United States

Dear Client, I am on vacation this week. Instead of our regular report, we are sending you a Special Report from my colleague Jonathan LaBerge. Jonathan explores the risks posed to commercial real estate and the banking system from work from home policies and the potential for urban flight towards less populated and more affordable areas. I hope you find his report insightful. Best regards, Peter Berezin, Chief Global Strategist Highlights Despite pronouncements that the “office is dead,” there are several arguments against the idea that working from home policies or urban flight will become broad-based and spell disaster for commercial real estate loans and the economy. However, the reality is that no one truly knows what the office environment will look like as a result of COVID-19. It is quite likely to be negative on balance for owners of office properties, but it is not yet clear whether it will be a marginal or catastrophic effect. Within the US, small banks clearly have more commercial real estate loan exposure than large banks. Applying the recent Dodd-Frank Act Stress Test (DFAST) to small US banks highlights that roughly 2/3rds of small banks might need to raise capital in the scenario modeled by the Fed, underscoring that forbearance and fiscal relief are essential to avoid a very widespread erosion in small bank capital. Still, of the nearly 5,100 banks included in our analysis, only 5 would see their equity capital wiped out by the simulated losses. Incorporating outsized, Work From Home (WFH)-driven CRE loan losses into our test of small banks highlights that WFH policies may act as a moderate “kicker” to severe pandemic-related bank loan losses were they to occur. But it is clear that the latter is by far the core risk facing both the US economy and its financial system. To the extent that the “white flight” phenomenon of the 1950s to 1970s is a reasonable historical analogue for large-scale urban flight today, the experience of Michigan in the 1960s suggests that it would not likely cause widespread problems in the housing market and/or systemic stress in the banking system. But even if large-scale urban flight does not initially occur due to time-saving WFH policies or health & safety concerns, there are some concerning parallels to the severe decay and decline of the city of Detroit that could play out over the coming few years in America’s cities if not prevented by policymakers. This could spur large-scale urban flight for reasons unrelated to WFH policies. The possibility of inadequate fiscal support is the chief risk to our positive cyclical stance towards risk assets and must be continually monitored by investors over the next several months. We expect large bank outperformance at some point over the coming year, reinforcing our positive stance towards value over growth. Feature Chart 1Rising CRE Prices And Deteriorating Fundamentals, Even Before The Pandemic Concern had already been growing among investors over the past few years about the potentially systemic implications of a possible crash in sky-high US commercial real estate (CRE) prices. Chart 1 highlights that overall CRE prices have doubled over the past decade, which has occurred alongside falling real rents (and thus deteriorating fundamentals) in most CRE subcategories. But the COVID-19 pandemic has introduced new risks for US CRE that many investors view as potentially acute. CMBS delinquency rates surged in May and June (but fell in July), led by accommodation and retail properties. And while multifamily and office delinquencies have so far remained low, many investors have questioned whether this can continue if recently enacted work from home policies become permanent and “urban flight” towards less populated and more affordable areas durably takes hold in major US cities. In this report we focus on the issue of WFH policies, the potential for urban flight, and the risk that these factors may pose to the CRE loans of small domestically-chartered US banks (sometimes informally referred to as “community banks”). There are arguments for and against the idea that work from home policies and/or migration out of city centers will have an extremely negative impact on office properties, but the truth is that it is currently a risk of largely unknown magnitude. It is not likely to be positive for owners of office properties, but it is yet unclear how negative it will be. As a result, we address the question as a “what if?” scenario, by stress testing small bank balance sheets. We conclude that the impact of potential WFH-driven CRE loan losses on the banking system is minor compared with the core risks facing the economy and its capital markets: The deeply negative impact of the COVID-19 pandemic on production and spending, and the risk that fiscal relief will fall short of what is required. Did COVID-19 Really Kill The Office? Chart 2Employers Found That Teleworking Worked Well In mid-to-late March, COVID-19 was spreading rapidly in industrialized economies. Following recommended or mandatory stay at home orders from governments, most office-based businesses rapidly shifted to WFH arrangements as an emergency response. However, in the month or two following the beginning of stay at home orders, several national US surveys found many office workers preferred the flexibility afforded by WFH arrangements. Many employers, correspondingly, found that the productivity of their employees did not suffer while working from home, or that it even improved (Chart 2). These findings led many in the business community to conclude that WFH policies are not, in fact, emergency measures that will ultimately be reversed and instead reflect the “new normal” for work. The arrangement ostensibly appears to be a win-win scenario for workers and firms: Employees save time and money not commuting to the office and gain more control over their work schedules, and businesses save money on the rental or purchase of office space. The conclusion for many in the marketplace has thus been that “the office is dead,” with the focus shifting to the potential investment implications. When thinking about the potential consequences that permanent and widespread WFH options may have, there are two distinct issues that must be considered. The first is the degree to which these policies will push up office property vacancy rates, and the second is whether the availability of WFH policies will cause significant urban flight towards less populated and more affordable areas. On the margin, we agree that both events will occur at least to some degree, and thus are likely to be highly unwelcome developments for owners of prime central business district real estate. This is in line with the conclusions of a recent Special Report by my colleague Garry Evans.1 But there are at least a few arguments against the idea that these trends will occur en masse, or that they will spell economic disaster on their own: While surveys show that many employees expect to continue to work remotely after the pandemic ends, these results likely reflect the desire to retain some flexibility afforded by WFH policies. In terms of office property utilization, there is a large difference between an employee never working from an office again and permanently working from home one day per week, and many surveys that have been conducted on the topic are not structured to distinguish between the two. Surveys that specifically ask how long employees expect it will take for them to return to the office and that include “never” as a possible answer imply a considerably lower impact on office space utilization than other surveys would suggest (Chart 3). If the percentage of never-returning workers shown in Chart 3 (5%-7%) is accurate and maps closely to the expected rise in the office vacancy rate, Chart 4 highlights that the corresponding increase in vacancy would not be unprecedented: It rose from roughly 8% in 2000 to 17% in 2003, without causing a disastrous collapse in office property prices (they fell, but not enormously). Today the vacancy rate would be rising from a much higher level than in 2000, but the point is that very significant changes have occurred in the vacancy rate before without substantially destabilizing the office property market. Chart 3Employers Found That Teleworking Worked Well For offices that reopen before the end of the pandemic, the need for physical distancing will act to at least somewhat restrain a rise in the vacancy rate over the coming several months, as it implies the need for more physical space per employee rather than less. Chart 4Large, Non-Impactful Rises In The Office Vacancy Rate Have Occurred Before Some surveys suggest that Americans are already starting to change their minds about their desire to move out of the city. In April and early-May, upwards of 35%-40% of people responding to a Harris poll said that the pandemic made them want to live either in a rural area more than 21 miles outside of a major city or a suburb within 10 miles of a major city. As of late-July / early-August, that number had fallen to 26% (Chart 5), with only 9% reporting that it is “very likely.” This suggests that the end or reduction of lockdown measures may have returned a sense of normality for many Americans, and that the ultimate degree of urban flight may end up being considerably smaller than some investors expect. Chart 5Few People Say It Is Very Likely They Will Move Due To COVID-19 Finally, the example set by Facebook in May suggests that employees who wish to work from home permanently and relocate to more affordable areas will experience salary reductions, as part of a plan to “localize employees' compensation.”2 If adopted on a widespread basis among firms offering their employees the option to permanently work from home, localized compensation will very likely erode some of the cost advantages of moving to a cheaper area, and thus is likely to result in even fewer employees choosing permanent WFH arrangements. However, even after considering these arguments, the bottom line for investors is that no one truly knows what the office environment will look like as a result of COVID-19, because it hinges both on the evolution/resolution of the pandemic as well as potentially ephemeral human sentiment and behavior – both of which are extraordinarily difficult to predict with high accuracy. It is quite likely to be negative on balance for owners of office properties, but it is not yet clear whether it will be a marginal or catastrophic effect. As such, we agree that the chance of a major and lasting shock to the holders of US commercial real estate loans warrants a thorough investigation, focused on its potential to affect the stability of the US financial system. We first present an overview of CRE exposure for all US banks, and then examine in detail the risk facing small domestically-chartered US banks. Reviewing US Bank CRE Exposure Table 1 presents an overview of CRE loan exposure for domestically-chartered US banks from the Fed’s H.8 data release (Assets and Liabilities of Commercial Banks in the United States), as well as a breakdown in exposure for large and small banks. Investors should note that different definitions of “large banks” exist in the US, and in the H.8 release they are defined as the top 25 domestically-chartered banks ranked by domestic assets. Table 1Most US Commercial Real Estate Loans Are Held By Small Banks Table 1 highlights two points. First, while CRE loans account for approximately 13% of total US domestically-chartered bank assets, exposure is clearly more concentrated for smaller banks than for the largest banks. CRE loans account for a full 1/4th of total assets for small banks, compared to just 6% for the top 25 domestic banks. Given this, the focus of our report will be on small rather than large bank exposure to CRE loans. Second, the table makes it clear that loans backed by nonfarm nonresidential structures account for just 2/3rds of total CRE exposure; the remaining exposure is to apartment buildings, construction and land development loans, and farmland. While not shown in Table 1, bank call reports also highlight that 1-4 family residential construction loans are included in the overall construction and land development category, accounting for up to 20% of those loans for small domestically-chartered banks. Chart 6Office Properties Make Up About 40% Of The Value Of Commercial Structures Unfortunately, it is difficult to break down small bank nonfarm nonresidential structure exposure by property type from a top-down perspective. Chart 6 highlights that office properties (including all financial buildings) make up approximately 37% of the current-cost net stock of US nonresidential commercial and health care structures, whereas office loans make up approximately 30%-40% of those included in US commercial mortgage-backed securities. For the purposes of our analysis, we assume that 40% of small domestically-chartered US banks’ nonfarm nonresidential property loans are secured by office properties. Stress Testing Small US Banks The first step in stress testing small US bank CRE exposure is to simply apply the recent Dodd-Frank Act Stress Test (DFAST) that was focused on large banks to the approximately 5,100 small banks in the US. We use Q1 bank call reports (which we use as a pre-COVID benchmark) sourced from the Federal Financial Institutions Examination Council (FFIEC) to test the breadth of the impact on small banks, and include essentially all US banks in our list except the top 25 banks by assets (those designated as “large” in the Fed’s H.8 release). The Federal Reserve recently released the 2020 DFAST results, which examined the impact on capital ratios of 33 large US banks in a “severely adverse” economic scenario. The scenario modeled by the Fed resulted in $553 billion in projected losses on loans and other positions for the banks included in the test over a 2-year period, of which $433 billion were from accrual loan portfolios (Table 2). These projected loan losses corresponded to a 6.3% loan portfolio loss rate; for comparison, Chart 7 highlights that this would represent even higher losses than what occurred during the worst two-year period following the global financial crisis (Q1 2009 – Q4 2010) by roughly one percentage point. Table 2The Fed’s Recent Stress Test Modeled A 6.3% Loan Loss Rate Over 2 Years Chart 7The Fed's Main Loss Scenario Is Meaningfully Worse Than What Occurred After 2008 In combination with additional provisioning, these assumed losses caused a 1.8% projected decline in the aggregate tier 1 capital ratio for the 33 firms participating in the stress test – from 13.6% to 11.8% – and a 1.7% projected decline in the common equity tier 1 capital ratio – from 12% to 10.3% (Table 3). While these declines are not trivial, they are far from a disastrous outcome for the US financial system. The capital ratios shown in Table 3 are relative to risk-weighted assets, and it is important to note that the projected change in capital ratios shown do not match the projected loan losses (plus provisioning) as a percent of risk-weighted assets. This is because projected losses are netted out against the banks’ projected pre-provision net revenue (“PPNR”) in the Fed’s exercise. In short, while the banks’ capital ratios declined roughly 2% in the DFAST scenario, simulated loan losses amounted to roughly 4% of risk-weighted assets and about 1/3rd of tier 1 common equity capital. Table 3Large Bank Capital Ratios Fell In The Stress Test, But Not Dramatically SoTo avoid the need to project PPNR for thousands of small US banks, we use these loan loss metrics (4% of risk-weighted assets and 33% of tier 1 common equity capital) from the 2020 DFAST to represent whether any individual small bank would likely have to raise capital. We also use the overall portfolio loan loss rate of 6.3% to stress small bank balance sheets, rather than a set of loan loss rates by loan type. Chart 8In The Fed’s Main Stress Test Scenario, Many Small Banks Would Likely Have To Raise Capital Chart 8 illustrates the number of small US banks that would “fail” the stated tier 1 common equity and risk-weighted asset thresholds given the DFAST assumptions. Roughly 64% of small banks would fail the equity test and 94% would fail the risk-weighted assets test. Weighting these results by bank assets rather than the number of banks does not generate a materially different result; instead, 63% and 97% of small bank assets would be held by banks failing the equity and risk-weighted assets tests, respectively. This exercise clearly highlights how much better capitalized large US banks are relative to smaller banks, and underscores that the existing forbearance programs and fiscal relief are essential to avoid a very widespread erosion in small bank capital. Still, of the nearly 5,100 banks included in our analysis, only 5 would see their equity capital wiped out by the simulated losses – meaning that while widespread capital raising and the accompanying tightening in lending standards would undoubtedly have a major impact on the economy and capital markets, the solvency of the US banking system is not in question in the scenario modeled by the Fed. Stress Testing Outsized CRE Losses As noted above, we employed the same average loan portfolio loss rate across all loan categories when testing the impact of the DFAST scenario on small banks, including commercial real estate loans. In order to gauge the specific risks facing commercial properties if recent WFH trends persist, we perform two additional exercises. First, we raise CRE loan losses beyond what was assumed in the DFAST scenario (see Box 1) while employing the same 6.3% loan loss rate on all other loan types to measure the incremental WFH effect on small bank balance sheets in a very negative economic scenario. Second, we examine a high CRE loan loss scenario alone, in order to isolate the potential impact of sustained WFH policies. Box 1Simulating Outsized CRE Loan Loss Rates The assumptions detailed in Box 1 result in an overall CRE loan loss estimate of 11.1%, versus the 6.3% assumed in the DFAST. Chart 9 replicates the DFAST scenario shown in Chart 8 but with our outsized CRE loss rate, whereas Chart 10 highlights the isolated impact (i.e., without any losses assumed for other loan categories). Chart 9Adding Outsized CRE Loans To The Stress Test Scenario Only Moderately Increases “Failure” Chart 10Big CRE Losses Alone, With No Other Loan Losses, Would Be A Relatively Minor Problem Two important observations emerge from Charts 9 and 10. First, despite the fact that small US banks carry disproportionately higher exposure to commercial real estate loans than large banks, it seems clear that the isolated effect of WFH policies on CRE loans, even in the extreme, do not amount to a major risk for the banking system. 80% of small US banks would pass our equity capital test, and 70% would pass the risk-weighted assets test, with absolutely devastating and unprecedented office and retail property losses but no losses outside of their commercial real estate portfolio. Second, while our outsized CRE losses would raise the number of banks that fail our equity capital test relative to the base DFAST scenario (from 64% to 74%), it is clear that this pales in comparison to the effect of the other loan losses assumed in the Fed’s stress test. The bottom line for investors is that while WFH policies may act as a “kicker” to severe pandemic-related bank loan losses were they to occur, it is clear that the latter is by far the core risk facing both the US economy and its financial system. Outsized Residential Real Estate Losses: The Elephant In The Room As noted above, the results shown in Charts 8 - 10 only include outsized losses on nonresidential CRE loans (excluding multifamily) in order to test the risk to bank balance sheets of widespread and continued use of highly permissive WFH policies and significantly reduced demand for office properties. On top of that, banks also face the risk of additional potential disruptions to residential real estate loans if the WFH phenomenon morphs into full-blown urban flight. In this scenario, migration out of densely-populated urban areas towards considerably cheaper suburbs and exurbs could possibly lead to significant house price declines in richly-valued metro-areas, leading in turn to defaults on underwater mortgages. Table 2 highlighted that the Fed’s base 2020 DFAST scenario assumed a 1.5% loan loss rate on first-lien mortgages, and a 3.1% loss rate on junior liens and HELOCs over a two-year period. Unfortunately for investors, it is exceedingly difficult to pinpoint the magnitude of urban migration that would be necessary to cause loss rates in line with the DFAST scenario or higher, forcing us to rely on an inferential approach based on historical example. Chart 11“White Flight” In The US: An Analogue For Urban Flight Today? The only meaningful historical analogue that we can identify for the idea of WFH-driven urban flight is the “white flight” phenomenon that occurred in the US from the 1950s to 1970s. During this period, many white middle-class Americans moved from increasingly racially mixed city centers to racially homogenous suburban or exurban areas. The city of Detroit is often cited as an example of the "white flight" phenomenon. Chart 11 shows Detroit’s white population over time, and highlights the sharp decline in the number of white residents that occurred during the 1950s and 1960s. The white share of Detroit’s population fell earlier, beginning after WWII, but this mostly reflected larger increases of the non-white population. Actual “white flight” occurred during the 50s and 60s, when several episodes of racial violence occurred in the United States. In Detroit, this was most clearly epitomized by the 12th Street Riot in 1967, which involved Federal troop deployment and resulted in over 40 deaths and the damage or destruction of over 2,500 businesses. Did “white flight” cause widespread problems for urban housing markets and/or systemic stress in the banking system? Table 4 and Chart 12 suggest that the answer is no. Table 4 highlights that the median real house price in Michigan rose in the 1960s, grew faster than nationwide house prices, and was modestly higher than the national average in 1970. While it is very likely that this reflects outsized suburban house price gains and that urban center prices fell, Chart 12 highlights that there was no noticeable uptick in US banking failures as a share of total depository institutions in the 1960s. Chart 13 also highlights that the late-1960s did not exhibit any particularly unusual behavior for bank stock prices, after considering interest rates and the state of the business cycle. Table 4Real Michigan Home Prices “Outperformed” The US In The 60s Chart 12No Uptick In Bank Failures In The 1960s Chart 13No Unusual Bank Underperformance In The 1960s     The US economy is very different today than it was in the 1960s, and it is possible that “white flight” serves as an insufficient analogue for potential urban flight today. It is also true that real house prices today are considerably higher than in the 1960s and thus have room to fall further. Nevertheless, based on the Detroit experience, our best inference (for now) is that urban flight does not pose a risk of outsized mortgage loan losses for banks. This is reinforced by the fact that mortgage interest rates have fallen to a record low and have the potential to fall even further based on their spread to 30-year Treasury yields (Chart 14), which may act to boost house prices outright or cushion any potential declines. Chart 14Low And Potentially Lower Mortgage Rates Will Help Cushion Any House Prices Declines Is The Real Risk To Cities Urban Flight, Or Urban Blight? In our view, the city of Detroit is a useful case study for two reasons. First, as noted above, it provides us with some sense of whether urban flight has the potential to pose a systemic threat to the financial system. But, second, it also serves as an example of another potential risk of the COVID-19 pandemic: urban “blight,” or decay. Chart 15Progressive Post-War Deindustrialization Hammered Cities Like Detroit The economic and sociological decay of the city of Detroit has taken place over several decades and has been caused by multiple factors whose relative importance is still debated today. But broadly-speaking, Detroit’s decline can be boiled down to three interacting and self-reinforcing sets of factors: Sociological factors: the general post-WWII trend towards suburbanization, rising levels of violent crime, the “white flight” phenomenon, and the outright decline in Detroit’s population that began in the 1950s; Economic factors: the progressive deindustrialization of the US economy that began in the early 1950s, as well as the debilitating effects of high inflation and energy prices in the 1970s and the double-dip recession of the early-1980s on manufacturing employment (Chart 15); Policy factors: the negative impact on city finances, tax competitiveness, and service quality from the previous two factors, as well as poor governance and outright corruption. Even if large-scale urban flight does not initially occur due to time-saving WFH policies or pandemic-related health & safety concerns, there are some worrying parallels to Detroit’s experience that could play out over the coming few years in America’s cities that could cause similarly self-reinforcing effects if not prevented by policymakers. On the economic front, very acute income and wealth inequality arrayed against stout house price gains over the past decade have made home ownership unaffordable for some, increasing the allure of urban flight even if localized compensation programs apply. In addition, the pandemic has most severely affected small retail businesses, raising the specter of a “hollowed out” or abandoned urban retail landscape which could push consumers to avoid shopping and travelling downtown. On the policy front, there is a clear risk that inadequate state & local government funding could contribute to a potential downward spiral of higher taxes, reduced city services, and economic decay – similar to what occurred in Detroit. Chart 16 highlights that the financial situation of state & local governments following the global financial crisis caused persistent fiscal drag for several years into the expansion that followed. This significant fiscal drag contributed importantly to the subpar nature of the expansion, and the odds that this will occur again without federal funding are high. Chart 16 shows that the contribution to real GDP growth from state & local government spending has again turned negative, and the US Center on Budget and Policy Priorities is currently forecasting state budget shortfalls of approximately $555 billion over state fiscal years 2020-2022 – in line with the $510 billion cumulative shortfall that occurred from 2009-2011.4 Finally, in this scenario, the sociological factor somewhat mimicking Detroit’s experience could be a significant rise in urban crime (especially if violent). This could cause urban flight for reasons totally unrelated to WFH policies, but if it occurred it would likely reinforce both the failure of urban center businesses and the deterioration in state & local government finances (risking a downward spiral). Chart 17 highlights that murders have already significantly increased this year in major American cities (by mid-year) relative to 2019, although other types of violent crimes have fallen.5 A trend of rising urban crime could also be sparked or accelerated if recent calls to cut police department funding in favor of other social services succeed, and if those newly funded initiatives fail to effectively prevent criminal activity. Chart 16Persistent State & Local Fiscal Drag Must Be Prevented This Time Chart 17Will US Cities Become Unsafe? While this scenario is far from our base case view, it underscores how urban flight and the accompanying second round effects on commercial real estate loans and the banking system could occur following the pandemic even if not triggered by WFH policies. It also underscores the great importance of Federal fiscal relief efforts: not only to households and businesses, but as well to state & local governments. Investment Conclusions Our analysis above points to three main investment conclusions: First, while there are arguments for and against the idea of significant CRE losses stemming from the widespread adoption of permanent WFH policies and the potential for large-scale urban flight, the uncertainty surrounding the question will likely linger for the coming few months, at a minimum. This suggests that the equity risk premium applied to bank stock prices may remain elevated in the near term. Chart 18Large US Banks Unduly Cheap Second, while large-cap banks may struggle to outperform in the near term due to this elevated risk premium, it is clear that large banks are far less susceptible than small banks to not only potential CRE loan losses, but also to the severely adverse economic scenario modeled in the Fed’s recent stress test. Our calculations suggest that large bank capital ratios would only marginally decline from the ending ratios shown in the DFAST scenario even with the outsized CRE loan loss scenarios that we used to stress test small bank balance sheets, and we highlighted how the Fed’s main stress test scenario involved 2-year loan losses in excess of what occurred in 2009-2010. Consequently, the collapse in large-cap bank valuation ratios seems unwarranted (Chart 18), and we would expect large banks to outperform the broad market at some point over the coming 6-12 months (and possibly even over the coming 0-3 months). This is also consistent with our expectation that value stocks are likely to outperform growth stocks at some point over the coming year.6 Third, while investors are often right to ask what risk they are “missing,” our analysis above highlights that the impact of potential WFH-driven CRE loan losses on the banking system is minor compared with the core risk facing the economy and its capital markets: The deeply negative impact of the COVID-19 pandemic on production and spending, and the risk that fiscal relief will fall short of what is required. This need for relief extends very significantly to state & local governments, and a failure to adequately resolve the substantial state budget shortfalls that will occur due to the pandemic and its aftermath would all but guarantee a repeat of the persistent fiscal drag that contributed to the subpar nature of the recent economic expansion. Our base case view remains that US policymakers will do what is necessary to avoid a very negative economic outcome and that the hiccup in congressional negotiations is temporary, but the possibility of inadequate fiscal support is the chief risk to our positive cyclical stance towards risk assets and must be continually monitored by investors over the next several months. Stay tuned! Jonathan LaBerge, CFA Vice President Special Reports Footnotes 1  Please see Global Asset Allocation / Global Investment Strategy Special Report, “The World After COVID-19: What Will Change, What Will Not?” dated August 7, 2020. 2 “Facebook employees could receive pay cuts as they continue to work from home,” USA Today, dated May 21, 2020. 3 Please see US Investment Strategy Special Report, “Mallpocalypse, Part 1: An Overnight Collapse Decades In The Making,” dated August 17, 2020 for the first of two reports presenting a detailed analysis of the challenges facing US retail properties. 4 Elizabeth McNichol and Michael Leachman, “States Continue to Face Large Shortfalls Due to COVID-19 Effects,” Center on Budget and Policy Priorities, Updated July 7, 2020. 5 Jeff Asher and Ben Horwitz, “It’s Been ‘Such a Weird Year.’ That’s Also Reflected in Crime Statistics.,” The New York Times, Updated August 24, 2020. 6 Please see Global Investment Strategy Weekly Report, “The Return Of Nasdog,” dated August 21, 2020. Global Investment Strategy View Matrix Current MacroQuant Model Scores  

Highlights US-China relations in 2020 consist of a gentleman’s agreement to keep the Phase One trade deal in place and aggressive maneuvering in every other policy area. Stimulus is unlikely to be curtailed in the US or China yet, which means brinkmanship will eventually lead to a negative surprise for markets. But it is just as unlikely to come after the election as before. Joe Biden would only initially benefit Chinese equities – trade and tech conflict is a secular trend. North Korea is not a red herring, but South Korea is still a geopolitical investment opportunity more than a risk, especially relative to Taiwan. Feature Chart 1US Power Struggle Raises Risk To Rally The “everything is awesome” rally continues, with US tech stocks unfazed by rising domestic and international risks. However, according to The Lego Movie 2, everything is not that awesome. The Treasury market smells trouble and long-dated yields remain subdued, despite a substantial new dose of monetary policy dovishness (Chart 1, top panel). In the near term we agree with the bears and remain tactically long 10-year Treasuries. Global policy uncertainty remains extremely elevated despite dropping off a bit from the heights of the pandemic lockdowns. US uncertainty, which is now rising relative to global, will climb through November and possibly all the way through Inauguration Day on January 20 (Chart 1, bottom panels). A contested election is not a low-probability event now that President Trump is making a comeback in the election race. President Trump’s comeback could generate a counter-trend bounce in the US dollar (Chart 2A). His comeback is not based in online betting odds but in battleground opinion polls (Chart 2B). Former Vice President Joe Biden is currently polling the same against Trump as Hillary Clinton did in 2016. Chart 2ATrump Staging A Comeback, But US Consumers Flagging Chart 2BTrump Staging A Comeback, But US Consumers Flagging Why should Trump be less negative for the greenback than Biden? First, Trump is a protectionist who would turn to aggressive foreign and trade policy when it became clear that most of his other legislative priorities would not make it past the Democratic House of Representatives. Unilateral, sweeping tariffs against China, and possibly the EU and various other nations, would weigh on global trade and economic recovery and hence support the dollar. Second, Trump’s populism means he would pursue growth at all costs, which means that US growth would increase relative to that of the rest of the world. Democrats, by contrast, would raise taxes and regulations that would have to be offset by new spending, weighing on growth at least at first. Thus Trump would inject animal spirits into the US economy while dampening those spirits abroad; Biden would do the opposite. The dollar may not rally sustainably, but it would be flat or fall less rapidly than if Biden and the Democrats reduced trade risks abroad while deterring domestic private investment. It is not yet clear that Trump’s comeback will have legs. The nation is still in thrall to the pandemic, recession, and social unrest, which undermine a sitting president. US consumer confidence has fallen, as anticipated (Chart 2, bottom panel). Trump should still be seen as an underdog despite his incumbent status. A Trump comeback could precipitate a counter-trend bounce in the US dollar. Nevertheless, our quantitative election model gives Trump a 45% chance of victory, up from 42% last month. Florida has shifted back into the Republican column – albeit as a “toss up” state with a roughly even chance of going either way (Chart 3). The shift reflects improvement in state leading economic indexes as a result of the V-shaped recovery in the economy thus far. Chart 3Trump Nearly Regains Florida In Our Quantitative Election Model, Odds Of Victory 45% Assuming Trump signs a new relief bill in September, which is working its way through Congress as we speak, we will upgrade our subjective odds from 35% to something closer to our quantitative model (and the market consensus). While Trump is less negative for the dollar than Biden, the dollar may fall anyway, at least beyond any near-term bounce. First, monetary policy is ultra-dovish. As we go to press, Fed Chairman Jerome Powell has given a sneak preview of the Fed’s strategic review of monetary policy at the Kansas City Fed’s annual Jackson Hole summit (this time hosted in cyberspace instead of Wyoming). Powell met expectations that the Fed will adopt average inflation targeting. Inflation will be allowed to overshoot the 2% inflation target to compensate for periods of undershooting. Maximum employment will be the goal rather than an attempt to prevent excessive deviation from the Fed’s estimates of neutral unemployment. This means US growth and inflation will push real rates lower and weaken the dollar. Moreover, as mentioned, Trump’s big spending would eventually drive investors away from the dollar, especially in the context of global economic recovery. Trump, like Biden, would refuse to impose fiscal austerity amid high unemployment. The one area where he would be able to compromise with House Democrats would be spending bills, as in his first term. The US budget deficit and trade deficit would remain very large, showering the world with dollar liquidity. Risk-on currencies will attract buyers in a new global business cycle. Republicans and Democrats have released their policy platforms following their national conventions. We will revisit these platforms in detail in a future report. The Democratic platform is the one that matters most because the Democrats are more likely to win full control of Congress and thus be capable of enacting their preferred policies. Their platform is reflationary, but in seeking to rebalance the economy to reduce financial and social disparities through more progressive tax policy it would offset some of the fiscal spending. Biden would also moderate foreign policy and trade policy, launching a new dialogue with China to manage tensions. The dollar would fall faster in this environment. Bottom Line: President Trump is staging a comeback in the election campaign. If the comeback receives a boost from fiscal stimulus, Trump could pull off a Harry Truman-style surprise victory. This would precipitate a bounce in the US dollar in the near term. Over the medium term, the dollar should continue falling due to US debt monetization and global recovery. The Trump-Xi Gentleman’s Agreement Has Two Months Left Financial markets have largely ignored US-China strategic tensions this year because the two countries are puffing themselves up with monetary and fiscal stimulus. Going forward, either the stimulus will falter, or the US-China conflict will escalate to the point of triggering a negative surprise for markets. Chart 4US-China: Embracing While Struggling China is unlikely to pull back on stimulus measures. It cannot do so when unemployment has spiked and the economy is experiencing the weakest growth in over 40 years. Authorities said as much during the annual July Politburo meeting on the economy (a meeting that has often marked turning points in policy), when they pledged to maintain accommodative policy and to speed up local government issuance of special bonds. Money supply is growing briskly. The market is validating the signal from China’s easy monetary policies and robust credit expansion. Our China Play Index – which consists of the Australian dollar, iron ore prices, Brazilian equities, and Swedish equities – continues to rally smartly, breaking above its 2019 peaks (Chart 4, top panel). The risk to this view is that the People’s Bank of China may not provide additional monetary easing in the near term, as the Politburo signaled that monetary policy would be more flexible and targeted in the second half of the year. The three-month Shanghai interbank rate has been rising since April. Politically, Chinese authorities would benefit from releasing negative news or statements that would undermine President Trump’s reelection campaign. However, Beijing would not make consequential moves merely to spite Trump. Its primary interest lies in its own stability. Credit growth will continue growing at its current clip through most of the rest of the year and fiscal spending will expand, particularly to support infrastructure projects. The US Congress is also likely to add more stimulus before the election, as noted above. Thus with both countries stimulating, the risk is that they escalate their strategic confrontation to the point that it causes a negative surprise in financial markets. Will this occur? The US-China relationship in 2020 has been characterized by (1) a gentleman’s agreement to adhere to the Phase One trade deal, which was reaffirmed by top negotiators this week; (2) an aggressive pursuit of national interest in every other policy area. Beijing accelerated its power grab in Hong Kong; the US accelerated up its ban on Chinese tech. Chinese imports of US commodities are naturally much weaker than projected due to economic reality but neither side has an interest in exiting the deal. The renminbi continues to appreciate against the dollar on the back of Chinese and global recovery (Chart 4, second and third panels). Nevertheless a new burst of stimulus will lower the hurdle to President Trump taking additional punitive measures against China. The administration could have paused after its major decision to finalize its ban on business with Huawei and other tech firms, which ostensibly even extends to foreign firms that use US-designed parts in sales to China. It did not. Trump is maintaining the pressure with new sanctions over China’s militarization of the South China Sea. Washington is also likely to kick Chinese companies off US stock exchanges if they fail to meet transparency and accounting standards. Trump is not only burnishing his “tough on China” credentials against Democratic candidate Joe Biden – the US’s recent measures are unlikely to be repealed under either president in the coming years. Chart 5China Faces Internal And External Political Pressures Therefore stimulus will enable US actions and Chinese reactions that will eventually trigger a pullback in financial markets. Chinese tech equities are reflecting this headwind. Equities ex-tech are likely to outperform (Chart 5, top panel). A Biden victory does not prevent Trump from taking punitive measures against China on his way out of office, to solidify his legacy as the Man Who Confronted China, so Chinese tech will remain at risk. Biden would be more favorable for emerging market equities because his administration would speed the dollar’s decline. A change of government in the US would temporarily disrupt the US’s overall policy assault against China. Biden’s foreign and trade policies would be more predictable and orthodox than Trump’s. Over a twelve month period, after a shot across the bow to warn that he is not a lightweight, Biden would probably attempt a diplomatic reset with China – a new round of engagement and dialogue that would support the Chinese equity rally. Eventually this reset would fail, however, and Biden would all the while be working up a coalition of democracies to pressure China to change its behavior – not only on trade but also on unions, carbon emissions, and human rights. Externally focused Chinese companies will remain exposed to the harmful secular trend of US-China power struggle regardless of the US election outcome. Coming out of the secretive leaders’ conclave at the Beidaihe resort in August, it is clear once again that Chinese domestic politics is not conducive to smooth US-China relations. Chinese political risk remains underrated. Our GeoRisk indicator is gradually picking up on this trend, and so are other quantitative political risk indicators such as that provided by GeoQuant (Chart 5, second panel). President Xi Jinping has been dubbed the “Chairman of Everything” due to his tendency to promote a neo-Maoist personality cult and thus shift Chinese governance from consensus-rule to personal rule. He is once again reportedly considering taking on the title of “Chairman” of the Communist Party, a position that only Mao Zedong has held.1 More importantly he is re-energizing his domestic anti-corruption campaign, i.e. political purge, this time against law enforcement. Xi had already seized control of China’s domestic security forces but controlling the police is even more critical in a period of high unemployment, slow growth, and social unrest (Chart 5, third panel). Xi’s attempt to re-consolidate power ahead of the Communist Party centennial in 2021 and especially the twentieth national party congress in 2022 is already under way. China’s domestic and international political environment is a risk for the renminbi, which we noted is rallying forcefully on the global rebound. We will not join this rally until the US election is decided at minimum. With the US posing a long-term threat, Beijing is speeding up its attempts to diversify away from the US dollar, both in trade settlements and foreign exchange reserves. Reliance on the dollar leaves Chinese banks and companies vulnerable to US financial sanctions, which have harmed US rivals like Russia and Iran. Over the long run there is a lot of upside for the yuan given its very low level of global penetration (about 2% of both SWIFT transactions and global foreign exchange reserves) and yet China’s very high share of global trade (about 15%). Cross-border settlements in RMB are recovering gradually after the steep drop-off following 2016. Beijing is also allowing foreign investors greater access to onshore financial markets where they will hold more and more RMB-denominated assets. However, the yuan will not become a reserve currency anytime soon given China’s state-controlled economy and closed capital account. We favor the euro, yen, and other G7 currencies as alternatives to the dollar. Hong Kong equities have suffered from the combination of Xi Jinping’s centralization of power and the US-China strategic conflict. The above analysis suggests that while they may get a temporary reprieve, the secular outlook is uninspiring. However, the Hong Kong monetary authorities are capable of managing the dollar peg. They have been able to manage dollar strength over the past decade, including the COVID-19 dollar run-up, and foreign exchange reserves are more than ample. By contrast, a sharp drop in the dollar can be handled even more easily by printing additional HKD. Eventually shifting to a trade basket, or a renminbi peg, is to be expected. The US election may support the Chinese equity rally if Biden wins, but tech equities should continue to underperform the rest of the bourse due to US grand strategy. Bottom Line: We prefer to play China’s growth recovery via outside countries that export into China, such as Sweden, Australia, and Brazil. The US election may support the Chinese equity rally if Biden wins, but tech equities should continue to underperform the rest of the bourse due to US grand strategy which will remain focused on constraining China’s tech ambitions. North Korea Is Not A Red Herring – But Taiwan Is Entirely Underrated The Taiwan Strait remains the chief geopolitical risk. Xi Jinping’s reassertion of Beijing’s supremacy within China’s sphere of influence has led to a backlash in Taiwanese politics and a confrontational posture across the Strait that is being expressed in saber-rattling and low-level economic sanctions that could easily escalate. Chart 6Taiwan Remains #1 Geopolitical Risk Military exercises and jingoistic rhetoric are also heating up, not only directly relating to Taiwan but also in the neighboring South China Sea, which is critical to national security for all geopolitical actors in Northeast Asia. On August 26 Beijing testing two anti-ship ballistic missiles known as “aircraft carrier killers” in the South China Sea (the DF-21D and the DF-26B). We have long argued that the lack of clarity over whether the US would uphold its defense obligations to Taiwan makes the situation ripe for misunderstandings. The US Naval Institute has recently confirmed the validity of fears about a full-scale conflict in the near term.2 Neither Beijing nor Taipei nor Washington has crossed a red line. But China’s imposition of legislative dependency on Hong Kong highlights the incompatibility of the Communist Party’s governing model with western liberalism. The “one country, two systems” formulation has become unacceptable to the Taiwanese people, who want to preserve their autonomy indefinitely. The US ban on doing business with Huawei extends to foreign companies that use US parts or designs, squeezing Taiwanese companies (Chart 6, top panel). War is possible, but our base case still holds that the mainland will first use economic means. In particular it will impose economic sanctions, either precipitating or in response to a Fourth Taiwan Strait Crisis. The market continues to underrate the enormous risk to the Taiwanese dollar, as captured by the low level of our risk indicators (Chart 6, second panel). We continue to recommend shorting Taiwan relative to emerging markets. Taiwan is a short relative to South Korea, in particular, which stands to benefit from any negative turn of events in cross-strait relations. Korean equities are finally perking up, though the US tech war with China is weighing on the South Korean tech sector (Chart 7, top panel). We see this as a geopolitical opportunity given that both China and the US will need South Korean companies as they divorce each other. Korean political risk, however, may also be shifting from adequately priced to underrated. The risk premium has trended upward since President Trump’s diplomatic overture to leader Kim Jong Un stopped making progress (Chart 7, second and third panels). We have largely dismissed concerns about North Korea since the reduction of tensions in late 2017 due to our assessment that diplomacy would remain on track throughout Trump’s first term. This has proved to be the case, but it is still possible that North Korea could prove globally relevant before the US election. Chart 7North Korea A Non-Negligible Risk The reason stems from rumors of Kim Jong Un’s health problems earlier this year. We noted at the time that it was suspicious that preparations for Kim’s sister, Kim Yo Jong, to take on greater responsibilities within the Politburo of the Worker’s Party seemed to predate reports of Kim Jong Un’s illness. For the North Korean state to continue to promote her implies that something may indeed be amiss. In fact, she has missed two Politburo meetings after her aggressive public relations campaign against South Korea was called off this summer. It is possible she got too much attention as the Number Two person in the regime. The South Korean National Intelligence Service is debating her status with the Defense Ministry and Unification Ministry. What is clear is that Kim Jong Un is preparing a new five-year economic plan, to be launched in January 2021, and that he is eager to share any blame for disastrous internal conditions in the country amid the global pandemic and recession. The market is typically correct not to hyperventilate over North Korean risks, but after 2016 North Korea is no longer a “red herring.” First, any domestic power struggle would occur at an immensely inconvenient time given the breakdown in US-China trust. Second, as the North manages any internal problems through its opaque and untested political process, it could be pressed into making a show of force that would either embarrass and antagonize President Trump, or provoke a forceful response from a future President Biden, given that North Korea in theory has the raw capability to deliver a crude nuclear weapon to the continental United States. If any US president makes a show of force, it will antagonize China and could lead to a major standoff. This would upset the markets at least temporarily. We are long Korean equities and would also look favorably on Korean tech. A geopolitical risk premium could temporarily undercut these stocks if North Korean diplomacy fails around the US election. But the risk is globally relevant only if Pyongyang somehow sparks a standoff between the US and China. Otherwise a major Korean peninsula crisis is far less of a concern than that of a crisis in the Taiwan Strait.   Matt Gertken Vice President Geopolitical Strategy mattg@bcaresearch.com   Footnotes 1Financial Times. 2 See Admiral James A. Winnefeld and Michael J. Morell, "The War That Never Was?" US Naval Institute Proceedings 146: 8 (August 2020), usni.org. Section II: GeoRisk Indicator China Russia UK Germany France Italy Canada Spain Taiwan Korea Turkey Brazil Section III: Geopolitical Calendar

Highlights The leading providers of fundamental oil data differ sharply in their estimates of demand destruction caused by the COVID-19 pandemic. This will keep uncertainty over the trajectory of prices elevated this year. Our forecast of demand destruction and those of the US EIA and the IEA are clustered around 8mm b/d for 2020, while OPEC’s most recent estimate exceeds 9mm b/d. The rebound in actual demand, which is apparent in the data, a weakening USD and strengthening of global trade in the wake of global fiscal and monetary stimulus support our expectation of lower demand destruction. As mentioned last month, we believe the odds of a COVID-19 vaccine are high by year-end or early 2021 (Chart of the Week). Against this, bloated floating storage levels – and their attendant port congestion – threaten to slow crude-oil demand growth in Asia into 4Q20, particularly if China follows through on putting 19 very large crude carriers (VLCCs) filled with oil from the US on the water over the coming months. We continue to see the balance of risk favoring the upside for prices. We are raising our 2H20 Brent forecast $2/bbl to $46/bbl, and keeping our 2021 expectation at $65/bbl. WTI will trade ~ $3/bbl below those levels.   Feature OPEC continues to see a higher level of demand destruction in the wake of the COVID-19 pandemic than BCA, the US Energy Information Administration (EIA) and the Paris-based International Energy Agency (IEA). OPEC continues to see a higher level of demand destruction in the wake of the COVID-19 pandemic than BCA, the US Energy Information Administration (EIA) and the Paris-based International Energy Agency (IEA). The cartel’s economists are estimating global oil-demand destruction would be ~ 9mm b/d year-on-year (y/y) in 2020. In their August projections, the EIA’s and IEA’s expectations for demand destruction are closer to ours at ~ 8mm b/d for this year. In the past, we focused more on OPEC’s output estimates for members of the cartel, particularly for its leader and top producer, the Kingdom of Saudi Arabia (KSA). In this month’s report, and in subsequent reports, we are incorporating OPEC’s demand estimates as a direct input to our price-forecasting models. For 2020, we are giving it an equal weight to the apparent consensus we share with the EIA and IEA. Chart of the WeekActual Oil Demand Continues Strong Recovery The immediate effect of this will be to temper the effect of the stronger demand growth expectations we share with the EIA and IEA in this year’s price forecast, which will put us at $46/bbl on average for 2H20.1 The improvement in actual demand is apparent in our base case model up to July, as seen in the Chart of the Week. Much of this recovery is the result of the massive fiscal and monetary stimulus deployed globally by governments and central banks, which will continue to support the demand this year and next.2 This stimulus also is visible in global trade data – particularly in EM imports, which we follow closely, given their high sensitivity to changes in income (GDP). Our modeling indicates this recovery will continue to year-end (Chart 2). Chart 2EM Imports Recovery Likely Continues Weaker USD Will Support Oil-Demand Recovery Speculators have crowded into the short-dollar trade, which augurs for a near-term correction in the USD DXY futures. We expect the USD to continue to weaken on the back of the Fed’s aggressive monetary accommodation, in line with our Global Investment and FX strategists.3 This will support the continued rally in crude oil prices we expect for the balance of this year and next. There are a number of short-term risks to our bearish USD view, however. These are mainly due to the marginal improvement of the US economy vis-à-vis Europe, which is evident in the manufacturing and services PMIs (Chart 3). Improving mobility data, which is coincident with the decline in its number of COVID-19 cases vs Europe, also is supportive of the USD (Chart 4). In the trading markets, speculators have crowded into the short-dollar trade, which augurs for a near-term correction in the USD DXY futures. Close to 60% of the DXY index is accounted for by the Euro (Chart 5). Lastly, while global economic policy uncertainty has fallen from its recent peak, taking the USD lower with it, it still is elevated and continues to represent a risk to the USD bear market (Chart 6). Chart 3USD Bear Market Could Stall All else equal, a weakening USD will continue to support Brent prices, and with that the rest of the global oil complex. As long as EM growth continues to improve, these short-term USD effects discussed above will affect the DXY more than the broad trade-weighted index (TWIB) for the USD, which has a Euro weight of 18% and is a more representative gauge of USD strength vis-à-vis trade. Chart 4DXY Could Rally Briefly As US Recovers Chart 5Specs Have Crowded Into The Short USD Trade Chart 6A Weaker USD Will Boost Oil Prices OPEC 2.0 Discipline, Capital Markets Will Restrain Supply While we expect some of this US production to come back on line as prices improve, overall output in the shales likely will continue to fall until 2H21. OPEC 2.0 production discipline largely is responsible for the 6.1mm b/d y/y decline in global oil production we estimate. The producer coalition’s putative leaders – KSA and Russia – continue to lead by example, having removed 460k b/d and 900k b/d y/y, respectively, from the market (Chart 7, top panel). We expect this to continue into next year (Table 1). Outside OPEC 2.0, US oil production is estimated to have fallen ~ 2mm b/d from its peak of 12.9mm b/d in 4Q19, in line with our expectation. This is largely the result of significantly reduced shale-oil output (Chart 7, bottom panel). While we expect some of this US production to come back on line as prices improve, overall output in the shales likely will continue to fall until 2H21. Chart 7OPEC 2.0, US Shales Output Will Remain Constrained Table 1BCA Global Oil Supply - Demand Balances (MMb/d, Base Case Balances) We continue to expect the combination of sustained demand growth and constrained supply to tighten balances globally, producing a physical deficit this year and next (Chart 8). As before, we expect this physical deficit to translate into lower inventories in the OECD, as refiners are forced to draw down stocks to meet demand (Chart 9). Chart 8Supply-Demand Balances Continue To Tighten Chart 9Expect OECD Inventories To Draw Asia’s Floating Storage Set To Increase Additional imports of US crude by China threaten to increase floating storage levels in Asia, which likely will reduce Chinese demand for light-sweet barrels sourced ex-US, and reverse the tightening in Brent and WTI forward curves. The inventory draw in the US – the largest component of OECD crude inventories – could accelerate if China follows through on its planned increase in imports of US crude oil, consistent with reports oil companies there recently chartered 19 VLCCs.4 This apparently was done to comply with the Phase 1 trade deal China and the US negotiated earlier this year. While these imports of US crude into China will lower US inventories – most of the tankers are shipping from the US Gulf – they could add to the swollen floating storage levels currently clogging Chinese harbors, and reduce demand for additional crude until these stocks are absorbed either by refinery demand or strategic reserves (Chart 10). In this context, we also are watching the evolution of forward curves, particularly for Brent, as this surge in Chinese imports could back out other light-sweet crudes similar to those produced in the US – and similar to Brent, for that matter – leaving them distressed on the water looking for a home. There already is early evidence the Brent forward curve’s flattening and expected return to backwardation has stalled (Chart 11). Given this pause in the tightening of the forward curve over the next year, we are closing our 4Q20 backwardation trades at tonight’s close. Chart 10China’s US Crude Imports Will Swell Asia’s Floating Storage Chart 11Brent Forward Curves Weaken Over the Next 12 Months Bottom Line: Global crude oil markets continue to tighten, as demand recovers, and supply discipline remains intact. However, additional imports of US crude by China threaten to increase floating storage levels in Asia, which likely will reduce Chinese demand for light-sweet barrels sourced ex-US, and reverse the tightening in Brent and WTI forward curves. This is prompting us to exit our 4Q20 backwardation trades at tonight’s close. The balance of price risks continues to favor the upside, in our estimation. We are raising our 2H20 Brent forecast slightly by $2/bbl to $46/bbl, and keeping our 2021 expectation at $65/bbl. WTI will trade ~ $3/bbl below those levels (Chart 12). Chart 12BCA Oil Price Trajectory Unchanged   Robert P. Ryan Chief Commodity & Energy Strategist rryan@bcaresearch.com Hugo Bélanger Associate Editor Commodity & Energy Strategy HugoB@bcaresearch.com   Commodities Round-Up Energy: Overweight US natgas prices were up this week as Hurricane Marco and Laura approached the Gulf of Mexico (GoM). Oil and gas producers evacuated part of their offshore facilities and shut 61% of their natural gas production from the region. Cheniere reported it suspended operation at its Sabine Pass LNG export terminal located in the GoM. Still, abundant oil and gas storage in the US limits the price increase. Separately, the latest estimate of US LNG cargo cancellations for October delivery – down to only 10 cancelled vessels – confirms natgas price spreads relative to Asia have reached levels sufficient to incentivize additional exports of US gas, supporting Henry Hub prices (Chart 13). Base Metals: Neutral Copper theft is rising in Chile and Africa, boosting LME copper prices and pushing treatment and refining charges lower. In Chile, “About 40 incidents were reported in the first half of this year, up from six in 2014,” according to mining-journal.com. Falling spot supplies in the wake of COVID-19 mining restrictions likely explain the thefts and increases in the cash-to-three-month copper spread on the LME to $17.25/MT earlier this week. Precious Metals: Neutral We closed our long gold recommendation for a 20% profit after reaching our $1,950/oz stop loss on August 11. We remain positive on gold strategically based on our view the dollar will continue depreciating and the Fed will keep rates low even as inflation and inflation expectations move up. As we go to press, markets await Jerome Powell’s speech at the annual Jackson Hole summit, where he is expected to discuss the Fed’s strategic review of its monetary policy strategy. This could push inflation breakevens slightly higher, and real yields lower. We are recommending a buy order at $1,875/oz as spec positioning remains stretch (Chart 14). Ags/Softs:  Underweight The USDA reported 69% of the US soybean crop was rated in good to excellent condition this week, up sharply from last year’s level of 55%. However, this was down from last week’s level of 72%, which was supportive of prices. Separately, the Sino-US Phase 1 trade deal is back in the news this week, with reports the two countries agreed to resume shipments of soybeans on a record scale – in the range of 40mm tons for 2020, which would be 10% above record purchases by China set in 2016, according to bloomberg.com. China had turned to Brazil earlier in the year as the trade deal became mired in tit-for-tat tariff spats. Chart 13Natgas Prices Supported By Hurricane Laura Chart 14Gold Vs. USD Spec Positioning Stretched   Footnotes 1     We also reduced the pass-through of the supply-demand difference into the oil price forecast in this month’s report, based on recent research we’ve completed, which also tempers the impact of the stronger growth expectations we share with the EIA and IEA. 2     Please see Low Vol, High Uncertainty Keeps Oil-Price Rally On Tenterhooks published June 18, 2020, for additional discussion of global fiscal and monetary stimulus vis-à-vis oil markets. 3    Please see The Dollar And The Budget Deficit: From Theory To Practice, a Special Report published by BCA Research’s Global Investment Strategy and Foreign Exchange Strategy on August 14, 2020.  It is available at gis.bcaresearch.com. 4    In July and August, China imported ~ 17mm barrels of US crude, according to S&P Global Platts.  Please see Crude moves higher amid China's US import boost ahead of key OPEC meet published by Platts August 17, 2020.  China reportedly charted 19 VLCCs to import ~ 37mm barrels of US crude beginning in August, according to worldoil.com.  Please see China charters 19 tankers for record U.S. crude oil shipment published August 21, 2020.    Investment Views and Themes Recommendations Strategic Recommendations Tactical Trades Trade Recommendation Performance In 2020 Q2 Commodity Prices and Plays Reference Table Trades Closed in 2020 Summary of Closed Trades

Highlights A weak dollar and low bond yields have pushed up the S&P 500 more than anticipated. Cyclical forces favor loftier stock prices in 12 months. Froth creates short-term vulnerabilities that higher yields could catalyze. The lack of yield curve control along with an improving economic outlook and a decline in deflationary risks indicate that Treasury yields will move toward 1% in the coming months. Long-term investors should begin to add small-cap stocks to their core US holdings. Feature The S&P 500 recent all-time high flies in the face of a long list of tactical indicators that flag an elevated risk of correction. The strength of the US equity market is a testament to the power of policy stimulus, the perceived invincibility of tech titans and the hopes that the powerful economic recovery will continue. Although equities will climb in the coming year, a move up in yields should transfer the leadership from tech and growth stocks to value and traditional cyclicals. While these shifts usually do not spell the end of bull runs, often they generate periods of elevated volatility, especially when the displaced leaders account for 40% of market capitalization. Small-cap stocks look increasingly attractive. A Post Mortem We have been cyclically bullish since late March,1 but on June 25th we warned that the S&P 500 would churn between 2800 and 3200 for the rest of the summer.2 This view did not materialize for several reasons. We underestimated the impact of a weak dollar, which has given a second life to the equity bull market. When expressed in euros, the S&P 500 has been flat since June 5 (Chart I-1). Relative to gold, the S&P 500 is down by 9% since June 8, which further highlights how equities have been supported by a weak US currency and a plentiful money supply. Meanwhile, the S&P 500 has outperformed the EURO STOXX 50 by 7.8% since June 5; however, when we factor in the effect of the strong euro, US equities have steadily underperformed the Eurozone benchmark since early May (Chart I-1, bottom panel). Low bond yields have also buttressed US equities. Near-zero interest rates have allowed the valuation of growth stocks to hit extraordinary levels. The NASDAQ trades at 32-times 2020 earnings and 27-times 2021 EPS. The S&P tech is valued at 29-times 2020 EPS and 25-times next year’s profits. In the most extreme cases, the five tech stocks that have accounted for 31.7% of market gains since March 23 (Apple, Amazon, Microsoft, Alphabet and Facebook) trade on average at 40-times 2020 EPS and 32-times 2021 earnings. Low bond yields have also buttressed US equities. Importantly, COVID-19 has had a positive influence on these same tech stocks. According to our European Investment Strategy colleagues, while spending on restaurant, entertainment and retail collapsed during the pandemic, outlays surged on Amazon, Apple products, Netflix subscriptions, etc.3 At the apex of the crisis, online retail sales expanded by 26.3% annually in the US, while bricks-and-mortar sales contracted by an unprecedented -17.7%. Meanwhile, global shipments of personal computers and servers are expanding by 11.2% and 21.5% annually, respectively (Chart I-2, top panel). Therefore, the largest sector of the S&P 500 is outperforming relative to the rest of the market (Chat I-2, bottom panel). As long as investors continue to expect COVID-19 to affect consumer behavior, they will pay a premium for tech stocks that benefit from the pandemic. Chart I-1The Weak Dollar Is Fueling The Recent Rally Chart I-2Earnings Have Supported Tech Stocks   Can Stocks Remain Unscathed? The outlook for stocks is positive, but near-term risks have not dissipated because short-term market conditions remain frothy. Watch for higher bond yields as the force to concretize the tactical risks. The following cyclical forces continue to act as crucial tailwinds for equities: The equity risk premium (ERP) remains low. Computations of ERP must factor in the expected expansion of earnings. To incorporate this alteration, we assume that long-term cash flows will grow in line with potential nominal GDP growth. However, we must also consider the absence of stability of the ERP’s mean. After this adjustment, the ERP is still consistent with significant additional gains for the S&P 500 (Chart I-3). Monetary policy is extraordinarily accommodative. Even when we account for the S&P 500’s elevated multiples, the exceptional jump in the BCA Monetary Indicator is large enough to push up equity prices (Chart I-4). Moreover, the strength of US housing activity indicators confirms that the Federal Reserve has pulled the right levers to boost domestic economic activity. For example, the NAHB Housing Market Index has reached a 22-year record, building permits in July grew at their fastest monthly rate in 30 years, and the Mortgage Applications Index for purchases rocketed to a 11-year high in August. Chart I-3A Low ERP Underpins Equities... Chart I-4...So Does Monetary Policy   The US economy continues to heal. For stocks to climb further on a cyclical basis, the market will need more than five tech giants leading the charge. Hence, earnings expectations for the rest of the market must also mount. Practically, the economy must recover its output loss and the pandemic must ebb. For now, the four-week moving average of initial unemployment claims is drifting lower, and the ISM New Orders-to-Inventories spread is consistent with a faster and more solid business cycle upswing. The ERP is still consistent with significant additional gains for the S&P 500. The global industrial sector outlook is brightening. Manufacturing and trade disproportionately contribute to fluctuations in global economic activity, therefore, they exert an outsized influence on the earnings of non-tech multinationals. The strength in Singapore’s electronics shipments indicates that our Global Industrial Activity Nowcast will accelerate (Chart I-5, top panel). Moreover, the rapid expansion in China’s credit flows points to a marked increase in Chinese imports, which will help industrial and commodity exporters around the world (Chart I-5, bottom panel). Core producer prices have bottomed. Core producer prices are a direct input in the corporate sector’s pricing power. A trough in this inflation gauge leads to stronger EPS and widening profit margins for the S&P 500 (Chart I-6). Chart I-5The Global Industrial Cycle Is Turning The Corner Chart I-6Easing Deflationary Pressures Will Help Profits   Investors should still wait to allocate new funds to the stock market. The stock market’s near-term outlook remains marked by short-term froth that dampens our cyclical optimism, especially because the market advance has been concentrated in a small group of equities. Chart I-7Tactical Froth The Exposure Index of the National Association of Active Investment Managers has hit 100.1 (Chart I-7). Such a lofty reading indicates that the price of stocks already incorporates optimistic expectations. From a contrarian perspective, this development boosts the probability that swing traders will face disappointments in the near future and will sell their equity holdings. Similarly, the put/call ratio is near a 10-year low, which confirms that traders have bought a lot of upside exposure to stocks without much protection against a pullback. This level of confidence is often a precursor to a significant correction. Finally, our Tactical Strength Indicator is 1.7-sigma above its mean. Historically, when this risk gauge has hit a reading above 1.3, there is a good probability that the S&P 500 will correct or move sideways (Chart I-8). A catalyst must emerge for those aforementioned vulnerabilities to morph into a correction. If Treasury yields move closer to 1%, then stocks will experience a significant pullback of 10% or more as the market rotates away from the leadership of growth stocks. This risk would be especially salient if real yields move up. As Chart I-9 illustrates, falling TIPS yields have been a pillar of the powerful rally of growth stocks. Moreover, low real yields are arithmetically necessary to justify the current level of market multiples exhibited by the S&P 500 (Chart I-9, bottom panel). Chart I-8The S&P 500 Is Vulnerable To A Correction Chart I-9Falling Real Yields Have Helped Growth Stocks   Growth and high-P/E ratio stocks are heavily represented in the tech and healthcare sectors, which together account for 42% of the S&P 500. This means that higher yields will likely temporarily drag down the entire market. Ultimately, leadership changes are painful events, but they rarely mark the end of bull markets. Can Yields Move Up? Chart I-10Positive Signs For Inflation It is time to tweak our bond market view because yields should soon move higher. For the past five months, we have written that yields offer minimal downside and that their asymmetric risk profile made government bonds an unappealing investment. We underweighted this asset class relative to stocks and recommended investors bet on higher inflation breakeven rates. However, forces are aligning to expect real rates to rise and thus, nominal yields should move up. The sequencing of the market’s response to QE increasingly favors lower bond prices. Our US Equity Strategy team recently highlighted that in 2009 stocks were the first asset to reflect the implementation of QE1 by the Fed.4 A weaker dollar followed. Bond yields started to perk up only after the USD deteriorated by enough, after stock prices had climbed by enough and after corporate spreads had narrowed by enough to ease financial conditions to stimulate the economy. So far, 2020 echoes the 2009 pattern and our Financial Conditions Index is more stimulatory than it was prior to the COVID-19 outbreak (see Chart III-36 in Section III). Chart I-11Commodities Point To Higher Yields... Inflation momentum confirms the risks to bonds. The apex of the deflationary shock has already passed. In July, core CPI excluding shelter rose by 0.84% month-on-month, which was the highest reading since 1981 when the Fed was combating the most violent inflation outbreak in generations. The upturn in core producer prices also warns that the annual inflation rate of core CPI should accelerate meaningfully by early 2021 (Chart I-10). The dollar’s weakness is another inflationary force. Import prices from China have already bottomed, which points to an escalation in goods inflation in the coming months. Firming commodity prices constitute another risk for yields. Our Commodities Advance/Decline line has recently broken out. This technical development is consistent with higher commodity prices and higher bond yields (Chart I-11). Rallying natural resources are inflationary, but they also indicate that the global economy is strengthening, which should put upward pressure on real interest rates. Strength in the housing sector also confirms that government bond yields have upside. As we highlighted above, a robust housing market is an important validation that monetary policy is very accommodative. By definition, the objective of loose policy is to boost future economic activity and eradicate deflationary pressures. The surge in lumber indicates bond prices are showing downside risk (Chart I-12). Additionally, the upswing in mortgage issuance is occurring as the Treasury and corporations boost their borrowings, which will generate more demand to use savings generated in the economy. The price of those savings will be higher real interest rates. Chart I-12...Especially Lumber The ebbing of COVID-19 also suggests that economic activity has scope to accelerate. Moreover, the House of Representatives reconvened to address the problems plaguing the US Postal Service ahead of the November elections. This early return to work gives Washington another opportunity to negotiate the stimulus bill that it failed to pass earlier this month. We still expect such a bill to ultimately become law because both Democrats and Republicans have too much to lose in November if the economy relapses in response of political paralysis. Declining infections and increased government support will bolster aggregate demand and put upward pressure on rates. The stock market’s near-term outlook remains marked by short-term froth that dampens our cyclical optimism. Market dynamics are also very negative for bonds. Our Valuation Index highlights that Treasurys are incredibly expensive (Chart I-13, top panel). Moreover, our Composite Technical Indicator remains overbought, though it has lost momentum. In this context, the lack of appetite for yield curve control or more QE demonstrated by the Federal Open Market Committee creates a genuine danger for bonds. Without these policies, bond yields will have trouble resisting the upward push created by our rising US Pipeline Inflation Pressures Index, our rebounding Nominal Cyclical Spending proxy (which is an average of the ISM Manufacturing headline index and Prices Paid component), and the uptick in the amount of liquidity sitting on commercial banks’ balance sheets (Chart I-14). Chart I-13Treasurys Are Expensive And Losing Momentum Chart I-14Building Cyclical Risks For Bonds   Thus, equities are at risk on a tactical basis because we anticipate that 10-year Treasury yields may climb towards 1%, including a rise in TIPS yields. The US election creates an additional near-term hurdle for stocks. As we wrote last month, President Trump will likely become more belligerent toward the US’s trading partners in the coming months. Moreover, Vice-President Joe Biden, who has a comfortable lead in the polls including in key swing states such as Florida, Michigan, Pennsylvania, and Wisconsin wants to cancel half of the 2017 tax cuts.5  Small Over Big Long-term investors should expect stocks to beat bonds on a 5- to 10-year horizon, but equities will generate paltry real returns compared with the past 40 years. Elevated valuations for US equities are consistent with long-term annualized real rates of return of only 0.5% (Chart I-15). Moreover, the long-term outlook for profit margins is poor. As we wrote three months ago, mounting populism will result in redistributive policies that will lift the share of wages relative to GDP.6 Moreover, the shift of the US population to the left on economic matters will push up corporate tax rates. Increased labor costs and corporate taxes are negative for profit margins. If profit margins normalize, then equities will probably underperform the uninspiring expected returns implied by current market multiples. The surge in lumber indicates bond prices are showing downside risk. Investors can still generate generous returns through geographical and sectoral selection. We have highlighted how value stocks, industrials and materials, and EM and European equities will likely beat US equities.7 This month we will explore how US small-cap equities are also well placed to best the dismal projected real returns offered by their large-cap counterparts. Our BCA Relative Technical Indicator shows that small-cap stocks are 1.8-sigma oversold when compared with the S&P 500, which indicates a capitulation among investors toward these equities. The bifurcation is even greater if we compare small-cap equities with the S&P 100’s mega-caps that have driven up the US market in recent years. Incorporating these influences, our Cyclical Capitalization Indicator has moved in favor of small-cap stocks, which suggests that small-cap stocks will be rerated if the yield curve can steepen further (Chart I-16). Equities are at risk on a tactical basis because we anticipate that 10-year Treasury yields may climb towards 1%. Chart I-15Valuations And Profit Margins Threaten Long-Term Stock Returns Chart I-16Indicators Favor Small Cap Stocks Chart I-17A Debt Turnaround Would Help Small Cap Stocks Debt dynamics could also increasingly beneficial to small-cap equities. In the past few years, the heavy debt-to-EBITDA of smaller firms created a major headwind for small-cap investors. The indebtedness of small-cap stocks often decreases relative to large-caps when an economic recovery begins. This shift in leverage portends an increase in small-caps’ relative future returns (Chart I-17). Our negative bias toward the dollar and our positive view on commodities also benefit small-cap stocks. Since the early 1990s, increasing real commodity prices and a falling Dollar Index have coexisted with a robust performance of small-cap firms (Chart I-18). The negative US balance-of-payment dynamics, coupled with escalating inflation risks, will continue to weigh on the dollar, especially as various large EM nations try to diversify their reserves and payment systems away from the dollar.8 Meanwhile, a declining dollar, expanding global growth, monetary debasement, populism, inflation and a lack of investment in supply, all will accentuate the appeal of natural resources. The sectoral bias of small-cap indices will capitalize on these trends. Chart I-18Small Is Beautiful Chart I-19Small Cap Stocks Like Higher Yields   Finally, cyclical timing is also moving in favor of small-cap firms. Since 2014, the Russell 2000 has outperformed the S&P 500 when real yields moved higher (Chart I-19). Small-cap firms display a more marked pro-cyclicality than large firms. Additionally, the S&P 500 growth bias implies that the US large-cap benchmark underperforms the small cap indices when real yields increase.   Mathieu Savary Vice President The Bank Credit Analyst August 27, 2020 Next Report: September 24, 2020   II. Global Semiconductor Stocks: A Hiatus Is Overdue In A Structural Bull Market The strength in global semiconductor sales in recent months has been due to one-off factors stemming from pandemic-related lockdowns. As the one-off demand surge subsides, global semiconductor sales will decline modestly toward the end of this year. In the near term, global semiconductor stock prices are vulnerable due to overbought conditions, excessive valuations and demand disappointment. The global semiconductor industry is at the epicenter of the US-China confrontation, and more US restrictions on chips sales to China are probable. This is another risk for this sector's share prices.   Nevertheless, the structural outlook for global semiconductor demand is constructive. Its CAGR may rise from 3% during 2014-2019 to 5% during 2020-2024.  Investor euphoria has taken hold of semiconductor stocks. Global semiconductor stock prices have skyrocketed by 68% from March lows and 96% from December 2018 lows. Meanwhile, global semiconductor sales during March-June rose only by 5% from a year ago. As a result, the ratio of market cap for global semiconductor stocks relative to global semiconductor sales has reached its highest level since at least the inception of data in 2003 (Chart II-1). Chart II-1Global Semi Sector: Market Cap-To-Sales Ratio Has Surged With semi equity multiples very elevated, their share prices have become even more sensitive to global semiconductor demand growth. Hence, the focus of this report is to try to gauge the strength of global semiconductor demand, both in the near term and structurally. The strength in global semiconductor sales in recent months has been due to one-off factors stemming from the lockdowns. Near-term semiconductor stock prices could disappoint due to weak chip demand from the smartphone sector and diminishing purchases of personal computers (PCs) and servers. However, structurally, we are positive on global semiconductor demand, which is underpinned by the continuing rollout of 5G networks and phones, a wider adoption of data centers, and further technological advancements in artificial intelligence (AI), cloud computing, edge computing and smaller nodes for chip manufacturing (Box II-1). Box II-1 Key Technologies Underpinning Potential Global Semiconductor Demand AI refers to the simulation of human intelligence in machines, for example, computers that play chess and self-driving cars. The goals of AI include learning, reasoning and perception. Cloud computing is the delivery of computing services – including servers, storage, databases, networking, software, analytics and intelligence – over the Internet (“the cloud”) to offer faster innovation, flexible resources and economies of scale. Edge computing is a form of distributed computing, which brings computation and data storage closer to where it is needed, to improve response times and save bandwidth. Technology node refers to the width of line that can be processed with a minimum width in the semiconductor manufacturing industry, such as technology nodes of 10 nanometers (nm), 7nm, 5nm and 3nm. The smaller the nodes are, the more advanced they are. Near-Term Headwinds Semiconductor demand worldwide grew by 6% year-on-year in the first half of this year. There has been a remarkable divergence between world semiconductor sales and the global business cycle (Chart II-2). The divergence between semiconductor sales and economic activity was most striking in the US and China. Semiconductor sales in China rose by 5% year-on-year in Q12020, and in the US they grew by 29% year-on-year in Q22020, despite a contraction in their aggregate demand during the same period. By contrast, Q2 annual growth of semiconductors sales was -2.2% for Japan, -17% for Europe and 1.8% for Asia ex. China and Japan (Chart II-3). Chart II-2World Semi Sales Diverged From The Global Business Cycle Chart II-3Strong Semi Sales In The US And China, But Not Elsewhere   The reasons why the US and China posted a surge in semiconductor demand while Europe and Japan experienced a contraction in domestic semiconductor sales are as follows: Most data center investment is occurring in the US and China. Chart II-4 shows that 40% of global hyperscale data centers are operating in the US, much larger than any other countries/regions. China, in turn, ranked second, with a global share of 8%. Chart II-4The US Has The Most Global Hyperscale Data Centers Demand contraction in Europe and Japan is due to semiconductor demand in these regions mainly originating from the automobile sector, where production was severely hit by the global pandemic. About 37% of European semiconductor sales were from last year’s automotive market. We believe the divergence between global economic activity and semiconductor sales, as demonstrated by Chart II-2 on page 3, has been due to one-off factors, as the global pandemic lockdowns have spurred semiconductor demand. Such a one-off demand boost will likely dissipate in the coming months. Traditional PCs and tablets: There has been a surge in demand for traditional PCs9 and tablets in the past six months. This was due to the significant increase in online activities, such as working from home, education, e-commerce, gaming and entertainment. Data from the International Data Corporation (IDC) has revealed that shipments of traditional PCs and tablets in volume terms had a strong year-on-year growth of 11.2% and 18.6%, respectively, in the period of April-June (Chart II-5). Looking forward, even renewed lockdowns will not lead to a similar rush to buy these products. Many households are already equipped to work from home and for other online activities. With many countries gradually opening their economies, such demand will diminish. The traditional PC and tablet sectors together account for about 13% of global chip demand (Chart II-6). Chart II-5Personal Computers Sales Have Surged Amid Lockdowns Server demand: Another major semiconductor demand contribution in Q2020 was from the server sector, which spiked by 21% year-on-year (Chart II-7). The surge in online activities triggered a strong demand for cloud services and remote work applications, both of which require computer servers to run on. Chart II-6The Breakdown Of Global Semiconductor Sales By Type Of Usage However, demand from the server sector is also set to diminish in 2H2020 and Q1 2021. Provided the inventories at major data center operators, including Microsoft, Google and Amazon, remain at high levels,10 global cloud service providers will likely reduce their orders of servers next quarter.11  Enterprises will also likely cut their investment in computer servers in 2H2020, as many of them had already increased their purchases of servers to prepare employees and business processes for remote working. We expect global server demand growth to soften in 2H2020. The Digitimes Research forecasted a 5.6% quarter-on-quarter contraction in 3Q2020 and a further cut in global sever shipment in the 4Q2020.10 The global server sector accounts for about 10% of global chip demand and, together with PCs and tablets, they make for 23% (please refer to Chart II-6 on page 5). Further, the smartphone sector – accounting for 27% of global semiconductor demand – will continue struggling in H2 this year. Chart II-7Server Sales Have Surged Amid Lockdowns Chart II-8Global Smartphone Shipments Will Likely Remain Weak In 2020H2   The global total smartphone demand has been hit severely, as households delayed their new smartphone purchases. According to Canalys’ data, global smartphone shipments dropped by 13% and 14% year-on-year in Q1 and Q2, respectively. We expect smartphone shipments to continue contracting over the next three-to-six months (Chart II-8). We believe global consumers will remain cautious in their spending on discretionary goods, such as smartphones, due to lowered incomes and increased job uncertainty. The IDC also forecasted that global smartphone shipments would not grow until 1Q2021.12 The Chinese smartphone sales showed a considerable weakness in July, with a 35% year-on-year contraction, which is much deeper than the 20% decline in H1 this year. 5G smartphone shipments also slowed last month, with a 21% drop from the previous month. The global semiconductor industry is at the epicenter of the US-China confrontation. Bottom Line: The strength in global semiconductor sales in recent months has been due to one-off factors stemming from the lockdowns. As this one-off demand subsides, global semiconductor sales will decline modestly toward the end of this year. Given the overbought conditions and the elevated equity valuations, global semiconductor stocks are currently vulnerable to near-term disappointments in semiconductor demand. At The Epicenter Of The US-China Rivalry Semiconductors are at the epicenter of the US-China confrontation. Ultimately, the US-China contention is about future technological dominance. That is access to technology and the capability to develop new technologies. China currently accounts for about 35% of the global semiconductor demand. US restrictions on semi producers worldwide to supply semiconductors to Chinese buyers constitute a major risk to semiconductor stock prices. On August 17, the US announced fresh sanctions that restrict all US and foreign semiconductor companies from selling chips developed or produced using US software or technology to Huawei, without first obtaining a license. In May, the US had already limited companies, such as the Taiwan Semiconductor Manufacturing Company (TSMC), from making and supplying Huawei with its self-designed chips. In addition, the US recently threatened bans on Chinese-owned apps TikTok and WeChat, and signaled that it could soon restrict Alibaba’s operations in the US. Chart II-9Global Semi Companies' Sales To China Are Substantial The global semiconductor sector is highly vulnerable to further escalation in the tension between these two superpowers. Major global semiconductor companies’ sales are heavily exposed to China, and their revenue from China ranges from 16% to 50% of total (Chart II-9). We have been puzzled why global semi share prices have been rallying in spite of US limitations on semiconductor shipments to Huawei and its affiliated entities. One explanation could be that the Chinese companies that are not affiliated with Huawei are able to import semiconductors and then supply them to Huawei. If this is true, the US will have no other choice but to limit all semiconductor sales to China. This will be devastating for global semi producers given their large exposure to China. In anticipation of US punitive policies limiting its access to semiconductors, China had boosted its semiconductor imports over the past 12 months (Chart II-10, top panel). Chinese imports of integrated circuits rose by 12% year-on-year in 1H2020, which is much higher than the 5% year-on-year increase in Chinese semiconductor demand during the same period (Chart II-10, bottom panel). This gap suggests the country had restocked its semiconductor inventories. China has particularly restocked its imports of non-memory chips with imports of processor & controller and other non-memory chips in H1, surging by 30% and 20%, respectively, in US dollar terms (Chart II-11). For memory chips, the contraction in Chinese imports was mainly due to a decline in global memory chip prices. Chart II-10China Had Likely Restocked Its Semi Inventories Chart II-11Strong Chinese Imports In Non-Memory Chips   Bottom Line: The global semiconductor industry is at the epicenter of the US-China confrontation, and more restrictions on sales to China are probable. In turn, the restocked semiconductor inventory in China raises the odds of weakening mainland semiconductor import demand in H2 of this year. Structural Tailwinds Table II-1Global Semiconductor Demand CAGR Forecast Over 2020-2024 By Device We are optimistic on structural global semiconductor demand. Its nominal CAGR may rise from 3% during 2014-2019 to 5% during 2020-2024 in US dollar terms. Table II-1 shows our demand growth forecasts for global chips in the main consuming sectors over the next five years. The major contributing sectors during 2020-2024 will be 5G smartphones, servers, industrials, electronics and automotive manufacturing. The underlying driving forces are the continuing rollout of 5G networks and phones, the development of data centers, and further technological advancements in AI, cloud computing and edge computing. Currently, the world is still in the early stages of 5G network development. AI, cloud computing and edge computing are constantly evolving. With increasing adoption of 5G smartphones, computer servers and IoT devices, global semiconductor demand is in a structural uptrend (Box II-2). Box II-2 Key Components For The Virtual World In Development Data centers and cloud computing allow data to be stored and applications to be running off-premises and to be accessed remotely through the internet. Edge computing allows data from Internet of things (IoT) devices to be analyzed at the edge of the network before being sent to a data center or cloud. IoT devices contain sensors and mini-computer processors that act on the data collected by the sensors via machine learning. The IoT is a growing system of billions of devices — or things — worldwide that connect to the internet and to each other through wireless networks. AI technology empowers cloud computing, edge computing and IoT devices. 5G is at the heart of the IoT industry transformation, making a world of everything connected possible. Chart II-125G Phone Shipments In China Will Continue To Rise 5G Smartphone Currently, China is the world’s largest 5G-smartphone consumer and the leading 5G-adopter in the world. According to Digitimes Research, global 5G smartphone shipments will reach over 250 million units in 2020, with 170 million (68%) in China and only 80 million units in the world ex. China. Looking forward, 5G smartphone shipments are set to accelerate worldwide over the coming years. The 5G phone shipments in China will continue to rise. The 5G phone sales penetration rate in China is likely to rise from 60% in July to 95% by the end of 2022. In such a case, we estimate that the monthly Chinese 5G phone shipments will increase from the current 16 million units to about 25-30 million units in 2022 (Chart II-12). In the rest of the world, the 5G smartphone adoption pace will also likely speed up over the next five years. The 5G phone selling prices in the world outside China will drop, as more models are introduced and become more affordable. 5G smartphone prices have already fallen in China and will inevitably fall elsewhere. Chinese 5G smartphone producers will ship their low-priced 5G phones overseas, putting pressure on other producers to lower their prices. The 5G infrastructure development is accelerating in China and will accelerate in the rest of the world. Both China and South Korea have been very aggressive in their respective 5G network development. As of the end of June, China's top three carriers: China Mobile, China Unicom, and China Telecom – which together serve more than 1.6 billion mobile users in the country – had installed 400,000 5G base stations against an annual target of 500,000. In comparison, as of April 2020, American carriers had only put up about 10,000 5G base stations.13  As the US is competing with China on the 5G front, the country will likely boost its investment in 5G network development aggressively over the next five years in order to catch up to, or even exceed, China. Importantly, the 5G smartphone has more silicon content than 4G smartphones. More silicon content means higher semiconductor value. Rising 5G smartphone sales and higher silicon content together will more than offset the loss in semiconductor sales due to falling global 4G smartphone shipments. Overall, global semiconductor stock prices have diverged from their sales and profits. Based on our analysis, we expect a CAGR growth of 4% in semiconductor demand from the global smartphone sector over the next five years, slightly lower than the 5% in previous five years (Table II-1 on page 10). This also takes into consideration that the 5G network will be more difficult and more expensive to develop than the 4G network. Servers Global server shipment growth will be highly dependent on both the pace and the scale of data center development (Box II-3). Data centers account for over 60% of global server demand.  Box II-3 Data Centers There are four main types of data centers – enterprise data centers, managed services data centers, colocation data centers, and cloud data centers. Data centers can have a wide range of number of servers. Corporate data centers tend to have either 200 (small companies), or 1000 servers (large companies). In comparison, a hyperscale data center usually has a minimum of 5,000 servers linked with an ultra-high speed, high fiber count network. Outsourcing and a move towards the cloud are driving the growth of the hyperscale data center. Instead of companies investing in physical hardware, they can rent server space from a cloud provider to both save their data and reduce costs. Amazon, Microsoft, Google, Apple and Alibaba are all top global cloud service providers. The more hyperscales to be built up, the higher the demand for servers. In 2019, about 13% of the total number of data centers in China were of the hyperscale and large-scale varieties. The plan of new infrastructure development announced earlier this year by Beijing was aiming to increase the number of hyperscale and large-scale data centers in China. Among current data centers either under construction or to be developed in the near future, 36% of them are hyperscale and large-scale data centers.   The future growth of data centers is promising. The global trend of data localization14 due to the concerns of data privacy and national security will also bolster a boom of data centers over the next five years. A growing number of countries are adopting data localization requirements, such as China, Russia, Indonesia, Nigeria, Vietnam and some EU countries. While the Chinese data center market is expected to expand by a CAGR of about 28% over 2020-2022,15 a report recently released by Technavio forecasted the global data center industry’s CAGR at over 17% during 2019-2023.  We forecast that the global semiconductor demand from servers will grow at a CAGR of 12% over 2020-2024. IoTs Technological advancements in AI, cloud computing and edge computing, in combination with 5G network development, will facilitate the IoTs adoption. According to the GSMA,16 46 operators in 24 markets had launched commercially available 5G networks by 30 January 2020. It forecasted that global IoT connections will be increased from 12 billion mobile devices in 2019 to 25 billion in 2025 with a CAGR at 13%.17   IoTs chips include the Artificial Intelligence of Things (AIoT) – a powerful convergence of AI and the IoT. IoTs is an interconnected network of physical devices. Every device in the IoT is capable of collecting and transferring data through the network. Looking forward, global demand of AI chips and IoT chips will have significant potential to grow with creation of “smarter manufacturing”, “smarter buildings”, “smarter cities”, etc. AI applications can be used in manufacturing processes to render them smarter and more automated. Productivity will be enhanced as machines achieve significantly improved uptime while also reducing labor costs. There are plenty of upsides in industrial semiconductor demand (Chart II-13). We expect the CAGR of industrial electronics to increase from 3.4% during 2014-2019 to 8% during 2020-2024. AI applications can create smart buildings by increasing connectivity across enterprise assets, enabling home network infrastructure (e.g., routers and extenders) and employing home-security devices (e.g., cameras, alarms and locks). AI applications can be used to create smart cities. A smart city is an urban area that uses different types of IoT electronic sensors to collect data. Insights gained from that data are used to manage assets, resources and services efficiently; in return, that data is used improve operations across the city. China has already developed about 750 trial sites of smart cities with different degrees of smartness in the past decade. As AI and 5G technology advances, the existing smart cities’ “smartness” will be upgraded and new trial smart cities will be implemented. Based on IDC data, China’s investment in smart cities will rise at a CAGR of 13.5% over 2020-2023 (Chart II-14). Globally, the U.S., Japan, European countries and other nations are also actively developing smart cities. According to a new study conducted by Grand View Research, the global smart cities market size is expected to grow at a CAGR of 24.7% from 2020 to 2027.18  Chart II-13Plenty Of Upside In Industrial Semi Demand Chart II-14China’s Investment In Smart Cities Will Continue To Grow   Automotive We expect the global automotive chip market to grow at a CAGR of 9% during 2020-2024, as in 2014-2019. The increase in consumption of semiconductors by the auto industry will continue to be driven by the market evolution toward autonomous, connected, electric and shared mobility. Most new vehicles now include some level of advanced driver assist systems (ADAS), such as adaptive cruise control, automatic brakes, blind spot monitoring, and parallel parking. The whole industry is progressing toward fully autonomous vehicles in the coming years. Increasing adoption of automotive chips and recovering car sales will revive automotive chip sales. In addition, rising penetration of new energy vehicles (NEVs) is beneficial to semiconductor sales, as NEVs contain higher semiconductor content than conventional vehicles. Conventional vehicles contain an average of a $330 value of semiconductor content while hybrid electric vehicles can contain up to $1,000 and $3,500 worth of semiconductors.19 Regarding other sectors, we are also positive on structural demand of storage and consumer electronics. AI applications generate vast volumes of data — about 80 exabytes per year, which is expected to increase by about tenfold to 845 exabytes by 2025.20 In addition, developers are now using more data in AI and deep learning (DL) training, which also increases storage requirements. With massive potential demand for storage, we estimate a CAGR of 7% over 2020-2024 (Table II-1 on page 10). A recent report from ABI Research predicts that the COVID-19 pandemic will increase global sales of wearables (such as a Fitbit or Apple Watch) by 29% to 30 million shipments of the devices this year. With contribution from wearables, we expect global semiconductor demand from the consumer sector to grow at a CAGR of 3% over 2020-2024, the same rate as in the previous five years. Bottom Line: Continuing rollout of 5G networks and phones, development of data centers, and further technological advancements in AI and cloud computing will provide tailwinds to structural global semiconductor demand, accelerating its CAGR growth from 3% during 2014-2019 to 5% during 2020-2024. Valuations And Investment Conclusions Most global semiconductor stocks are currently over-hyped. Critically, both DRAM and NAND prices have been deflating since January, reflecting weak demand for memory chips. Yet, share prices of memory producers have rallied (Chart II-15). Overall, global semiconductor stock prices have diverged from their sales and profits (Chart II-16). Chart II-15Falling Memory Prices Pose Risk To Memory Stocks Chart II-16Global Semiconductor Stocks Have Deviated From Profits   Consequently, the multiples of semiconductor stocks have spiked to multi-year highs (Chart II-17).  Even after adjusting for negative US real bond yields, valuations of semiconductor stocks are not cheap. Chart II-18 illustrates the equity risk premium for global semiconductor stocks is at the lower end of its range of the past 10 years. The ERP is calculated as forward earnings yield minus 10-year US TIPS yields. Chart II-17Global Semi Stocks: Elevated Valuations Chart II-18Equity Risk Premium For Global Semi Stocks Is Historically Low   It is impossible to time a correction or know what the trigger would be (US-China tensions have been our best guess). Nevertheless, we do not recommend chasing semiconductor stocks higher due to their overstretched technicals and valuations on the one hand and potential weakening demand in H2 on the other. In addition, the ratio of global semi equipment stock prices relative to the semi equity index correlates with absolute share prices of global semi companies. This is because equipment producers are higher-beta as they outperform during growth accelerations and underperform during growth slumps. The basis is that semi manufacturers have to purchase equipment if there is actual strong demand coming up and vice versa. The recent underperformance by global semi equipment stocks relative to the semi equity index might be an early sign of a potential reversal in semi share prices in absolute terms (Chart II-19). Chart II-19A Signal Of A Potential Reversal In Semi Share Prices Meanwhile, we believe the subsector- memory chip stocks - will outperform the overall semiconductor index amidst the potential correction, because they have lagged and are less over-extended. Finally, we remain neutral on Taiwanese and Korean bourses within the EM equity space for now. Escalation in US-China confrontation, as well as their exposure to semiconductors, put these bourses at near-term risk. That said, we are reluctant to underweight these markets because fundamentals in EM outside North Asia remain challenging.   Ellen JingYuan He Associate Vice President Emerging Markets Strategy III. Indicators And Reference Charts We continue to favor stocks at the expense of bonds, but equities are increasingly vulnerable because short-term sentiment and positioning measures are growing increasingly stretched. Three forces can prompt a correction. First, a rebound in yields toward 1% would cause turbulence for the S&P 500, because the index is dominated by growth stocks that are highly sensitive to fluctuations in the risk-free rate. Second, a dollar bounce would hurt the S&P 500 because a depreciating USD has fueled the US stock market rally since June. Finally, the US presidential election is drawing nearer; hence, the risk of potentially damaging political headlines is growing.  Despite these short-term risks, the main pillar supporting the rally remains intact: global monetary conditions are highly accommodative and the chance of inflation moving high enough to spook central bankers is minimal in the near future. Additionally, the fiscal spigots are open and governments around the world will ultimately continue to support their economies. Hence, any correction in the S&P 500 is unlikely to move beyond 15% or a level of 2900. Our cyclical indicators confirm the positive backdrop for stocks. While our Valuation Indicator has reached overvalued territory, our Monetary Indicator remains extremely accommodative. Moreover, our Technical Indicator is now flashing a clear buy signal. Putting all those forces together, our Intermediate-Term Indicator continues to support equities. Finally, our Revealed Preference Indicator strongly argues in favor of staying invested in equities. That being said, our Speculation Indicator has surged back up, thus the volatility of the rally should increase. Bonds remain extremely unappealing. Our Bond Valuation Index shows Treasurys as prohibitively expensive and our Composite Technical Indicator continues to lose momentum. So far, government bond yields have managed to remain stable at very low levels even if they have not declined further. Nonetheless, bonds have underperformed equities, which is a trend that will remain in place for many more quarters. Moreover, the pick-up in commodity prices and in various gauges of the business cycle suggests that bond yields should soon move higher, especially because the Fed is far from enthused at the concept of yield curve control. Our Cyclical Bond Indicator has turned higher and will soon flash an outright sell signal. The dollar continues to weaken after its recent breakdown. For now, the USD’s weakness has been concentrated among DM currencies. For the dollar to weaken further, EM currencies must begin to rally more markedly than they have until now, especially in Latin America. The firmness of the CNY is a good sign for the EM complex, but another clear up-leg in global growth must emerge before EM currencies can fully blossom. As a result, we are likely to have entered a temporary period of consolidation for the US dollar. The extremely oversold nature of our Dollar Composite Technical Indicator supports the idea that the dollar needs to digest its recent losses before its poor fundamentals force it lower once again. Finally, commodities have been a prime beneficiary of the weakness in the dollar and the combination of stable yields and improving economic activity. Our Composite Technical Indicator is now well into overbought territory which makes natural resource prices vulnerable to a pullback. A move up in yields as well as a short-term rebound in the dollar will likely catalyze any underlying technical risks to commodities. Gold will be particularly vulnerable to any such pullback, especially if higher real yields are the cause of the correction in natural resource prices. Despite these short-term worries, the outlook for commodities remains bright. As a result, we would use any correction to add exposure to the commodity complex. EQUITIES: Chart III-1US Equity Indicators Chart III-2Willingness To Pay For Risk Chart III-3US Equity Sentiment Indicators   Chart III-4Revealed Preference Indicator Chart III-5US Stock Market Valuation Chart III-6US Earnings Chart III-7Global Stock Market And Earnings: Relative Performance Chart III-8Global Stock Market And Earnings: Relative Performance   FIXED INCOME: Chart III-9US Treasurys And Valuations Chart III-10Yield Curve Slopes Chart III-11Selected US Bond Yields Chart III-1210-Year Treasury Yield ComponentsChart III-13US Corporate Bonds And Health Monitor Chart III-14Global Bonds: Developed Markets Chart III-15Global Bonds: Emerging Markets   CURRENCIES: Chart III-16US Dollar And PPP Chart III-17US Dollar And Indicator Chart III-18US Dollar Fundamentals Chart III-19Japanese Yen Technicals Chart III-20Euro Technicals Chart III-21Euro/Yen Technicals Chart III-22Euro/Pound Technicals   COMMODITIES: Chart III-23Broad Commodity Indicators Chart III-24Commodity Prices Chart III-25Commodity Prices Chart III-26Commodity Sentiment Chart III-27Speculative Positioning   ECONOMY: Chart III-28US And Global Macro Backdrop Chart III-29US Macro Snapshot Chart III-30US Growth Outlook Chart III-31US Cyclical Spending Chart III-32US Labor Market Chart III-33US Consumption Chart III-34US Housing Chart III-35US Debt And Deleveraging   Chart III-36US Financial Conditions Chart III-37Global Economic Snapshot: Europe Chart III-38Global Economic Snapshot: China   Mathieu Savary Vice President The Bank Credit Analyst Footnotes 1 Please see The Bank Credit Analyst "April 2020," dated March 26, 2020, available at bca.bcaresearch.com 2 Please see The Bank Credit Analyst "July 2020," dated June 25, 2020, available at bca.bcaresearch.com 3 Please see European Investment Strategy "An Economy Without Mouths Or Noses Will Lose 10 Percent Of Jobs," dated July 30, 2020, available at eis.bcaresearch.com 4 Please see US Equity Strategy "Inversely Correlated," dated August 25, 2020, available at uses.bcaresearch.com 5 Please see The Bank Credit Analyst "August 2020," dated July 30, 2020, available at bca.bcaresearch.com 6 Please see The Bank Credit Analyst "June 2020," dated May 28, 2020, available at bca.bcaresearch.com 7 Please see The Bank Credit Analyst "August 2020," dated July 30, 2020, available at bca.bcaresearch.com 8 Diversifying away from the dollar does not mean that the USD will lose its reserve status. However, a return to the share of FX reserves that prevailed in the first half of the 1990s will hurt the dollar, especially because the US net international investment position has fallen from -4.6% of GDP in 1992 to -57% today. 9 Traditional PCs are comprised of desktops, notebooks, and workstations. 10 Global server shipments to contract 5.6% sequentially in 3Q2020, says Digitimes Research 11 Global server shipments forecast to increase by 5% this year: TrendForce 12 IDC Expects Worldwide Smartphone Shipments to Plummet 11.9% in 2020 Fueled by Ongoing COVID-19 Challenges 13 America does not want China to dominate 5G mobile networks 14 “Data localization” can be defined as the act of storing data on a device that is physically located within the country where the data was created. Data localization requirements are governmental obligations that explicitly mandate local storage of personal information or strongly encourage local storage through data protection laws that erect stringent legal compliance obligations on cross-border data transfers. 15 The big data center industry ushered in another outbreak 16 The GSMA represents the interests of mobile operators worldwide, uniting more than 750 operators with almost 400 companies in the broader mobile ecosystem, including handset and device makers, software companies, equipment providers and internet companies, as well as organizations in adjacent industry sectors. 17 GSMA: 5G Moves from Hype to Reality – but 4G Still King 18 Smart Cities Market Size Worth $463.9 billion By 2027 19 The Automotive Semiconductor Market – Key Determinants of U.S. Firm Competitiveness 20 AI is data Pac-Man. Winning requires a flashy new storage strategy.