” (2017), “Civil War Lite” (2019), and “Peak Polarization” (2020), as well as in our top five “Black Swans” report for this year. Our updated “Great Gatsby Curve” shows countries with high levels of income inequality and social immobility. The US is right in the danger zone, joined by other countries that have had unrest or political disruptions (Argentina, Chile, UK, Italy) or will soon (China) (Chart 4). African Americans suffer the worst of these ills and also have long-running grievances with the criminal justice system. Chart 4The US Is In The Danger Zone For Populism, Unrest Unrest was an easy prediction even before the pandemic and recession, which made matters worse. The US ranks last, among developed markets, just below Greece, in our COVID-19 Unrest Index (Table 1). This index combines four factors – economic fundamentals, vulnerability to COVID-19, household grievances, and governance indicators – to rank countries according to their susceptibility to social unrest. US unemployment has soared higher than that of other countries as it has less generous automatic stabilizers. Table 1US Ranks Worst In Our COVID-19 Social Unrest Rankings When it comes to the virus, the US is not any harder hit than most of its European peers (Chart 5). And the black community is not much harder hit than whites, although both have suffered more than their population share would imply, and more than the Hispanic community (Chart 6). Chart 5US No Different Than Western Europe On COVID-19 Deaths   Chart 6COVID-19 Least Deadly For Hispanics However, the lockdowns have caused the unemployment rate to soar and exacted a greater toll on the least educated and lowest paid members of society. The election is enflaming the situation. President Trump’s economy has now performed little better for households than President Obama’s economy, assuming they suffer an income and wealth shock at least equal to that of 2008-09 (Chart 7). Chart 7Households Suffer Massive Income Shock Given the collapsing economy, Trump is doubling down on “law and order,” taking an aggressive stance against rioting and looting and thus provoking a backlash. The media is also in a feeding frenzy as the pandemic and economic reopening narratives lose traction and yet Trump perseveres. Polarization is intensifying as a result. Trump’s rhetoric has been egregious as always. His threat to invoke the Insurrection Act of 1807 is not. President George Bush Sr invoked the act to suppress the LA riots in 1992. The act’s provisions, as well as the specific exceptions to the posse comitatus laws and norms, give the president broad discretion in matters precisely like these. The real constraint is not legal but political: any popular backlash from Trump and his advisers in trying to “dominate the battlespace” when it comes to civilians at home. Rioting and looting are also unpopular, so a larger crackdown could easily happen if more unrest takes place. Since the riots are driven by structural factors, they could still escalate, especially if another incident of police brutality occurs. Bottom Line: US unrest is driven by structural and cyclical factors and thus we are in for another “long, hot summer” like 1967. Negative surprises should be expected. The larger risks have to do with the impact on the election and sentiment. Trump’s Polling Was Dropping Even Before The Riots Trump’s approval rating has fallen to the lowest level this year and diverged from the historic average (Chart 8). This increases the risk that the market experiences volatility either in expectation of “regime change” in November or in reaction to Trump’s attempts to regain the initiative. Trump’s deviation from President Obama’s approval at this stage in 2012 is a warning sign (Chart 9). Chart 8Trump’s Polling Drops Below Average Chart 9Trump Falls Off Obama’s Pathway To Reelection Chart 10Trump’s Pandemic Bounce Turns Negative, Unlike Others Trump and the Republican Party received a smaller polling bounce from the pandemic – and year-to-date the bounce is not only gone but has turned negative, comparable only to Vladimir Putin and United Russia (Chart 10). At its peak it was smaller than that of previous US presidents in crisis situations (Table 2, see Appendix). These data come from before the George Floyd incident which will make matters worse for Trump, given that initial polls suggest 35% approve and 52% disapprove of his response to it. The presumptive Democratic nominee Joe Biden is narrowly leading in all major swing states (Chart 11A). Trump has dropped off in critical swing states of Florida, Wisconsin, and Arizona (Chart 11B). Biden is closer to Trump than he should be in states like Ohio and even Texas. Chart 11ATrump Trailing Biden In Swing States Chart 11BTrump Loses Critical Support In FL, WI, AZ Chart 12Biden Polling Better Than Clinton Did Against Trump Biden is tentatively outperforming Hillary Clinton’s showing in 2016 in head-to-head polls against Trump, including in swing states (Chart 12). He has not been on voters’ minds much during the crises. But he has strong support among African American voters, who primarily handed him the party’s nomination, so he may be able to exploit the unrest. Voters indicate they favor him on race relations as well as the coronavirus, though they still favor Trump on the economy. Bottom Line: Trump’s polling was deteriorating before the social unrest. It will suffer more in the near term. But there are still five months until the election. The Market Now Recognizes That Trump Is An Underdog Now, with the country’s biggest cities ablaze, the market is waking up to the fact that Trump and the Republicans have a much greater chance of entirely losing control of the government in just five months. Online gamblers have recently upgraded Biden and the Democrats substantially (Chart 13). Opinion polling has shown weakness but now it is likely to seep into the financial industry’s consciousness that US domestic political risks could still go higher. Policy uncertainty will not fall as sharply as otherwise expected during the economic reopening. Unrest typically reflects negatively on the ruling party, suggesting the status quo is unacceptable and driving voters to vote for change. This is one of the 13 keys to the presidency under the scheme of Professor Allan J. Lichtman, at American University, who has predicted every popular vote outcome since 1984. If one accepts this thesis, then at least five of the keys have now turned against Trump and the GOP. If the economy somehow continues to shrink in the third quarter, or if GDP per capita falls harder than estimated in Chart 7 above, Lichtman’s model will turn against Trump (Table 3, see Appendix). Our own argument has been that a health crisis and surge in unemployment alone are enough to undercut him given his thin margins of victory four years ago and low approval rating. The George Floyd incident reinforces this logic. Not only is voter turnout correlated with the change in unemployment over the president’s term in office, but the correlation holds in swing states and among African Americans. Here is where the devastating impact of COVID-19 among blacks may be relevant (Chart 14). Chart 13Online Bookies Now See Trump Is Underdog Chart 14Hardship For Blacks In Swing States Chart 15Unemployment Pushes Up Voter Turnout (For Blacks And All) If the pandemic and unemployment did not already provide sufficient motivation, then the George Floyd incident might rally this core Democratic Party constituency to turn up at the ballot box (Chart 15). That is a threat to President Trump given that Barack Obama is not on the ballot, so black turnout is unlikely to reach 2008 or 2012 levels. Bottom Line: An increase in African American voter turnout due to unemployment and poor race relations would broaden the electoral pathway to a Democratic victory in November. A Risk To The View: The Silent Majority Could the unrest help Trump? Possibly. Once the peaceful protests turned violent, the possibility emerged that Trump could benefit. The Democrats are not in a strong position whenever they link themselves to economic lockdowns and rioting and looting. It is clear from the police killings and unrest of 2014-15 that more and more people have lost confidence in police treating blacks and whites equally (Chart 16), but they do not make up a majority. Chart 16Over Time, Voters Losing Confidence In Police Fairness Chart 17Majority Sees Racism As Individual, Not Institutional Moreover, two-thirds of citizens, two-thirds of Hispanics, and almost half of blacks believed at that time that racism and discrimination stem from individual actions rather than institutional factors (Chart 17). Confidence and institutional trust will fall during today’s crisis moments but the above polls suggest limits to the protest movement. Generally Americans are satisfied with the work of their local police departments (Chart 18). This includes 72% of blacks. Only about a quarter of Americans report being harassed by the police at any time, according to a Monmouth University poll. Chart 18Silent Majority? Most Americans Satisfied With Local Police Almost 80% of people believe police funds should be increased or kept the same, versus 21% who agree with defunding the police. Only 39% of blacks support such a proposal (Chart 19). If House Democrats pass legislation characterized as taking funds away from police it will hurt them. Chart 19Silent Majority? Americans Don’t Want To Cut Police Funding Finally, regarding the use of the military, 58% of Americans approve of the US military supplementing city police forces, while 30% oppose (Chart 20). George Bush Sr deployed troops in a similar predicament, the LA riots of 1992, albeit with an invitation from the California governor. Chart 20Silent Majority? Americans Mostly Support Military Aid To Police Amid Unrest Legal constraints on Trump’s use of the military are low. Given that the political constraint is also low, a resurgence in violence will likely lead to a crackdown. Trump could benefit if it is managed successfully, but the risk of a bloody mistake that harms or kills civilians would also go up. Bottom Line: Trump could benefit from his pitch as the candidate of law and order if unrest continues, violence worsens, and his actions are deemed to restore order. We will upgrade Trump’s reelection odds if his polling improves and the stock market and economy continue to rebound. Investment Takeaways Historic bouts of unrest show that market volatility occurred in the wake of the 1965-69 disturbances, the 1992 LA riots, the breakdown of order in New Orleans after Hurricane Katrina in 2005, and the protests and riots against police brutality in 2014-15. Unrest did not prevent the market from rallying in all of these cases, but it did in some, and pullbacks also followed unrest periods. In every case presidential approval suffered – and in 1968, 1992, 2006, and 2014 the ruling party suffered losses in the election (Charts 21 A-D). Chart 21AThe ‘Long, Hot Summer’ Saw Inflation, Volatility Chart 21BLA Riots Saw Unemployment, Volatility Chart 21CKatrina Saw Volatility, Presidential Approval Drop Chart 21DFerguson Saw Volatility Amid Falling Unemployment Chart 22Confidence Suffers Amid Social Unrest Furthermore, consumer and business confidence generally suffered in these periods (Chart 22). Trump’s reelection bid could fail to recover, which would make him a lame duck and heighten political risks dramatically. Our longstanding view that the party that wins the White House will also win the senate is reinforced by this year’s polls. The market is reacting to stimulus now but policies look to turn a lot tougher on business. The election puts a self-limiting factor into the equity rally. Either the market sells off in the short run to register the currently likely victory of Joe Biden, who will hike taxes, wages, and regulation, or the market rallies all the way till the election, increasing the chances of President Trump’s reelection, which would revolutionize the global system, especially on trade, and would require a selloff around December. The US dollar faces near-term headwinds as global growth recovers and uncertainty related to COVID-19 abates, but the near term is murky, whereas the major headwinds are over a cyclical time horizon. Our theme of “peak polarization” in the US contrasts starkly with our theme of “European integration” and implies that the euro can continue to advance. However, we are unlikely to reinitiate our long EUR-USD trade until the US election cycle is complete. The risk of a Trump victory is still substantial and we view Europe as a marginal loser in that scenario. We still expect investors to flee to the dollar in the event of any global crisis, even if it originates in the United States.   Matt Gertken Vice President Geopolitical Strategist mattg@bcaresearch.com Appendix Table 2Trump’s Crisis Polling Bounce Compared To Previous Presidential Bounces Table 3Lichtman’s 13 Keys To The Presidency Likely Turning Against Trump … Economy Critical   Footnotes 1     Please see BCA US Equity Strategy Insight Report, “Housekeeping” dated June 4, 2020, available at uses.bcaresearch.com. 2    Please see BCA US Equity Strategy Weekly Report, “There’s No Limit” dated May 26, 2020, available at uses.bcaresearch.com. 3    Please see BCA Geopolitical Strategy Weekly Report, “Michelle, Amash, Trump, Biden” dated May 15, 2020, available at gps.bcaresearch.com

Highlights If policymakers can neutralize default pressures arising from the lockdowns, the lasting impacts of this recession may not be so bad: As Jay Powell put it on 60 Minutes several weeks ago, policymakers just have to keep people and businesses out of insolvency until health professionals can gain the upper hand over the virus. Fiscal spending caused income and savings to spike, … : Generous transfer payments have left the majority of the unemployed better off than they were when they were working, and April household income and savings soared accordingly. … allowing consumers to meet nearly all of their obligations … : April’s income and savings gains showed up in reduced delinquencies across all categories of consumer loans and in solid April and May rent collections. May’s employment gains suggest that the private sector may not be too far away from taking the baton from Congress: The May employment report blew away expectations and sent risk assets surging, but the positive surprise may derail plans for further fiscal support. Feature Since March, investors have been presented with a simple choice: believe their eyes or believe in the government. They could either focus on horrendous economic data illustrating the crippling effects of widespread lockdowns, or they could trust in policymakers’ ability to shield most citizens and businesses from lasting damage. Our base case has been that policymakers would succeed, for the most part, provided they didn’t have to contend with acute COVID-19 pressures for more than six months. There are as many guesses about the virus’ future path as there are commentators, but it seems reasonably conservative to estimate that the most onerous restrictions will be eased by October. Chart 1DC To The Rescue In our view, preventing defaults is the key to mitigating the effects of the virus. If newly vulnerable debtors can be kept from defaulting until the economy can return to something resembling normal, a negatively self-reinforcing dynamic will not take hold, the infection will not spread to the financial system and creditworthy individuals’ and viable businesses’ temporary liquidity issues will not morph into solvency issues. Banking system data to confirm or disprove our thesis will not be available until August, however, as Fed and FDIC data are quarterly, and the shutdowns only began in late March. The unemployment safety net has turned into a trampoline; ... In this report, we have turned to a range of other sources for higher-frequency insights into what is happening in real time. We start with an academic paper showing that most laid-off workers are eligible for benefits comfortably exceeding their previous income, a conclusion reinforced by the April personal income data (Chart 1). We then look at April delinquency data from TransUnion, one of the major credit reporting agencies, and April and May rent-collection data from an apartment trade organization and large-cap publicly traded apartment REITs. We also review the Fed’s Survey of Consumer Finances to get a sense of household indebtedness across the income and wealth spectrums. For now, the data support the conclusion that policymakers have successfully defused credit distress pressures. What Comes In … Unemployment benefits typically fall far short of workers’ regular compensation, averaging about 40% of the median worker’s wage. To cushion the blow of unemployment from COVID-19, the CARES Act included a federal supplement to unemployment benefit payments distributed by the individual states. Added onto the average $400 weekly state benefit, the $600 federal supplement would make the average worker whole (mean earnings are a little less than $1,000 a week). As income inequality has intensified, the compensation distribution for all American workers has come to exhibit a pronounced rightward skew. That skew has pulled mean compensation (the average of all Americans’ earnings) well above median compensation (the earnings of the worker at the exact middle of the earnings distribution).1 By targeting mean compensation, the CARES Act opened the door for a lot of lower-income workers to make more money in unemployment than they did when they were working. According to a recent paper from three Chicago professors, 68% of unemployed workers are eligible to receive benefits that exceed their previous income, while 20% of unemployed workers are eligible for benefits that will at least double it. Overall, they calculate that the median worker is eligible to receive benefits amounting to 134% of his/her previous income.2 ... instead of keeping laid-off employees' incomes from falling below 40 cents on the dollar, it's launched them to $1.30. We offer no judgments about the policy merits of a 134% median replacement rate, but unusually generous benefits should help reduce the drag from unemployment that would otherwise ensue with a 40% replacement rate. Thanks to lower-income households’ higher marginal propensity to consume, consumption should rise at the margin (once activity resumes). Thanks to increased income, lower-income households should be better positioned to meet their financial obligations. We suspect the marginal consumption boost may be hard to see with the naked eye, but auto, credit card and mortgage delinquencies should be appreciably lower than any regression model not adjusted to reflect record replacement rates would predict. … And What Goes Out The Personal Income and Outlays data for April reflected the significant impact on household income of the up-to-$1,200 stimulus checks (economic impact payments) and the supplemental unemployment benefits. Despite an annualized $900 billion decline in employee compensation, personal income rose by nearly $2 trillion in April, thanks to a $3 trillion increase in transfer payments. De-annualizing the components, $250 billion in transfer payments offset a $75 billion decrease in compensation. At about $220 billion, the economic impact payments accounted for the bulk of the transfer payments, and they will fall sharply in May. The IRS did not disclose the amount of economic impact payments it had disbursed by April 30, but it appears that around 80% of the distributions have been made, leaving approximately $55 billion yet to be disbursed. Unemployment insurance receipts will rise in May on an extra week of benefits and an increase in the weekly sums of initial and continuing unemployment claims. We project that employee compensation rose about 3% in May, based on a 2% gain in employment and a 1% increase in average weekly earnings. Aggregating the February-to-May changes, it appears that May personal income ought to exceed February (Table 1). Absent another round of stimulus checks, however, personal income will slide below its pre-shutdown level beginning in June. Table 1May Personal Income Should Exceed Its Pre-Pandemic Level Income is not the sole driver of households’ capacity to service their debt, however. Assets matter, too, and even if the surge in cash flow was a one-off event, it left behind an elevated stock of cash as households slashed consumption in both March and April. Real personal consumption expenditures have fallen 19% from February’s all-time high and are now back to a level they breached in January 2012 (Chart 2). Households saved 33% of their April disposable income, and on a level basis, April savings were up nearly fivefold from their 2019 average. They were a whopping 20 times April interest payments, ex-mortgages (Chart 3). Chart 2Eight Years Of Spending Undone In Two Months Chart 3Consumers' Interest Coverage Ratios Have Soared Household Borrowers Are Staying Current … Table 2Consumer Borrowers Are Hanging In There It is possible to make too much of the April income and outlays data. We had been expecting another round of stimulus checks, but lawmakers’ comments even before the blockbuster employment report suggested one may not be forthcoming. Some of the savings activity was forced on homebound consumers, and some pent-up demand will surely be unleashed as the economy re-opens. Households amassed a mighty savings war chest across March and April, however, and it has left them better-positioned to service their debt obligations going forward. Despite an unemployment rate not seen since FDR, households made their scheduled payments in April. According to TransUnion, delinquency rates fell month-over-month across every major consumer loan category and delinquency rates for mortgages and unsecured personal loans declined on a year-over-year basis (Table 2). The TransUnion data comes from its inaugural Monthly Industry Snapshot, intended to provide a higher-frequency read on headline consumer credit metrics than its typical quarterly releases. In addition to crunching the delinquency numbers, the report noted that forbearance programs have helped ease consumer liquidity pressures, consumers have reduced their outstanding credit card balances and credit scores have slightly improved. None of the factors is decisive on its own, but they contribute to a marginally improved consumer credit outlook. … And Apartment Tenants Are Paying Their Rent It is more common for households in the lower half of the income and net worth distributions to rent their residence than own it. Just one in every five households in the bottom two quintiles of the income distribution (Chart 4, top panel), and one in four in the bottom half of the net worth distribution (Chart 4, bottom panel), have a mortgage. Rent is the single largest recurring expense for these households and the shutdowns made paying it a concern. Several newspaper stories have highlighted the plight of distressed renters while discussing grassroots rent-strike movements, but the National Multifamily Housing Council’s (NMHC) Rent Payment Tracker tells a different story.3 Chart 4Households In The Lower Half Of The Income And Wealth Distributions Rent Their Homes The Rent Payment Tracker distills the results of a national survey covering over 11 million professionally managed apartment units. Through May 27th, it reported that 93.3% of renters had made full or partial payments for the month of May. The share of paying tenants was down just 150 basis points year-over-year, and up 160 basis points month-over-month. The six apartment REITs in the S&P 500 reported April and May rent collections that were better than the NMHC data. By the end of May, the REITs had collected 94-99% of the April rent they were due, and 93-96% of their May rents (Table 3). (Equity Residential (EQR) reported its April collections through April 7th and did not provide an end-of-month update; on June 1st, it reported that its May collections through May 7th were in line with April’s.) Essex Property Trust (ESS), which owns a portfolio of apartments in southern California, the Bay Area and greater Seattle, provided a table showing how the economic impact payments and the supplemental unemployment benefit would affect the income of unemployed California and Washington state couples without children. Table 4 expands it to cover four income scenarios, illustrating just how far up the income distribution CARES Act relief stretches. Table 3Residential Tenants Are Paying Their Rent Table 4The CARES Act For Essex Property Trust Renters Who Borrows: Evidence From The Survey Of Consumer Finances Helping the households in the bottom half of the income distribution won’t materially limit credit distress across the economy if those households don’t have access to credit. The latest edition of the Fed’s triennial Survey of Consumer Finances, published in 2017, makes it clear that they do. Those households may be much less likely to carry mortgage debt (Chart 5), but they make up for it by borrowing via other channels. 64% of households in the bottom two quintiles have some debt, and the share grows to 70% when the middle quintile, which qualified for the full $1,200 economic impact payment, is included (Chart 6). Chart 5The Homeownership Income Divide Chart 6Households In The Lower Two Quintiles Have Debt To Service, Too Investment Implications The discussion above focused solely on the consumer, as we discussed the Fed’s efforts to assist lenders and business borrowers in a joint Special Report with our US Bond Strategy colleagues in April.4 Record corporate bond issuance in March and April – before the Fed bought a single corporate bond – testifies to the effectiveness of the Fed’s measures. Its corporate credit facilities bazooka was so large that it was able to soothe the roiled corporate issuance market without firing a single shot. Spreads have narrowed across the spread product spectrum and the primary and secondary markets are once again able to function normally. Too much economic improvement could be self-limiting, and the S&P 500 is trading at an ambitious multiple. We remain equal weight equities over the tactical three-month timeframe. The foregoing review of consumer performance reinforces our view that the SIFI banks should be overweighted relative to the S&P 500. The ongoing data indicate that the SIFI banks will not have to build up their reserves for loan losses as much as investors feared. Our conviction that the SIFI banks are unlikely to face material book value declines has only increased. It has become possible that second- and third-quarter reserve builds may be even less than our optimistic two-times-the-first-quarter view, but the virus will have the final say. The SIFI banks remain our favorite long idea. At the asset allocation level, we remain equal weight equities over the tactical three-month timeframe. We are encouraged by the green shoots visible in the employment report, but stocks are generously valued and the virus outlook is still unclear. The improvement on the ground could prove to be self-limiting if it kills the momentum for further fiscal assistance, or if it encourages officials and individuals to let their guard down regarding the social distancing measures that have been effective in lowering COVID-19 infection rates.   Doug Peta, CFA Chief US Investment Strategist dougp@bcaresearch.com Footnotes 1 According to the Census Department’s annual Current Population Survey, mean household income ($90,000) exceeded median household income ($63,000) by 42% in 2018. 2 Ganong, Peter, Noel, Pascal J., Vavra, Joseph S. "US Unemployment Insurance Replacement Rates During the Pandemic," NBER Working Paper No. 27216. 3https://www.nmhc.org/research-insight/nmhc-rent-payment-tracker/ Accessed June 1. 4 Please see the April 14, 2020 US Investment Strategy/US Bond Strategy Special Report, "Alphabet Soup: A Summary Of The Fed’s Anti-Virus Measures," available at usis.bcaresearch.com.

Our reinstated long S&P oil & gas exploration & production (E&P)/short global gold miners pair trade is up again near the 20% mark. This parabolic rise compels us to re-institute a 10% rolling stop in order to protect gains. Importantly, neither the macro backdrop nor relative profit fundamentals have changed. A rising number of states and countries are setting the groundwork to reopen their economies. This should absorb some of the excess oil supply and help to further steepen the yield curve. Taken together, this will cement the handoff from liquidity to growth and thus further propel the pair trade (see chart). In addition, the Fed’s determination to quash volatility was another reason underpinning this intra-commodity pair trade. The lower the VIX falls, the higher the share price ratio goes. Bottom Line: Institute a 10% rolling stop in the reinstated long S&P oil & E&P/short global gold miners pair trade, today. For a full discussion on the rationale behind the trade, please refer to the following Weekly Report.