BCA Indicators/Model

Highlights The EU’s €750 billion fiscal package, along with another round of US stimulus likely exceeding $1 trillion, will support global oil demand. On the supply side, OPEC 2.0’s production discipline likely holds, and US shale output will remain depressed. These fundamentals, along with a weakening USD, will continue to support Brent prices, which are up 129% from their lows in April. China’s record-setting crude-oil-import surge during the COVID-19 pandemic – averaging 12.7mm b/d in 1H20, up 28.5% y/y – is at risk of slowing in 2H20, as domestic storage fills. Supply-side risks are acute: Massive OPEC 2.0 spare capacity – which could exceed 6mm b/d into 2021 – will tempt producers eager to monetize these to boost revenue. On the demand side, COVID-19 infection rates are surging in the US. Progress on vaccines notwithstanding, politically intolerable public-health risks in big consuming markets could usher in demand-crushing lockdowns again. Economic policy uncertainty remains elevated globally, but the balance of risks continues to favor the upside: We expect 2H20 Brent prices to average $44/bbl, and 2021 prices to average $65/bbl, unchanged from last month’s forecast. Feature We are marginally lifting our forecast of average 2020 Brent prices to $43/bbl, with 2H20 expected to average $44/bbl, and $65/bbl next year, unchanged from June. Marginal improvements to preliminary supply and demand estimates earlier in the COVID-19 pandemic support the thesis that fundamentals will not derail the massive oil-price rally that lifted Brent 129% from its April 21 low of $19.30/bbl. A weakening US dollar, and the expectation this trend will continue, also is supportive to commodities in general, oil in particular. As a result, we are marginally lifting our forecast of average 2020 Brent prices to $43/bbl, with 2H20 expected to average $44/bbl, and $65/bbl next year, unchanged from June (Chart of the Week). The three principal oil-market data providers – the US EIA, IEA and OPEC – raised demand estimates at the margin for 1H20, particularly for 2Q20, the nadir for global oil consumption. The EIA’s estimate for 2Q20 demand shows an upward revision of 550k b/d from last month’s estimate. On the supply side, the EIA estimates global output fell -8.1mm b/d in 2Q20, a -300k b/d downward revision vs. its estimate from last month (Chart 2). Chart of the WeekOil Price Rally Remains Intact Oil Price Rally Remains Intact Oil Price Rally Remains Intact Chart 2OPEC 2.0, US Shale Production Cuts Deepen OPEC 2.0, US Shale Production Cuts Deepen OPEC 2.0, US Shale Production Cuts Deepen We continue to expect the drawdown in storage levels to flatten – and then backwardate – the forward curves for Brent and WTI. After accounting for this better-than-expected fundamental performance, we now expect global supply to fall 5.9mm b/d in 2020 and to increase 4.2mm b/d in 2021. On the demand side, we now expect 2020 demand to fall 8.1mm b/d vs. 8.9mm b/d last month, and for 2021 demand to rise 7.8mm b/d vs 8.5mm b/d in June (Chart 3). This will keep the physical deficit we’ve been forecasting for 2H20 and 2021 in place, allowing OECD storage to fall to 3,026mm barrels by year-end and to 2,766mm barrels by the end of next year (Chart 4). Chart 3Supply-Demand Balances Tighten ... Supply-Demand Balances Tighten ... Supply-Demand Balances Tighten ... Chart 4... Leading To Deeper Storage Draws ... ... Leading To Deeper Storage Draws ... ... Leading To Deeper Storage Draws ... We continue to expect the drawdown in storage levels to flatten – and then backwardate – the forward curves for Brent and WTI (Chart 5). One caveat, though: We are watching floating storage levels closely, particularly in Asia: The current structure of the Brent forwards does not support carrying floating inventory, but it’s been slow moving lower (Chart 6). This could reflect a slowing in China’s crude-oil import surge, which hit record levels in May and June. Chart 5... And More Backwardation In Brent And WTI Forwards ... ... And More Backwardation In Brent And WTI Forwards ... ... And More Backwardation In Brent And WTI Forwards ... Chart 6… Even As Floating Storage In Asia Remains Elevated Balance Of Oil-Price Risk Remains To The Upside Balance Of Oil-Price Risk Remains To The Upside China’s Crude-Import Binge Ending? There is a non-trivial risk China’s crude-buying binge during the COVID-19 pandemic, which supported prices during the brief Saudi-Russian market-share war in March and the collapse in global demand in 2Q20, may have run its course (Chart 7).1 At the depths of the global pandemic in 2Q20, China’s year-on-year (y/y) crude imports surged 15%. According to Reuters, China’s crude oil imports totaled 12.9mm b/d in June, a record level for the second month in a row.2 Much of this was converted to refined products – chiefly gasoline and diesel fuel – as China’s demand recovered from the global pandemic (Chart 8). China’s 208 refineries can process 22.3mm b/d of crude, according to the Baker Institute at Rice University in Houston.3 Refinery runs in June were estimated at just over 14mm b/d by Reuters. Chart 7China's Crude Import Binge Stalls China's Crude Import Binge Stalls China's Crude Import Binge Stalls Chart 8China's Refiners Lift Runs As Imports Surge China's Refiners Lift Runs As Imports Surge China's Refiners Lift Runs As Imports Surge A reduction in China’s crude imports would force barrels to either remain on the water until refiners find a need for it, or demand for refined products increases in the region. China imports its oil into 59 port facilities, which can process ~ 16mm b/d. Storage is comprised of 74 crude oil facilities holding ~ 706mm barrels, and 213 refined-product facilities with capacity to hold ~ 357mm barrels of products (Map 1). By Reuters’s count, ~ 2mm b/d of crude went into storage in the January-May period, while close to 2.8mm b/d was stored in June. Official storage data is a state secret, so it is not possible to determine whether China’s crude and product storage is full. However, if crude oil imports remain subdued – and floating storage in Asia remains elevated – we would surmise the Chinese storage facilities are close to full. Additionally, any sharp and sustained increase in refined product exports would indicate storage is brimming. Map 1Baker Institute China Oil Map Balance Of Oil-Price Risk Remains To The Upside Balance Of Oil-Price Risk Remains To The Upside A reduction in China’s crude imports would force barrels to either remain on the water until refiners find a need for it, or demand for refined products increases in the region. We expect the latter condition to obtain, in line with our expectation of a global recovery in demand, even though China remains out of sync with the rest of the world presently. China was the first state to confront the pandemic and first to emerge out of it; its trading partners still are in various stages of recovery (Chart 9). Chart 9China's Demand Recovery Likely Will Be Choppy China's Demand Recovery Likely Will Be Choppy China's Demand Recovery Likely Will Be Choppy OPEC 2.0’s Remains Sensitive To Demand Fluctuations OPEC 2.0’s leaders – the Kingdom of Saudi Arabia (KSA) and Russia – also managed to secure additional “compensation” cuts from members that have missed their targets in previous months. The asynchronous recovery in global oil demand poses a unique problem for OPEC 2.0 this year and next. OPEC 2.0 will be easing production curtailments to 7.7mm b/d beginning in August from 9.6mm b/d in July, on the advice of its Joint Ministerial Monitoring Committee (JMMC). This is a decision that will be closely monitored, amid rising concern over the speed of demand recovery in the US and EM economies, due to mounting COVID-19 cases (Chart 10). The surge in US infections relative to its trading partners is of particular concern, given the size of US oil demand (Chart 11). In 2H20, we expect US demand will account for close to 20% of global demand, much the same level it was prior to the pandemic (Table 1). Chart 10COVID-19 Infections Surge In The US Balance Of Oil-Price Risk Remains To The Upside Balance Of Oil-Price Risk Remains To The Upside Chart 11US COVID-19 Infections Are A Risk To Global Commodity Demand Balance Of Oil-Price Risk Remains To The Upside Balance Of Oil-Price Risk Remains To The Upside Table 1BCA Global Oil Supply - Demand Balances (MMb/d, Base Case Balances) Balance Of Oil-Price Risk Remains To The Upside Balance Of Oil-Price Risk Remains To The Upside OPEC 2.0’s leaders – the Kingdom of Saudi Arabia (KSA) and Russia – also managed to secure additional “compensation” cuts from members that have missed their targets in previous months, bringing the actual increase in production closer to 1-1.5mm b/d. Together, Iraq, Nigeria, Kazakhstan, and Angola, over-produced versus their May and June targets by ~ 760k b/d. In our balances estimates, as is our normal practice, we haircut these estimates and use a lower compliance level that those stated in the official OPEC 2.0 agreement. In the case of these producers, we assume they will compensate for ~ 70% of their overproduction, bringing the adjusted cuts to ~ 8.3mm b/d. This should be sufficient to maintain the current supply deficit in oil markets that continues to support Brent prices above $40/bbl. However, the reliance on laggards’ extra cuts to balance markets adds instability. There is a lot of supply on the sidelines from the OPEC 2.0 cuts and the restart of the Neutral Zone shared by Saudi Arabia and Kuwait. The JMMC is continually assessing supply-demand balances and remains focused on making sure the totality of the cuts does not fall on a small group of countries. It reiterated its position that “achieving 100% conformity from all participating Countries is not only fair, but vital for the ongoing rebalancing efforts and to help deliver long term oil market stability.” In June, OPEC 2.0’s overall compliance was 107% – mostly reflecting over-compliance from KSA, the UAE, and Kuwait.4 There is a lot of supply on the sidelines from the OPEC 2.0 cuts and the restart of the Neutral Zone shared by Saudi Arabia and Kuwait. The US EIA estimates that within the original OPEC cartel spare capacity will average close to 6mm b/d this year, the first time since 2002 that it has exceeded 5mm b/d. On top of this, there’s the looming downside risk of a new Iran deal if Democrats win the White House and Congress in US elections in November, and a possible restart of Libyan exports this year. Watch The DUCs In The US With WTI prices averaging $41/bbl so far in July, we continue to expect part of previously shut-in US production to come back on line in July, August and September. Nonetheless, the negative effect of the multi-year low rig count will be felt heavily in 4Q20 and 1Q21 and will push production lower. The rig count appears to be bottoming but is not expected to increase meaningfully until WTI prices move closer to $45-50/bbl. On average it takes somewhere between 9-12 months for the signal from higher prices to result in new oil production flowing to market in the US. As the rig count moves back up in 2021, its effect on production will be apparent only in late-2021. However, the massive inventory of drilled-but-uncompleted (DUC) wells in the main US tight-oil basins will provide a source of cheaper new supply, if WTI prices remain above $40/bbl. DUCs are 30-40% cheaper to complete compared to drilling a new well from start. We expect DUCs completion will begin adding to US crude output in 1Q21, and that this will continue to be a source of supply beyond 2021. Bottom line: Global economic policy uncertainty remains elevated, albeit off its recent highs (Chart 12). We expect this uncertainty to continue to wane, which will allow the USD to continue to weaken. This will spur global oil demand, and will augment the fiscal and monetary stimulus to the COVID-19 pandemic undertaken globally. Chart 12Global Policy Uncertainty Remains High, Which Could Support USD Demand Balance Of Oil-Price Risk Remains To The Upside Balance Of Oil-Price Risk Remains To The Upside Nonetheless, the global recovery remains out of sync, which complicates OPEC 2.0’s production management, and markets’ estimation of supply-demand balances. Uneven success in combating the pandemic keeps the risk of lockdowns on the radar in the US. Policy is driving oil production at present, and, given the temptation to monetize spare capacity, the supply side remains a risk to prices. We continue to see upside risk dominating the evolution of prices and are maintaining our expectation Brent prices will average $44/bbl in 2H20 – lifting the overall 2020 average to $43/bbl – and $65/bbl next year. Our expectation WTI will trade $2-$4/bbl below Brent also remains intact.   Robert P. Ryan Chief Commodity & Energy Strategist rryan@bcaresearch.com Hugo Bélanger Associate Editor Commodity & Energy Strategy HugoB@bcaresearch.com Fernando Crupi Research Associate Commodity & Energy Strategy FernandoC@bcaresearch.com     Commodities Round-Up Energy: Overweight Canadian oil production averaged 4.6mm b/d in 2Q20 vs. 5.5mm b/d in 2Q19, based on EIA estimates. The lack of demand from US refiners – crude imports from Canada fell by 420k b/d y/y during the quarter – and close to maxed-out local storage facilities pushed prices below cash costs, forcing the shut-ins of more than 1mm b/d of crude production. Canadian energy companies started releasing their 2Q20 earnings this week and analysts expect the results to be one of the worst ever recorded, reflecting the extent of the pain producers felt during the COVID-19 shock. Base Metals: Neutral High-grade iron ore prices (65% Fe) were trading above $120/MT this week, on the back of forward guidance from the commodity’s top exporter, Brazilian miner Vale, which suggested exports will be lower than had been previously estimated this year, according to Fastmarkets MB, a sister service of BCA Research. This is in line with an Australian Department of Industry, Science, Energy and Resources analysis in June, which noted, “The COVID-19 pandemic appears to have affected both sides of the iron ore market: demand disruptions have run up against supply problems localised in Brazil, where COVID-19-related lockdowns have derailed efforts to recover from shutdowns in the wake of the Brumadinho tailings dam collapse” (Chart 13). Precious Metals: Neutral Our long silver position is up 17.5% since it was recommended July 2. We are placing a stop-loss on the position at $21/oz, our earlier target, given the metal was trading ~ $22/oz as we went to press. The factors supporting gold prices – chiefly low real rates in the US, a weakening dollar and global monetary accommodation, also support silver prices. However, silver also will benefit from the recovery in industrial activity and incomes we anticipate in the wake of global fiscal and monetary stimulus, which will drive demand for consumer products (Chart 14). Ags/Softs:  Underweight Lumber prices have more than doubled since April lows. The uncertainty brought by the COVID-19 health emergency altered the perception of future housing demand and, by extension, lumber demand, to the point that mills responded by substantially decreasing capacity utilization rates. However, in the wake of global monetary and fiscal stimulus, housing weathered the storm better than expected. Furthermore, a surge in DIY projects from individuals working from home at a time of reduced supply contributed to the current state of market shortage. Chart 13Lower Supply Supports Iron Ore Prices Lower Supply Supports Iron Ore Prices Lower Supply Supports Iron Ore Prices Chart 14Silver Favored Over Gold Silver Favored Over Gold Silver Favored Over Gold         Footnotes 1     In our reckoning, a non-trivial risk is something greater than Russian roulette odds – i.e., a 1-in-6 chance of an event occuring. Re the ever-so-brief Saudi-Russian market-share war, please see KSA, Russia Will Be Forced To Quit Market-Share War, which we published March 19, 2020. It is available at ces.bcaresearch.com. 2     Please see COLUMN-China's record crude oil storage flies under the radar: Russell published by reuters.com July 20, 2020. 3    The Baker Institute’s Open-Source Mapping of China's Oil Infrastructure was last updated in March 2020. The map is “a beta version and is likely missing some pieces of existing infrastructure. The challenge of China’s geographic expanse — it is roughly the same area as the U.S. Lower 48 — is compounded by a lack of transparency on the part of China’s government,” according to the Baker Institute. 4    In our supply-side estimates, we used IEA estimates of cuts for June this month. This doesn’t change the overall estimate of cuts from our earlier analysis; however, it slightly changes how the 9.7mm b/d was split between OPEC 2.0 members. the official eased cuts are 7.7mm b/d from 9.7mm b/d in May-June-July, but it actually is closer to 8.3mm b/d accounting for the compensation from the countries mentioned above.   Investment Views and Themes Recommendations Strategic Recommendations Tactical Trades Trade Recommendation Performance In 2020 Q2 Balance Of Oil-Price Risk Remains To The Upside Balance Of Oil-Price Risk Remains To The Upside Commodity Prices and Plays Reference Table Trades Closed in 2020 Summary of Closed Trades Balance Of Oil-Price Risk Remains To The Upside Balance Of Oil-Price Risk Remains To The Upside

  Highlights Q2/2020 Performance Breakdown: Our recommended model bond portfolio outperformed the custom benchmark by +11bps during the second quarter of the year. Winners & Losers: The government bond side of the portfolio outperformed by +8bps, led by overweights in the US (+4bps), Canada (+4bps) and Italy (+3bps). Spread product generated a small outperformance (+3bps), with overweights in US investment grade (+43bps) offsetting underweights in emerging market debt (-35bps). Scenario Analysis For The Next Six Months: We are sticking close to benchmark on overall duration and spread product exposure, focusing more on relative value between countries and sectors to generate outperformance amid economic uncertainties caused by the growing spread of COVID-19. We continue favoring markets where there is direct buying from central banks, but we are also increasing our recommended exposure to EM USD-denominated debt versus US investment grade corporates. Feature The first half of 2020 has been one of rapid market moves and regime shifts for global fixed income markets. In the first quarter, developed market government debt provided the best returns as bond yields plunged with central banks racing to support collapsing economies through rate cuts and liquidity injections. In Q2, corporate credit delivered the top returns, as economies started to emerge from the COVID-19 lockdowns and, more importantly, the Fed and other major central banks delivered direct support to frozen credit markets through asset purchases. Now, even as an increasing number of global growth indicators are tracing out a "V"-shaped recovery, new cases of COVID-19 are surging though the southern US and major emerging economies like Brazil and India. This raises new challenges for investors for the second half of 2020. A second wave of the coronavirus could jeopardize the nascent global economic recovery, even after the massive easing of monetary and fiscal policies, at a time when valuations on many risk assets appear stretched. In this report, we review the performance of the BCA Research Global Fixed Income Strategy (GFIS) model bond portfolio during the second quarter of 2020. We also present our recommended portfolio positioning for the next six months. Given the lingering uncertainties from the renewed spread of COVID-19, we continue to take a more measured approach in our portfolio allocations. That means focusing more on relative value between countries and sectors while staying closer to benchmark on overall global duration and spread product exposure versus government bonds (Table 1). Table 1GFIS Model Bond Portfolio Recommended Positioning For The Next Six Months GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism As a reminder to existing readers (and to new clients), the model portfolio is a part of our service that complements the usual macro analysis of global fixed income markets. The portfolio is how we communicate our opinion on the relative attractiveness between government bond and spread product sectors. We do this by applying actual percentage weightings to each of our recommendations within a fully invested hypothetical bond portfolio. Q2/2020 Model Portfolio Performance Breakdown: Slight Outperformance For Both Sovereigns And Credits Chart 1Q2/2020 Performance: Modest Gains From Relative Positioning Q2/2020 Performance: Modest Gains From Relative Positioning Q2/2020 Performance: Modest Gains From Relative Positioning The total return for the GFIS model portfolio (hedged into US dollars) in the second quarter was 3.22%, modestly outperforming the custom benchmark index by +11bps (Chart 1).1 In terms of the specific breakdown between the government bond and spread product allocations in our model portfolio, the former generated +8bps of outperformance versus our custom benchmark index while the latter outperformed by +3bps. That government bond return includes the small gain (+2bps) from inflation-linked bonds, which we added as a new asset class in our model portfolio framework on June 23.2 In a world of very low bond yields (Table 2), our preference for the higher-yielding government bond markets in the US, Canada, the UK and Italy was the main source of outperformance, delivering a combined excess return of +13bps (including inflation-linked bonds). Our underweight in Japan delivered a surprising positive excess return of +4bps as longer-dated JGB yields – which do not fall under the Bank of Japan’s yield curve control policy – rose during the quarter. Underweights in the low-yielding core euro area countries of Germany and France were a drag on the portfolio (a combined -10bps), particularly the latter where longer-maturity French bonds enjoyed a very strong rally in Q2. Table 2GFIS Model Bond Portfolio Q2/2020 Overall Return Attribution GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism In spread product, our overweights in US investment grade corporates (+43bps), UK investment grade corporates (+7bps) and US commercial MBS (+5bps) squeezed out a combined small gain versus underweights in emerging markets (EM) USD-denominated credit (-35bps), euro area high-yield (-8bps) and lower-rated US high-yield (-6bps). In a world of very low bond yields (Table 2), our preference for the higher-yielding government bond markets in the US, Canada, the UK and Italy was the main source of outperformance. That modest outperformance of the model bond portfolio versus the benchmark is in line with our cautious recommended stance on what are always the largest drivers of the portfolio returns: overall duration exposure and the relative allocation between government debt and spread product. We have stuck close to benchmark exposures on both, eschewing big directional bets on bond yields or credit spreads while focusing more on relative opportunities between countries and sectors. This conservative approach is how we are approaching what we have dubbed “The Battle of 2020” between the opposing forces of coronavirus contagion (which is bullish for government bonds and bearish for credit) and policy reflation (vice versa).3 The bar charts showing the total and relative returns for each individual government bond market and spread product sector are presented in Charts 2 & 3. Chart 2GFIS Model Bond Portfolio Q2/2020 Government Bond Performance Attribution GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism Chart 3GFIS Model Bond Portfolio Q2/2020 Spread Product Performance Attribution By Sector GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism The most significant movers were: Biggest Outperformers Overweight US investment grade industrials (+28bps) Overweight US investment grade financials (+12bps) Overweight UK investment grade corporates (+7bps) Overweight US CMBS (+5bps) Underweight Japanese government bonds with maturity greater than 10 years (+5 bps) Biggest Underperformers Underweight EM USD denominated corporates (-24bps) Underweight EM USD denominated sovereigns (-10bps) Underweight EUR high-yield corporates (-8bps) Underweight French government bonds with maturity greater than 10 years (-5bps) Underweight US B-rated high-yield corporates (-4bps) Chart 4 presents the ranked benchmark index returns of the individual countries and spread product sectors in the GFIS model bond portfolio for Q2/2020. Returns are hedged into US dollars (we do not take active currency risk in this portfolio) and adjusted to reflect duration differences between each country/sector and the overall custom benchmark index for the model portfolio. We have also color coded the bars in each chart to reflect our recommended investment stance for each market during Q2/2020 (red for underweight, dark green for overweight, gray for neutral).4 Ideally, we would look to see more green bars on the left side of the chart where market returns are highest, and more red bars on the right side of the chart were returns are lowest. Chart 4Ranking The Winners & Losers From The GFIS Model Bond Portfolio In Q2/2020 GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism The top performing sectors in our model bond portfolio universe in Q2 were all spread product: EM USD-denominated sovereign (+12.9% in USD-hedged terms, duration-matched to the custom model portfolio benchmark index), EM USD-denominated corporate debt (+12.6%), UK investment grade corporates (+11.3%), US investment grade corporates (+10.9%), and high-yield corporates in the euro area (+6.7%) and US (+5.6%). The top performing sectors in our model bond portfolio universe in Q2 were all spread product. During the quarter, we maintained relative exposures to those sectors within an overall small above-benchmark allocation to global spread product – overweight US and UK investment grade versus underweight emerging market credit, neutral overall US high-yield (favoring Ba-rated debt) versus underweight euro area high-yield. Those allocations were motivated by our theme of “buying what the central banks are buying”, like the Fed purchasing US investment grade corporates. Importantly, we had limited exposure to the worst performing sectors during Q2: underweight government bonds in Japan (index return of -0.47% in USD-hedged, duration-matched terms) and Germany (+0.47%), a neutral allocation to Australian sovereign debt (-0.07%) and an underweight in US Agency MBS (+0.20%). The latter two positions came after we downgraded US MBS to underweight in early April and cut our long-held overweight in Australia to neutral in mid-May. Bottom Line: Our model bond portfolio modestly outperformed its benchmark index in the second quarter of the year by +11bps – a positive result driven by our relative positioning that favored higher yielding government debt and spread product sectors directly supported by central bank purchases. Future Drivers Of Portfolio Returns Chart 5Overall Portfolio Allocation: Slightly Overweight Credit Vs Governments GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism Typically, in these quarterly performance reviews of our model bond portfolio, we make return forecasts for the portfolio based off scenario analysis and quantitative predictions of various fixed income asset classes. However, the current environment is unprecedented because of the COVID-19 outbreak. Not only is there now elevated economic uncertainty, but central banks are running extreme monetary policies in response - including direct intervention in markets through purchases of both government bonds and spread product. Thus, we are reluctant to rely on historical model coefficients and correlations to estimate expected fixed income returns. Instead, we will focus on the logic behind our current model portfolio allocations and the expected contribution to overall portfolio performance over the next six months. At the moment, the main factors that will drive the performance of the model bond portfolio over the next six months are the following: Our recommended overweight stance on relatively higher-yielding sovereigns like the US, Canada and Italy versus low-yielders like Germany, France and Japan; Our allocation to inflation-linked bonds out of nominal government debt in the US, Italy and Canada; Our recommended overweight stance on spread product backstopped by central bank purchases - US investment grade corporates, US Agency CMBS, US Ba-rated high-yield, and UK investment grade corporates; Our recommended underweight stance on riskier spread product - euro area high-yield, US B-rated and Caa-rated high-yield, and EM USD-denominated corporates and sovereigns. The portfolio currently has a small aggregate overweight allocation to spread product relative to government bonds, equal to three percentage points (Chart 5). We feel that is an appropriate allocation to credit versus sovereigns in an environment that is still highly uncertain concerning the spread of COVID-19 and how global growth will evolve over the next 6-12 months. This also leaves room to increase the spread product allocation should the news on the virus and the global economy take a turn for the better. We also remain neutral on overall portfolio duration exposure. Our Global Duration Indicator, which contains growth data like our global leading economic indicator and the global ZEW expectations index, has rebounded sharply and is signaling that bond yields should bottom out in the second half of 2020 (Chart 6). A rise in yields will take longer to develop, however, with virtually all major central banks signaling that policy rates will stay near 0% for an extended period. Chart 6Our Global Duration Indicator Says Bond Yields Will Bottom Out In H2/2020 Our Global Duration Indicator Says Bond Yields Will Bottom Out In H2/2020 Our Global Duration Indicator Says Bond Yields Will Bottom Out In H2/2020 Chart 7Within Governments, Overweight Inflation-Linked Bonds Vs. Nominals Within Governments, Overweight Inflation-Linked Bonds Vs. Nominals Within Governments, Overweight Inflation-Linked Bonds Vs. Nominals The recent moves in developed market government bonds are interesting in terms of the underlying drivers of yields – real yields and inflation expectations. Longer-maturity inflation breakevens – the spread between the yields of nominal and inflation-linked government debt – have drifted higher since late March after major central banks began rapidly easing monetary conditions. At the same time, the actual yields on inflation-linked bonds, i.e. real yields, have moved lower and largely offset the gains in inflation breakevens (Chart 7). Nominal yields have been stuck in very narrow ranges as a result. We do not see that dynamic changing, at least in the near term. Inflation breakevens are too low on our models across all developed markets, and are likely to continue inching higher in the coming months on the back of a pickup in global growth and rising energy prices. At the same time, central banks will be staying on hold for longer while continuing to buy large quantities of nominal bonds, helping push real yields lower. Given these opposing forces on nominal government bond yields, we think it is far too soon to contemplate reducing overall duration – even with equity and credit markets having rallied sharply off the lows and global economic indicators rebounding. Thus, we are maintaining an overall duration exposure close to benchmark in the model portfolio (Chart 8). At the same time, we are playing for wider breakevens and lower real bond yields through allocations to markets where our models indicate better value in being long breakevens: US TIPS, Italian inflation-linked BTPs, and Canadian Real Return Bonds. Within the government bond side of the model bond portfolio, we continue to recommend focusing more on country allocation to generate outperformance. That means concentrating exposures in relatively higher yielding markets like the US, Canada and Italy while maintaining underweights in low-yielding core Europe and Japan. Turning to spread product allocations, we continue to recommend focusing more on policymaker responses to the COVID-19 recession, and its uncertain recovery, rather than the downturn itself. The now double-digit year-over-year growth in global central bank balance sheets - which has led global high-yield and investment grade excess returns by one year in the years after the Global Financial Crisis (Chart 9) – is pointing to additional global corporate bond market outperformance versus governments over the next 6-12 months. Chart 8Overall Portfolio Duration: Close To Benchmark Overall Portfolio Duration: Close To Benchmark Overall Portfolio Duration: Close To Benchmark In other words, we are focusing on global QE rather than global recession, while maintaining a modest recommended overall weighting on global spread product. That allocation could be larger, but we suggest picking the lowest hanging fruit in the credit universe rather than going for the highest beta credit markets like Caa-rated US high-yield that have already seen significant spread compression relative to higher-rated US junk bonds (bottom panel). Chart 9Global QE Supporting Credit Markets Global QE Supporting Credit Markets Global QE Supporting Credit Markets Chart 10Overall Credit Allocation: Keep Buying What The Central Banks Are Buying Overall Credit Allocation: Keep Buying What The Central Banks Are Buying Overall Credit Allocation: Keep Buying What The Central Banks Are Buying We continue to focus our recommended spread product allocations on the parts of global credit markets where central banks are directly buying. We continue to focus our recommended spread product allocations on the parts of global credit markets where central banks are directly buying (Chart 10). In the US, that means overweighting US investment grade corporate bonds (particularly those with maturities of less than five years), US Ba-rated high-yield that the Fed can hold in its corporate bond buying program, US Agency CMBS that is also supported by Fed programs, and UK investment grade corporate bonds that the Bank of England is buying. We also put Italian government bonds into this category, with the ECB buying greater amounts of BTPs as part of its COVID-19 monetary support efforts. What about emerging market debt? We have expressed reservations in recent months about upgrading EM USD-denominated sovereign and corporate debt, even within our portfolio theme of being “selectively opportunistic” about recommended spread product allocations. We have long felt that the time to buy those markets would be when the US dollar had clearly peaked and global growth had clearly bottomed. The latter condition now appears to be in place, and the strong upward momentum in the US dollar is starting to weaken. This forces us to reconsider our stance on EM debt in the model portfolio. Even after the powerful Q2 rally in EM corporate and sovereign debt, EM credit spreads still look relatively attractive using one of our favorite credit valuation metrics – the percentile rankings of 12-month breakeven spreads. Those breakeven spreads are calculated, as the amount of spread widening that would make the return of EM credit equal to duration-matched US Treasuries over a 12-month horizon. We then compare those spreads to their own history to determine how attractive current spread levels are now on a “spread volatility adjusted” basis. Current 12-month breakeven spreads for EM USD-denominated sovereigns and corporates are in the upper quartile of their own history. This compares favorably to other spread products in our model bond portfolio universe, particularly US investment grade corporates where the 12-month breakevens are now just below the long-run median (Chart 11). Chart 11A Comparison Of Credit Sectors Using 12-Month Breakeven Spreads GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism The current Bloomberg Barclays EM corporate benchmark index option-adjusted spread (OAS) is around 300bps above that of the US investment grade corporate index OAS. That spread still has room to compress further if global growth continues to rebound and the US dollar softens versus EM currencies. Leading growth indicators like the China credit impulse, which has picked up sharply as Chinese authorities have ramped up economic stimulus measures, are now back to levels last seen in 2016 when EM credit strongly outperformed US investment grade corporates (Chart 12). Chart 12Upgrade EM Credit Versus US Investment Grade Upgrade EM Credit Versus US Investment Grade Upgrade EM Credit Versus US Investment Grade Chart 13Overall Portfolio Yield: Close To Benchmark Overall Portfolio Yield: Close To Benchmark Overall Portfolio Yield: Close To Benchmark This week we are upgrading our weighting on EM USD-denominated corporates and sovereigns to neutral, from underweight, in our model bond portfolio. Although we acknowledge that the EM story has been made more complicated by the rapid spread of COVID-19 through the major EM economies, an underweight stance – particularly versus US investment grade credit – is increasingly unwarranted. Therefore, this week we are upgrading our weighting on EM USD-denominated corporates and sovereigns to neutral, from underweight, in our model bond portfolio (see the updated table on pages 17-18). That new allocation will be “funded” by reducing our overweight in US investment grade corporates. Model bond portfolio yield and tracking error considerations Importantly, the selective global government bond and credit allocations we have just outlined do not come at a cost in terms of forgone yield. The portfolio yield after our upgrade of EM debt will be slightly above that of the custom benchmark index (Chart 13), indicating no “negative carry” even when avoiding parts of the US and euro area high-yield markets. Chart 14Overall Portfolio Risk: Moderate Overall Portfolio Risk: Moderate Overall Portfolio Risk: Moderate Finally, turning to the risk budget of the model portfolio, we are aiming for a “moderate” overall tracking error, or the gap between the portfolio’s volatility and that of the benchmark index. The portfolio volatility has fallen dramatically from the surge seen during the global market rout in March, moving lower alongside realized market volatility. The tracking error now sits at 64bps, well below our self-imposed limit of 100bps and within the 50-70bps range we are targeting as a “moderate” level of overall portfolio risk (Chart 14). Bottom Line: We are sticking close to benchmark on overall duration and spread product exposure, focusing more on relative value between countries and sectors to generate outperformance amid economic uncertainties caused by the growing spread of COVID-19. We continue favoring markets where there is direct buying from central banks. We are also increasing our recommended exposure on EM USD-denominated debt to neutral, funded by a reduced allocation to US investment grade corporates where valuations are less attractive.   Robert Robis, CFA Chief Fixed Income Strategist rrobis@bcaresearch.com   Ray Park, CFA Research Analyst ray@bcaresearch.com Footnotes 1 The GFIS model bond portfolio custom benchmark index is the Bloomberg Barclays Global Aggregate Index, but with allocations to global high-yield corporate debt replacing very high quality spread product (i.e. AA-rated). We believe this to be more indicative of the typical internal benchmark used by global multi-sector fixed income managers. 2 Please see BCA Global Fixed Income Strategy Weekly Report, "How To Play The Revival Of Global Inflation Expectations'", dated June 23 2020, available at gfis.bcaresearch.com. 3 Please see BCA Global Fixed Income Strategy Weekly Report, "Contagion Vs. Reflation: The Battle Of 2020 Rages On", dated June 30, 2020, available at gfis.bcaresearch.com. 4 Note that sectors where we made changes to our recommended weightings during Q2/2020 will have multiple colors in the respective bars in Chart 4. Recommendations The GFIS Recommended Portfolio Vs. The Custom Benchmark Index GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism GFIS Model Bond Portfolio Q2/2020 Performance Review & Current Allocations: Selective Optimism Duration Regional Allocation Spread Product Tactical Trades Yields & Returns Global Bond Yields Historical Returns

Highlights Energy Bond Model: This report presents models for both investment grade and high-yield Energy bond excess returns. The models are based on overall corporate bond index spreads and the oil price. They can be used to generate Energy bond excess return forecasts for investment horizons up to 12 months. IG Energy Bonds: Our model suggests that investment grade Energy bond excess returns will be strong during the next 12 months under likely economic scenarios. We recommend an overweight allocation to investment grade Energy bonds.  HY Energy Bonds: Our models imply positive excess return outcomes for high-yield Energy bonds, but we remain concerned about near-term default risk for lower-rated issuers. We advise a cautious (neutral) allocation for now. Part 2 of this Special Report, to be published next week, will dig further into the high-yield Energy index on an issuer-by-issuer basis. Feature Table 1Energy Bond Excess Return* Scenarios (12-Month Investment Horizon) The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns During the past couple of months we’ve published several reports that take more detailed looks at specific industry groups within both the investment grade and high-yield corporate bond markets. So far, we’ve published reports on: Banks1 Healthcare & Pharmaceuticals2 Technology3 This week and next week, we continue our series with a deep dive into Energy bonds that is split between two Special Reports. This week’s report develops a model for Energy bond excess returns based on overall corporate bond index excess returns and the oil price. In next week’s report, we look more deeply into the characteristics of the investment grade and high-yield Energy indexes. We also consider the outlooks for the five sub-categories of Energy debt: Independent, Integrated, Oil Field Services, Refining and Midstream. A Model Of Energy Bond Excess Returns A good starting point for modeling the excess returns of any corporate bond sector is to combine the sector’s Duration-Times-Spread (DTS) ratio with the excess returns of the overall corporate bond index.4 Please note that “excess returns” refers to returns relative to a duration-matched position in Treasury securities. The DTS-only model explains 86% of the variance in monthly investment grade Energy excess returns. Considering only a sector’s DTS ratio, we can define the following model for monthly investment grade Energy excess returns: EXSENRG = (DTSENRG / DTSCORP) * EXSCORP Where: EXSENRG = Monthly investment grade Energy excess returns versus duration-matched Treasuries (DTSENRG / DTSCORP) = The investment grade Energy sector’s DTS ratio EXSCORP = Monthly investment grade corporate index excess returns versus duration-matched Treasuries For example, the current DTS for the investment grade Energy sector is 18. The DTS for the overall corporate index is 12. This means that the DTS ratio for the Energy sector is 18/12 = 1.5. According to our simple model, we would expect Energy sector excess returns to be 1.5 times corporate index excess returns in any given month. It turns out that our simple model performs quite well. Chart 1 shows monthly investment grade Energy sector excess returns versus our model’s prediction. Our sample period spans from 1997 to the present. Specifically, we find that our model explains 86% of the variance in monthly investment grade Energy excess returns. Chart 1Investment Grade Energy Monthly Excess Returns*: DTS-Only Model** The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns The simple (DTS-only) model’s performance is admirable, but we can do slightly better if we also incorporate the oil price. Chart 2 shows a statistically significant relationship between the residual from the DTS-only model and the monthly change in the Brent crude oil price. Chart 2Residual From DTS-Only Model* Versus Oil Price The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns Combining the models shown in Charts 1 and 2, we get a model for investment grade Energy monthly excess returns based on both corporate index excess returns and the oil price: EXSENRG = (DTSENRG / DTSCORP) * EXSCORP + (376.84 * ∆ ln Oil) – 1.0587 Where excess returns are measured in basis points and (∆ ln Oil) = the monthly change in the natural logarithm of the Brent crude oil price. Chart 3 shows the historical performance of this complete model. Note that the model now explains 91% of the historical variance of investment grade Energy excess returns, 5% more than the initial DTS-only model. Chart 3Investment Grade Energy Monthly Excess Returns*: Complete Model (DTS & Oil)** The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns Robustness Checks We performed the same analysis for 3-month, 6-month and 12-month excess returns and found very consistent results (Table 2). The oil price adds significant explanatory power to the model in each case, but the bulk of variation in investment grade Energy excess returns is determined by trends in the overall corporate index spread. Table 2Investment Grade Energy Excess Returns*: Model Results Using Different Return Frequencies (1997 - Present) The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns We also find consistent results when looking at high-yield Energy returns (Table 3). Once again, the bulk of excess return variation is explained by multiplying the DTS ratio and the benchmark index’s excess returns. The oil price also adds a statistically significant amount of extra explanatory power. Table 3High-Yield Energy Excess Returns*: Model Results Using Different Return Frequencies (1997 - Present) The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns One final observation is that oil explains a greater proportion of the variation in Energy sector excess returns if we limit our sample period to the past few years. Specifically, we re-ran the monthly iterations of both the investment grade and high-yield models from July 2014 to present. We found that the DTS component of the model explains the same amount of excess return variation as it did for the full sample. However, we also found that the oil price has a much greater impact if the sample is limited to the past six years (Table 4). Table 41-Month Excess Return* Models: Full Sample (1997 - Present) Versus Recent Sample (2014 - Present) The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns The Outlook For Energy Bonds Part 1: A Model Of Energy Bond Excess Returns Energy Excess Return Scenarios Finally, using our 12-month excess return models for investment grade and high-yield Energy, we can project likely outcomes for Energy excess returns versus Treasuries for the next 12 months. All we have to do is assume different outcomes for the overall benchmark index spread (either the investment grade or High-Yield index, depending on the model) and the oil price.5 The results of this scenario analysis are shown in Table 1. Starting with investment grade Energy, we see that all scenarios where the investment grade corporate index spread tightens lead to positive Energy excess returns. This is true even in a scenario where the oil price falls by $20 during the next year. Our model also suggests that a $10-$20 increase in the oil price during the next 12 months will keep Energy excess returns positive, even in a modest “risk off” scenario where the corporate index spread widens by 25 bps. All scenarios where the investment grade corporate index spread tightens lead to positive Energy excess returns. The story is similar in high-yield, though returns are much more variable. For example, high-yield Energy is projected to lose money relative to Treasuries in a scenario where the junk index spread tightens 50 bps and the oil price falls by $20. There are no scenarios where benchmark index spread tightening coincides with negative Energy excess returns in the investment grade model. Chart 4Watch For Falling Inventories Watch For Falling Inventories Watch For Falling Inventories In terms of likely scenarios for the next 12 months, we anticipate further spread tightening for corporate bonds rated Ba & above. But we also view B-rated and lower spreads as too tight given the default outlook for the next 12 months and the fact that these lower-rated issuers usually can’t access the Fed’s emergency lending facilities.6 With that in mind, we would confidently bet on investment grade index spread tightening during the next 12 months, but can envision high-yield spread widening driven by the lower credit tiers. On oil, our Commodity & Energy Strategy service forecasts an average Brent crude oil price of $65 in 2021, a sizeable increase relative to the current price of $43.27.7 Our strategists expect a significant supply contraction in the second quarter of this year that will cause the oil market to enter a physical deficit in the second half of 2020. Investors can look for falling storage levels in the coming months to confirm whether that forecast is playing out (Chart 4). Escalating tensions between the US and Iran pose an additional near-term upside risk to oil prices. This risk increased during the past few weeks as a string of mysterious explosions struck several Iranian military and economic facilities.8 However, with major oil producers now operating significantly below capacity, any net impact on oil prices from a supply disruption in the Persian Gulf would likely be short-lived. Investment Conclusions All in all, our bullish outlook for both investment grade corporate bond spreads and the oil price makes us inclined to overweight investment grade Energy bonds on a 12-month horizon. Within high-yield, our model also suggests that we should have a bullish bias toward Energy, but we remain concerned about default risk for lower-rated (B & below) Energy issuers during the next few months. We will dig into the high-yield Energy index on an issuer-by-issuer basis in Part 2 of this report, to be published next week. For now, we advise a more cautious stance toward high-yield Energy.   Ryan Swift US Bond Strategist rswift@bcaresearch.com Footnotes 1 Please see US Bond Strategy Weekly Report, “Negative Oil, The Zero Lower Bound And The Fisher Equation”, dated April 28, 2020, available at usbs.bcaresearch.com 2 Please see US Bond Strategy Weekly Report, “Assessing Healthcare & Pharma Bonds In A Pandemic”, dated June 9, 2020, available at usbs.bcaresearch.com 3 Please see US Bond Strategy Weekly Report, “Take A Look At High-Yield Technology Bonds”, dated June 23, 2020, available at usbs.bcaresearch.com 4 Duration-Times-Spread (DTS) is a simple measure that is highly correlated with excess return volatility for corporate bonds. The DTS ratio is the ratio of a sector’s DTS to that of the benchmark index. It can be thought of like the beta of a stock. A DTS ratio above 1.0 signals that the sector is cyclical (or “high beta”), a DTS ratio below 1.0 signals that the sector is defensive or (“low beta”). For more details on the DTS measure please see: Arik Ben Dor, Lev Dynkin, Jay Hyman, Patrick Houweling, Erik van Leeuwen & Olaf Penninga, “DTS (Duration-Times-Spread)”, Journal of Portfolio Management 33(2), January 2007. 5 We translate changes in benchmark index spread into 12-month excess returns using the formula: excess return = option-adjusted spread – (duration * change in option-adjusted spread) 6 Please see US Bond Strategy Weekly Report, “No Holding Back”, dated June 16, 2020, available at usbs.bcaresearch.com 7 Please see Commodity & Energy Strategy Weekly Report, “Low Vol, High Uncertainty Keeps Oil-Price Rally On Tenterhooks”, dated June 18, 2020, available at ces.bcaresearch.com 8 Please see Geopolitical Strategy Special Alert, “Cyber-Rattling In The Middle East”, dated July 10, 2020, available at gps.bcaresearch.com

Highlights Our quantitative US election model suggests Trump has a 44% chance of re-election. This presents a risk to our formal subjective view that he has a 35% chance. We are sticking with our subjective odds for now, as Trump is beset with a reviving COVID-19 outbreak, a recession, social unrest, and execution risks for the next round of fiscal stimulus. But we may increase his chances in August if his circumstances improve. In the worst case, the devastated economy will lead to a landslide in which Trump even loses Iowa. But peak political polarization makes that unlikely and suggests that the race will tighten from here. Uncertainty and volatility will rise from here through November and possibly beyond. Feature The BCA Geopolitical Strategy presidential election model was first introduced to our readers in November 2019 in order to predict and quantify the Electoral College vote outcome of the 2020 US presidential election. The election model is a state-by-state model that uses both economic and political variables in order to predict the probability of the incumbent party winning the Electoral College votes in each of the 50 states. We favored predicting the Electoral College vote over the popular vote since the winner of the presidential election is determined by the Electoral College. There have been five cases in history where the popular vote did not determine the outcome and two in recent history (George W. Bush in 2000 and Donald Trump in 2016). The college imposes a significant (and deliberate) constraint on popularity and mass movements. Our sample size includes nine elections over the period 1984-2016, across 50 states, netting 450 observations. One of our four explanatory variables, the Federal Reserve Bank of Philadelphia State Leading Index, was suspended indefinitely amid the COVID-19 crisis. Hence we needed a replacement variable that could capture a similar impact on the predicted outcome, and one that was readily available on a state-by-state basis. Enter our replacement variable: 1. The Federal Reserve Bank of Philadelphia State Coincident Index. The state leading index in our previous election model was an estimate of the six-month growth rate in the state coincident index. Therefore the state coincident index is the natural replacement variable as it will essentially proxy the state leading index, albeit without the forward-looking element. The coincident index for each state combines four of the state’s indicators to summarize current economic conditions in a single statistic. The four indicators are nonfarm payroll employment; average hours worked in manufacturing by production workers; the unemployment rate; and wage and salary disbursements plus proprietors' income deflated by the consumer price index (US city average). We applied several transformations to the data to obtain meaningful results in the modeling process. Transformations included three-month, six-month, and twelve-month changes in the state coincident indexes. Ultimately we decided to use the three-month change of the state coincident index in our updated Version 2 (V2) election model. As before, we took a weighted average of the three-month change of all the monthly state coincident indexes in the presidential term preceding the election. Later months are weighted heavier than earlier months. A significant difference from the first version of our model is that, unlike the state leading indexes, the state coincident indexes do not have leading properties that give a forward-looking “view” on what the economic environment will look like going into Q1 of the post-election year. We acknowledge that past, current, and future economic conditions are likely to weigh on voters’ minds when casting their vote, but we also note the difficulties in accurately weighting one expectation more than another. We assume that prevailing economic conditions matter most to voters (as people’s assessment of their current situation inevitably affects their future expectations, and vice versa), and this bolsters our rationale in using a 3-month change of the state coincident index. Our final calculation of three-month changes to the state coincident indexes occurs in September of the election year, given that most voters make their decision at least one month in advance of the election, as we have previously shown. The October data release will arrive too late in November for inclusion in the election forecast anyway. Our remaining explanatory variables for V2 of our model update: 2. The incumbent party’s margin of victory in the previous presidential election in each state. Same as our original model. 3. A “time for change” variable – a categorical variable indicating whether the incumbent party has been in the White House for one or more terms. Same as our original model. 4. The range of the incumbent president’s job approval rating. Our original model used the level of approval. Our V2 model excludes the average approval level of the incumbent president in July of the election year as it was found to be statistically insignificant at widely accepted significance levels (1%, 5% and 10%) when estimated with the state coincident index (as opposed to the state leading index), no matter the transformation applied to that index. This does not mean we exclude Trump’s approval data from our modeling process. Instead, we include the range of the incumbent president’s job approval rating. This was the only transformed variation in presidential job approval rating data that showed statistical significance when combined with the variables above. For V2 of our model, the range is computed as the maximum monthly average of various job approval polls less the minimum monthly average of such polls throughout a president’s term. Despite Trump’s job approval being low relative to previous presidents, he has maintained consistency. Hence the range of Trump’s job approval is fairly tight relative to previous presidents and should not be ignored in affecting the election outcome. Upside Risk To Trump’s Re-Election Odds? Chart 1 below depicts our revised prediction of November’s presidential election. Chart 1Trump Is Slated To Lose Re-election With 259 Electoral College Votes Updating Our Quantitative Election Model Updating Our Quantitative Election Model As it stands, Trump is slated to lose the election with 259 Electoral College votes (45 less than his 2016 victory). This is just ten votes shy of our previous prediction in March this year, but several swing states that were narrowly in Trump’s camp in March are now far less likely to go his way. Our previous prediction, which of course did not account for COVID-19’s economic shock, had Trump tied with the presumptive democrat nominee at the time. But the latest results still point to a tight race come November. Our updated quantitative model gives Trump a 44% chance of winning. The collapse of the state economies is overwhelming Trump’s re-election bid. Poor economic conditions hardly ever favor a sitting president up for reelection. But note that the three-month change in the state indexes will be the first to register the economic rebound this summer and fall (should it continue). This would improve Trump’s probability of victory. Under our V2 model, New Hampshire, Pennsylvania, and Wisconsin are no longer toss-up states. Rather, Florida is the only toss-up state, with a 52% probability of staying with the incumbent party. Minor negative changes to the state indexes could result in more toss-up states, even throwing traditionally red states into toss-up territory. States that are expected to turn from Republican in 2016 to Democratic in 2020 are Michigan, Pennsylvania, and Wisconsin – the entire “Blue Wall” that delivered Trump his surprise victory four years ago. On the whole, the model gives Trump a 44% chance of retaining the White House. Do we uncritically accept these results? No. As with all of our analysis, we provide a qualitative judgment in addition to our quantitative indicators and models. In general the findings make sense. We agree that Florida, Arizona, and North Carolina remain in Trump’s camp at present, if narrowly. Our qualitative estimate, since March, has given Trump a 35% chance of winning, in keeping with the historical win rate of incumbent parties when recessions occurred during the election year (Table 1). Online political betting markets have recently converged to this view (Chart 2). Thus our quant model suggests that the risk to our view, and the new consensus, is a Trump comeback. Table 1Recessions Weigh On Incumbent Win Rates Updating Our Quantitative Election Model Updating Our Quantitative Election Model Chart 2A Democratic Victory Is The New Consensus Updating Our Quantitative Election Model Updating Our Quantitative Election Model We will not formally upgrade Trump’s odds until we are convinced that his freefall has been reversed. We are concerned about the rise in deaths from COVID-19 in key swing states, including Florida, Arizona, and Texas and the potential for another major economic setback. We also would want to see Trump get the next round of fiscal stimulus passed in order to turn more optimistic on his chances. Therefore we will stick to our 35% odds and will reassess in late August when the Republican and Democratic party conventions are held. Model Performs Well In Back Tests Our V2 model performs well during in-sample back testing when comparing actual Electoral College vote outcomes for each election since 1984. On balance, V2 correctly predicts all election outcomes over our sample period (Chart 3). Chart 3Our Model Predicts All Election Outcomes In Our Sample … Updating Our Quantitative Election Model Updating Our Quantitative Election Model The same can be said of V2 during out-of-sample back testing, correctly predicting election outcomes from 2000 - 2016 (Chart 4). Chart 4… And During Out-Of-Sample Back Testing Updating Our Quantitative Election Model Updating Our Quantitative Election Model As mentioned, we cannot ignore the impact that Trump’s job approval may have on his re-election. Since no other transformation of Trump’s approval data test significantly in our V2 model, what if we transform the state coincident index by a longer frequency? What would the predicted outcome be? Trump would maintain his current level of predicted Electoral College votes of 259. The major change is that the state of Florida would no longer be a toss-up. Instead New Hampshire would become the only toss-up, with Trump having only a 45% chance of winning it. Transforming the state coincident index by a longer frequency is more favorable for Trump. Florida moves out of toss-up territory and New Hampshire moves in. But no change in Electoral College votes are recorded as neither party flips a state in this scenario. What if we were to exclude Trump’s approval range as a variable entirely – how would Trump fare? This “barebones” or economic-focused variation is the least favorable for Trump, allocating just 180 Electoral College votes. Arizona and – surprisingly – Iowa would become toss-up states with probabilities of Trump victory at 47% and 49%, respectively (Table 2). Table 2The Economy Is Weighing Down On Trump’s Odds Of Re-Election Updating Our Quantitative Election Model Updating Our Quantitative Election Model It should be noted that models including Trump’s approval range as an explanatory variable exhibited higher over/under estimation during the sample period when compared to models that excluded Trump’s approval range entirely. Despite larger errors in some election years, these models also predicted two elections with almost no error (1988 and 2004), and one election with zero error (2008). These results suggest that Trump’s job approval should not be ignored. Peak Polarization Chart 5Peak Polarization Updating Our Quantitative Election Model Updating Our Quantitative Election Model An interesting takeaway from our V1 model was that it produced a new measure of American political polarization, a phenomenon widely observed by scholars. The model showed that many states would be won or lost with extreme certainty (0% or 100%), i.e. that they are not even competitive. We take this finding as an indication of polarization, in which group loyalty overcomes all other variables. Results of in-sample predictions from our V2 model corroborate this finding (Chart 5). They are virtually the same as in V1, except that they show a higher degree of polarization in 2020, which now matches the previous peak in 2012. This is intuitive and corroborates other evidence that US polarization is reaching or exceeding recent highs. Polarization may or may not rise higher in the next election cycle, but we suspect that we are witnessing peak polarization from a historical point of view. Over five to ten years, polarization should fall. Generational change in the US will produce more domestic policy consensus, while geopolitical struggle with China will unify the nation against a common enemy for the first time since the cold war. Expect uncertainty and market volatility ahead of the election and in the aftermath. Thus the US may continue to export political instability to the rest of the world in the near-term. But eventually it will find an internal equilibrium and external sources of instability will become the bigger geopolitical risk for investors. So What? Our V2 US presidential election model predicts Trump will lose the November reelection, only amassing 259 Electoral College votes. The model implies that Trump has an overall probability of 44% in taking the White House. Florida is the only toss-up state in the latest prediction, with a 52% probability of staying with the incumbent party. Florida accounts for 29 Electoral College votes. Should the states of Michigan, Pennsylvania, and Wisconsin switch back to Republican, Trump would score an additional 46 Electoral College votes. But if Trump has Florida then he only needs to win one of these three states to win the election. Should the states of Michigan, Pennsylvania, and Wisconsin switch back to Republican, Trump would score an additional 46 Electoral College votes which would hand him the win in November. Conversely, the Democrats are expected to win in November with 279 Electoral College votes. As it stands, the Democrats have a 55% chance of victory. For now, we will maintain our subjective 35/65 odds. But the model shows that the risk is to the upside for Trump and that the race will likely tighten from here. We will likely increase his odds in late August if the renewed virus outbreak in Sunbelt swing states gets under control and Congress passes another major stimulus bill by August 10, as we expect. These findings reinforce our long-held view that the election will come down to narrow margins in the swing states. The deluge of bad news for Trump makes it less likely that the election will be narrow. But the fundamentals, as captured in our V2 model, suggest that Florida, at minimum, will still be an extremely tight race. Thus we would reiterate that this election may feature contested results, vote recounts, and Supreme Court interventions, like the year 2000. Investors should prepare for uncertainty and market volatility to rise between now and November 3, and possibly beyond.   Guy Russell Research Analyst GuyR@bcaresearch.com   Statistical Appendix Some clients may be curious about how our V2 election model differs from our V1 model. We discuss the salient differences herein. Chart A1Our Updated Model Offers Reduced Error Updating Our Quantitative Election Model Updating Our Quantitative Election Model 1. The modeling method remains the same Firstly, our V1 model was based off a probit regression, where the dependent variable is stated as 1 = incumbent party wins all Electoral College votes in a given state, or 0 = incumbent party does not win any Electoral College votes in this state.1 The probit regression allows us to assign probabilities of the incumbent party winning each state, given that the inverse of the probability is modeled as a linear combination of the model’s predictors. This modeling technique is maintained in V2 of our model. 2. Variable replacement In V1 of our model, we relied on the Federal Reserve Bank of Philadelphia State Leading Index as an economic variable. In V2, due to the state leading index being discontinued, we adopt the Federal Reserve Bank of Philadelphia State Coincident Index. V1 of our model also used the average approval level of the incumbent president in July of the election year. Since this transformation of job approval data proved statistically insignificant, we tested and included the range of the incumbent president’s job approval rating. The approval range variable showed statistical significance at 5% and 10% levels. 3. Predicted error Assessing the predicted error by each election outcome shows that our V2 model, on balance, trends well with our V1 model (Chart A1), and offers reduced error, on balance, post the 2000 election. Our V2 model also has a lower absolute error when compared to our V1 model. Note, and as we pointed out earlier, our V2 model suffers from some large errors mid-way through the sample period but V2’s predictability improves notably over time. Comparing the error of our V2 model with alternative models that we highlighted in Table A1 also shows just how closely they trend together, despite offering some differing results pertaining to Electoral College votes and toss-up states. Table A1Variations Of Our Model Offer Similar Classified Predicted Outcomes Updating Our Quantitative Election Model Updating Our Quantitative Election Model Our V2 model has a lower predicted error in the 2012 and 2016 election than an alternative V2 in which the state coincident indicator is transformed by a six-month change (Chart A2). This warrants our decision in choosing V2 as our preferred model. Chart A2Three-Month Change In State Coincident Indicators Reduces Model Error Updating Our Quantitative Election Model Updating Our Quantitative Election Model Chart A3Including Trump’s Approval Data Improves The Model’s Robustness Updating Our Quantitative Election Model Updating Our Quantitative Election Model Our V2 model versus the “barebones” V2 model (which excludes the approval range variable and thus can be seen as a purely economic model) has higher predicted error in the elections of 1992 and 1996, but lower error from 2000 onwards (Chart A3). Whilst our V2 model does have a higher absolute error in contrast to the “barebones” model, we believe minimizing a model’s error while still including an element of Trump’s approval data provides us with the most robust election model. Model Diagnostics Regression diagnostics for V2 of our model and other variations that we highlighted in Table A1 above, but do not use, show that our updated model correctly classifies predicted outcomes at a rate of 88.21%. The “barebones” model classifies predicted outcomes marginally better, but we take confidence in the fact that predicted error in our V2 model trends lower as we move further into our sample period, and in the lead up to the 2020 election, bolstering our preferred model choice. The V2 model, if we apply a six-month change to the coincident indicator, classifies predicted outcomes the lowest at 87.43%. Summary Our V2 model shows areas of improved robustness when compared to V1. We keep to the same modeling technique as we did in V1 of our model, a probit regression. We replaced the Federal Reserve Bank of Philadelphia State Leading Index with the Coincident Index and through statistical testing. We opted to drop the average approval level of the incumbent president in July and replace it with the range of the incumbent president’s job approval rating. With mostly lower error for election outcomes from 2000-2016, and lower absolute error and higher correctly classified outcomes, V2 is an adequate model in predicting the upcoming presidential election. Footnotes 1 Two states, Maine and Nebraska, do not have a “winner takes all” distribution of Electoral College votes. Instead they give two Electoral College votes to the winner of the statewide election, plus additional Electoral College votes to the winner within each congressional district. Maine has two congressional districts, Nebraska has three. Nebraska’s second district voted for President Obama in 2008 while Maine’s second district voted for President Trump in 2016.

Highlights Butterflies & Yield Curve Models: With bond market volatility now back to the subdued levels seen prior to the COVID-19 market turbulence earlier in 2020, it is a good time to update our global yield curve valuation models to look for attractive butterfly trade ideas. Valuations: The models generally indicate that flattener trades offer better value across all countries. Our medium-term strategic bias, however, is towards steeper yield curves with policy rates on hold and depressed global inflation expectations likely to continue drifting higher over the latter half of the year. Yield Curve Trades: We are initiating the first set of yield curve trades within our rebooted Tactical Trade Overlay: going long a 7-year bullet vs. a 5-year/10-year barbell in the US; long a 2-year/30-year barbell vs. a 5-year bullet in France; long a 5-year/30-year barbell vs. a 10-year bullet in Italy; and long a 3-year/20-year barbell vs. a 10-year bullet in the UK. Feature In a Special Report published back in February of this year, we dusted off our model-based framework to find value in trades focused on the shape of government bond yield curves.1 By comparing the market-implied short-term interest rate expectations extracted from our curve models to our own macro views, we are able to come up with actionable buy or sell signals across the yield curve in nine developed markets: the US, Germany, France, Italy, Spain, the UK, Japan, Canada, and Australia. Table 1Most Attractive Butterfly Trades Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies Given the extreme market turbulence around the time we published that report, as the full scope of the COVID-19 pandemic was becoming evident, we chose not to recommend any curve trades from our models until global volatility subsided to acceptable levels. The vigorous action from central banks to manipulate bond yields since then - quantitative easing, aggressive forward guidance, outright yield curve control in Japan and Australia, and other unconventional monetary policy measures - introduced another layer of difficulty in implementing successful curve trades using models estimated in more normal times. With global bond market volatility now back down to pre-COVID levels, we feel that the time is right to use our curve models to help identify opportunities. Specifically, we are implementing new recommended yield curve trades in the US, France, Italy, and the UK. Table 1 shows the most attractive butterfly trades across all the markets covered in this analysis. Note that three of the four trades we are initiating include very long-dated bonds where yields are less susceptible to direct central bank influence. The only exception is our US long 7-year bullet vs. 5-year/10-year barbell trade, the reasoning for which we outline later in this report. Three of the four trades we are initiating include very long-dated bonds where yields are less susceptible to direct central bank influence. The only exception is our US long 7-year bullet vs. 5-year/10-year barbell trade. Before delving into our analysis proper, a quick note: in the interest of brevity, we will limit ourselves to a simple explanation of butterfly strategies and our yield curve models in this report. For those interested in a deeper explanation of the curve modeling framework, please refer to our February 25, 2020 Special Report. A Recap On Butterflies And An Update On Our Yield Curve Models A butterfly fixed income strategy involves two main components: a barbell (a weighted combination of long-term and short-term bonds) and a bullet (a medium-term bond that sits within the yield curve segment selected in the barbell). To implement a butterfly strategy, a bond investor would go long (short) the barbell while simultaneously going short (long) the bullet. By weighting the combination of the long- and short-term bonds in the butterfly such that the weighted sum of their duration equals the duration of the medium-term bond in the bullet, we achieve immunization to parallel shifts in the yield curve. At the same time, due to the relatively higher duration of the longer-term component of the butterfly, we get exposure to specific changes in the slope of the yield curve. In general, the barbell will outperform the bullet in a flattening yield curve environment, and vice-versa. Chart of the WeekButterfly Spreads & Yield Curves Butterfly Spreads & Yield Curves Butterfly Spreads & Yield Curves To actually decide how, and on which parts of the yield curve, to implement our butterfly strategies, we make use of our yield curve models. These models rely on the positive relationship typically observed between the butterfly spread and the slope of the yield curve. When the curve steepens, the butterfly spread widens, and vice-versa (Chart of the Week). This has to do with mean reversion: as the curve steepens, it increases the odds that the curve will flatten in the future since it cannot steepen indefinitely. Consequently, investors will ask for greater compensation to enter a curve steepener trade when the curve is already steepening. As a result, we can create simplified models of the yield curve by regressing any butterfly spread on its corresponding curve slope. Deviations from these fair value models indicate which butterfly strategies are cheap or expensive. However, the model output does not by itself constitute a buy or sell signal and must be integrated with our macro view on the slope of the curve. For example, a butterfly strategy with an expensive bullet implies that there is already a certain amount of steepening discounted in the yield curve. If the yield curve flattens, or even steepens by an amount smaller than what is discounted in the yield curve over the investment horizon, the barbell will outperform, as expected. However, if we see more steepening than is discounted in the yield curve, the bullet will outperform, even though it was already at relatively expensive levels. Therefore, it is crucial to integrate our macro view on how much the curve will steepen or flatten over the investment horizon into our curve trade selection framework. In recent reports, we have emphasized our high-conviction view that global inflation expectations will drift higher in the coming months, driven by reflationary fiscal and monetary policy and a continued rebound in global commodity prices (most notably, oil).2 However, a rise in inflation expectations does not necessarily translate to a “one-to-one” rise in nominal yields if it is offset by a compression in real bond yields. To disentangle this, we look at the 3-year rolling betas of nominal 10-year government bond yields to the corresponding 10-year breakeven inflation rates using inflation-linked bonds (Chart 2). The data suggest a currently weaker relationship between inflation expectations and nominal yields, with all betas well below their post-crisis maxima. Our overall macro bias is towards a global steepening in yield curves, but given our strong belief in a rebound in inflation expectations, we would be more willing to enter steepener trades in higher-beta regions such as Germany, Canada, the US, and Australia where it is more likely that a rise in inflation expectations will translate to higher nominal yields. Conversely, we are less hesitant to enter flatteners in the lower-beta regions such as the UK, France, Italy, and Japan. Chart 2The Link Between Nominal Yields And Inflation Expectations Has Weakened Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies When we said earlier this year that we were “dusting off” our yield curve models, that was not just a figure of speech. The models date back originally to 2002, meaning that they are old enough to vote—perhaps even for a popular rapper. Even though we have been refining and updating it along the way, one of our concerns was that this model was estimated for a pre-crisis sample period before near-zero rates became ubiquitous in developed markets. Our overall macro bias is towards a global steepening in yield curves, but given our strong belief in a rebound in inflation expectations, we would be more willing to enter steepener trades in higher-beta regions such as Germany, Canada, the US, and Australia. To test that the curve relationships within our models are maintained when global central banks are pinning policy rates near 0%, we have re-estimated all the regressions for the post-financial crisis period from 2009 to 2017 when most central banks kept rates near the zero bound. Chart 3 shows the results for the representative 2-year, 5-year and 10-year portions of the yield curve. On the whole, the coefficients are weaker but still positive with the exception of Japan, where many years of zero rates and quantitative easing have caused the 2-year/5-year/10-year butterfly spread to become largely unmoored from the 2-year/10-year slope. Chart 3Looking For Structural Shifts In Our Yield Curve Models Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies Therefore, we still see value in our curve modeling approach, even in the current environment where central banks are likely to be on hold for a period measured in years, not months. Bottom Line: Butterfly strategies are an effective way to position for changes in the slope of the yield curve without exposure to shifts in the curve. Our current strategic bias is to expect steepening of developed market yield curves through rising longer-term inflation expectations, but our global yield curve models indicate better value in most flattening trades. Thus, we need to be extremely selective in recommending trades based on the results of our yield curve models. Yield Curve Models And Trades By Region In the remaining pages of this report, we present the current read-outs from of our yield curve models for each of the major developed markets. More specifically, we provide the deviations from fair value for different combinations of bullets and barbells and highlight the most attractive butterfly strategy. The deviations from fair value shown in Tables 2-10 are standardized to facilitate comparisons between the different butterfly combinations. In addition, for each country we provide a quick assessment of the performance of these butterfly strategies over time by applying a simple mechanical trading rule. Every month, we enter the most attractive butterfly strategy, i.e. the one with the highest absolute standardized deviation from its model fair value. The overall message from the models is that barbells appear attractive relative to bullets across all the countries shown. However, we will only initiate trades in cases where the model output and our macro outlook complement each other. US Looking solely at our model output, US Treasury curve flatteners appear most attractive, with the long 3-year/30-year barbell vs. 5-year bullet trade displaying the greatest deviation from fair value with a residual of -1.55 (Table 2). However, we are inclined to agree with our colleagues at BCA Research US Bond Strategy on how to interpret Treasury curve valuation in the current environment. They argue that even though steepeners in the US are currently expensive, valuations can become even more overstretched with the Fed signaling no rate increases for at least the next two years and the market priced for an extended period of near-zero rates.3 Table 2US: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies Our fundamental bias is towards US Treasury curve steepening, with the Fed locking down the front end of the curve and rising inflation expectations putting upward pressure on longer-term yields. Thus, we are entering into the long 7-year bullet vs. 5/10 barbell trade which has a small but positive model residual of +0.17. That represents a better valuation starting point than the other US butterfly spreads, and is therefore a more efficient and profitable way to position for steepeners becoming even more expensive going forward. As highlighted earlier, nominal yields in the US are also more sensitive to rising inflation expectations—another reason to enter into a curve steepener. The specific securities used to execute this trade, as well as the weights for the barbell component used to the make both legs of the trade duration-equivalent, can be found on Page 27 within our Tactical Trade Overlay table. Nominal yields in the US are also more sensitive to rising inflation expectations—another reason to enter into a curve steepener.  The 7-year bullet appears just 1bp cheap according to our model and would only underperform its counterpart given a flattening in the 5-year/10-year Treasury slope greater than 22bps, which we believe is unlikely given the reasons outlined above (Chart 4A). Chart 4AUS 5/7/10 Spread Fair Value Model US 5/7/10 Spread Fair Value Model US 5/7/10 Spread Fair Value Model Chart 4BUS Butterfly Strategy Performance US Butterfly Strategy Performance US Butterfly Strategy Performance Following the mechanical trading rule has delivered steady returns with only a few periods of negative year-over-year returns (Chart 4B). Germany The most attractively valued butterfly combination on the German yield curve is going long the 1-year/30-year barbell and shorting the 5-year bullet, which is almost one standard deviation above its model-implied fair value, with a standardized residual of -0.97 (Table 3). Table 3Germany: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies The 5-year bullet appears 29bps expensive according to our model and would only outperform its counterpart given a steepening in the 1-year/30-year German curve slope greater than 50bps (Chart 5A). Chart 5AGermany 1/5/30 Spread Fair Value Model Germany 1/5/30 Spread Fair Value Model Germany 1/5/30 Spread Fair Value Model Chart 5BGermany Butterfly Strategy Performance Germany Butterfly Strategy Performance Germany Butterfly Strategy Performance Following the mechanical trading rule has been quite profitable, delivering consistently positive year-over-year returns for all but the initial period of our sample (Chart 5B). France The most attractively valued butterfly combination on the French OAT yield curve is going long the 2-year/30-year barbell and shorting the 5-year bullet (Table 4). This combination is a little less than one standard deviation over its model-implied fair value with a standardized residual of -0.84. Nominal yields in France are also relatively less correlated with inflation expectations, which makes this a prime candidate for a flattener trade. The specific securities used to execute this trade, as well as the weights for the barbell component used to the make both legs of the trade duration-equivalent, can be found on Page 27 within our Tactical Trade Overlay table. Table 4France: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies The 5-year bullet appears 21bps expensive according to our model and would only outperform its counterpart given a steepening in the 2-year/30-year French curve slope greater than 48bps (Chart 6A). Chart 6AFrance 2/5/30 Spread Fair Value Model France 2/5/30 Spread Fair Value Model France 2/5/30 Spread Fair Value Model Chart 6BFrance Butterfly Strategy Performance France Butterfly Strategy Performance France Butterfly Strategy Performance As with Germany, following the mechanical trading rule in the French OAT market has also been profitable, with only three periods of negative year-over-year returns in our sample period (Chart 6B). Italy And Spain In Italy, the most attractively valued butterfly combination is going long the 5-year/30-year barbell and shorting the 10-year bullet – a combination with a standardized residual of -0.79 (Table 5). In Spain, going long the 3-year/30-year barbell and short the 5-year bullet seems most attractive with a standardized residual of -0.83 (Table 6). Of the two peripheral euro area countries, we are choosing to put on a trade in the relatively larger and more liquid Italian government bond market. As with France, Italian nominal yields also display a relatively low beta to inflation breakevens. The specific securities used to execute this trade, as well as the weights for the barbell component used to the make both legs of the trade duration-equivalent, can be found on Page 27 within our Tactical Trade Overlay table. Table 5Italy: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies Table 6Spain: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies In Italy, the 10-year bullet appears 22bps expensive according to our model and would only outperform its counterpart given a steepening in the 5-year/30-year Italian curve slope greater than 153bps (Chart 7A). Following the mechanical trading rule in Italy has yielded strong excess returns, with only one very short period of negative year-over-year returns in our sample period (Chart 7B).  As with Italy, following the mechanical trading rule in Spain has yielded some of the strongest excess returns on a cumulative and year-over-year basis. Chart 7AItaly 5/10/30 Spread Fair Value Model Italy 5/10/30 Spread Fair Value Model Italy 5/10/30 Spread Fair Value Model Chart 7BItaly Butterfly Strategy Performance Italy Butterfly Strategy Performance Italy Butterfly Strategy Performance In Spain, the 5-year bullet appears 14bps expensive according to our model and would only outperform its counterpart given a steepening in the 3-year/30-year Spanish curve slope greater than 47bps (Chart 8A). As with Italy, following the mechanical trading rule in Spain has yielded some of the strongest excess returns on a cumulative and year-over-year basis (Chart 8B). Chart 8ASpain 3/5/30 Spread Fair Value Model Spain 3/5/30 Spread Fair Value Model Spain 3/5/30 Spread Fair Value Model Chart 8BSpain Butterfly Strategy Performance Spain Butterfly Strategy Performance Spain Butterfly Strategy Performance UK On the UK Gilt yield curve, the most attractive butterfly combination is holding a 3-year/20-year barbell versus a 10-year bullet, which currently displays a standardized residual of -1.08 (Table 7). As with France and Italy, not only is this flattener trade attractively valued, the UK is also one of the countries where inflation breakevens are relatively less correlated with nominal yields, making this another excellent candidate for our Tactical Trade Overlay. The specific securities used to execute this trade, as well as the weights for the barbell component used to the make both legs of the trade duration-equivalent, can be found on Page 27. Table 7UK: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies The 10-year bullet appears 13bps expensive according to our model and would only outperform its counterpart given a steepening in the 3-year/20-year Gilt curve slope greater than 52bps (Chart 9A). Chart 9AUK 3/10/20 Spread Fair Value Model UK 3/10/20 Spread Fair Value Model UK 3/10/20 Spread Fair Value Model Chart 9BUK Butterfly Strategy Performance UK Butterfly Strategy Performance UK Butterfly Strategy Performance Following the mechanical trading rule in the UK has produced consistent returns on a year-over-year basis (Chart 9B). Canada The most attractively valued butterfly combination on the Canadian yield curve is favoring the 5-year/30-year barbell versus the 7-year bullet, which currently displays a standardized residual of -1.41 (Table 8). Table 8Canada: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies The 7-year bullet appears 7bps expensive according to our model and would only outperform its counterpart given a steepening in the 5-year/30-year Canadian curve slope greater than 42bps (Chart 10A). Chart 10ACanada 5/7/30 Spread Fair Value Model Canada 5/7/30 Spread Fair Value Model Canada 5/7/30 Spread Fair Value Model Chart 10BCanada Butterfly Strategy Performance Canada Butterfly Strategy Performance Canada Butterfly Strategy Performance Following the mechanical trading rule in Canada has historically been a good strategy, but we do note two periods of minor losses in 2013 and 2019 (Chart 10B). Japan The most attractively valued butterfly combination on the JGB yield curve is the 5-year/20-year barbell versus the 7-year bullet, which currently has a standardized residual of -1.03 (Table 9). As we noted earlier, however, valuations in the JGB market are likely distorted due to the Bank of Japan’s long-running programs of quantitative easing, zero policy rates and Yield Curve Control that aims to keep the 10-year JGB yield around 0%. Table 9Japan: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies The 7-year bullet appears 6bps expensive according to our model and would only outperform its counterpart given a steepening in the 5-year/20-year Japan curve slope greater than 23bps (Chart 11A). Following our mechanical trading rule has produced decent returns, especially given the dormant nature of the JGB market, with only a couple minor periods without positive year-over-year returns. Chart 11AJapan 5/7/20 Spread Fair Value Model Japan 5/7/20 Spread Fair Value Model Japan 5/7/20 Spread Fair Value Model Chart 11BJapan Butterfly Strategy Performance Japan Butterfly Strategy Performance Japan Butterfly Strategy Performance Following our mechanical trading rule has produced decent returns, especially given the dormant nature of the JGB market, with only a couple minor periods without positive year-over-year returns (Chart 11B). Australia The most attractively valued butterfly combination on the Australian yield curve is going long the 2-year/10-year barbell versus the 7-year bullet, displaying a standardized residual of -1.73 (Table 10). Table 10Australia: Butterfly Strategy Valuation: Standardized Residuals Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies The 7-year bullet appears 15bps expensive according to our model and would only outperform its counterpart given a steepening in the 2-year/10-year Australian curve slope greater than 101bps (Chart 12A). Chart 12AAustralia 2/7/10 Spread Fair Value Model Australia 2/7/10 Spread Fair Value Model Australia 2/7/10 Spread Fair Value Model Chart 12BAustralia Butterfly Strategy Performance Australia Butterfly Strategy Performance Australia Butterfly Strategy Performance Compared to the other markets in our analysis, following the mechanical trading rule in Australia has not produced stellar returns (Chart 12B). However, excess returns on a year-over-year basis have been positive barring two periods.   Shakti Sharma Research Associate ShaktiS@bcaresearch.com Footnotes 1 Please see BCA Research Global Fixed Income Strategy Special Report, "Global Yield Curve Trades: Follow The Butterflies", dated February 25, 2020, available at gfis.bcaresearch.com. 2 Please see BCA Research Global Fixed Income Strategy Weekly Report, "How To Play The Revival Of Global Inflation Expectations", dated June 23, 2020, available at gfis.bcaresearch.com. 3 Please see BCA Research US Bond Strategy Weekly Report, "Take A Look At High-Yield Technology Bonds", dated June 23, 2020, available at usbs.bcaresearch.com. Recommendations The GFIS Recommended Portfolio Vs. The Custom Benchmark Index Global Yield Curve Trades: Netting Returns With Butterflies Global Yield Curve Trades: Netting Returns With Butterflies ​​​​​​​ Duration Regional Allocation Spread Product Tactical Trades Yields & Returns Global Bond Yields Historical Returns

Highlights Our intermediate-term timing models suggest the US dollar is broadly overvalued.  We are maintaining a modest procyclical currency stance (long NOK, GBP and SEK), but also have a portfolio hedge (short USD/JPY). Go long a basket of petrocurrencies versus the euro. Stay short the gold/silver ratio. Feature Our fundamental intermediate-term timing models (FITM) are one of the toolkits we use in currency management. These simple models enable us to time shifts in developed-market currencies using two key variables. Real Interest Rate Differentials: G10 currencies tend to move with their real rate differentials. Under interest rate parity, if one country is expected to have high interest rates versus another, its currency will rise today so as to gradually depreciate in the future and nullify the interest rate advantage. Risk factor: The ebb and flow of risk aversion affects the path of currencies, as it does their domestic capital markets. Procyclical currencies tend to perform better during risk-on periods. We use high-yield spreads and/or commodity prices as a gauge for risk. For all countries, the variables are highly statistically significant and of the expected signs. These models help us understand in which direction fundamentals are pushing the currencies we look at. These models are more useful as timing indicators on a three-to-nine month basis, as their error terms revert to zero quickly. For the most part, our models have worked like a charm. On a risk adjusted-return basis, a dynamic hedging strategy based on our models has outperformed all static hedging strategies for all investors with six different home currencies since 2001.1  The US Dollar Chart I-1USD Is Overvalued By 4.4% USD Is Overvalued By 4.4% USD Is Overvalued By 4.4% The dollar is a sell, according to the model, with a fair value that is falling much faster than the DXY index itself. Going forward, the Federal Reserve’s dovish stance should keep real interest rate differentials moving against the dollar. This will especially be the case if the authorities move to some form of yield curve control. The wildcard is how risk aversion gyrates as we navigate the volatile summer months, especially given rising geopolitical tensions and the potential for an equity market correction (Chart I-1). One of the factors holding up the dollar is that US domestic growth has been relatively strong, with the Citigroup economic surprise index at the highest level since the inception of the series. For the dollar to decline meaningfully, these positive surprises will need to be repeated abroad. On the data front this week, pending home sales rose 44.3% month-on-month in May, following a 21.8% decline the previous month. House prices are rebounding, to the tune of 4%. The ISM manufacturing index broke out to 52.6 in June from 43.1 the prior month. Job gains for the month of June came in at 4.8 million versus expectations of 3.23 million, pushing the unemployment rate down to 11.1%. These strong numbers provide a high hurdle that non-US growth will need to overcome in order for dollar weakness to continue. The Euro Chart I-2EUR/USD Is Undervalued By 3.8% EUR/USD Is Undervalued By 3.8% EUR/USD Is Undervalued By 3.8% The euro is not excessively undervalued versus the US dollar (Chart I-2). Usually, strong buy signals for the euro have been triggered at a discount of about 10% or so relative to the greenback. That said, the euro can still bounce towards 1.16, or about 3%-4% higher, to bring it back to fair value. The biggest catalyst for the euro remains that interest rate differentials with the US are quite wide and can continue to mean revert. The Treasury-bund spread peaked at 2.8%, and has since lost around 1.7%. Yet, a gap of 100 basis points remains wide by historical standards. On the data front, the CPI numbers from the euro area this week were quite instructive. German inflation came in at +0.8% versus a decline of -0.3% in Spain. In a general sense, inflation in Germany has been outperforming that in the periphery for a few months now, which is a sea-change from the historical trend in eurozone inflation, where both the core and periphery have seen CPI tied at the hip. If rising competitiveness in the periphery is a key driver, then the fair value of the Spanish “peseta” is rapidly catching up to that of the German “Deutsche mark,” which is positive for the euro. The Yen Chart I-3USD/JPY Is Overvalued By 10.3% USD/JPY Is Overvalued By 10.3% USD/JPY Is Overvalued By 10.3% The yen’s fair value has benefited tremendously from the plunge in global bond yields, making rock-bottom Japanese rates relatively attractive from a momentum standpoint (Chart I-3). This has pushed the yen to undervalued levels, supporting our tactically short USD/JPY position. The data out of Japan this week suggest that deflationary forces remain quite strong, which will continue to boost real rates and support the yen. The jobs-to-applicants ratio, a key barometer of labor market health, plunged to 1.20 in May from a cycle high of 1.63. Industrial production fell 25.9% year-on-year in May, the worst since the financial crisis. Meanwhile, the second quarter all-important Tankan survey suggests small businesses will continue to bear the brunt of the economic slowdown.  With most of the increase in the Bank of Japan’s balance sheet coming from USD swaps with the Fed rather than asset purchases, it suggests little ammunition or appetite for more stimulus. Fiscal policy remains the wild card that could help lift domestic demand.   The British Pound Chart I-4GBP/USD Is Undervalued By 5.9% GBP/USD Is Undervalued By 5.9% GBP/USD Is Undervalued By 5.9% Our model shows the pound as only slightly undervalued, putting our long cable position at risk. The drop in UK real rates since the Brexit referendum has prevented our model from flagging the pound as being much cheaper. Given the potential for added volatility this summer, we are looking to book modest profits on long cable (Chart I-4). Data out of the UK remains grim. Mortgage approvals fell to 9.3K in May, well below expectations. Consumer credit is falling much faster than during the depths of the financial crisis, suggesting all the BoE’s liquidity measures are still not filtering down to certain pockets of the economy. Meanwhile, the trend in the trade balance suggests that the pound has not yet started to reflate the economy.   The Canadian Dollar Chart I-5USD/CAD Is Overvalued By 8.1% USD/CAD Is Overvalued By 8.1% USD/CAD Is Overvalued By 8.1% The Canadian dollar is undervalued by about 8% (Chart I-5). Going forward, movements in the Canadian dollar will be largely dictated by interest rate differentials and crude oil prices, which remain supportive for now. We are going long a petrocurrency basket today, one that includes the Canadian dollar. Canadian data have been slowly improving, with housing starts up 20.2% month-on-month in May and existing home sales up 56.9% month-on-month. House prices have also remained resilient. More importantly, foreign investors have used the plunge in oil prices to deploy some fresh capital into Canadian assets. International security transactions in April stood at C$49 billion, the highest on record, and will likely continue to improve as oil prices recover.   The Swiss Franc Chart I-6USD/CHF Is Undervalued By 20.6% USD/CHF Is Undervalued By 20.6% USD/CHF Is Undervalued By 20.6% Our models suggest the Swiss franc is tactically at risk (Chart I-6). The main reason is that the franc has remained strong, despite the pickup in risk sentiment since March. Even if strength in the franc is sniffing market turbulence ahead, the yen remains a better and cheaper hedge. The Swiss National Bank continues to intervene in the foreign exchange market, but this week’s data shows that growth in sight deposits is rolling over. This is happening at a time when the economy remains weak. The June PMI came in at 41.9, well below expectations. Deflation has returned to Switzerland, with the CPI print for June at -1.3%, in line with the May number. While this is boosting real rates, the strength in the franc is an unnecessary headache for the SNB, especially against the euro.    The Australian Dollar Chart I-7AUD/USD Is Undervalued By 7.3% AUD/USD Is Undervalued By 7.3% AUD/USD Is Undervalued By 7.3% Despite the 20% rally in the Aussie dollar since March, it still remains 7%-8% cheap, according to our FITM (Chart I-7). Typical reflation indicators such as commodity prices and industrial share prices are showing nascent upturns. This suggests that so far, policy stimulus in China has been sufficient to lift commodity demand. Meanwhile, 10-year Aussie government bonds sport a positive spread vis-à-vis 10-year Treasurys. Recent data in Australia have been holding up. The private sector is slowly releveraging, the CBA manufacturing PMI went to 51.2 in June, and the trade balance continues to sport a healthy surplus, at A$8 billion for the month of May. Meanwhile, LNG is a long-term winner from China’s shift away from coal and will continue to benefit Australian terms of trade. We are currently in an LNG glut due to Covid-19, but should electricity generation in China, Japan, and other Asean countries recover to pre-crisis peaks, this will ease the glut. The New Zealand Dollar Chart I-8NZD/USD Is Overvalued By 4.9% NZD/USD Is Overvalued By 4.9% NZD/USD Is Overvalued By 4.9% Unlike the AUD, our FITM for the NZD is in expensive territory. This favors long positions in AUD/NZD (Chart I-8). The New Zealand economy will certainly benefit from having put Covid-19 mostly behind it. Both the ANZ business confidence and activity outlook indices continue to rebound strongly from their lows, with the final print for June released this week. However, the hit to tourism will still impact national income. Meanwhile, the adjustment to housing, especially given the ban to foreign purchases, will continue to constrain domestic spending, relative to its antipodean neighbor. In terms of trading, long CAD/NZD and AUD/NZD remain attractive positions. The Norwegian Krone Chart I-9USD/NOK Is Overvalued By 16.9% USD/NOK Is Overvalued By 16.9% USD/NOK Is Overvalued By 16.9% Our fundamental model for the Norwegian krone shows it as squarely undervalued. This favors long NOK positions, which we have implemented via multiple crosses in our bulletins (Chart I-9). The Norwegian economy remains closely tied to oil, and the negative oil print in April probably marked a structural bottom in prices. With inflation near the central bank’s target and our expectation for oil prices to grind higher, the Norwegian currency will likely fare better than a lot of its G10 peers. In terms of data, the unemployment rate ticked higher in April, but at 4.8%, it remains much lower than other developed economies. Our bet is that once the global economy stabilizes, the Norges Bank might find itself ahead of the pack, in any hiking cycle. The Swedish Krona Chart I-10USD/SEK Is Overvalued By 10.6% USD/SEK Is Overvalued By 10.6% USD/SEK Is Overvalued By 10.6% Like its Scandinavian counterpart, the Swedish krona is also quite cheap and is one of our favorite longs at the moment (Chart I-10). Meanwhile, since the Fed extended its USD swap lines, SEK has lagged the bounce in AUD, NZD, and NOK, suggesting some measure of catch up is due. The export-driven Swedish economy was hit hard by Covid-19, despite no widespread lockdowns being implemented. As such, the Riksbank expanded its QE program this week, boosting asset purchases from SEK 300 billion to SEK 500 billion, until June 2021. In September, it will start purchasing corporate bonds in addition to government, municipal, and mortgage bonds. While the repo rate was left unchanged at zero, interest rates on the standing loan facility were slashed 10 basis points and on weekly extraordinary loans by 20 basis points. These measures should provide sufficient liquidity to allow Sweden to recover as economies open up across the globe.     Chester Ntonifor Foreign Exchange Strategist chestern@bcaresearch.com Footnotes 1 Please see Foreign Exchange Strategy / Global Asset Allocation Strategy Special Report titled, "Currency Hedging: Dynamic Or Static? – A Practical Guide For Global Equity Investors (Part II)", dated October 13, 2017.   Trades & Forecasts Forecast Summary Core Portfolio Tactical Trades Limit Orders Closed Trades

The GAA DM Equity Country Allocation model is updated as of June 30, 2020.  The model has added another 6 points to the US overweight at the expense of the euro area, mainly Germany, Netherlands, and Spain. The driving force for this change is from the relatively favorable momentum and liquidity indicators, despite an unfavorable valuation indicator. Now the top four overweight countries are the US, Spain, Australia, and Sweden, while the biggest four underweight countries remain Japan, the UK, France, and Switzerland, as shown in Table 1.  Table 1Model Allocation Vs. Benchmark Weights GAA Quant Model Updates GAA Quant Model Updates As shown in Table 2 and Charts 1, 2 and 3, the overall model outperformed the MSCI World benchmark in June by 49 bps. The Level 2 model outperformed its benchmark by 162 bps, thanks largely to the underweight in Japan and the UK, as well as the overweight in Australia and Spain. The Level 1 model underperformed slightly by 3 bps due to the slight overweight in the US. Since going live, the overall model has outperformed its MSCI World benchmark by 260 bps, with 463 bps of outperformance from the Level 2 model, and 29 bps of outperformance from the Level 1 model. Table 2Performance (Total Returns In USD %) GAA Quant Model Updates GAA Quant Model Updates Chart 1GAA DM Model Vs. MSCI World GAA DM Model Vs. MSCI World GAA DM Model Vs. MSCI World Chart 2GAA US Vs. Non US Model (Level 1) GAA US Vs. Non US Model (Level 1) GAA US Vs. Non US Model (Level 1)     Chart 3GAA Non US Model (Level 2) GAA Non US Model (Level 2) GAA Non US Model (Level 2) For more on historical performance, please refer to our website https://www.bcaresearch.com/site/trades/allocation_performance/latest/G…. For more details on the models, please see Special Report, “Global Equity Allocation: Introducing The Developed Markets Country Allocation Model,” dated January 29, 2016, available at https://gaa.bcaresearch.com. Please note that the overall country and sector recommendations published in our Monthly Portfolio Update and Quarterly Portfolio Outlook use the results of these quantitative models as one input, but do not stick slavishly to them. We believe that models are a useful check, but structural changes and unquantifiable factors need to be considered as well when making overall recommendations. GAA Equity Sector Selection Model The GAA Equity Sector Model (Chart 4) is updated as of June 30, 2020. Chart 4Overall Model Performance Overall Model Performance Overall Model Performance The model’s relative tilts between cyclicals and defensives have changed compared to last month. The model maintains its cyclical stance driven by an improvement in its global growth proxy. The model reversed its overweight position on the only defensive sector where it was previously overweight, Healthcare, given a deterioration in its momentum component. Over the past month, the model outperformed its benchmark by 42 basis points. Year-to-date, the model has outperformed its benchmark by 109 basis points, and 108 basis points since inception.   Table 3Overall Model Performance GAA Quant Model Updates GAA Quant Model Updates Table 4Current Model Allocations GAA Quant Model Updates GAA Quant Model Updates The model’s global growth proxy improved – driven by EM currencies and rising metal prices, and therefore continues to remain positive on cyclical sectors. Global monetary easing and low rates should keep the liquidity component favoring a mixed bag of cyclical and defensive sectors. The valuation component remains muted across all sectors except Energy. However, multiple sectors continue to be near the expensive and cheap zones – mainly Info Tech and Consumer Discretionary (expensive), and Real Estate and Consumer Staples (cheap). The model awaits confirming momentum signals to change recommendations for those sectors.  The model is now overweight four cyclical sectors in total. These are Information Technology, Consumer Discretionary, Communication Services, and Materials. For more details on the model, please see the Special Report “Introducing the GAA Equity Sector Selection Model”, dated July 27, 2016, as well as the Sector Selection Model section in the Special Alert “GAA Quant Model Updates,” dated March 1, 2019, available at https://gaa.bcaresearch.com.    Xiaoli Tang Associate Vice President xiaoliT@bcaresearch.com Amr Hanafy  Senior Analyst amrh@bcaresearch.com  

  Highlights In the short run, extreme policy uncertainty is problematic for risk assets. In the long run, gargantuan fiscal and monetary stimulus continues to support cyclical trades. Equity volatility always increases in the lead-up to US presidential elections. Trump has a 35% chance of reelection. The US-China trade deal is intact for now but the risk of a strategic crisis or tariffs is about 40%. Our Turkish GeoRisk Indicator is lower than it should be based on Turkey’s regional escapades. Feature US equities fell back by 2.6% on June 24 as investors took notice of rising near-term risks to the rally. With gargantuan global monetary and fiscal stimulus, we expect the global stock-to-bond ratio to rise over the long run (Chart 1). However, we still see downside risks prevailing in the near term related to the pandemic, US politics, geopolitics, and the rollout of additional stimulus this summer. Chart 1Risk-On Phase Continues - But Risks Mounting Risk-On Phase Continues - But Risks Mounting Risk-On Phase Continues - But Risks Mounting Chart 2Policy Uncertainty Hitting Extremes Policy Uncertainty Hitting Extremes Policy Uncertainty Hitting Extremes Global economic policy uncertainty is skyrocketing – particularly due to the epic the November 3 US election showdown. Yet Chinese policy uncertainty remains elevated and will rise higher given that the pandemic epicenter now faces an unprecedented challenge to its economic and political order. China’s economic instability will increase emerging market policy uncertainty (Chart 2). Only Europe is seeing political risk fall, yet Trump’s threats of tariffs against Europe this week highlight that he will resort to protectionism if his approval rating does not benefit from stock market gains, which is currently the case. The COVID-19 outbreak is accelerating in the US in the wake of economic reopening and insufficient public adherence to health precautions and distancing measures. The divergence with Europe is stark (Chart 3). Authorities will struggle to institute sweeping lockdowns again, but some states are tightening restrictions on the margin and this will grow. Chart 3US COVID-19 Outbreak Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) The divergence between daily new infection cases and new deaths in the US, as well as countries as disparate as Sweden and Iran, is not entirely reassuring. The US is effectively following Sweden’s “light touch” model. Ultimately COVID is not much of a risk if deaths are minimized – but tighter social restrictions will frighten the markets regardless (Chart 4). President Trump’s election chances have fallen under the weight of the pandemic – followed by social unrest and controversy over race relations. But net approval on handling the economy is holding up well enough (Chart 5). Chart 4Divergence In New Cases Versus New Deaths Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) Chart 5Trump’s Lifeline Is The Economy Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) Our subjective 35% odds of reelection still seem appropriate for now – but we will upgrade Trump if the financial and economic rebound is sustained while his polling improves. His approval should pick up in the face of a collapse of law and order, not to mention left-wing anarchists removing or vandalizing historical monuments to America’s Founding Fathers and some great public figures who had nothing to do with the Confederacy in the Civil War. Equity volatility will increase ahead of the US election. Chart 6Volatility Always Rises Before US Elections Volatility Always Rises Before US Elections Volatility Always Rises Before US Elections Equity volatility always increases in the lead up to modern American elections (Chart 6) and this year’s extreme polarization, high unemployment, and precarious geopolitical environment suggest that negative surprises could be worse than usual, notwithstanding the tsunami of stimulus. So far this year the S&P 500 is tracing along the lower end of its historical performance during presidential election years. This is consistent with a change of government in November, unless it continues to power upward over the next four months – typically a change of ruling party requires a technical correction on the year. Our US Equity Strategist, Anastasios Avgeriou, also expects the market to begin reacting to political risk – and he precisely timed the market’s peak and trough over the past year (Chart 7). We suspect that the positive correlation between the S&P and the Democratic Party’s odds of a full sweep of government is spurious. The reason the S&P has recovered is because of the economic snapback from the lockdowns and the global stimulus. The reason the odds of a Blue Wave election have surged is because the pandemic and recession decimated Trump and the Republicans. Going forward, the market needs to do more to discount a Democratic sweep. At 35%, this scenario is underrated in Chart 8, which considers all possible presidential and congressional combinations. Standalone bets put the odds of a Blue Wave at slightly above 50%. We have always argued that the party that wins the White House in 2020 is highly likely to take the Senate. Chart 7Market At Risk Of Election Cycle Market At Risk Of Election Cycle Market At Risk Of Election Cycle Chart 8Market Will Soon Worry About 'Blue Wave' Market Will Soon Worry About 'Blue Wave' Market Will Soon Worry About 'Blue Wave' True, the US is monetizing debt and this will push risk assets higher regardless over the long run. But if former Vice President Joe Biden wins the presidency, he will create a negative regulatory shock for American businesses, and if his party takes the Senate, then corporate taxes, capital gains taxes, federal minimum wages, liability insurance, and the cost of carbon (implicitly or explicitly) will all rise. The market must also reckon with the possibility that Trump is reelected or that he becomes firmly established as a “lame duck” and thus takes desperate measures prior to the election. His threat to impose tariffs on Europe this week underscores our point that if Trump’s approval rating stays low, despite a rising stock market, then the temptation to spend financial capital in pursuit of political capital will rise. This will involve a hard line on immigration and trade. Bottom Line: Tactically, there is more downside. Strategically, we remain pro-cyclical. Stimulus Hiccups This Summer One reason we have urged investors to buy insurance against downside risks this month is because of hurdles in rolling out the next round of fiscal stimulus. The four key drivers of the global growth rebound are liquidity, fiscal easing (Chart 9), an enthusiastic private sector response, and the large cushion of household wealth prior to the crisis. This is according to Mathieu Savary – author of our flagship Bank Credit Analyst report. Mathieu argues that it will be harder for investors to overlook policy uncertainty after the stimulus slows, i.e. the second derivative of liquidity turns negative. Chart 9Gargantuan Fiscal Stimulus Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) The massive increase in budget deficits and the quick recovery in activity amid reopening have reduced politicians’ sense of urgency. We fear that the stock market will have to put more pressure on lawmakers to force them to provide more largesse. Ultimately they will do so – but if they delay, and if delay looks like it is turning into botching the job, then markets will temporarily panic. Why are we confident stimulus will prevail? In the United States, fiscal bills have flown through Congress despite record polarization. Democrats cannot afford to obstruct the stimulus just to hurt the economy and the president’s reelection chances. Instead they have gone hog wild – promoting massive spending across the board to demonstrate their fundamental proposition that government can play a larger and more positive role in Americans’ lives. Their latest proposal is worth $3 trillion, plus an infrastructure bill that nominally amounts to $500 billion over five years. President Trump, for his part, was always fiscally profligate and now wants $2 trillion in stimulus to fuel the economic recovery, thus increasing his chances of reelection as voters grow more optimistic in the second half of the year. He also wants $1 trillion in new infrastructure spending over five years. Yet Republican Senators are dragging their feet and offering only a $1 trillion package. In the end they will adopt Trump’s position because if they do not hang together, they will all hang separately in November. The debate will center on whether the extra $600 in monthly unemployment benefits will be continued (at a cost of $276bn in the previous Coronavirus Aid, Relief, and Economic Security Act). Republicans want to tie benefits to returning to work, since this generous subsidy created perverse incentives and made it more economical for many to stay on the dole. There will also be a debate over whether to issue another round of direct cash checks to citizens ($290bn in the CARES Act). Republicans want to prioritize payroll tax cuts, again focusing on reducing unemployment (Chart 10). Chart 10US Fiscal Stimulus Breakdown Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) Our US bond strategist, Ryan Swift, has shown that the cash handouts present a substantial fiscal “cliff.” Without the original one-time stimulus checks, real personal income would have fallen 5% since February, instead of rising 9% (Chart 11). If Republicans refuse to issue a new round of checks, yet the extra unemployment benefits stay, then over $1 trillion in income will be needed to fill the gap so that overall personal income will end up flat since February. In other words, an ~8% increase in income less transfers from current levels is necessary to prevent overall personal income from falling below its February level. China and the EU will eventually provide more largesse. Republican Senators will capitulate, but the process could be rocky and the market should see volatility this summer. China may also be forced to provide more stimulus in late July at its mid-year Politburo meeting – any lack of dovishness at that meeting will disappoint investors. European talks on the Next Generation recovery fund could also see delays (though they are progressing well so far). Brexit trade deal negotiations pose a near-term risk. There is also a non-negligible chance that the German Constitutional Court will raise further obstructions with the European Central Bank’s quantitative easing programs on August 5. European risks are manageable on the whole, but the market is not discounting much (Chart 12). Chart 11Will Congress Takeaway The Money Tree? Will Congress Takeaway The Money Tree? Will Congress Takeaway The Money Tree? Bottom Line: We expect the S&P 500 to trade in a range between 2800 and 3200 points during this period of limbo in which risks over pandemic response and political risks will come to the fore while the market awaits new stimulus measures, which may not be perfectly timely. Chart 12European Risks Are Getting Priced European Risks Are Getting Priced European Risks Are Getting Priced Has The Phase One China Deal Failed Yet? President Trump’s threat this week to slap Europe with tariffs, if it imposes travel restrictions on the US over the coronavirus, points to the dynamic we have highlighted on the more consequential issue of whether Trump hikes broad-based tariffs on China, and/or nullifies the “Phase One” trade deal. Our sense is that if Trump is doing extremely poorly, or extremely well, in terms of opinion polls and the stock market, then the roughly 40% odds of sweeping punitive measures of some kind will go up (Diagram 1). Cumulatively we see the chance of a major tariff hike at 40%. Diagram 1Decision Tree: Risk Of Significant Trump Punitive Measures On China In 2020 Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) White House trade czar Peter Navarro’s comments earlier this week, suggesting that the Phase One trade deal was already over, prompted Trump to tweet that he still fully supports the deal. Negotiations between Secretary of State Mike Pompeo and Chinese Politburo member Yang Jiechi also nominally kept the lid on tensions. However, China may need to depreciate the renminbi to ease deflationary pressures on its economy – and this would provoke Trump to retaliate (Chart 13). Chart 13Chinese Depreciation Would Provoke Trump Chinese Depreciation Would Provoke Trump Chinese Depreciation Would Provoke Trump We have always argued against the durability of the Phase One trade deal. Investors should plan for it to fall apart. Judging by our China GeoRisk Indicator, investors are putting in a higher risk premium into Chinese equities (Chart 14). They are also doing so with Korean equities, which are ultimately connected with US-China tensions. Only Taiwan is pricing zero political risk, which is undeserved and explains why we are short Taiwanese equities. After China’s imposition of a controversial national security law in Hong Kong and America’s decision to prepare retaliatory sanctions, reports emerged that Chinese authorities ordered state-owned agricultural traders to halt imports of soybean and pork – and potentially corn and cotton. These reports were swiftly followed by others that highlighted that state-owned Chinese firms purchased at least three cargoes of US soybeans on June 1, in spite of China’s decision to stop imports.1 Thus this aspect of the deal has not yet collapsed. But we would emphasize that the constraints against a failure of the deal are not prohibitive this year. The $200 billion worth of additional Chinese imports over 2020-2021 promised in the deal included $32 billion worth of additional US farm purchases – with at least $12.5 billion in 2020 and $19.5 billion in 2021 over 2017 imports of $24 billion. However, to date, US agricultural exports to China suggest that China may not even meet 2017 levels (Chart 15). Chart 14GeoRisk Indicators Show Rising Risk GeoRisk Indicators Show Rising Risk GeoRisk Indicators Show Rising Risk Chart 15Trade Deal Durability Still Shaky Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) Soybeans account for roughly 60% of US agricultural exports to China. While Chinese imports are up so far this year relative to 2019, they remain well below pre-trade war levels. Although lower hog herds on the back of the African Swine Flu and disruptions caused by COVID-19 may be blamed, they are not the only cause of subdued purchases. The share of Chinese soybean imports coming from the US is also still below pre-trade war levels (Chart 16). Chart 16China Still Substituting Away From US Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) New Chinese regulation requiring documents assuring food shipments to China are COVID-19 free adds another hurdle – China already banned poultry imports from Tyson Foods Inc. plants. Although the US’s share of China’s pork imports has picked up significantly, it will not go far toward meeting the trade deal requirements. China’s pork purchases from the US were valued at $0.3 billion in 2017, while soybean imports came in at $14 billion. Bottom Line: Trump’s only lifeline at the moment is the economy which pushes against canceling the US-China deal. But if he becomes a lame duck – or if exogenous factors humiliate him – then all bets are off. The passage of massive stimulus in the US and China removes economic constraints to conflict. Will Erdogan Overstep In Libya? We have long been bearish on Turkey relative to other emerging markets due to President Tayyip Erdogan’s populist policies, which erode monetary and fiscal responsibility and governance. Turkey’s intervention in Libya has marked a turning point in the Libyan civil war. The offensive to seize Tripoli on the part of General Khalifa Haftar of the Tobruk-based Libyan National Army (LNA) has been met with defeat (Map 1). Map 1Libya’s Battlefront Is Closing In On The Oil Crescent Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) Foreign backing has enabled the conflict. Egypt, the UAE, Saudi Arabia, and Russia are the Libyan National Army’s main supporters, while Turkey and Qatar support Prime Minister Fayez al-Sarraj of the UN recognized Government of National Accord (GNA). The GNA’s successes this year can be credited to Turkey, which ramped up its intervention in Libya, even as oil prices collapsed, hurting Haftar and his supporters. Now the battlefront has shifted to Sirte and the al-Jufra airbase – the largest in Libya – and is closing in on the eastern oil-producing crescent, which contains over 60% of Libya’s oil. The victor in Sirte will also have control over the oil ports of Sidra, Ras Lanuf, Marsa al-Brega, and Zuwetina. With all parties eying the prize, the conflict is intensifying. Tripoli faces greater resistance as its forces move east. Egyptian President Abdel Fattah al-Sisi’s June 6 ceasefire proposal, dubbed the Cairo Initiative, was rejected by al-Sarraj and Turkey. Instead, the Tripoli-based government wants to capture Sirte and al-Jufra before coming to the table. The recapturing of oil infrastructure would bring back some of Libya's lost output (Chart 17). Nevertheless, OPEC 2.0 is committed to keeping oil markets on track to rebalance, reducing the net effect of a Libyan production increase on global supplies. However, the GNA’s swift successes in the West may not be replicable as it moves further East, where support for Haftar is deeper and where the stakes are higher for both sides. This is demonstrated by the GNA’s failed attempt to capture Sirte on June 6. The battlefront is now at Egypt’s red line – GNA control of al-Jufra would pose a direct threat to Egypt and is thus considered a border in Egypt’s national security strategy. A push eastward risks escalating the conflict further by drawing in Egypt militarily. In a televised speech on June 20, al-Sisi threatened to deploy Egypt’s military if the red line is crossed. The statement was interpreted by Ankara as a declaration of war, raising the possibility that Egypt will go to war with Turkey in Libya. On paper, Egypt’s military is up to the task. Its recent upgrades have pulled up its ranking to ninth globally according to the Global Fire Power Index, surpassing Turkey’s strength in land and naval forces (Chart 18). However, while Turkey’s military has been active in other foreign conflicts such as in Syria, Egypt’s army is untested on foreign soil. Its most recent military encounter was the 1973 Yom Kippur War. Even after years of fighting, it has yet to declare victory against terrorist cells in the Sinai Peninsula. Thus Egypt’s rusty forces could face a protracted conflict in Libya rather than a swift victory. Chart 17GNA/Turkish Success Would Revive Libyan Oil Production Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) Chart 18Egypt Is Militarily Capable … On Paper Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) Other constraints may also deter al-Sisi from following through on his threat: Other Arab backers of the Libyan National Army – the UAE and Saudi Arabia – are unlikely to provide much support as their economies have been hammered by low oil prices. Egypt’s own economy is in poor shape to withstand a protracted war, with public debt on an unsustainable path. Not coincidentally, Egypt faces another potential military escalation to its south where it has been clashing with Ethiopia over the construction of the Grand Ethiopian Renaissance Dam on the Blue Nile. The dam will control Egypt’s water supply. The latest round of negotiations failed last week. While Cairo is hoping to obtain a bilateral agreement over the schedule for filling the dam, Addis Ababa has indicated that it will begin filling the dam in July regardless of whether an agreement is reached. Al-Sisi’s response to the deadlocked situation has been to request an intervention by the UN Security Council. However, as the July filling date nears, the Egypt-Ethiopia standoff risks escalating into war. For Egypt, there is an urgency to secure its future water supplies now before Ethiopia begins filling the dam. And while resolving the Libyan conflict is also a matter of national security – Egypt sees the Libyan National Army as a buffer between its porous western border and the extremist elements of the GNA – the risks are not as pressing. Thus a military intervention in Libya would distract Egypt from the Ethiopian conflict and risk drawing it into a war on two fronts. Moreover, Egypt generally, and al-Sisi in particular, risk losing credibility in case of a defeat. That said, Egypt has high stakes in Libya. A GNA defeat could annul the recent Libya-Turkey maritime demarcation agreement – a positive for Egypt’s gas ambitions – and eliminate the presence of unfriendly militias on its Western border. Thus, if the GNA or GNA-allied forces kill Egyptian citizens, or look as if they are capable of utterly defeating Haftar on his own turf, then it would be a prompt for intervention. Meanwhile Turkey’s regional influence and foreign policy assertiveness is growing – and at risk of over-extension. Erdogan’s interests in Libya stem from both economic and strategic objectives. In addition to benefitting from oil and gas rights and rebuilding contracts, Ankara’s strategy is in line with its pursuit of greater regional influence as set out in the Mavi Vatan, its current strategic doctrine.2 There are already rumors of Turkish plans to establish bases in the recently captured al-Watiya air base and Misrata naval base. This would be in addition to Ankara’s bases in Somalia and in norther Iraq. Erdogan is partly driven into these foreign policy adventures to distract from his domestic challenges and keep his support level elevated ahead of the 2023 general election (Chart 19). However, his growing assertiveness threatens to alienate European neighbors and NATO allies, which have so far played a minimal role in the Libyan conflict yet have important interests there. For now, the western powers seem focused on countering Russian intervention in Libya and the broader Mediterranean. Prime Minister al-Sarraj and General Stephen Townsend, head of US Africa Command (AFRICOM), met earlier this week and reiterated the need to return to the negotiating table and respect Libyan sovereignty and the UN arms embargo, with a focus on stemming Russian interference. However, Turkish relations with the West may take a turn for the worse if Erdogan oversteps. Turkey continues to threaten Europe with floods of refugees and immigrants if its demands are not met. This pressure will grow due to the COVID-19 crisis, which will ripple across the Middle East, Africa, and South Asia. Ankara also continues to press territorial claims in the Mediterranean Sea, ostensibly for energy development.3 Turkey has recently clashed with Greece and France on the seas. In sum, the Libyan conflict is intensifying as it moves into the oil crescent. The Turkey-backed GNA will face greater resistance in Sirte and al-Jufra, even assuming that Egypt does not follow through on its threat of intervening militarily. Erdogan’s foreign adventurism will provoke greater opposition in Libya and elsewhere among key western powers, Russia, and the Gulf Arab states. Bottom Line: The implication is that a deterioration in Turkey’s relationship with the West, military overextension, and continued domestic economic mismanagement will push up our Turkey GeoRisk Indicator, which is a way of saying that it will weigh on the currency (Chart 20). Chart 19Erdogan’s Fear Of Opposition Drives Bold Policy Volatility And Mediterranean Quarrels (GeoRisk Update) Volatility And Mediterranean Quarrels (GeoRisk Update) Chart 20Foreign And Domestic Factors Will Push Up Turkish Risk Foreign And Domestic Factors Will Push Up Turkish Risk Foreign And Domestic Factors Will Push Up Turkish Risk Stay short our “Strongman Basket” of emerging market currencies, including the Turkish lira. Investment Takeaways We entered the year by going strategically long EUR-USD, but closed the trade upon the COVID-19 lockdowns. We have resisted reinitiating it despite the 5% rally over the past three months due to extreme political risks this year, namely the US election and trade risks. Trump’s threat of tariffs on Europe this week highlights our concern. We will wait until the election outcome before reinstituting this trade, which should benefit over time as global and Chinese growth recover and the US dollar drops on yawning twin deficits. Throughout this year’s crisis we have periodically added cyclical and value plays to our strategic portfolio. We favor stocks over bonds and recommend going long global equities relative to the US 30-year treasuries. We are particularly interested in commodities that will benefit from ultra-reflationary policy and supply constraints due to insufficient capital spending. This month we recommend investors go long our BCA Rare Earth Basket, which features producers of rare earth elements and metals that can substitute for Chinese production (Chart 21). This trade reflects our macro outlook as well as our sense that the secular US-China strategic conflict will heat up before it cools down. Chart 21Position For An Escalation In The US-China Conflict Position For An Escalation In The US-China Conflict Position For An Escalation In The US-China Conflict   Matt Gertken Vice President Geopolitical Strategist mattg@bcaresearch.com   Roukaya Ibrahim Editor/Strategist Geopolitical Strategy RoukayaI@bcaresearch.com Footnotes 1 Please see Karl Plume et al, "China buys U.S. soybeans after halt to U.S. purchases ordered: sources," Reuters, June 1, 2020. 2 The Mavi Vatan or “Blue Homeland Doctrine” was announced by Turkish Admiral Cem Gurdeniz in 2006 and sets targets to Turkish control in two main regions. The first region is the three seas surrounding it – the Mediterranean Sea, Aegean Sea, and Black Sea with the goal of securing energy supplies and supporting Turkey’s economic growth. The second region encompasses the Red Sea, Caspian Sea and Arabian Sea where Ankara has strategic objectives. 3 Ankara’s gas drilling activities off Cyprus have been a form of frequent provocation for Greece and Cyprus. Ankara has also stated that it may begin oil exploration under a controversial maritime deal with Libya as early as August. Section II: Appendix : GeoRisk Indicator China China: GeoRisk Indicator China: GeoRisk Indicator Russia Russia: GeoRisk Indicator Russia: GeoRisk Indicator UK UK: GeoRisk Indicator UK: GeoRisk Indicator Germany Germany: GeoRisk Indicator Germany: GeoRisk Indicator France France: GeoRisk Indicator France: GeoRisk Indicator Italy Italy: GeoRisk Indicator Italy: GeoRisk Indicator Canada Canada: GeoRisk Indicator Canada: GeoRisk Indicator Spain Spain: GeoRisk Indicator Spain: GeoRisk Indicator Taiwan Taiwan: GeoRisk Indicator Taiwan: GeoRisk Indicator Korea Korea: GeoRisk Indicator Korea: GeoRisk Indicator Turkey Turkey: GeoRisk Indicator Turkey: GeoRisk Indicator Brazil Brazil: GeoRisk Indicator Brazil: GeoRisk Indicator Section III: Geopolitical Calendar

Highlights In this Weekly Report, we present our semi-annual chartbook of the BCA Central Bank Monitors. All of the Monitors are now below the zero line, indicating the need for continued easy global monetary policy to help mitigate the COVID-19 recession (Chart of the Week). Central bankers have already responded in an intense and rapid fashion to the crisis, delivering a series of rate cuts, increased asset purchase programs and measures to support bank lending to businesses suffering under quarantines. All of these vehicles have helped trigger a powerful rally in global bond markets that helped revitalize risk assets as well. After the coordinated global easing response of the past few months, the optimal policy choices now differ from country to country. This creates opportunities to benefit from country allocation decisions even in a world of puny government bond yields. The overall signal from our Central Bank Monitors is still bond bullish, however – at least over the next few months until there is evidence of how fast global growth is rebounding from the COVID-19 lockdowns. An Overview Of The BCA Central Bank Monitors Chart of the WeekUltra-Accommodative Monetary Policies Are Still Required Ultra-Accommodative Monetary Policies Are Still Required Ultra-Accommodative Monetary Policies Are Still Required Chart 2A Bond-Bullish Message From Our CB Monitors A Bond-Bullish Message From Our CB Monitors A Bond-Bullish Message From Our CB Monitors The BCA Central Bank Monitors are composite indicators designed to measure the cyclical growth and inflation pressures that can influence future monetary policy decisions. The economic data series used to construct the Monitors are not the same for every country, but the list of indicators generally measure the same things (i.e. manufacturing cycles, domestic demand strength, commodity prices, labor market conditions, exchange rates, etc). The data series are standardized and combined to form the Monitors. Readings above the zero line for each Monitor indicate pressures for central banks to raise interest rates, and vice versa. Through the nexus between growth, inflation, and market expectations of future interest rate changes, the Monitors do exhibit broad correlations to government bond yields in the Developed Markets (Chart 2). All of the Monitors are indicating intense pressure to maintain very easy monetary policies in response to the global COVID-19 recession. While the bad economic and inflation news is largely discounted in the depressed level of bond yields worldwide, there are still opportunities to position country allocations within a government bond portfolio based on the message from our Monitors (overweighting the US, the UK and Canada, underweighting Germany and Japan). All of the Monitors are indicating intense pressure to maintain very easy monetary policies in response to the global COVID-19 recession.  In each BCA Central Bank Monitor Chartbook, we include a new chart for each country that we have not shown previously. In this edition, we show the components of the Monitors, grouped into those focusing on economic growth and inflation, plotted alongside our estimate of the appropriate level of central bank policy interest rates derived using a Taylor Rule. Fed Monitor: Policy Must Stay Accommodative Our Fed Monitor has collapsed below the zero line to recessionary levels (Chart 3A) in response to the coronavirus crisis. The Fed has already delivered a series of aggressive policy responses since March to help support an economy ravaged by the virus, including: interest rate cuts; quantitative easing (QE), including buying corporate and municipal debt; and setting up lending schemes for small businesses. The lockdown of almost the entire country has helped “flatten the curve” of the spread of COVID-19, but at a painful economic cost. The unemployment rate rose to 14.7% in April, the highest level since the Great Depression, and is expected to peak at levels above 20%. The result is unsurprising: a massive increase in spare economic capacity with a threat of deflation as headline CPI inflation plummeted to 0.3% in April (Chart 3B). Chart 3AUS: Fed Monitor US: Fed Monitor US: Fed Monitor Chart 3BUS Realized Inflation Flirting With 0% US Realized Inflation Flirting With 0% US Realized Inflation Flirting With 0% Within the components of our Fed Monitor, weakening growth has been the main driver of the decline (Chart 3C). Our Taylor Rule estimate suggests a deeply negative fed funds rate is “appropriate”, although the Fed is likely to pursue other avenues of easing like yield curve control before ever attempting a sub-0% policy rate. Chart 3CNegative Rates Are 'Required' In The US, But The Fed Has Other Options Negative Rates Are 'Required' In The US, But The Fed Has Other Options Negative Rates Are 'Required' In The US, But The Fed Has Other Options The fall in US Treasury yields over the past few months has been in line with the decline in our Fed Monitor (Chart 3D). While the US economy is slowly awakening from lockdowns, consumer and business confidence are likely to remain fragile given the numerous risks from a second wave of COVID-19, worsening US-China relations and, more recently, social unrest. Thus, we continue to recommend an overweight strategic allocation to the US within global government bond portfolios. The fall in US Treasury yields over the past few months has been in line with the decline in our Fed Monitor Chart 3DTreasury Yields Fully Reflect Pressure For More Fed Easing Treasury Yields Fully Reflect Pressure For More Fed Easing Treasury Yields Fully Reflect Pressure For More Fed Easing BoE Monitor: Negative Rates On The Horizon? Our Bank of England (BoE) Monitor has collapsed to the lowest level in its history on the back of the severe COVID-19 recession (Chart 4A). The BoE already cut the Bank Rate to 0.1% in March, ramped up asset purchases, and introduced a Term Funding scheme to support business lending. Any additional easing from here might entail negative policy rates, which markets are already discounting. The UK unemployment rate is expected to peak around 8%, with the BoE projecting the economy to shrink by -14% this year, which would be the worst recession in modern history. Inflation has dropped sharply on the back of the dual collapse of energy prices and economic growth, ending a period of currency-fueled inflation increases (Chart 4B). Chart 4AUK: BoE Monitor UK: BoE Monitor UK: BoE Monitor Chart 4BUK Realized Inflation Is Slowing Rapidly UK Realized Inflation Is Slowing Rapidly UK Realized Inflation Is Slowing Rapidly The components of our BoE Monitor fully reflect the dire economic situation (Chart 4C), with weak growth – led by sharp falls in business confidence – driving the collapse of the Monitor more than falling inflation pressures. Our Taylor Rule estimate of the policy rate is not yet calling for negative rates, but that is because we are using the New York Fed’s estimate of r* as the neutral real rate, which is a relatively high 1.4% (by comparison, r* in the US is estimated to be 0.5%). Chart 4CNegative Rates Are Not Yet Required In The UK Negative Rates Are Not Yet Required In The UK Negative Rates Are Not Yet Required In The UK The sharp fall in the BoE Monitor suggests that Gilt yields will remain under downward pressure in the coming months (Chart 4D). New BoE Governor Andrew Bailey has stated that a move to negative rates is not imminent, but markets will continue to flirt with the notion of sub-0% interest rates until the economy and inflation stabilize. We maintain an overweight stance on UK Gilts. Chart 4DBoE Monitor Suggests Continued Downward Pressure On Gilt Yields BoE Monitor Suggests Continued Downward Pressure On Gilt Yields BoE Monitor Suggests Continued Downward Pressure On Gilt Yields ECB Monitor: Continued Monetary Support Is Needed Our European Central Bank (ECB) Monitor is now well below the zero line, signaling a strong need for easier monetary policy to fight the COVID-19 downturn (Chart 5A). The ECB has delivered multiple measures to ease monetary conditions, including a new €750bn bond-buying vehicle and liquidity operations to help banks maintain lending to European businesses.​​​​​​​​​​​​​​ The recession has hit the region hard, with real GDP declining by -3.8% in Q1, the sharpest fall since records began in 1995. Unemployment rates have climbed higher, although to much lower levels than seen in the US thanks to more generous government labor support programs that have helped to limit layoffs. The sharp downturn has resulted in both a surge in spare economic capacity and plunge in headline inflation to 0.3% in April (Chart 5B). Chart 5AEuro Area: ECB Monitor Euro Area: ECB Monitor Euro Area: ECB Monitor Chart 5BEurope Is On The Edge Of Deflation Europe Is On The Edge Of Deflation Europe Is On The Edge Of Deflation Within the individual components of our ECB Monitor, both weaker growth and near-0% inflation have both contributed to the Monitor’s decline (Chart 5C). Our Taylor Rule measure shows that the ECB’s current stance of having policy rates modestly below 0% is appropriate. Chart 5CThe ECB Needs To Keep Its Foot On The Monetary Accelerator The ECB Needs To Keep Its Foot On The Monetary Accelerator The ECB Needs To Keep Its Foot On The Monetary Accelerator Despite the ECB’s easing measures, and in contrast to the message from our ECB Monitor, the downward momentum in core European bond yields has been fading (Chart 5D). With the ECB reluctant to push policy rates deeper into negative territory, and with reliable cyclical indicators like the German ZEW and IFO surveys showing signs that euro area growth is starting to recover from the lockdowns, the case for even lower core European yields in the coming months is not strong. We maintain our recommended underweight stance on German and French government bonds. We maintain our recommended underweight stance on German and French government bonds. Chart 5DNo Pressure For Higher German Bund Yields No Pressure For Higher German Bund Yields No Pressure For Higher German Bund Yields BoJ Monitor: What More Can Be Done? Our Bank of Japan (BoJ) Monitor has fallen further below zero, indicating easier policy is required (Chart 6A). The BoJ has already introduced additional easing measures in the past couple of months: extending forward guidance (inflation is projected to remain below the BoJ’s 2% target for the next three years), increasing asset purchases and enhancing loan programs to small and medium sized companies. New cases of COVID-19 have slowed sharply in Japan, prompting an end to the national state of emergency last week. Importantly, the virus did not hit Japan's labor market as severely as in other developed countries. The unemployment rate did reach a two-year high in April, but is still only 2.6% (Chart 6B). Fiscal stimulus and measures to protect job losses have played a major role in preventing a bigger spike in joblessness. Even with those measures, growth remains weak and realized inflation is heading back towards deflation. Chart 6AJapan: BoJ Monitor Japan: BoJ Monitor Japan: BoJ Monitor Chart 6BJapan Nearing Deflation Once Again Japan Nearing Deflation Once Again Japan Nearing Deflation Once Again Looking at the components of our BoJ Monitor, contracting growth, more than weakening inflation pressures, is the bigger driver of the fall in the Monitor below zero (Chart 6C). However, our Taylor Rule estimate does not suggest that the current level of the policy rate is out of line. Chart 6CBoJ Needs More Easing (Somehow) Until The Economy Revives BoJ Needs More Easing (Somehow) Until The Economy Revives BoJ Needs More Easing (Somehow) Until The Economy Revives The BoJ’s current combined policies of negative rates, QE and yield curve control are keeping JGB yields at near-0% levels. Those policies are also suppressing yield volatility and preventing an even bigger fall in JGB yields (with larger capital gains) as suggested by our BoJ Monitor (Chart 6D). We continue to recommend a maximum underweight in Japanese government bonds in a yield-starved world. Chart 6DJGB Yields Will Be Anchored For Some Time JGB Yields Will Be Anchored For Some Time JGB Yields Will Be Anchored For Some Time BoC Monitor: Deflationary Pressures Intensifying Our Bank of Canada (BoC) Monitor has collapsed into “easier policy required” territory, reaching levels last seen during the 2009 recession (Chart 7A). The central bank has already introduced several easing measures to help boost the virus-stricken economy, including cutting the Bank Rate to a mere 0.25% and starting a QE program to buy government bonds for the first time ever. Before the COVID-19 outbreak, some softening of the economy was already underway. Now, after the imposition of nationwide lockdowns to limit the spread of the virus, the unemployment rate has spiked to 13% - a level last seen in the early 1980s. The result is a massive deflationary output gap has opened up (Chart 7B), with realized headline CPI inflation printing at -0.2% in April. Chart 7ACanada: BoC Monitor Canada: BoC Monitor Canada: BoC Monitor Chart 7BOutright Headline CPI Deflation In Canada Outright Headline CPI Deflation In Canada Outright Headline CPI Deflation In Canada The fall in our BoC Monitor has been driven by both collapsing economic growth and weakening inflation pressures (Chart 7C). Our Taylor Rule estimate suggests that one of new BoC Governor Tiff Macklem’s first policy decisions may need to be a move to negative interest rates. Macklem and other BoC officials have not played up the possibility of cutting rates below 0%. However, the fact that the BoC provided no economic growth forecasts in the most recent Monetary Policy Report highlights the extreme uncertainties surrounding the economic impact from COVID-19 – even with the Canadian government providing a large fiscal response to the pandemic. Chart 7CBoC Monitor Plunging Due To High Unemployment & Low Inflation BoC Monitor Plunging Due To High Unemployment & Low Inflation BoC Monitor Plunging Due To High Unemployment & Low Inflation We upgraded our recommended stance on Canadian government debt to overweight back in March, and the collapse of the BoC Monitor suggests continued downward pressure on Canadian yields (Chart 7D). Stay overweight. The collapse of the BoC Monitor suggests continued downward pressure on Canadian yields.  Chart 7DCanadian Yield Momentum In Line With The BoC Monitor Canadian Yield Momentum In Line With The BoC Monitor Canadian Yield Momentum In Line With The BoC Monitor RBA Monitor: Rate Cutting Cycle Is Done Due to a slump in export demand and a weakening housing market, our Reserve Bank of Australia (RBA) monitor has been consistently calling for rate cuts since April 2018 (Chart 8A). Australia began its easing cycle early, having delivered a total of 125bps of stimulus since June 2019, with the two most recent cuts coming directly in response to the COVID-19 crisis. As in other developed markets, the unemployment gap in Australia has widened dramatically, owing to job losses concentrated in tourism, entertainment, and dining out (Chart 8B). Although inflation briefly breached the low end of the RBA’s 2-3% target band in Q1, this will not be a lasting development. The RBA sees headline CPI deflating by -1% year-on-year in Q2/2020 and, even as far as 2022, only sees it growing at 1.5%. Chart 8AAustralia: RBA Monitor Australia: RBA Monitor Australia: RBA Monitor Chart 8BInflation Will Remain Stuck Below RBA 2-3% Target Inflation Will Remain Stuck Below RBA 2-3% Target Inflation Will Remain Stuck Below RBA 2-3% Target Although both the growth and inflation components of our RBA Monitor are below zero, the former drove the most recent decline (Chart 8C) led by consumer confidence almost touching the 2008 lows. The RBA has already responded by cutting rates to near 0%, well below the Taylor Rule implied estimate, and initiating yield curve control with a cap on 3-year government bond yields at 0.25%. Chart 8CNo Pressure For The RBA To Go To Negative Rates No Pressure For The RBA To Go To Negative Rates No Pressure For The RBA To Go To Negative Rates Overall, Australian bond yields have accurately priced in the dovish signal from our RBA Monitor (Chart 8D). With COVID-19 relatively well contained in Australia, there is less pressure on the RBA to ease further. Governor Lowe has also ruled out negative rates, which will put a floor under yields. Owing to these factors, we confidently reiterate our neutral stance on Australian government debt within global fixed income portfolios. Australian bond yields have accurately priced in the dovish signal from our RBA Monitor. Chart 8DAustralian Bond Yields Are Unlikely To Move Much Lower Australian Bond Yields Are Unlikely To Move Much Lower Australian Bond Yields Are Unlikely To Move Much Lower RBNZ Monitor: Cause For Concern After a resurgence late last year, our Reserve Bank of New Zealand (RBNZ) Monitor has declined to a level slightly below zero (Chart 9A). The RBNZ responded to the pandemic by delivering a massive -75bps cut in March, but has since left the policy rate untouched, preferring to deliver further stimulus by doubling the size of its QE program. Forward guidance is signaling that the policy rate will remain at 0.25% until 2021, but the central bank has not ruled out negative rates in the future. ​​​​​​​Although the actual unemployment numbers do not yet capture the impact of the pandemic, both consensus and RBNZ forecasts call for a blowout in the unemployment gap (Chart 9B). The RBNZ expects the steady improvement in inflation seen up to Q1/2020 to be wiped out, with headline CPI projected to remain below the 1-3% target range until mid-2022. Chart 9ANew Zealand: RBNZ Monitor New Zealand: RBNZ Monitor New Zealand: RBNZ Monitor Chart 9BRealized NZ Inflation Was Drifting Higher, Pre-Virus Realized NZ Inflation Was Drifting Higher, Pre-Virus Realized NZ Inflation Was Drifting Higher, Pre-Virus Surprisingly, the inflation component of our RBNZ Monitor is actually calling for tighter monetary policy, owing to significant strength in the housing market (Chart 9C). However, this trend is likely to reverse - the RBNZ foresees a -9% decline in house prices over the remainder of 2020. Meanwhile, growth components such as consumer confidence and employment will remain depressed, holding down our RBNZ monitor. Chart 9CGrowth, Now Inflation, Has Driven The RBNZ Monitor Lower Growth, Now Inflation, Has Driven The RBNZ Monitor Lower Growth, Now Inflation, Has Driven The RBNZ Monitor Lower Overall, the momentum in New Zealand bond yields seems to have overshot the message from our RBNZ Monitor (Chart 9D). However, with so much uncertainty about business investment and cash flows from key sectors such as tourism and education, it is too early to bet on an improvement in yields. We therefore maintain a neutral recommendation on NZ sovereign debt. Chart 9DNZ Bond Yields Are Unlikely To Move Lower NZ Bond Yields Are Unlikely To Move Lower NZ Bond Yields Are Unlikely To Move Lower Riksbank Monitor: Worries For The Coronavirus Mavericks Amid the global pandemic, our Riksbank Monitor has collapsed to all-time lows (Chart 10A). In its April monetary policy decision, the Riksbank opted for continued asset purchases and liquidity measures to support bank lending to companies over a move to negative rates. One of the primary concerns for the Riksbank is headline CPI inflation, which fell into mild deflation (-0.4% year-over-year) in April on the back of lower energy prices and weaker domestic demand (Chart 10B). This could spill over into a lasting decline in long-term inflation expectations if the economy does not quickly improve. Chart 10ASweden: Riksbank Monitor Sweden: Riksbank Monitor Sweden: Riksbank Monitor Chart 10BSwedish Realized Inflation Back To 0% Swedish Realized Inflation Back To 0% Swedish Realized Inflation Back To 0% Both the growth and inflation components of our Riksbank Monitor are calling for further easing, with the growth component now at post-crisis lows (Chart 10C). The collapse on the growth side can be attributed to historic falls in retail confidence, the manufacturing PMI and employment while the inflation component remains depressed due to low headline numbers and inflation expectations. Chart 10CThe Riksbank Hates Negative Rates, But Could Still Need Them If The Economy Worsens The Riksbank Hates Negative Rates, But Could Still Need Them If The Economy Worsens The Riksbank Hates Negative Rates, But Could Still Need Them If The Economy Worsens The sharp downward move in our Riksbank Monitor suggests Swedish bond yields should remain under downward pressure in the coming months (Chart 10D). The key factor for yields will be the effect of the relatively lax measures implemented by Sweden to combat the pandemic. Sweden saw positive GDP growth in Q1/2020 due to fewer restrictions on the economy. However, infection and mortality rates are much higher in Sweden than in neighboring countries and, as a result, Denmark and Norway excluded Sweden from their open border agreement. Continued restrictions of the sort are bearish for growth – and bullish for bonds – in this trade-dependent economy. Chart 10DSwedish Bond Yields Will Remain Under Downward Pressure Swedish Bond Yields Will Remain Under Downward Pressure Swedish Bond Yields Will Remain Under Downward Pressure   Robert Robis, CFA Chief Fixed Income Strategist rrobis@bcaresearch.com Ray Park, CFA Research Analyst ray@bcaresearch.com ​​​​​​​Shakti Sharma Research Associate ShaktiS@bcaresearch.com Recommendations The GFIS Recommended Portfolio Vs. The Custom Benchmark Index BCA Central Bank Monitor Chartbook: Collapse BCA Central Bank Monitor Chartbook: Collapse ​​​​​​​ Duration Regional Allocation Spread Product Tactical Trades Yields & Returns Global Bond Yields Historical Returns

The GAA DM Equity Country Allocation model is updated as of May 29, 2020.  The model has not made any significant change this month.  It has kept the same order for the top four overweight countries (Spain, Australia, Sweden, and the US) as well as the four large underweight countries (Japan, the UK, France, and Switzerland), as shown in Table 1.  Table 1Model Allocation Vs. Benchmark Weights GAA Quant Model Updates GAA Quant Model Updates As shown in Table 2 and Charts 1, 2 and 3, the overall model outperformed the MSCI World benchmark in May by 29 bps. The Level 1 model outperformed 2 bps because of the overweight in the US. The Level 2 model outperformed by 85 bps thanks to the overweight of Sweden, Germany and the Netherlands, as well as the underweight in the UK and Switzerland. Since going live, the overall model has outperformed its MSCI World benchmark by 180 bps, with 246 bps of outperformance from the Level 2 model, and 33 bps of outperformance from the Level 1 model. Table 2Performance (Total Returns In USD %) GAA Quant Model Updates GAA Quant Model Updates Chart 1GAA DM Model Vs. MSCI World GAA DM Model Vs. MSCI World GAA DM Model Vs. MSCI World Chart 2GAA US Vs. Non US Model (Level 1) GAA US Vs. Non US Model (Level 1) GAA US Vs. Non US Model (Level 1)     Chart 3GAA Non US Model (Level 2) GAA Non US Model (Level 2) GAA Non US Model (Level 2) For more on historical performance, please refer to our website https://www.bcaresearch.com/site/trades/allocation_performance/latest/G…. For more details on the models, please see Special Report, “Global Equity Allocation: Introducing The Developed Markets Country Allocation Model,” dated January 29, 2016, available at https://gaa.bcaresearch.com. Please note that the overall country and sector recommendations published in our Monthly Portfolio Update and Quarterly Portfolio Outlook use the results of these quantitative models as one input, but do not stick slavishly to them. We believe that models are a useful check, but structural changes and unquantifiable factors need to be considered as well when making overall recommendations.   GAA Equity Sector Selection Model Chart 4Overall Model Performance Overall Model Performance Overall Model Performance The GAA Equity Sector Model (Chart 4) is updated as of May 29, 2020. The model’s relative tilts between cyclicals and defensives have changed compared to last month. The model reversed its defensive stance implemented throughout March and April and is now tilted towards cyclical sectors. However, the semi-defensive tilt led the model to outperform its benchmark by 21 basis points during May. Year-to-date, the model has outperformed its benchmark by 88 basis points, and 86 basis points since inception. The model’s global growth proxy improved – mostly driven by EM currencies and commodity prices, and therefore turned positive on various cyclical sectors and reversed its defensive stance implemented in March. Global monetary easing and low rates should keep the liquidity component favouring a mixed bag of cyclical and defensive sectors. The valuation component remains muted across all sectors except Energy. However, multiple sectors are approaching expensive and cheap territories – mainly Info Tech (expensive), and Real Estate (cheap). The model awaits confirming momentum signals to change recommendations for that component. The model is now overweight five sectors in total, four cyclical sectors versus one defensive sectors. These are Information Technology, Consumer Discretionary, Communication Services, Materials and Health Care.  Table 3Overall Model Performance GAA Quant Model Updates GAA Quant Model Updates For more details on the model, please see the Special Report “Introducing the GAA Equity Sector Selection Model”, dated July 27, 2016, as well as the Sector Selection Model section in the Special Alert “GAA Quant Model Updates,” dated March 1, 2019 available at https://gaa.bcaresearch.com. Table 4Current Model Allocations GAA Quant Model Updates GAA Quant Model Updates Xiaoli Tang Associate Vice President xiaoliT@bcaresearch.com Amr Hanafy Senior Analyst amrh@bcaresearch.com