BCA Indicators/Model

Highlights 2020 Model Bond Portfolio Positioning: Translating our 2020 global fixed income Key Views into recommended positioning within our model bond portfolio comes up with the following conclusions: target a moderately aggressive level of overall portfolio risk, with below-benchmark duration exposure alongside meaningful overweight allocations to global corporate credit. Country Allocations: The cyclical improvement in global growth heralded by leading indicators should put upward pressure on overall global bond yields in 2020. With central banks likely to maintain accommodative policy settings to try and boost depressed inflation expectations, government bond allocations should reflect each country’s “beta” to global yield changes. That means favoring lower-beta countries (Japan, Germany, Spain, Australia, the UK) over higher-beta countries (the US, Canada, Italy). Spread Product: Better global growth, combined with stimulative monetary conditions, will provide an ideal backdrop for growth-sensitive spread product like corporate bonds to outperform government debt this year. We are maintaining an overweight stance on US high-yield credit, while increasing overweights to euro area corporates (both investment grade and high-yield). With the US dollar likely to soften as 2020 evolves, emerging market hard currency debt, both sovereign and corporate, is poised to outperform – we are upgrading both to overweight. Feature Welcome to our first report of the New Year. Just before our holiday break last month, we published our 2020 “Key Views” report, outlining the thematic implications of the BCA 2020 Outlook for global bond markets.1 In this follow-up report, we turn those themes into specific investment recommendations for the next twelve months. We will also make any necessary changes to the allocations in the Global Fixed Income Strategy (GFIS) model bond portfolio to reflect our themes. The main takeaway is that 2020 will be a much different year than 2019, when virtually all global fixed income classes delivered solid absolute returns. The unusual combination of rapidly falling government bond yields and stable-to-narrowing spreads on the majority of credit products – especially in developed market corporate debt – will not be repeated in 2020. Absolute returns from fixed income will be far lower than in 2019, forcing bond investors to focus on relative returns across maturities, countries and credit sectors to generate outperformance. With global monetary policy to remain stimulative, alongside improved global growth, market volatility should remain subdued over the next 6-12 months. Being more aggressive on overall levels of portfolio risk, particularly through higher allocations to markets like high-yield corporates and emerging market (EM) credit, is a solid strategy in a world of low risk-free interest rates and tame volatility. Top-Down Bond Market Implications Of Our Key Views As a reminder, the main fixed income investment themes from our 2020 Key Views report were the following: Global growth will rebound in 2020, led by the US and China, putting upward pressure on global bond yields. Central banks will stay dovish until policy reflation has clearly turned into inflation, limiting how high bond yields can climb in 2020 but sowing the seeds for a far more bond-bearish backdrop in 2021. Accommodative monetary policy and faster growth will delay the peak in the aging global credit cycle. Returns on global fixed income will be far lower in 2020 than in 2019, given rich valuation starting points. Country and sector selection will be more important in driving fixed income outperformance. We now present the specific fixed income investment recommendations that derive from those themes, described along the following lines: overall portfolio risk, overall duration exposure, country allocations within government bonds, yield curve allocations within countries, and corporate credit allocations by country and credit rating. Overall Portfolio Risk: MODERATELY AGGRESSIVE Global growth is in the process of bottoming out after the sharp manufacturing-driven slowdown in 2019. The cumulative lagged impact of monetary easing by central banks last year, led by the US Federal Reserve cutting rates and the European Central Bank (ECB) restarting its Asset Purchase Program, is a main reason why growth is set to rebound. Reduced trade uncertainty between the US and China should augment the impact of easier monetary policy through improved business confidence. Our global leading economic indicator (LEI), which has increased for nine consecutive months, is already heralding this improvement in the global economy. Our global LEI diffusion index – which measures the number of countries with a rising LEI and is itself a leading indicator of the LEI – suggests more gains ahead as 2020 progresses. The LEI diffusion index is also a reliable leading indicator of bond market volatility, with the former signaling that the latter will remain quiescent in 2020 (Chart 1). At the same time, cross-asset correlations across fixed income sectors should drift a bit higher alongside a more broad-based upturn in global economic growth and expanding monetary liquidity via central bank asset purchases (Chart 2). This pickup in correlations suggests that there is scope for markets that lagged the 2019 global credit rally, like EM USD-denominated sovereign debt, to make up for that underperformance in 2020. Chart 1Improving Global Growth Will Keep Volatility Subdued Chart 2Cross-Asset Correlations Should Increase In 2020 The combination of better growth, stable volatility – but with only a mild rise in correlations – is a good backdrop to take a somewhat more aggressive investment stance in fixed income portfolios in 2020.  The combination of better growth, stable volatility – but with only a mild rise in correlations – is a good backdrop to take a somewhat more aggressive investment stance in fixed income portfolios in 2020. We prefer to take that additional risk by adding to our recommended overweight to global credit, rather than further reducing our below-benchmark overall duration exposure. Overall Portfolio Duration Exposure: BELOW BENCHMARK Chart 3Global Bond Yields Poised To Move Higher The improvement in global growth that we are anticipating in 2020 would normally be expected to put upward pressure on the real component of global government bond yields (Chart 3, top panel). This would initially manifest itself through asset allocation shifts out of bonds into equities and, later, through expectations of rate hikes and tighter monetary policy. However, with all major developed market central banks now expressing a desire to keep policy as easy as possible to try and boost inflation expectations, the cyclical move higher in real yields is likely to be more muted in 2020. However, given our expectation that the US dollar is likely to see a moderate decline, as global capital flows move into more growth-sensitive markets in EM and Europe, there is scope for global bond yields to rise via higher inflation expectations – especially with global oil prices likely to move a bit higher, as our commodity strategists expect (bottom two panels). We recommend only a moderate below-benchmark overall duration exposure in global fixed income portfolios in 2020, given that real yields will likely stay relatively muted. Investors should maintain core allocations to inflation-linked bonds, however, to benefit from the pickup in inflation expectations that is likely to occur this year. We recommend only a moderate below-benchmark overall duration exposure in global fixed income portfolios in 2020, given that real yields will likely stay relatively muted. Investors should maintain core allocations to inflation-linked bonds, however, to benefit from the pickup in inflation expectations that is likely to occur this year. Government Bond Country Allocation: UNDERWEIGHT HIGHER-BETA MARKETS, OVERWEIGHT LOWER-BETA MARKETS At the country level, we would typically let our expectations of monetary policy changes guide our recommended allocations. Yet in 2020, we see very little potential for any change in monetary policy outside of Australia (where rate cuts can happen early in the year) and Canada (where a rate hike may be possible later in the year). Thus, we think that a more useful framework for making fixed income country allocation decisions in 2020 is to rely on the “yield betas” of each country to changes in the overall level of global bond yields. Chart 4 shows the three-year trailing yield betas for 10-year government bonds of the major developed markets. Changes in the 10-year yields are compared to the yield of the 7-10 year maturity bucket of the Bloomberg Barclays Global Treasury Index (as a proxy for the unobservable “global bond yield”). On that basis, the higher-beta markets are the US, Canada and Italy, while the lower-beta markets are Japan, Germany, France, Spain, Australia and the UK. Thus, we want to maintain underweight positions in the former group and overweight positions in the latter group. At the moment, we already have most of those tilts within our model bond portfolio, with two exceptions: we are currently neutral (benchmark index weight) in the UK and Canada. For the UK, Brexit uncertainty has made it difficult to take a strong view on the direction of Gilt yields - a problem now compounded further with Andrew Bailey set to take over from Mark Carney as the new Governor of the Bank of England. Staying neutral, for now, still seems like the best strategy until all the policy uncertainties are fully resolved. Canadian bond yields are more likely to maintain their “higher-beta” status as global yields rise, as we discussed in a recent report.2 Thus, this week, we move our recommended allocation for Canadian government bonds to underweight from neutral. For Canada, the growth and inflation data continue to print strong enough to keep the Bank of Canada on a relatively more hawkish path than the other developed market central banks. This suggests that Canadian bond yields are more likely to maintain their “higher-beta” status as global yields rise, as we discussed in a recent report.2 Thus, this week, we move our recommended allocation for Canadian government bonds to underweight from neutral. Applying Our Global Golden Rule To Government Bond Allocations In September 2018, we published a Special Report introducing a government bond return forecasting methodology called the “Global Golden Rule.”3 This is an extension of a framework introduced by our sister service, US Bond Strategy, that links US Treasury returns (versus cash) to changes in the fed funds rate not already discounted in the US Overnight Index Swap (OIS) curve. That relationship also holds in other developed market countries, where there is a clear correlation between the level of bond yields and our 12-month discounters, which measure the change in policy rates over the next year priced into OIS curves (Chart 5). Chart 4Favor Lower-Beta Government Bond Markets In 2020 In Table 1, we show the expected returns generated by the Global Golden Rule (shown hedged into US dollars) for the countries in our model bond portfolio universe, based on monetary policy scenarios that we deem to be most plausible for 2020. Chart 5Monetary Policy Expectations Will Remain Critical For Bond Yields In Table 2, we show the returns on a duration-adjusted basis (expected total return divided by duration). We then rank the return scenarios for overall country indices, aggregating the returns of the individual yield curve maturity buckets shown in those two tables, in Table 3. Table 1Global Golden Rule Forecasts For 2020 The results in Table 1 show that expected returns are still expected to be positive across most countries, although this is largely due to the gains from hedging into higher-yielding US dollars. The duration-adjusted returns shown in Table 2 look most attractive at the front-end of yield curves across all the countries. This is somewhat consistent with our view, discussed in the 2020 Key Views report, that investors should expect some “bear-steepening” of global yield curves over the course of this year as inflation expectations drift higher (Chart 6). Table 2Global Golden Rule Duration-Adjusted Forecasts For 2020   Chart 6Expect A Mild Reflationary Bear Steepening Of Global Yield Curves Table 3Ranking The 2020 Return Scenarios The results in Table 3 show that the best expected returns would come in rate cutting scenarios – an unsurprising outcome given that there is very little change in policy rates currently discounted in OIS curves in all countries in our model bond portfolio universe. We see rates more likely to remain stable across all countries, however, making the “rates flat” scenarios in the middle of Table 3 more likely in 2020. After our downgrade of Canada this week, our recommended country allocations now reflect both yield betas and the results of our Global Golden Rule. Spread Product Allocation: OVERWEIGHT GLOBAL CORPORATES VERSUS GOVERNMENT BONDS, IN THE US, EURO AREA AND EM Chart 7Stay Overweight US High-Yield Turning to credit markets, the shift of global central banks to a more accommodative stance – with global growth improving – has opened a window for another year of outperformance versus sovereign bonds in 2020. With market volatility likely to remain low, as discussed earlier, there is a strong case to increase credit allocations relative to government debt as 2020 begins. Turning to credit markets, the shift of global central banks to a more accommodative stance – with global growth improving – has opened a window for another year of outperformance versus sovereign bonds in 2020. We already have a recommended overweight allocation to US high-yield corporate debt within our model bond portfolio. As we discussed in a recent report, the conditions that would lead us to become more cautious on US junk bonds – deteriorating corporate health, restrictive Fed policy and tightening bank lending standards – are currently not in place (Chart 7).4 If US economic growth starts to rebound in the first half of 2020, as we expect, then the case for US junk bond outperformance is even stronger. We are maintaining only a neutral allocation to US investment grade corporates, however, but this is part of a relative value view versus US Agency mortgage backed securities, which look more attractive on a volatility-adjusted basis.5 We are comfortable with our US credit views, but we are making the following changes this week to raise the credit allocation in our model bond portfolio: Upgrade EM USD-denominated sovereign and corporate debt to overweight. The two conditions that typically must be in place before EM hard currency debt can outperform – a softer US dollar and improving global growth – are now both in place. The two conditions that typically must be in place before EM hard currency debt can outperform – a softer US dollar and improving global growth – are now both in place (Chart 8). The momentum in the US dollar has clearly rolled over and even in level terms, the trade-weighted dollar is peaking. Add in the improvement in both our global LEI and the global manufacturing PMI (and the China PMI, most importantly) and the case for upgrading EM hard currency debt for 2020 is a strong one. Increase the size of overweights to euro area investment grade and high-yield corporate debt. We already have a modest overweight stance on euro area corporate bonds in our model bond portfolio, based on our expectations that the ECB will maintain a highly-accommodative stance – which could include buying more corporate debt in its Asset Purchase Program. Yet with an increasing body of evidence highlighting that the sharp downturn in European growth seen in 2019 is bottoming out, the argument for raising euro area corporate bond allocations for this year is compelling. This is especially true for euro area high-yield, where the backdrop looks even more constructive (Chart 9) compared to US junk bonds using the same metrics described above – corporate health (not deteriorating), monetary policy (not restrictive) and lending standards (not tightening). Chart 8Upgrade EM Credit To Overweight Chart 9Increase Overweights To European Credit Summing It All Up: Our Model Bond Portfolio Adjustments To Begin 2019 The outlook described in our 2020 Key Views report, and in this week’s report, lead us to make several adjustments to our model bond portfolio weightings seen in the table on Pages 15 and 16. The results of those changes are the following: Duration: We are maintaining an overall portfolio duration of 6.5 years, which is 0.5 years below that of our custom benchmark portfolio index (Chart 10). Credit Allocation: We are increasing the allocation to EM USD-denominated debt, funded by reducing exposure to US Treasuries. We are also increasing the size of the overweight positions in euro area investment grade and high-yield corporate debt, funded by cutting allocations to German and French government bonds. The net effect of these changes is to increase our total spread product weighting to 57% of the portfolio (Chart 11), which represents an overweight tilt versus the benchmark of +15% (versus a +8% overweight prior to this week’s changes). Chart 10Stay Below-Benchmark On Duration Exposure Chart 11A Larger Recommended Allocation To Spread Product For 2020 Country Allocation: We are cutting the Canadian government bond allocation to underweight, while making additional modest adjustments to yield curve positioning in the US, Japan, and the UK to reflect the output from our Global Golden Rule. The net result of these changes, combined with the increased allocation to corporate bonds, is to boost the overall portfolio yield to 3%, which represents a positive carry of +43bps versus our benchmark index (Chart 12). Chart 12Greater Portfolio Yield To Begin 2020 Chart 13Move To A Moderately Aggressive Level Of Portfolio Risk Overall Portfolio Risk: All of the above changes represent an increase in the usage of the “risk budget” of our model bond portfolio, which is now running a tracking error (or excess volatility versus that of the benchmark) of 73bps (Chart 13). This is higher than the 58bps prior to this week’s changes, but is still below the maximum allowable tracking error of 100bps that we have imposed on the model portfolio since its inception. This is consistent with our view that investors should maintain a “moderately aggressive” level of risk in fixed income portfolios in 2020.   Robert Robis, CFA Chief Fixed Income Strategist rrobis@bcaresearch.com Footnotes 1 Please see BCA Global Fixed Income Strategy Weekly Report, “2020 Key Views: Delay Of Reckoning”, dated December 12th 2019, available at gfis.bcarsearch.com. 2 Please see BCA Research Global Fixed Income Strategy Weekly Report, “How Sweet It Is”, dated November 6, 2019, available at gfis.bcaresearch.com. 3 Please see BCA Research Global Fixed Income Strategy Special Report, “The Global Golden Rule Of Bond Investing”, dated September 25th 2018, available at gfis.bcaresearch.com. 4 Please see BCA Research Global Fixed Income Strategy Weekly Report, “The Lowdown On Low-Rated High-Yield”, dated November 27, 2019, available at gfis.bcaresearch.com. 5 Please see BCA Research Global Fixed Income Strategy Weekly Report, “Big Mo(mentum) Is Turning Positive”, dated October 29, 2019, available at gfis.bcaresearch.com. Recommendations The GFIS Recommended Portfolio Vs. The Custom Benchmark Index Duration Regional Allocation Spread Product Tactical Trades Yields & Returns Global Bond Yields Historical Returns

GAA DM Equity Country Allocation Model Update The GAA DM Equity Country Allocation model is updated as of December 31, 2019.  The model made two significant changes to its allocations this month. First, the allocation to the US is now neutral from underweight previously; second, Australia becomes the second largest overweight (from underweight previously), largely due to an improvement in liquidity conditions. Japan, the UK and France remain the three large underweight countries, as shown in Table 1.  Table 1Model Allocation Vs. Benchmark Weights As shown in Table 2 and Charts 1, 2 and 3, the overall model outperformed the MSCI World benchmark in December by 7 bps, driven by the outperformance from the Level 2 model (17 bps), especially the overweight of Spain and underweight of Japan. The Level 1 model also generated one basis point of outperformance from the slight underweight in the US. Since going live, the overall model has outperformed by 64 bps, with 270 bps of outperformance by the Level 2 model, offset by 58 bps of underperformance from Level 1. Table 2Performance (Total Returns In USD %) Chart 1GAA DM Model Vs. MSCI World Chart 2GAA US Vs. Non US Model (Level 1) Chart 3GAA Non US Model (Level 2) For more on historical performance, please refer to our website http://gaa.bcaresearch.com/trades/allocation_performance. For more details on the models, please see Special Report, “Global Equity Allocation: Introducing The Developed Markets Country Allocation Model,” dated January 29, 2016, available at https://gaa.bcaresearch.com. Please note that the overall country and sector recommendations published in our Monthly Portfolio Update and Quarterly Portfolio Outlook use the results of these quantitative models as one input, but do not stick slavishly to them. We believe that models are a useful check, but structural changes and unquantifiable factors need to be considered as well when making overall recommendations. GAA Equity Sector Selection Model The GAA Equity Sector Model (Chart 4) is updated as of December 31, 2019. The model’s relative tilts between cyclicals and defensives have changed compared to last month. The global growth proxies used in our model have turned bullish, based on rising metal prices and EM currencies appreciating relative to the US dollar. This in turn led the model to reverse its defensive overweight it had instated last month on Consumer Staples and favor more cyclical sectors. The valuation component remains muted across all sectors except Energy. The accommodative stance likely to be implemented by global central banks will continue to lead the model to favor a mixed bag of cyclical and defensive sectors. The model is now overweight five sectors in total, four cyclical versus one defensive sectors. These are Consumer Discretionary, Information Technology, Financials, Materials, and Health Care.     Chart 4Overall Model Performance Table 3Overall Model Performance   For more details on the model, please see the Special Report “Introducing the GAA Equity Sector Selection Model”, dated July 27, 2016, as well as the Sector Selection Model section in the Special Alert “GAA Quant Model Updates”, dated March 1, 2019 available at https://gaa.bcaresearch.com.   Table 4Current Model Allocations   Xiaoli Tang Associate Vice President xiaoliT@bcaresearch.com Amr Hanafy Research Associate amrh@bcaresearch.com  

Special Report

Highlights US politics are the chief source of global geopolitical risk over the coming year – and likely beyond. President Trump’s reelection remains our base case – the sitting president rarely loses if the economy is expanding. Yet the risk of a Democratic victory is high – Trump’s low approval rating, impending impeachment trial, and various policy troubles threaten his reelection bid. Trump’s tactics and the Democrats’ turn to the progressive left pose threats to BCA Research’s cyclically bullish house equity view. Feature If a time-traveler had accosted you in the fall of 2014 and told you that Donald Trump, the host of the reality TV show The Apprentice, would be the next American president, would you have believed him? What if the time-traveler had gone on to say that President Trump’s unconventional behavior would get him into hot water and that in 2020 he would become the first president in US history to be impeached and removed from office? Granting the premise, the second proposition is easier to imagine. And yet Trump is highly unlikely to be removed from office. He is in fact favored to be reelected. Just as his victory in 2016 proved more likely than the consensus held at the time, so his reelection in 2020 is more likely than the consensus holds today. The reason comes down to political constraints. First, the bar for removal in the Senate is very high. Second, it is easier for a sitting president to get reelected than it is for the opposition to convince voters to start over with something entirely different. Especially if the economy is in decent shape. In what follows we present our quantitative 2020 election model and our qualitative, constraints-based analysis of the election and likely market responses. Trump's fate is only one factor. But US politics is the chief source of market-relevant global political risk over the next 12-24 months. Not A Lame Duck (Yet) After a harrowing year in which global manufacturing slumped due to China’s tight credit policy and Trump’s trade war, the probability of a US recession is now – tentatively – subsiding (Chart 1). This is good news for Trump, whose presidency is hanging by a thread. Chart 1Recession Averted? Or Trump's Death Knell? Chart 2Bookies Expect A Democrat Victory Betting markets like PredictIt.org suggest that Democrats are slightly more likely than Republicans to win the White House next November (Chart 2). The narrow spread is appropriate given that the balance of evidence is fairly even. However, if there is to be a tilt, it should go the opposite way, i.e. toward Republicans as the incumbent party. The history of US elections since 1860 shows a strong tendency for the incumbent party to hold the White House when the sitting president is running at the head of the ticket. This is especially true when there has not been a recession during the president’s four-year term. It is even true when the ruling party has lost seats in preceding congressional elections, as occurred in 2018 and as is often the case (Chart 3). Other than recession, the biggest exception to the sitting president’s victory – especially in modern times – is when a major scandal has occurred, as with Gerald Ford in 1976. This is clearly relevant to today. In these rare cases the incumbent president’s and incumbent party’s historic reelection rates are both 50/50. The implication of Chart 3 is that Trump’s odds, from a historical point of view, are slightly above 50%. Of course, history does not afford an example of a first-term president being impeached, acquitted, and running for election again.1 Yet this is the most likely outcome today, as there is not an overwhelming popular demand to remove Trump from office. Despite the revelations and public hearings in the impeachment inquiry so far, support for removal stands at 47%, while opposition to removal stands at 45% (Chart 4). In other words, there is no majority in favor of removal, but only a narrow plurality. Removal – nullifying an election result – requires more. Chart 3History Says Trump More Likely To Win Than Not Chart 4No Consensus On Removal From Office The spread is conspicuously close to the 46%-to-48% popular vote spread for Trump and Hillary Clinton, respectively, in 2016. The impeachment is not a tsunami of public opposition to the administration. It is a bare-knuckle power struggle: Trump tried to have his top rival investigated and tarred with corruption allegations, the Democrats are retaliating by trying to remove Trump prior to the election. Support for removal will fluctuate, but it will take more than 47% of the population to generate a 67-vote supermajority against Trump in a Republican-held Senate. Republican senators would be taking a grave risk in voting against their base when they have the option of deferring to voters in just 11 months’ time. Both Richard Nixon and Bill Clinton were in their second terms when Congress began moving articles of impeachment: the public had no other recourse in the event that they committed “high crimes and misdemeanors.” Trump is in his first term and is due for the public’s verdict shortly. Nixon resigned when it became clear that grassroots Republicans had lost faith in him and the Senate would not acquit. Trump’s political base has not yet lost faith – his approval among Republicans is still 90%, higher than the average of Republican presidents and at the high end of his term in office (Chart 5). When it comes to the final vote, some Republican senators may defect, but it would take 20 to remove Trump from office. This will require a Nixon-like hemorrhage of support. Remarkably Trump’s general approval rating has not been affected by the impeachment inquiry (Chart 6). His approval rating is still comparable to President Barack Obama’s rating at this stage in his first term (as well as Ronald Reagan’s). While Trump is highly unlikely to break above 50%, he is emphatically not a lame duck … at least not yet. Presidential approval tends to rise as the opposition nomination is settled and the election approaches. If Trump’s approval revives to the 46% of the popular vote he won in 2016, then he remains competitive in the swing states where the election will be fought and won. Chart 5Trump’s Political Base Geared Up For Battle Chart 6A Precarious Approval Rating What about the Republicans’ heavy losses in the midterm elections and special elections since 2016? Haven’t national voting trends already condemned Trump and the Republicans to a loss in 2020? Not necessarily. Democrats lost elections more dramatically in 2009-11 than Republicans lost in 2017-19 – both in voter support and turnout (Table 1) – and yet President Obama secured the victory in 2012. Presidential elections are a different beast. Table 1Democrats Suffered More Post-2008 Than Republicans Post-2016 … Yet Obama Won Reelection Chart 7GOP Governorships At Low End Of Rising Trend The same goes for Republican losses in recent gubernatorial races. In Kentucky the incumbent governor was a Republican and lost; in Louisiana the incumbent governor was a Democrat and won. The catch is that the number of Republican governors was extremely elevated prior to 2018. Recent losses have merely brought the Republicans back to the bottom of their upward channel as a share of the nation’s 50 governors (Chart 7). Thus while the interim elections are a warning sign to Trump and the GOP, they are not a death knell – as long as the economy rebounds and President Trump’s approval rises as the election approaches. Bottom Line: Trump is not a lame duck yet. His administration is embattled and the impeachment process could permanently damage his standing. But so far his general approval rating and the specific impeachment polling suggest that he will stay in office and remain competitive in the 2020 race. If the election were today he would almost surely lose, but a lot can change in 12 months. If the economy avoids recession, then investors should take reelection as their base case. Cyclical Constraints Will Prevail A recession is the surest way to render a president a lame duck. It does not have to be a technical recession. The contraction in the manufacturing sector – and corresponding cutbacks in lending in the manufacturing-heavy and electorally vital Midwest – are extremely threatening to a president who promised to revive manufacturing and trade (Chart 8). Incumbency, economic growth, failed impeachment, and partial policy victory are enough to win the key swing states. Having declared that “trade wars are good and easy to win,” President Trump will not be able to hide from a deeper slowdown in the industrial heartland. State-level wage growth is positive, but swing states, particularly Trump swing states, are seeing a sharp drop-off from the highs prior to the trade war (Chart 9). The solution is the trade ceasefire being pursued with China. Trump is now in the position of the Federal Reserve Chairman: he can no longer afford to hike (tariff) rates, and the equity market may force him to cut, as long as he can reasonably hope to improve the economy. If the economy is lost, the trade war is back on. Chart 8An Urgent Need For A Trade Ceasefire Chart 9Trump Swing States Took A Hit From The Trade War Chart 10Buttigieg And Warren More Favorable Than Others Are incumbency, economic growth, failed impeachment, and partial policy victories enough to get Trump over the line in the key swing states?2 Subjectively, we think so. The Democrats have to win all of the states they won in 2016 plus Michigan and Florida (or two other states in place of Florida, such as Wisconsin and Pennsylvania). President Trump can afford to lose Michigan and one other state (but not Florida). This assessment has little to do with the Democratic presidential nominee – as yet unknown – and everything to do with whether the incumbent president or party has been fundamentally discredited. Democratic candidates like Senator Elizabeth Warren and Mayor Pete Buttigieg are generally more competitive than consensus holds. Warren, for instance, is one of the few candidates in recent elections who has a net positive favorability rating (Chart 10). But her favorability is not enough to overturn a sitting president – that will most likely require a shock that renders the status quo intolerable. The cyclical constraints on Trump and his opponents are thus clear. What of the structural constraints? Trump’s 2016 victory is often attributed to long-running structural trends in the US such as deindustrialization, immigration, and racial attitudes. The Democrats’ “blue wall” in the Rust Belt crumbled because Trump courted the working-class voter there and/or stoked racial anxieties. The implication, however, is that Trump still has an advantage in these swing states. Older voters and especially white voters have drifted toward Republicans for several years – the trend was interrupted only by the Great Recession, which saw a surge in Democratic support that has now subsided (Chart 11). Chart 11Old And White People Drifting To GOP Over Time ... Excepting The Great Recession While the white share of the swing states is falling over time, that trend is not sufficient to prevent Trump from winning the Electoral College in the year 2020. Instead the rapidly changing racial and ethnic composition of society should be seen as motivating the attitudes that Trump exploits. Trump’s electoral strategy of maximizing white turnout and support for the Republican Party, which we dubbed “White Hype” in 2016, is still the only way for him to achieve a popular vote victory in 2020, and hence the clearest pathway for him to achieve an Electoral College victory (Chart 12). Needless to say, tensions and controversies over race and immigration will swell in the coming year. Chart 12Electoral College Scenarios Show Trump Win Still Possible Chart 13Swing State Turnout Follows Unemployment By the same token, demographic change means that the Democrats can theoretically win by performing no better than they did in 2016 in terms of voter turnout and support rates (see the “Status Quo” scenario in Chart 12). This is a low hurdle for Democrats – suggesting once again that the election will be extremely close, that Trump can win only through the Electoral College (not the popular vote), and that the election outcome will ultimately swing on the cyclical factors outlined above, particularly the state of the economy. A final word about voter turnout. The greatest electoral risk to President Trump is an increase in voter turnout among traditionally low turnout groups that heavily favor the Democratic Party, such as young people and minorities. Given the surge in turnout for the 2018 midterm elections, and the extremely controversial and heated environment surrounding Trump’s presidency, there is considerable reason to suspect that 2020 will be a high-turnout election. Other things being equal, this would likely penalize Trump’s reelection prospects. However, it is important to recognize that voter turnout in swing states is fairly well correlated with the unemployment rate (Chart 13). Depending on the state, surges in turnout occurred in 1992, in the wake of recession; 2004, in the wake of recession, terrorism and war; and 2008, in the wake of the great financial crisis. The exception is Pennsylvania, where a surge in white voter turnout helped Trump pull off a surprise win in the state. Turnout is the hardest political variable to predict, so it is not clear whether Trump’s scandals and impeachment will do the trick. But an increase in the unemployment rate would virtually destroy Trump’s bid, being negatively correlated with presidential approval and positively correlated with voter turnout. Bottom Line: Trump’s executive powers give him the potential to achieve some additional policy victories that could boost his approval rating – namely a trade ceasefire with China that simultaneously improves the economic outlook. Meanwhile structural factors such as demographics do not forbid Trump from winning the Electoral College – on the contrary, aging and the decline in the white share of the population mean that Trump’s electoral strategy could succeed again in 2020, but will be much harder to pull off after 2020. Introducing … BCA’s Geopolitical Strategy 2020 US Presidential Election Model The BCA Geopolitical Strategy Presidential Election Model is a state-by-state model that uses political and economic variables to predict the Electoral College vote. What differentiates our model from that of others is that it attempts to predict the probability of the incumbent party winning the Electoral College votes in each of the 50 states. The model would have predicted the past five elections correctly on an out-of-sample basis, even the controversial win of George W. Bush over Al Gore in 2000. Why do we predict the electoral vote rather than the popular vote? First, the winner of the presidential election is determined by the Electoral College, not the popular vote. Second, in recent history, two candidates who lost the popular vote (George W. Bush in 2000 and Donald Trump in 2016) won the election. It is possible that we will see a similar result in 2020, given President Trump’s low national popularity yet distinctive policy pitch for the Midwestern states (e.g. economic patriotism, hardline on immigration). With only minor exceptions, electoral votes are allocated based on a winner-take-all process, as opposed to proportionately to the popular vote. Hence the best way to forecast the presidential election winner is to predict the probability of winning each state, i.e. receiving all the electoral votes assigned to each state.3 Due to the data availability of our input variables, our sample size includes nine elections (1984 to 2016) across 50 states, making for a total of 450 observations. We designed the model to be as succinct as possible. It includes four explanatory variables: A weighted average of the Federal Reserve Bank of Philadelphia State Leading Index, from the beginning of the previous presidential term until September of the election year. The state leading indexes predict the 6-month growth rate of the state coincident indexes, which include nonfarm payroll employment, average hours worked in manufacturing by production workers, the unemployment rate, and wage and salary disbursements deflated by the consumer price index (U.S. city average).4 Chart 14Voters Make Up Their Minds Ahead Of Time We use a weighted average of all the monthly forecasts in the presidential term preceding an election, where later months are weighted more heavily than earlier months. Our sample includes 6-month growth rates up to and including September of the election year, which means it includes a rough forecast of the direction of the state’s economy in Q1 of the new president’s term. Since we weigh recent months more heavily, our model assigns more importance to forward-looking factors. It is sufficient to end our calculations of the average state leading indexes in September of the election year. First, the October data comes out in early November, just days before the election, which would be an insufficient lead-time for our final forecast. Second, most voters make their decision at least one month in advance of the election and last-minute changes in economic forecasts will likely not influence their decision (Chart 14). The incumbent party’s margin of victory in the previous presidential election in each state. This is measured as the incumbent party vote share minus the non-incumbent party vote share. Simply put, if the incumbent party failed to secure a solid win in a given state in the previous election, the probability of securing a solid win in the current election is much smaller. Average national approval level of the incumbent president in July of the election year. We tested the correlation between presidential approval in every month leading up to the election versus the election outcome and found that July approval levels have the second-highest correlation with the popular vote and Electoral College vote (Chart 15). Average October approval levels have slightly higher correlation with election outcomes, but not sufficiently so to sacrifice three months of lead-time. A “time for change” variable. This is a categorical variable indicating whether the incumbent party has been in the White House for one or more terms. Academic literature shows that a party that has occupied the White House for two terms or more is much less likely to win an election than a party that is running for a second term.5 Chart 15Voters Mostly Decided By July The output of our model is the probability of an incumbent win in each state. There are two ways of aggregating these probabilities to produce a national-level outcome: Allocate the number of Electoral College votes won by the incumbent proportionally to their probability of victory in each state, and then sum them up across all states. This method would smooth out potential errors in our forecast. The Republican Party is expected to win with 279 Electoral College votes in 2020. Assume a probability threshold of 50%: any state with an incumbent win that is at least 50% likely is fully assigned to the incumbent. While this method could significantly sway our forecast towards one of the parties because of small changes in probability, it is closer to the political reality. Even the smallest majority in a given state will (usually) result in the winning candidate getting all of the state’s Electoral College votes. We therefore adopt this method in our aggregation.6 Our model performs well in back tests: it correctly predicted every election in in-sample tests and every election from 2000 to 2016 in out-of-sample tests (Chart 16). Chart 16BCA Research Geopolitical Strategy Election Model: Back Tests Accurate Chart 17 shows our initial 2020 prediction. Overall, the Republican Party is expected to win 279 Electoral College votes, a 25-vote decrease from its 2016 result. Chart 17Trump Narrowly Slated To Win 2020 With 279 Electoral College Votes As of the latest available data, our model predicts that the Republicans will lose Michigan and Wisconsin (critical victories in 2016). Wisconsin, Pennsylvania, and New Hampshire become borderline or “toss-up” states: the probability of a Republican win in these states is 48.77%, 50.17%, and 46.90%, respectively. Even the smallest change in our inputs can shift these states to either party. The two inputs that can affect our forecast are the state leading index and President Trump’s approval level, since the other two inputs – the time for change variable and last election’s margin of victory – are fixed. Table 2 shows the predicted Electoral College votes for the Republican Party for various scenarios of these two variables. According to the model, President Trump is currently at the lowest level of approval and weakest state-by-state economy that he can afford. If one of these factors stabilizes below today’s level, Trump will lose his reelection bid. Table 2Small Decline In State Economies Could Ruin Trump’s 2020 Bid In the worst-case scenario for Trump – if his approval and the state leading indexes drop to the lowest levels they have touched in Trump’s presidency – the Republican Party will only manage to secure 230 Electoral College votes. The opposite, optimistic scenario would see them winning with 329 votes. An interesting takeaway from our model is that it captures the increase in American political polarization that has been widely observed by scholars. The 2020 forecast shows that many states will be won or lost by the incumbent party with extreme certainty (0% or 100%). Results of in-sample predictions show that this trend has been increasing since 1992 (Chart 18, top panel), which is also in line with our own measure of polarization (Chart 18, bottom panel). Since the results are based on in-sample estimations, the coefficients remain constant, so the differences in the results can be attributed to the underlying data. The impression of ever-intensifying polarization in the US is correct. What does this mean for Trump? He cannot be written off simply because he has a relatively low approval rating. Structural political factors that propelled him to the White House are still in place. His approval and the economy must deteriorate to change this base case. The chief risk to our model is the accuracy and interpretation of presidential approval polling. While polling data always has a margin of error, it is possible that approval polling is underestimating Trump’s support, particularly on the state level, as was witnessed in 2016 (Chart 19). Chart 18Rising Polarization – It’s Empirical Chart 19State-Level Polling Still A Risk We have a high degree of confidence in professional pollsters, who have also made improvements since 2016.   But asking Americans whether they “approve” of the unorthodox Trump may be a different proposition than in the past, disguising voting intentions to some degree. By choosing the level of Trump’s approval in our model (see Appendix), we are guarding against overstating his support and not allowing much room for any dampening effects or self-censorship, which is thus a risk to our model. Bottom Line: Quantitative modeling, entirely independent of our qualitative assessment, suggests that Trump is favored to win the 2020 election. However, he is skating on very thin ice with regard to key cyclical variables such as state-level economic performance and popular approval rating. If his approval level suffers from a slowing economy, or scandal and impeachment, then he will lose the critical toss-up states and the White House. Investment Conclusions In this report we have outlined a case where President Trump, despite his extreme unorthodoxy in general, and acute vulnerability at this moment in time, is still the most likely winner of the 2020 election. Elections are a Bayesian process in which investors should establish a clear prior, or starting place, and update their probabilities according to reliable data streams. This report establishes our prior and our key data streams. So what? Does it matter if Trump is reelected? Is it relevant to investors? From a bird’s eye view, Trump has made a few decisions that clearly distinguish his term in office from that of previous presidents. First, Trump replaced Janet Yellen with Jerome Powell at the Federal Reserve. It is debatable whether or how this affected the normalization of monetary policy. What is clear is that Trump made a change at the helm while pushing through highly stimulative fiscal policy. Fed hikes contributed to a rise in bond yields and an increase in market volatility, and the Fed was ultimately forced to adjust. Trump has vociferously criticized the Fed and demanded ever-lower rates. Second, by embracing sweeping Republican tax reform, Trump initiated pro-cyclical fiscal stimulus that widened the US’s monetary and economic divergence from the rest of the world, while exacerbating the US’s long-term fiscal woes. Third, by adopting protectionist trade policy to confront China’s mercantilism, Trump rattled global sentiment and contributed to a manufacturing recession. As long as our view remains correct, investors will have a base case that is cyclically bullish. Of these three macro developments, the only one that the election could substantially change is trade policy – and yet the Democrats are also taking a more hawkish approach to China. On the fiscal front, the Democrats will raise taxes, but they will not impose austerity – instead they propose large expansions of entitlements that the populace increasingly demands. Populist social spending combined with geopolitical struggle with China ensures that the deficit/GDP ratio will go up regardless of the party in power. From a market point of view, the historical record suggests that presidential elections – specifically elections that lead to gridlock between the White House and Congress, since we do not expect the Democrats to lose the House of Representatives – usually see a rising US stock market beforehand and a higher degree of volatility afterwards (Chart 20). Relative to developed market equities, US stocks typically underperform, and only resume their rise in the second half of the following year (i.e. 2021). Comparing Trump to other first-term presidents, it is clear that his “pluto-populism” (populism plus tax cuts for the rich) has exerted a reflationary effect on the equity market (Chart 21). As long as the data show that he has a fair chance of reelection, investors will have a base case that is cyclically bullish, despite the volatility to come from the Democrats’ taxation and regulation proposals. Chart 20Equity Outcomes Surrounding US Presidential Votes Chart 21Trump A Reason To Be Bullish What is most striking about Trump’s presidency is the low real total return on US Treasuries. This is despite his aggressive foreign and trade policy, which has motivated safe-haven flows into Treasuries this year (Chart 22). The bottom line is that the output gap is closed, the labor market is tight, and fiscal policy is expansive, putting upward pressure on yields. Given that Trump needs to cultivate a China ceasefire and economic improvement for reelection, this trend should continue until the next recession looms. Chart 22Trump Marks End Of Bull Market In Bonds The risk, however, is that Trump’s precarious China negotiations fall through, or that his scandals cause a permanent downshift in his approval rating, rendering him a lame duck. Not only would this free him of the election constraint that currently forces him to pursue pro-market policies, but it would also make a Democratic victory more likely. The Democratic nomination, meanwhile, could easily produce a progressive populist in the figure of Elizabeth Warren, who is still a frontrunner in the Democratic nomination. A bear market could develop quite easily if a normal equity market correction, which improves the odds of a Democratic victory becomes entangled in expectations that Warren is set to win the nomination. If the opposition can summon enough votes to unseat an incumbent president, chances are that the circumstances will include a “blue wave” that also sees the Democrats take the Senate. This would institute another sweeping change to American policy, this time in a direction that is unfriendly to corporate profits. As the probability of such a scenario rises, the equity market will have to discount it. Expectations of a Trump victory will spur the market upward – but investors should be wary. If this very long bull market has continued all the way to November 3, 2020, and President Trump is confirmed in office, the positive stock market reaction will likely provide an excellent time for booking profits and reducing risk. In a second term, Trump will be unshackled from his electoral constraints – very much unlike a first-term Democrat. This would free him to pursue his trade wars with fewer inhibitions – against China but also likely against Europe. A continuation of the trade war has important impacts across the full slate of global assets, as outlined in Chart 23, which depicts the movement of assets on days in which US equities reacted negatively to trade war developments. Chart 23A Trump Second Term Means Trade War With Fewer Constraints With 11 months to go, we are a world away from the election. The party nomination process, or third-party candidates, could overturn all expectations. But if there is one certainty, it is that polarization and political risk will rise in the coming 12-24 months. The losing side of the population will have deep heartburn. A crisis of legitimacy could easily haunt the next administration. There could be hanging chads, vote recounts, faithless electors, or contested results. The outcome of the election could turn upon unprecedented developments in the Electoral College, Supreme Court, or even in cyberspace. If the Democrats win, redistribution will amplify partisanship. If Trump wins, inequality will rise. There is no easy way forward for the United States.   Matt Gertken Vice President Geopolitical Strategist mattg@bcaresearch.com   Ekaterina Shtrevensky Research Analyst ekaterinas@bcaresearch.com Appendix 1: The Approval Question: Level Or Change? Chart 24Trump’s Historically Low Approval Rating The chief risk to our model is the interpretation of the presidential approval rating and its impact on the election. President Trump’s approval rating is notoriously low compared to the average president (Chart 24). While many authors use approval rating (or popularity) in their models, some argue that it is not the approval level, but the change in approval leading up to the election that matters.7 Consider the following: if President Trump’s approval increases from today’s level of 43% by 5%, he would be at the same level of approval as the average president if their approval were to drop by 5%. A model based on approval level would place these two presidents equally, while a model based on the change in approval would favor Trump. So which one is correct? We compare the incumbent’s popular vote in post-WWII elections with four different “variations” of incumbent president approval: the average level in July of the election year (as in our model); the deviation of the average October level from the election-year average, the change during the last two years of the term; and the range throughout the entire term. Directionally, the results are as expected. Level and change in approval are positively correlated with the popular vote, while a less stable approval (higher range) is negatively correlated (Chart 25A). We also find that approval level has the best fit with the election outcome, followed by the change in approval in the two years leading up to the election. However, if we restrict the sample size to the range of elections used in our model, 1984 to 2016, we find that the change in approval has a much better fit than the level (Chart 25B). In other words, in modern elections the presidential candidate’s momentum matters more in the final outcome. Chart 25AHigh, Rising, And Stable Approval Ratings … Chart 25B… Help Presidents Win Elections We tested each variation of approval as an input in our model instead of the July approval level. Table 3 summarizes the results. Trump wins in all four versions. Table 3All Measures Of Approval Favor Trump In 2020 Our current model penalizes Trump the most, while the model based on approval range favors him. This makes sense, given that President Trump’s approval is relatively low but very stable (Chart 26). Chart 26Trump Approval Very Low … And Very Stable We will continue to use approval level in our model to generate updated predictions, given that this measure has the best long-term historical fit with the election outcome. However, given that President Trump is performing relatively well on these other measures of approval, there is upside risk to his 2020 performance. Appendix 2: A Word About The Probit Model Table 4 presents the regression coefficients of our model. Since this is a probit model, the coefficients cannot be directly interpreted as they would in an ordinary regression. The coefficients in a probit regression model measure the change in the Z-score associated to each independent variable for a one-unit change in that variable. Table 4BCA 2020 US Presidential Election Model Statistics The sign of the coefficient corresponds to the direction of change in probability. So increases in the state leading index, presidential approval, or the incumbent’s margin of victory in the last election increase the probability of the incumbent winning a state. Of course, the latter variable is fixed and will not change until the election. At the same time, having occupied the White House for two terms or more decreases the probability of an incumbent win. But this is not the case in the current election. Footnotes 1 Andrew Johnson, the first to be impeached, did not run in 1868; Ulysses Grant bowed out after two terms in 1876, amid the “Great Barbecue” scandal; Warren Harding died before the election of 1924, amid the infamous “Teapot Dome” scandal; Harry Truman stepped down amid scandal after two terms in 1952; Richard Nixon resigned before the election of 1976; Bill Clinton was impeached and hit the two-term limit before the election of 2000. For these examples, and the electoral impact of great scandals in general, please see Allan J. Lichtman, Predicting The Next Presidency: The Keys To The White House 2016 (Rowman and Littlefield, 2016). 2 Trump’s policy record contains one major legislative victory, the Tax Cut and Jobs Act of 2017, along with a number of works in progress. The Republicans’ failed attempt to repeal and replace the Affordable Care Act (Obamacare) exacted an opportunity cost: it deprived Trump and the GOP Congress of time needed to legislate a southern border wall, while mobilizing the opposition for all subsequent elections. As for other policies, the renegotiation of NAFTA is only a partial success as the USMCA has not been ratified. The promised infrastructure package will become a campaign pledge for the second term. We expect some kind of North Korea deal. 3 To this end, we use a probit model, where the dependent variable is stated as 1 = incumbent party won all Electoral College votes in this state, or 0 = incumbent party did not win any Electoral College votes in this state. This model allows us to measure the probability that a state with certain characteristics will fall into one of these two categories. 4 “The leading index for each state predicts the six-month growth rate of the state’s coincident index. In addition to the coincident index, the models include other variables that lead the economy: state-level housing permits (1 to 4 units), state initial unemployment insurance claims, delivery times from the Institute for Supply Management (ISM) manufacturing survey, and the interest rate spread between the 10-year Treasury bond and the 3-month Treasury bill.” See the Federal Reserve Bank of Philadelphia, www.philadelphiafed.org. 5 Alan I. Abramowitz, “Forecasting the 2008 Presidential Election with the Time-for-Change Model,” Political Science and Politics, Vol. 41, No. 4 (Oct., 2008), pp. 691-695. 6 We also assume that the Democrats always win the District of Columbia. 7 Please see Michael S. Lewis-Beck, Charles Tien, “Forecasting presidential elections: When to change the model,” International Journal of Forecasting, Volume 24, Issue 2, April–June 2008, Pages 227-236, and Mark Zandi, Dan White, Bernard Yaros, “2020 Presidential Election Model,” Moody’s Analytics, September 2019.

GAA DM Equity Country Allocation Model Update The GAA DM Equity Country Allocation model is updated as of November 29, 2019.  The model has not made any directional change in its allocations this month. In terms of magnitude, however, the underweight of the US and the UK are both reduced slightly at the expense of other countries, as shown in Table 1.  As shown in Table 2 and Charts 1,  2 and 3, the overall model underperformed the MSCI World benchmark in November by 22 bps, caused by the underperformance from both the Level 1 (11 bps) and the Level 2 (27 bps) models. Four out of the five underweights worked well, especially the large underweight in Japan. However, none of the seven overweights panned out, especially the large overweight in Spain and Italy. Since going live, the overall model has outperformed by 51 bps, with 237 bps of outperformance by the Level 2 model, offset by 58 bps of underperformance from the Level 1. Table 1Model Allocation Vs. Benchmark Weights Table 2Performance (Total Returns In USD %) Chart 1GAA DM Model Vs. MSCI World Chart 2GAA U.S. Vs. Non U.S. Model (Level 1) For more on historical performance, please refer to our website http://gaa.bcaresearch.com/trades/allocation_performance. For more details on the models, please see Special Report, “Global Equity Allocation: Introducing The Developed Markets Country Allocation Model,” dated January 29, 2016, available at https://gaa.bcaresearch.com. Please note that the overall country and sector recommendations published in our Monthly Portfolio Update and Quarterly Portfolio Outlook use the results of these quantitative models as one input, but do not stick slavishly to them. We believe that models are a useful check, but structural changes and unquantifiable factors need to be considered as well when making overall recommendations. Chart 3GAA Non U.S. Model (Level 2) GAA Equity Sector Selection Model The GAA Equity Sector Model (Chart 4) is updated as of November 29, 2019. The model’s relative tilts between cyclicals and defensives have changed compared to last month. The global growth proxies used in our model have turned slightly bearish, reflecting concerns about the rebound. This in turn led the model to reverse a few of the overweights it had instated last month on sectors such as Industrials and Consumer Discretionary. The valuation component remains muted across all sectors except Energy. The model is now overweight three sectors in total, one cyclical versus two defensive sectors. These are Consumer Staples, Health Care, and Information Technology. Chart 4Overall Model Performance Table 3Overall Model Performance For more details on the model, please see the Special Report “Introducing the GAA Equity Sector Selection Model”, dated July 27, 2016, as well as the Sector Selection Model section in the Special Alert “GAA Quant Model Updates”, dated March 1, 2019 available at https://gaa.bcaresearch.com.   Table 4Current Model Allocations Xiaoli Tang Associate Vice President xiaoliT@bcaresearch.com Amr Hanafy Research Associate amrh@bcaresearch.com  

Highlights The US-China trade talks will continue despite Hong Kong. The UK election will not reintroduce no-deal Brexit risk – either in the short run or the long run. European political risk is set to rise from low levels, but Euro Area break-up risk will not. There is no single thread uniting emerging market social unrest. We remain constructive on Brazil. Feature Chart 1Taiwan Indicator To Rise Despite Ceasefire President Trump signed the Hong Kong Human Rights and Democracy Act into law on November 27. The signing was by now expected – Trump was not going to veto the bill and invite the Senate to override him with a 67-vote at a time when he is being impeached. He does not want to familiarize the Senate with voting against him in supermajorities. The Hong Kong bill will not wreck the US-China trade talks, but it is a clear example of our argument that strategic tensions will persist and cast doubt on the durability of the “phase one trade deal” being negotiated. It is better to think of it as a ceasefire, as Trump’s electoral constraint is the clear motivation. Trump is embattled at home and will contend an election in 11 months. He will not impose the tariff rate hike scheduled for December 15. A relapse into trade war would kill the green shoots in US and global growth, which partly stem from the perception of easing trade risk. Only if Trump’s approval rating collapses, or China stops cooperating, will he become insensitive to his electoral constraint. Will China abandon the talks and leave Trump in the lurch? This is not our base case but it is a major global risk. So far China is reciprocating. Xi Jinping’s political and financial crackdown at home, combined with the trade war abroad, has led to an economic slowdown and an explosion in China’s policy uncertainty relative to America’s. A trade ceasefire – on top of fiscal easing – is a way to improve the economy without engaging in another credit splurge. The US and China will continue moving toward a trade ceasefire, despite the Hong Kong bill. The move toward a trade ceasefire will probably keep our China GeoRisk Indicator from rising sharply over the next few months. However, our Taiwan indicator, which we have used as a trade war proxy at times, may diverge as it starts pricing in the heightened political risk surrounding Taiwan’s presidential election on January 11, 2020 (Chart 1). Sanctions, tech controls, Hong Kong, Taiwan, North Korea, Iran, the South China Sea, and Xinjiang are all strategic tensions that can flare up. Yes, uncertainty will fall and sentiment will improve on a ceasefire, but only up to a point. China’s domestic policy decisions are ultimately more important than its handling of the trade war. At the upcoming Central Economic Work Conference authorities are expected to stay focused on “deepening supply-side structural reform” and avoiding the use of “irrigation-style” stimulus (blowout credit growth). But this does not mean they will not add more stimulus. Since the third quarter, a more broad-based easing of financial controls and industry regulations is apparent, leading our China Investment Strategy to expect a turning point in the Chinese economy in early 2020. This “China view” – on stimulus and trade – is critical to the outlook for the two regions on which we focus for the rest of this report: Europe and emerging markets. Assuming that China stabilizes, these are the regions where risk assets stand to benefit the most. Europe is a political opportunity; the picture in emerging markets is, as always, mixed. United Kingdom: Will Santa Bring A Lump Of Coal? The Brits will hold their first winter election since 1974 on December 12. Prime Minister Boris Johnson’s Conservative Party has seen a tremendous rally in opinion polls, although it has stalled at a level comparable to its peak ahead of the last election in June 2017 (Chart 2). Another hung parliament or weak Tory coalition is possible. Yet the Tories are better positioned this time given that the opposition Labour Party is less popular than two years ago, while the Liberal Democrats are more capable of stealing Labour votes. The Tories stand to lose in Scotland, but the Brexit Party of Nigel Farage is not contesting seats with them and is thus undercutting Labour in certain Brexit-leaning constituencies. Markets would enjoy a brief relief rally on a single-party Tory majority. This would enable Johnson to get his withdrawal deal over the line and take the UK out of the EU in an orderly manner by January 31. The question would then shift to whether Johnson feels overconfident in negotiating the post-Brexit trade agreement with the EU, which is supposed to be done by December 31, 2020. This date will become the new deadline for tariff increases, but it can be extended. Johnson is as unlikely to fly off the cliff edge next year as he was this year, and this year he demurred. Negotiating a trade agreement is easier when the two economies are already integrated, have a clear (yet flexible) deadline, and face exogenous economic risks. Our political risk indicator will rise but it will not revisit the highs of 2018-19 (Chart 3). The pound’s floor is higher than it was prior to September 2019. Chart 2Tories Look To Be Better Positioned For A Single Party Majority Chart 3UK Risk Will Rise, But Not To Previous Highs Bottom Line: A hung parliament is the only situation where a no-deal Brexit risk reemerges in advance of the new Brexit day of January 31. The market is underestimating this outcome based on our risk indicator. But Johnson himself prefers the deal he negotiated and wishes to avoid the recession that would likely ensue from crashing out of the EU. And a headless parliament can prevent Johnson from forcing a no-deal exit, as investors witnessed this fall. We remain long GBP-JPY. Germany: The Risk Of An Early Election Germany is wading deeper into a period of political risk surrounding Chancellor Angela Merkel’s “lame duck” phase, doubts over her chosen successor, and uncertainty about Germany’s future in the world. The federal election of 2021 already looms large. Our indicator is only beginning to price this trend which can last for the next two years (Chart 4). On October 27 Germany’s main centrist parties suffered a crushing defeat in the state election of Thuringia. For the first time, the Christian Democratic Union (CDU) not only lost its leadership position, but also secured less vote share than both the Left Party and the right-wing Alternative für Deutschland (AfD) (Chart 5, top panel). Chart 4Germany Is Heading Toward A Period Of Greater Political Risk The AfD successfully positioned itself with the right wing of the electorate and managed to capture more undecided voters than any other party (Chart 5, bottom panel). Chart 5The Right-Wing AfD Outperformed In Thuringia … While the rise of the AfD (and its outperformance over its national polling) may seem alarming, Germany is not being taken over by Euroskeptics. Both support for the euro and German feeling of being “European” is near all-time highs (Chart 6). The question is how the centrist parties respond. Merkel’s approval rating is at its lower range. Support for Annegret Kramp-Karrenbauer (AKK), Merkel’s chosen successor, is plummeting (Chart 7). Since AKK was confirmed as party chief, the CDU suffered big losses in the European Parliament election and in state elections. Several of her foreign policy initiatives were not well received in the party.1 In October 2019, the CDU youth wing openly rejected her nomination as Merkel’s successor. At the annual CDU party conference on November 22-23, she only narrowly managed to avoid rebellion. She is walking on thin ice and will need to recover her approval ratings if she wants to secure the chancellorship. Meanwhile the CDU will lose its united front, increasing Germany’s policy uncertainty. Chart 6... But Euroskeptics Will Not Take Over Germany Germany’s other major party – the Social Democratic Party (SPD) – is also going through a leadership struggle. Chart 7The CDU Party Leader Is Walking On Thin Ice Chart 8A Return To The Polls Would Result In A CDU-Green Coalition In the first round of the leadership vote, Finance Minister Olaf Scholz and Klara Geywitz (member of the Brandenburg Landtag) secured a small plurality of votes with 22.7%, just 1.6% more than Bundestag member Saskia Esken and Norbert Walter-Borjans (finance minister of North Rhine-Westphalia from 2010-17). The latest polling, and Scholz’s backing by the establishment, implies that he will win but this is uncertain. The results of the second round will be published on November 30, after we go to press. What does the SPD’s leadership contest mean for the CDU-SPD coalition? More likely than not, the status quo will continue. Scholz is an establishment candidate and supports remaining in the ruling coalition until 2021. Esken is calling for the SPD to leave the coalition, but Walter-Borjans has not explicitly supported this. An SPD exit from the Grand Coalition would likely lead to a snap election, not a favorable outcome for stability-loving Germans. A return to the polls would benefit the Greens and AfD at the expense of the mainstream parties, and would likely see a CDU-Green coalition emerge (Chart 8). Given that a majority of voters want the SPD to remain in government (Chart 9), and that new elections would damage the SPD’s prospects, we believe that the SPD is likely to stay in government until 2021, even if the less established Esken and Walter-Borjans win. The risk is the uncertainty around Merkel’s exit. October 2021 is a long time for Merkel to drag the coalition along, so the odds of an early election are probably higher than expected. Chart 9Germans Prefer The SPD Remains In Government Chart 10Climate Spending Closest Germany Gets To Fiscal Stimulus (For Now) Chart 11There Is Room For More Fiscal Stimulus In Germany, If Needed What would a Scholz win mean for the great debate over whether Germany will step up its fiscal policy? If the establishment duo wins the SPD leadership, the Grand Coalition remains in place, and the economy does not relapse, we are unlikely to see additional fiscal stimulus in the near future. Scholz argues that additional stimulus would not be productive, as the slowdown is due to external factors (i.e. trade war).2 The recently released Climate Action Program 2030 is the closest to fiscal stimulus that we will see. This package will deliver additional spending worth 9bn euro in 2020 and 54bn euro until 2023 (Chart 10). We are unlikely to see additional fiscal stimulus from Germany in the near future. Bottom Line: Germany is wading into a period of rising political uncertainty. In the event of a downward surprise in growth, there is room to add more fiscal stimulus (Chart 11). But there is no change in fiscal policy in the meantime, e.g. no positive surprise. France: Macron Takes Center Stage While Merkel exits, President Emmanuel Macron continues to position himself as Europe’s leader – with a vision for European integration, reform, and political centrism. But in the near term he will remain tied down with his ambitious domestic agenda. France is trudging down the path of fiscal consolidation. After exiting the Excessive Deficit Procedure in 2018, and decreasing real government expenditures by 0.3% of GDP, France’s budget deficit is forecasted to decline further (Chart 12). Macron’s government is moving towards balancing its budget primarily by reducing government expenditures to finance tax cuts and decrease the deficit. Macron’s reform efforts following the Great National Debate – tax cuts for the middle class, bonus exemptions from income tax and social security contributions, and adjustment of pensions for inflation – have paid off.3 His approval rating is beginning to recover from the lows hit during the Yellow Vest protests (Chart 13). These reforms will be financed by lower government expenditures and reduced debt burden as a result of accommodative monetary policy. Chart 12Fiscal Consolidation In France Chart 13Macron's Reform Efforts Have Paid Off Overall, France has proven to a very resilient country in light of a general economic slowdown (Chart 14, top panel). Business investment and foreign direct investment, propped up by gradual cuts in the corporate income tax rate, have remained steady, and confidence remains strong (Chart 14, bottom panels). France is consumer driven and hence somewhat protected from storms in global trade. Chart 14French Economy Resilient Despite Global Slowdown Chart 15Ongoing Strikes Will Register In French Risk Indicator Bottom Line: France stands out for remaining generally stable despite pursuing structural reforms. Strikes and opposition to reforms will continue, and will register in our risk indicator (Chart 15), but it is Germany where global trends threaten the growth model and political trends threaten greater uncertainty. On the fiscal front France is consolidating rather than stimulating.   Italy: Muddling Through This fall’s budget talks caused very little political trouble, as expected. The new Finance Minister Roberto Gualtieri is an establishment Democratic Party figure and will not seek excessive conflict with Brussels over fiscal policy. Italy’s budget deficit is projected to stay flat over 2019 and 2020. The key development since the mid-year budget revision was the repeal of the Value Added Tax hike scheduled for 2020, a repeal financed primarily by lower interest spending.4 Equity markets have celebrated Italy’s avoidance of political crisis this year with a 5.6% increase. Our own measure of geopolitical risk has dropped off sharply (Chart 16). But of course we expect it to rise next year given that Italy remains the weakest link in the Euro Area over the long run. The left-leaning alliance between the established Democratic Party and the anti-establishment Five Star Movement hurt both parties’ approval ratings. In fact, the only parties that have seen an increase in approval in the last month are the League, the far-right Brothers of Italy, and the new centrist party of former Prime Minister Matteo Renzi, Italia Viva (Chart 17). We expect to see cracks form next year, particularly over immigration, but mutual fear of a new election can motivate cooperation for a time. Chart 16Decline In Italian Risk Will Be Short Lived Chart 17The M5S-PD Alliance Damaged Their Approval Bottom Line: Italy’s new government is running orthodox fiscal policy, which means no boost to growth, but no clashing with Brussels either. Spain: Election Post Mortem Chart 18A Gridlocked Parliament In Spain The Spanish election produced another gridlocked parliament, as expected, with no party gaining a majority and no clear coalition options. The Spanish Socialist Workers’ Party (PSOE) emerged as the clear leader but still lost three seats. The People’s Party recovered somewhat from its April 2019 defeat, gaining 23 seats. The biggest loser of the election was Ciudadanos, which lost 47 seats after its highly criticized shift to the right, forcing its leader Alberto Rivera to resign. The party’s seats were largely captured by the far-right Vox party, which won 15.1% of the popular vote and more than doubled its seats (Chart 18). Socialist leader Pedro Sanchez has arranged a preliminary governing agreement with Podemos leader Pablo Iglesias, but it is unstable. Even with Podemos, Sanchez falls far short of the 176 seats he needs to govern. In fact, there are only three possible scenarios in which the Socialists can reach the required 176 seats and none of these scenarios are easy to negotiate (Chart 19). The first – a coalition with the People’s Party – can already be ruled out. The other two require the support of the smaller pro-independence party, which will be difficult for Sanchez to secure, given that he hardened his stance on Catalonia in the days leading up to the election. Chart 19No Simple Way To A Majority Government The next step for Sanchez is to be confirmed as prime minister in an “investiture” vote, likely on December 16.5 He would need 176 votes in the first round (or a simple majority in the second round) to gain the confidence of Congress. He looks to fall short (Chart 20).6 If he fails to be confirmed, Sanchez will have another two months to form a government or face the possibility of yet another election. Chart 20Sanchez Set To Fall Short In Investiture Vote Spain’s indecision is leading to small conflicts with Brussels. Last week, the European Commission placed Spain under the preventative arm of the Stability and Growth Pact, stating that the country had not done enough to reach its medium-term budget objective.7 The European Commission’s outlook on Spain is slightly more pessimistic than that of the Spanish government (Chart 21). Deficit projections could worsen if a left-wing government takes power that includes the anti-austerity Podemos – which means that Spain is the only candidate for a substantial fiscal policy surprise. Chart 21A Fiscal Policy Surprise In Spain? Chart 22Spanish Risk Will Keep Rising We expect our Spanish risk indicator to keep rising (Chart 22). The silver lining is that Spain’s turmoil – like Germany’s – poses no systemic risk to the Euro Area. Spain could also see an increase in fiscal thrust. Stay long Italian government bonds and short Spanish bonos. Bottom Line: We remain tactically long Italian government bonds and short Spanish bonos. Italian bonds will sell off less in a risk-on phase and rally more in a risk-off phase, and relative political trends reinforce this trade. Emerging Markets: Global Unrest Civil unrest is unfolding across the world, grabbing the attention of the global news media (Chart 23). The proximate causes vary – ranging from corruption, inequality, governance, and austerity – but the fear of contagion is gaining ground. Chart 23Pickup In Civil Unrest Raising Fear Of Contagion A country’s vulnerability to unrest can be gauged by two main factors: political voice and underlying economic conditions. • Political Voice: The Worldwide Governance Indicators, specifically voice and accountability, corruption, and rule of law, provide proxies for political participation (Chart 24). The aim is to assess whether there is a legitimate channel for discontent to lead to change. Countries with low rankings are especially at risk of experiencing unrest when the economy is unable to deliver. Chart 24Greater Risk Of Unrest Where Political Voice Is Absent • Economic Conditions: Last year’s tightening monetary conditions, the manufacturing and trade slowdown, the US-China trade war, and a strong US dollar have weighed on global growth this year. This is challenging, especially for economies struggling to pick up the pace of growth (Chart 25). It translates to increased job insecurity, in some cases where insecurity is already rife (Chart 26). The likelihood that economic deterioration spurs widespread unrest depends on both the level and change in these variables. The former political factor is a structural condition that becomes more relevant when economic conditions deteriorate. Chart 25The Global Slowdown Weighed On Growth In Regions Already Struggling … Chart 26… And Raise Job Insecurity Chart 27Brazilian Risk Unlikely To Reach Previous Highs BCA Research is optimistic on global growth as we enter the end game of this business cycle. Nevertheless risks to this view are elevated and emerging market economies are still reeling from the past year’s slowdown. This makes them especially sensitive to failures on the part of policymakers. As a result, policymakers will be more inclined to ease monetary and fiscal policy and less inclined to execute structural reforms. Brazil is a case in point. Our indicator is flagging a sharp rise in political risk (Chart 27). This reflects the recent breakdown in the real – which can go further as the finance ministry has signaled it is willing to depreciate to revive growth. Meanwhile the administration has postponed its proposals to overhaul the country’s public sector, including measures to freeze wages and reduce public sectors jobs. On the political front, President Jair Bolsonaro’s recent break from the Social Liberal Party and launch of a new party, the Alliance for Brazil, threatens to reduce his ability to get things done. This move comes at a time when Brazil’s political landscape is being shaken up by former president Luiz Inacio Lula da Silva’s release from jail, pending an appeal against his corruption conviction. The former leader of the Worker’s Party lost no time in vowing to revive Brazil’s left. Our risk indicator might overshoot due to currency policy, but we doubt that underlying domestic political instability will reach late-2015 and mid-2018 levels. Brazil has emerged from a deep recession, an epic corruption scandal, and an impeachment that led to the removal of former president Dilma Rousseff. It is not likely to see a crisis of similar stature so soon. Bolsonaro’s approval rating is the lowest of Brazil’s recent leaders, save Michel Temer, but it has not yet collapsed (Chart 28). An opinion poll held in October – prior to Lula’s release – indicates that Bolsonaro is favored to win in a scenario in which he goes head to head against Lula (Chart 29). Justice Minister Sergio Moro, who oversaw the corruption investigation, is the only candidate that would gain more votes when pitted against Bolsonaro. He is working with Bolsonaro at present and is an important pillar of the administration. So it is premature to pronounce Bolsonaro’s presidency finished. Chart 28Bolsonaro’s Approval, While Relatively Low, Has Not Collapsed Chart 29Bolsonaro Not Yet Finished The problem, as illustrated in Charts 25 and 26, is that Brazil still suffers from slow growth and an uninspiring job market – longstanding economic grievances. This will induce the administration to take a precautionary stance and slow the reform process. The result should be reflationary in the short run but negative for Brazil’s sustainability over the long run. There is still a positive path forward. Unlike the recently passed pension cuts and the public sector cuts that were just postponed – both of which zap entitlements from Brazilians – the other items on the reform agenda are less controversial. Privatization and tax reform are less politically onerous and will keep the government and economy on a positive trajectory. Meanwhile the pension cuts are unlikely to be a source of discontent as they will be phased in over 12-14 years. Thus, while the recent political events justify a higher level of risk, speculation regarding the likelihood of mass unrest in Brazil – apart from the mobilization of Worker’s Party supporters ahead of the municipal elections next fall – is overdone. Bottom Line: The growth environment in emerging markets is set to improve in 2020. US-China trade risk is falling and China will do at least enough stimulus to be stable. Moreover emerging markets will use monetary and fiscal tools to mitigate social unrest. This will not prevent unrest from continuing to flare. But not every country that has unrest is globally significant. Brazil is a major market that has recently emerged from extreme political turmoil, so a relapse is not our base case. Otherwise one should monitor Hong Kong’s impact on the trade deal, Russia’s internal stability, and the danger that Iranian and Iraqi unrest could cause oil supply disruptions. In the event that the global growth rebound does not materialize we expect Mexico and Thailand – which have better fundamentals – to outperform. Our long Thai equity relative trade is a strategic defensive trade.   Matt Gertken Vice President Geopolitical Strategist mattg@bcaresearch.com Ekaterina Shtrevensky Research Analyst ekaterinas@bcaresearch.com Roukaya Ibrahim Editor/Strategist Geopolitical Strategy RoukayaI@bcaresearch.com Footnotes 1 Please see “Merkel’s Successor Splits German Coalition With Rogue Syria Plan,” dated October 22, 2019 and “Merkel's Own Party Wants Outright Huawei Ban From 5G Networks,” dated November 15, 2019, available at bloomberg.com. 2 Please see “Scholz Says No Need for German Stimulus After Dodging Recession,” dated November 14, 2019, available at bloomberg.com. 3 Please see “France: Draft Budgetary Plan For 2020,” dated October 15, 2019, available at ec.europa.eu. 4 Please see “Analysis of the Draft Budgetary Plan of Italy,” dated November 20, 2019, available at ec.europa.eu. 5 Please see “Investiture calendar | Can a government be formed before Christmas?” dated November 14, 2019, available at elpais.com. 6 If Sanchez convinces PNV, BNG, and Teruel Exists to vote in his favor for both rounds of the vote, he would need ERC and Eh Bildu to abstain in order to win. However, given that the PSOE has stated that it will not even negotiate with Eh Bildu, it is likely that this party will vote against Sanchez, giving the opposition 168 votes. In this case, Sanchez would not only need PNV, BNG, and Teruel in his favor, but also the support of either CC or ERC, both unlikely scenarios. 7 Please see “Commission Opinion on the Draft Budgetary Plan of Spain,” dated November 20, 2019, available at ec.europa.eu. Appendix Germany: GeoRisk Indicator France: GeoRisk Indicator Italy: GeoRisk Indicator Spain: GeoRisk Indicator UK: GeoRisk Indicator Canada: GeoRisk Indicator China: GeoRisk Indicator Taiwan: GeoRisk Indicator Korea: GeoRisk Indicator  Russia: GeoRisk Indicator Brazil: GeoRisk Indicator Turkey: GeoRisk Indicator What's On The Geopolitical Radar? Section III: Geopolitical Calendar

Special Report

Highlights Net inflows into US assets have been rolling over since the beginning of 2019, given that the repatriation associated with the 2017 tax cuts was a one-off effect. Besides, fading interest rate differentials are making US Treasuries less attractive, which is a headwind for the greenback.  A trade war ceasefire between the US and China should improve the balance of payments dynamics for export-oriented nations. We maintain a pro-cyclical stance.  A revival in oil demand and curbs on supply should underpin oil prices through 2020, which could lift the trade balances of Norway and Canada. However, we expect the Canadian dollar to underperform, weighed by pipeline constraints and the divergence between WCS and WTI prices. Stay short CAD/NOK. Feature The balance of payments is one of the key indicators we watch on a regular basis to gauge the direction of exchange rates. While the power of BoP on currency moves differs from one country to another, it provides a big picture view of a country's transactions with other nations. Generally speaking, persistent surpluses are usually associated with appreciation in currencies, and vice versa. Ongoing trade disputes since early 2018 have caused some fluctuation in current account balances globally. Political uncertainties and rising protectionism have also limited foreign investments in some countries. Going forward, should global growth stabilize amid a possible trade détente, export-oriented regions will have more scope to improve their balance of payments dynamics. In what follows we present balance of payments across G10 through five categories: the trade balance, the current account balance, foreign direct investment, the basic balance, and lastly, portfolio investment. United States Chart 1US Balance Of Payments The US trade deficit has been more or less flat, lingering around 3% of GDP. The trade deficit mostly comes from manufactured goods. On the positive side, the US has been producing and exporting more petroleum and related products, which has decreased oil demand from abroad. Meanwhile, exports of pharmaceutical products are on the rise. The current account is at a smaller deficit of 2.5% of GDP, thanks to a positive net international investment position. Foreign direct investment had been increasing due to repatriation by US companies since the 2017 Trump tax cuts. If this one-off tax break was a source of US dollar strength in 2018, that support is now gone. Meanwhile, dollar strength since the beginning of 2018 may have made US assets less attractive to foreign investors. Since the beginning of 2019, net inflows into US assets have been rolling over, and have fallen to 0.9% of GDP. This has brought the US basic balance down to -1.6% of GDP. In terms of portfolio investment, US bond markets are still appealing to foreign investors, but interest rate differentials are moving against the greenback. Total foreign purchases of US Treasury bonds have been negative this year, of which official purchases stand at US$350 billion of net outflows. In short, the path of least resistance for the US dollar is down, due to a widening current account deficit, waning foreign direct investment, fading interest rate differentials and increasing dollar liquidity. Euro Area Chart 2Euro Area Balance Of Payments The slowdown in global trade has hit European exports, but the trade balance is still sporting a “healthy” surplus of 1.7% of GDP, albeit far below its peak. As a result, the current account as of September 2019 was still at a healthy level of 2.7% of GDP. Should a US-China "phase one" deal be finalized, the trade balance in the euro area is likely to rebound going into 2020. Foreign direct investment has been increasing to the point of being at its highest level over the past 20 years, or 1% of GDP. This has been aided in part by the peripheral countries, further evidence that we are getting a convergence in competitiveness across Eurozone countries. The cheap euro and lower cost of capital have helped. As a result, the basic balance for the euro area reached a new high of 3.8% of GDP in September 2019. Portfolio investment into the euro area has stopped deteriorating since the beginning of 2017 and is now sporting net inflows of 0.8% of GDP. European purchases of both foreign equities and foreign bonds are falling, probably a sign that domestic assets are becoming more attractive. For example, ETF inflows are accelerating. The restart of the European Central Bank’s asset purchase program will continue to act as an anchor for spread convergence in the euro area. Meanwhile, a rally in European equities will be another signal of recovery in the euro area. A healthy current account balance and improving foreign investments both signal a higher euro going forward. Japan Chart 3Japanese Balance Of Payments The trade slowdown has dealt a small blow to Japan’s current account balance. The trade deficit widened further in 2019, reaching -0.5% of GDP in Q3. Exports have been falling for a 10th consecutive month, weighed down in part by lower sales of auto parts and semiconductor equipment. But these will pick up should a trade truce be reached. Among its major trading partners, sales to the US, China and other Asian countries have fallen, but have risen in the Middle East and Western Europe. That said, Japan’s large net international investment position has helped keep the current account surplus at an elevated level of 3.4% of GDP. Foreign direct investment in Japan has been dismal for many years due to an offshoring of industrial production. Net FDI is currently standing at -4% of GDP, which has brought the basic balance below zero for the first time since 2016. The recent deceleration is further evidence that corporate Japan needs structural reforms. Portfolio investment remains in negative territory mostly due to Japanese residents' large purchases of foreign long-term bonds. Going forward, fund inflows to Japan could face more headwinds with the proposed change to the Foreign Exchange and Foreign Trade Act. The change aims to lower the minimum stake for foreign investors without government approval from 10% to 1%. Other changes include requiring foreign directors to seek permission before becoming a board member. That said, Japan’s large net international investment position, which produces a high current account surplus, will continue to make the yen a safe haven amid global uncertainties. United Kingdom Chart 4UK Balance Of Payments So far, a cheap pound has not yet staunched the deterioration in UK balance of payments. The UK trade deficit remained wide at 7% of GDP in the third quarter. Among its major trading partners, the trade deficit comes mainly from Germany and China, offset by a smaller surplus from the US, the Netherlands and Ireland. Net receipts are positive, but the current account balance is still in negative territory at -5% of GDP. The Brexit imbroglio has led to an exodus of foreign direct investment. Many international companies are fleeing the UK, but to the extent that we get a quick resolution after the December elections, the uncertainty is likely to subside. Portfolio investment in the UK has been volatile over the past few years and has not really helped dictate any discernable trend in the UK basic balance. More recently, inflows into UK gilts have been £19 billion in the second quarter, while flows into equities are also improving. Relative interest rate differentials are also likely to move in favor of the UK, especially if reduced uncertainty provides scope for the Bank of England to hike interest rates. At a minimum, compared with other European nations, gilts remain appealing to international investors. We remain positive on the pound and are long GBP/JPY in our portfolio. Canada Chart 5Canadian Balance Of Payments The Canadian trade deficit has been hovering near -1% of GDP over the past few years. The goods trade deficit narrowed this year, led mostly by an increase in energy exports and lower imports of transportation equipment. Further improvement in energy product sales will require an improvement in pipeline capacity and a smaller gap between WCS and Brent crude oil prices. The current account deficit has been narrowing, now standing at -2% of GDP, the smallest since 2008. This is helped by net receipts, especially driven by a rise in direct investment income. FDI has been the bright spot in Canadian BoP dynamics. FDI inflows have been in part helped by increased cross- border M&A activities. Net FDI into Canada now accounts for 2.7% of GDP. This has brought the basic balance back above zero for the first time since 2015. Portfolio investment is positive on a net basis, but the trend looks quite worrisome. Foreign entities are fleeing Canada. In the meantime, Canadian investment in foreign securities is on the rise, reaching C$6 billion in Q3. Profitability, liquidity concerns and a global push towards sustainable investing are making Canadian energy and mining companies unappealing for foreign capital. Moreover, with elevated house prices and depressed interest rates, the outlook for banking profitability is also concerning. A drop in the US dollar will help the loonie in the short term. Over the longer term, however, we prefer to be underweight the Canadian dollar, especially via the Australian dollar and the Norwegian krone, which have a better macro outlook. Australia Chart 6Australian Balance Of Payments Australia has seen the best balance of payments improvement among the G10. The Australian trade balance soared this year and now stands at 2.5% of GDP, the highest in several years. Terms of trade, which have increased by 45% since their 2016 bottom, have been one of the main drivers. Exports of iron ore and concentrates increased by 64% year-on-year in September 2019, adding to the positive trade balance. Ergo, Australia is benefitting from both a price and volume boost. Trade has lifted the current account to be on track to post its first surplus since the ‘70s. Going forward, we expect Australian trade to continue improving amid the US-China trade détente. Foreign direct investment dipped slightly in 2019, but from very elevated levels. At present, it still stands at 3.5% of GDP. This has allowed for a very healthy basic balance surplus of 2.9% of GDP. The largest sources of Australian foreign direct investment are the US and the UK. The FDI inflows tend to be concentrated in the mining and manufacturing sectors and generate a negative income balance for Australia. This has been part of the reason behind the country’s chronic current account deficit, but it is impressively becoming less and less important. Portfolio investment in Australia plunged in 2019, and now stands at -4.2% of GDP. This has been driven by an exodus from the bond market. The repatriation of capital back to the US probably helped exacerbate this trend. The Australian dollar is likely to rebound from a contrarian perspective. We are playing Aussie dollar strength via the New Zealand and Canadian dollars. New Zealand Chart 7New Zealand Balance Of Payments New Zealand is also benefitting from a terms-of-trade boost. The trade deficit marginally narrowed to -1.7% of GDP in the third quarter. Exports rose by 4% year-on-year in the third quarter, while imports rose by 3.6% year-on-year. Terms of trade increased in 2019, mainly driven by a rise in dairy and meat prices. It appears the pork crisis in China is benefitting New Zealand exports. As a result, the current account deficit narrowed slightly to 3.4% of GDP. Foreign direct investment in New Zealand rose sharply to 3.1% of GDP, partly driven by reinvestment in the banking sector. This almost brought the basic balance back into positive territory. If this trend continues, it will be the first time the basic balance is in positive territory in two decades. Portfolio investment in New Zealand has been deteriorating, with net outflows of $6.2 billion in the second quarter. This is almost 4% of GDP on an annualized basis. The withdrawal of equity and investment fund shares by foreign entities, as well as divestment of debt securities by the general government, are some of the reasons behind falling portfolio investment. In a nutshell, increased portfolio investment in New Zealand will be predicated on a terms-of-trade shock that boosts margin growth for agricultural exporters, or a policy shift that boosts domestic return on capital. We like the kiwi versus the dollar, but are underweight against its pro-cyclical peers, namely the Australian dollar and the Swedish krona. Switzerland Chart 8Swiss Balance Of Payments The Swiss trade balance has been in a structural surplus, and hugely underpins the nation’s large current account surplus. The improvement this year, a rebound to 5.4% of GDP in the third quarter, is notable. The increase in exports has been partly driven by higher sales of chemical and pharmaceutical products, jewelry, and metals. Combined with income inflows from its large net international investment position, this has produced a current account balance of 10.7% of GDP. The slowdown in foreign direct investment has eased sharply from a record-low of -16% to -8% of GDP. Tax breaks from the US Jobs Act in 2017 allowed for favorable divestment of FDI in Switzerland and repatriation back to the US. This was a one-off that is now behind us, which explains why the basic balance is shifting back into surplus territory, to the tune of 2.5% of GDP.  Portfolio investment has been gradually improving and now stands at 0.3% of GDP. Swiss paper and equities (which are defensive) have benefitted from increased safe-haven demand this year. The Swiss franc is likely to continue its slow structural appreciation in the years to come, interspersed with bouts of volatility. In the short-term, however, the Swiss National Bank is likely to use the currency to fight deflationary pressures. This suggests the EUR/CHF has upside tactically. Sweden Chart 9Swedish Balance Of Payments The Swedish trade balance has been in structural decline since 2004 and turned negative in 2016. A large component of Swedish exports are machinery and automobiles which have suffered stiff competition from other global giants. The good news is that the weak krona is starting to help. The third-quarter trade balance shifted to a surplus for the first time since 2016 and is currently standing at 0.2% of GDP. Combined with inflows from Sweden’s external investments, this has nudged the current account balance to 3.3% of GDP. Despite net FDI inflows falling to -2.1% of GDP, the basic balance still managed to remain stable at 1.2% of GDP due to the improvement in the current account balance. The recent decline in Swedish FDI has mirrored those in other countries. However, Swedish exports will benefit from a trade détente as well as from a broader improvement in global growth. This should stem FDI outflows. Net portfolio investment in Sweden has been volatile in recent years, but our expectation is for improvement. A weak krona has typically helped the manufacturing sector with a lag of 12 months. Moreover, with the krona trading at a large discount to its long-term fair value, foreign investors will likely benefit from both equity and currency returns, should cyclical stocks continue to outperform defensives. In summary, Sweden’s basic balance should recover to levels that have prevailed over the past few years. Norway Chart 10Norwegian Balance Of Payments The bottom in oil prices since 2016 has gone a long way towards improving Norway’s trade balance. Net trade has fallen marginally this year due to lower exports of oil and natural gas, but still stands at 7.2% of GDP. The trade balance is the primary driver of the current account balance, and the latter now stands at 6.4% of GDP. Norway has seen an exodus of foreign capital from both direct and portfolio investment.  Net FDI and portfolio investment stand at -3% and -4% of GDP, respectively. Declining oil production in the North Sea has been partly responsible for falling FDI. On the portfolio side of the equation, it has been mainly due to increased purchases of foreign equities and bonds, especially via the Oil Fund. Concerns around sustainable investing have also likely diverted investors away from Norwegian assets. Despite this, Norway still sports a basic balance surplus of 3.4% of GDP. Eventually, this basic balance will move from being supported by trade to income inflows from Norway’s large net international investment position. The Norwegian krone is cheap on many metrics, and is one of our favorite petrocurrencies at the moment. Should global growth stabilize, which will revive oil demand, inflows into Norway should improve.   Kelly Zhong Research Analyst kellyz@bcaresearch.com Chester Ntonifor Foreign Exchange Strategist chestern@bcaresearch.com Footnotes Trades & Forecasts Forecast Summary Core Portfolio Tactical Trades Limit Orders Closed Trades

Special Report

Highlights Building on a previous special report focused on the investable market, in this report we construct and present models designed to predict the odds of Chinese domestic equity sector outperformance. BCA Research's China Investment Strategy service will aim to use our newly developed sector outperformance probability models to better understand the drivers of performance at any given moment, and to make more active equity sector recommendations in the future. Episodes of domestic equity sector outperformance over the past decade appear to be more idiosyncratic (or sector specific) than has been the case for the investable market, suggesting that periods of “abnormal” relative sector performance may occur more frequently than in the investable universe. Among the predictors included in our model, our Li Keqiang leading indicator (based on monetary conditions, money, and credit growth) has been the most important. Our base case view argues in favor of domestic cyclicals over defensives over the coming year, but recent sector performance suggests that domestic consumer discretionary and tech should be favored within a cyclical equity portfolio over energy, materials, and industrials barring a surge in oil prices or a capitulation by Chinese policymakers in favor of “flood irrigation-style” stimulus. Over the long-term, we argue that investors have a good reason to favor domestic defensives over cyclicals until the latter demonstrates meaningfully better earnings performance. Feature We examined China’s investable equity sector performance in detail in our October 30 Special Report,1 with a particular emphasis on understanding the specific macroeconomic or equity market factors that have historically predicted relative sector performance. In today’s report, we extend our approach to China’s A-share market. Our research focused on constructing and presenting models that quantify a checklist-based approach to determining the odds of equity sector performance. The aim is to use these models to better understand the drivers of performance at any given moment, and to make more active equity sector recommendations in the future. These recommendations will not mechanically follow the models; rather, we plan to use them as a stand in for what typically would be expected given the macro and financial market environment, and as a basis to investigate “abnormal” relative performance. We find that episodes of domestic equity sector outperformance over the past decade appear to be more idiosyncratic (or sector specific) that has been the case for the investable market, suggesting that periods of “abnormal” relative sector performance may occur more frequently than in the investable universe. Among the macroeconomic and equity market factors that we found to be important predictors, our Li Keqiang leading indicator was the most significant. This confirms that China’s domestic market is more sensitive to monetary conditions, money, and credit growth than its investable peer. We also note the sharp difference in the relative performance of cyclicals versus defensives in the domestic market compared with the investable market, and what this means for investors over the coming 6-12 months. Finally, we argue that investors should maintain a structural bias towards defensive stocks in the domestic market until cyclicals demonstrate meaningfully better earnings performance, and point to an existing position in our trade book for investors interested in strategically allocating to the A-share market. Detailing Our Approach In our effort to better understand historical periods of domestic sector performance, we have chosen to model the probability of outperformance of each level 1 GICS sector (plus banks) based on a set of macro and equity market variables. Specifically, we use an analytical tool called a logistic regression, which forecasts the probability of a discrete event rather than forecasting the value of a dependent variable. We utilized this approach when building our earnings recession model for China (first presented in our January 16 Special Report).2 The “events” that we modeled are historical periods of individual Chinese investable sector outperformance from 2010 to 2018, relative to the MSCI China index (the “broad market”). We find that episodes of domestic equity sector outperformance over the past decade appear to be more idiosyncratic (or sector specific) than has been the case for the investable market. Chart I-1A and Chart I-1B illustrate these periods with shading in each panel. We then attempt to explain these episodes of outperformance with the following macro predictors: Chart I-1AThis Report Builds Models ##br##Aimed At... Chart I-1B...Predicting The Shaded Regions Of These Charts Periods of accelerating economic activity, represented by our BCA's China Activity Index Periods of rising leading indicators of economic activity, represented by our BCA Li Keqiang (LKI) Leading Indicator Episodes of tight monetary policy, defined as periods where China’s 3-month interbank repo rate is rising Periods of accelerating inflation, measured both by headline and core inflation We also include several equity market variables: uptrends in relative sector earnings, periods of rising broad market stock prices, uptrends in broad market earnings, and episodes of extreme technical conditions and relative over/undervaluation for the sector in question. In the case of energy stocks, we also include oil prices as a predictor. Chart I-2A and Chart I-2B illustrate these periods as well as the macro & market variables that we have included as predictors. Chart I-2AWe Use These Macroeconomic And Equity Market Factors... Chart I-2B...To Predict Periods Of Equity Sector Outperformance Our approach also accounts for the existence of any leading or lagging relationships between the macro and market variables we have used as predictors and sector relative performance. In most cases the predictors lead relative sector performance, but in some cases it is the opposite. In the case of the latter, we have limited the lead of any variable in our models to three months in order to reduce the need to forecast. Finally, our approach also limits the extent to which we consider a leading relationship between our predictors and relative sector performance, in order to avoid picking up overlapping economic cycles. This issue, and the evidence supporting the existence of a 3½-year credit cycle in China, is detailed in Box I-1 of our October 30 Special Report (please see footnote 1). Key Drivers Of Sector Performance: Domestic Versus Investable Pages 11-22 present the results of each sector’s outperformance probability model, along with a list of factors that were found to be useful predictors and a summary of the results. The importance of the factors included in the models is shown in each of the tables at the top right of pages 11-22 by a score of 1-3 stars, (loosely representing key levels of statistical significance) as well as each factor’s optimal lead or lag. A minus sign shows that the predictor leads sector relative performance, whereas a plus sign shows that it lags. Following a review of our domestic equity sector outperformance models, differences in the results from those presented in our previous report can be organized into three distinct elements: 1) the breadth of macro & equity market factors in predicting sector performance, 2) the relative importance of our LKI leading indicator, and 3) the difference between domestic/investable cyclical versus defensive performance. The Breath Of Predictive Factors Chart I-3In The Domestic Market, The Breadth Of Predictive Factors Is Narrower Compared with the models for investible sector performance that we detailed in our previous report, our work modeling domestic equity sector performance highlights that the breadth of predictive factors is narrower, particularly among cyclical sectors (Chart I-3). Our model for domestic materials (shown on page 12) is one exception to this rule, but we found that our models for energy, industrial, and consumer discretionary relative performance were all focused on fewer predictors than is the case for the investable market. In addition, our domestic utilities model has considerably worse predictive power than our model for investable utilities. The case of industrials is particularly notable: our model for investable industrials highlighted the importance of tight monetary policy, rising core inflation, rising broad market stock prices & earnings, and overbought and oversold technical conditions in explaining past periods of industrial sector outperformance. By contrast, our domestic industrials model is quite simple: the sector has been more likely to outperform, with a lag, when our BCA China Activity Index and LKI leading indicator have been rising, and underperform following periods of extreme overvaluation. One of the core conclusions of our previous report was that investors should view the relative performance of investable industrials versus consumer staples as a reflationary barometer, given the strong sensitivity of both sectors to tight monetary policy. We explained this sensitivity by pointing to the substantial difference in corporate health between the two sectors: industrial firms are heavily debt-laden and thus experience deteriorating operating performance and an environment of rising interest rates. In comparison, food and beverage firms appear to have the strongest balance sheets among the sub-sectors that we have examined, suggesting that they would benefit less from easier monetary conditions than firms in other industries. Our leading indicator for Chinese economic activity has been considerably more important in predicting domestic equity sector outperformance than in the investable market. However, these dynamics appear to be completely absent in influencing performance in China’s domestic equity market. Not only has domestic industrial sector relative performance not been negatively linked to periods of tight monetary policy, but our model for consumer staples (shown on page 15) highlights that periods of staples performance have been driven by two simple factors: the relative trend in staples EPS  (positive sign), and the trend in broad market EPS (negative sign). The Relative Importance Of Monetary Conditions, Money, And Credit Growth Chart I-4 summarizes the significance of the factors in predicting sector performance in general, by summing up each predictor’s number of stars across all of the models. The chart shows that our LKI leading indicator is the most important signal of sector performance that emerged from our analysis, followed by rising core inflation, rising broad market stock prices, rising economic activity, and oversold technical conditions. The ranking of results shown in Chart I-4 is fairly similar to those that we listed for the investable market, with two exceptions. First, for the domestic market, periods of tight monetary policy were considerably less important than in the investable market as an important predictor of relative sector performance. Instead, our LKI leading indicator was by far the most important predictor, which underscores a point that we have made in previous reports: domestic stocks appear to be much more sensitive to the trend in monetary conditions, money, and credit growth than for the investable market. This increased sensitivity has helped explain the difference in performance this year between the investable and domestic market, underscoring that the former has more catch-up potential than the latter in a trade truce scenario. Chart I-4Monetary Conditions, Money, & Credit Growth Drive A-Share Performance Second, in the investable market, episodes of significant overvaluation had essentially no power to predict future episodes of equity market underperformance. But this factor was an important or very important contributor to our domestic industrials, health care, and tech models. This finding is consistent with our May 23 Special Report, which noted that value stocks have outperformed in China’s domestic equity market over the past five years and underperformed in the investable market (Chart I-5). Chart I-5Value Has Been A More Successful ##br##Factor In The Domestic Market   Major Differences In The Performance Of Cyclicals Versus Defensives The results of our models for domestic equity sector performance did not change the cyclical & defensive labels that we applied in our previous report. The signs of the predictors shown in the tables on pages 11-22 clearly highlight that the domestic energy, materials, industrials consumer discretionary, and information technology sectors are cyclical sectors, and that consumer staples, health care, financials, telecom services, utilities, and real estate are defensive. What is striking, however, is that there is a major difference in the relative performance of equally-weighted domestic cyclicals versus defensives compared with what has occurred in the investable market over the past decade. Chart I-6A and Chart I-6B illustrate the different relative performance trends, along with their corresponding trends in relative P/E and relative EPS. Whereas the relative performance of investable cyclicals versus defensives has had somewhat of a stable mean over the past decade, domestic cyclicals have badly underperformed since early-2011. The charts also make it clear that this underperformance has been driven by a downtrend in relative EPS, not due to trend differences in relative valuation. Chart I-6ACyclicals/Defensives Somewhat Mean-Reverting In The Investable Market... Chart I-6B...But Not So In The Domestic##br## Market Digging further, it appears that this discrepancy can be largely explained by the significant difference in performance between investable and domestic tech over the past decade (Chart I-7). Whereas the former has outperformed the overall investable index by roughly 4-5 times since 2010, the relative performance of the latter has only very modestly risen. In effect, Charts I-6 and I-7 highlight that Chinese cyclical sectors have been structurally impaired over the past decade and have only been “saved” in the investable market by massive outsized outperformance of the tech sector. The fact that investable tech sector performance itself has been largely driven by 2 extremely successful firms underscores how narrowly based the investible cyclical versus defensives performance trend has been. Chart I-7A Huge Gap In Tech Explains Domestic Cyclical Underperformance Investment Conclusions There are three conclusions that investors can draw from our analysis. First, our research shows that episodes of domestic equity sector outperformance over the past decade appear to be more idiosyncratic (or sector specific) that has been the case for the investable market. This does not mean that domestic sector performance is not significantly impacted by macro and top down equity market factors, but it suggests that periods of “abnormal” relative sector performance may occur more frequently than in the investable universe. As such, investors should be prepared to include episode-specific investigation of abnormal performance as a regular part of their domestic equity sector allocation decisions. Investors should favor domestic cyclicals over the coming year, with exposure focused on consumer discretionary and tech. Second, the fact that our LKI leading indicator is in an uptrend suggests that investors should favor domestic cyclicals over defensives over the coming year, with a caveat. We have noted in several previous reports that our indicator is in a shallow uptrend, and the slower pace of money and credit growth than during previous economic upswings suggests that the bar may be higher for some cyclical sectors to outperform. We would advise investors to watch closely over the coming 3-6 months for signs of a technical breakout in all cyclical sectors. But sector performance in Q1 of this year, when the overall A-share market rose sharply versus global stocks, suggests that domestic consumer discretionary and tech should be favored within a cyclical equity portfolio over energy, materials, and industrials barring a surge in oil prices or a capitulation by Chinese policymakers in favor of “flood irrigation-style” stimulus (Chart I-8). Within resources, we prefer the investable energy sector to its domestic peer, due to a sizeable valuation advantage. Chart I-8Favor Select Domestic Cyclical Sectors Over The Coming Year As a third and final point, abstracting from our bullish outlook for select cyclical sectors over the coming year, Charts 6 and 7 clearly argue for investors to maintain a structural bias towards defensive stocks in the domestic market until cyclicals demonstrate meaningfully better earnings performance. In the May 23 Special Report that we referred to above, we noted that an A-share portfolio formed of industry groups with above-median return on equity and below-median ex-post beta has significantly outperformed over the past decade. Table I-1 presents the current industry group weights of this portfolio, and shows that overweight exposure is concentrated in the health care, consumer staples, and real estate sectors (all of which are defensive), and a heavy underweight towards industrials. Table I-1Current High ROE / Low Beta Factor Industry Group Portfolio Weights* For clients who are interested in strategically allocating to the A-share market, we maintain a long position in this portfolio relative to the MSCI China A Onshore index in our trade book, and plan to continue to update the performance of the trade on a weekly basis. Energy Chart II-1 Table II-1 Similar to the investable energy sector, periods of domestic energy sector outperformance are strongly positively related to rising oil prices and rising headline inflation in China. We noted in our previous report that this is a behavioral relationship, rather than a fundamental one. Domestic energy stocks are negatively associated with rising broad market stock prices, unlike their investable peers. This largely reflects the fact that the relative performance of domestic energy stocks has been in a structural downtrend over the past decade. From 2010 to mid-2016, this decline was caused by a persistent underperformance in earnings. Since mid-2016, domestic energy sector EPS have been rising in relative terms, meaning that more recent underperformance has been due to multiple contractions. While not as relatively cheap as their investable peers, domestic energy stocks are heavily discounted versus the broad domestic market based on both the price/earnings ratio and the dividend yield. Consequently, it is possible that domestic energy stocks may at some point begin to outperform in a rising broad equity market environment. For now, our model argues for an underweight stance towards domestic energy due to the lack of a clear uptrend in oil prices. As a pure value play, investable energy stocks maintain a dividend yield of nearly 6.5%, and are thus more attractive than their domestic peers. Materials Chart II-2 Table II-2 Our model for the domestic materials highlights that the sector’s performance has been related to strengthening economic activity and strongly related to a rising Li Keqiang leading indicator. Among the equity market variables that we tested, materials outperformance has been positively associated with rising relative EPS, rising broad market EPS, and prior oversold technical conditions. Similarly, the investable materials sector, these results show that domestic materials are a strong play on accelerating Chinese economic activity. The factors included in our domestic materials sector model are similar to those included in our investable material, except that relative material earnings have also been a significant predictor of sector relative performance. In addition, the macro & equity market predictors included in our domestic materials model have done a better job of leading material sector performance. The odds of domestic materials outperformance rose twice above the 50% mark this year according to our model, without any corresponding improvement in relative stock prices. The spikes in the model occurred largely because domestic materials became significantly oversold; technical conditions for the sector have only twice been weaker over the past decade. This underscores that investors should be watching domestic materials closely in Q1 of next year for signs of a relative rebound. Industrials Chart II-3 Table II-3 The results of our model for domestic industrial sector outperformance are interesting, as they imply that the drivers of performance are different between the domestic and investable markets. In the investable index, we found that industrials were heavily sensitive to monetary policy, rising core inflation, relative sector earnings, and periods of rising broad market stock prices. Our domestic model is considerably simpler: industrials outperform, with a lag, when our activity index and Li Keqiang leading indicator are rising. Periods of strong overvaluation have also been significant in predicting future episodes of domestic industrial sector underperformance. It is not clear to us why the drivers of relative performance for domestic industrials have been different than in the investable equity index, But the good news is that the relative simplicity of the model makes the investment decision making process for domestic industrials considerably easier. Today, domestic industrials are significantly undervalued, and our Li Keqiang leading indicator is in a shallow uptrend. This suggests that domestic industrials are likely to begin outperforming at some point in early-2020 following a bottoming in Chinese economic activity, unless policymakers are quick to tighten once activity begins to improve (which would be contrary to our expectations). Consumer Discretionary Chart II-4 Table II-4 Our domestic consumer discretionary model highlights that the sector’s relative performance is positively associated with a rising Li Keqiang leading indicator, rising core inflation, and rising broad market stock prices. Similar to its investable peers, domestic consumer discretionary stocks are cyclical, and positive relationship with core inflation may reflect improved pricing power for the sector. Unlike investable consumer discretionary, the domestic consumer discretionary has not been meaningfully impacted by the December 2018 changes to the global industry classification standard. Hence, our model does not exclude the internet & direct marketing retail sector as we did in our previous report on investable sectors. For now, our model suggests that the domestic consumer discretionary sector is likely to continue to underperform, given decelerating core inflation and the lack of a clear uptrend in the broad domestic equity index. However, as a cyclical sector, we will be watching closely for an upside breakout in domestic consumer discretionary performance in the first quarter as a signal to increase exposure to the sector. Consumer Staples Chart II-5 Table II-5 Our domestic consumer staples model is significantly different than that shown in our previous report for investable staples. This reflects sizeable differences in investable/domestic staples relative performance over the past decade, particularly from mid-2015 to late-2017 (where domestic staples outperformed significantly and investable staples languished). Of the two predictors found to be significant in explaining historical periods of domestic staples performance, a negative relationship with the trend in broad market EPS has been the most important. This underscores that staples are defensive sector. The trend in staples relative earnings has closely followed in importance, showing that the tremendous outperformance in domestic consumer staples over the past several years has, at least in part, been driven by fundamentals. Still, domestic consumer staples are currently priced at 34x earnings per share, compared with 15x for the overall domestic market. While our model currently argues for continued staples outperformance, the risk of a valuation mean reversion next year, against the backdrop of an improving economy, is above average. Over the coming 6-12 months, investors should be closely monitoring domestic staples for signs of waning earnings momentum and/or a major technical breakdown as potential signals to reduce domestic staples exposure. Health Care Chart II-6 Table II-6 Over the past decade, periods of domestic health care outperformance have been negatively associated with rising economic activity, rising core inflation, and rising broad market stock prices. Oversold technical conditions and periods of overvaluation have also helped predict future episodes of health care relative performance. These factors clearly point to the defensive nature of domestic health care, similar to health care stocks in the investable index. However, one clear difference between investable and domestic health care is that the former appears to have leading properties and the latter does not. We noted in our previous report that periods of investable health care underperformance appeared to lead, on average, our BCA Activity Index, periods of rising core inflation, and uptrends in the broad investable index. By contrast, domestic health care lags the Activity Index and core inflation by just over a year, and also lags the trend in broad market EPS. Our model points to further health care outperformance, but we would expect domestic health care stocks to underperform at some point next year following an improvement in economic activity and a resumed uptrend in broad domestic EPS. Financials Chart II-7 Table II-7 Our outperformance probability model for domestic financials highlights that the sector is countercyclical: periods of outperformance have been negatively related to our LKI leading indicator, rising core inflation, and rising broad market stock prices. Similar to the case of the investable index and unlike the case globally, financials are clearly defensive. Investable financials have exhibited atypical performance this year according to the model presented in our previous report. By contrast, domestic financials have performed in line with what our model has suggested: our LKI leading indicator is in a shallow uptrend, and the relative performance of domestic financials has trended flat-to-down since late-2018. Barring a major shift by the PBoC towards a hawkish stance in the coming year (which we do not expect), our base case view for the Chinese economy implies that domestic financials are likely to continue to underperform. Banks Chart II-8 Table II-8 Our model for domestic banks is similar to that of financials, with some important differences. In addition to being sensitive to our LKI leading indicator, domestic bank performance is negatively related to our Activity Index. Oversold technical conditions have also been quite important in predicting future episodes of domestic bank outperformance. The model is currently forecasting domestic bank underperformance, although it was late in predicting the selloff in bank stocks that began late last year. Similar to the case for domestic financials, our baseline view for the Chinese economy implies that domestic bank are likely to continue to underperform over the coming year. Information Technology Chart II-9 Table II-9 Our model for the domestic technology sector is different than that of investable tech, which reflects the vast difference in performance between the two sectors. While the relative performance of domestic tech has trended sideways over the past decade, investable tech stock prices have risen fourfold relative to the broad investable index. This difference is largely accounted for by the absence of the BAT stocks (Baidu, Alibaba, Tencent) from the domestic market. Similar to investable tech, domestic technology stocks are negatively related to tight monetary policy, and positively linked with a pro-cyclical economic variable (a rising LKI leading indicator). However, strangely, domestic tech has been strongly and negatively related to rising headline inflation, a finding with no clear fundamental basis. The model has been less successful in predicting domestic tech performance over the past year than in the past, which appears to be linked to the inclusion of headline inflation in the model. Rising headline inflation has been clearly associated with three major episodes of domestic tech underperformance since 2010, but over the past year domestic tech has outperformed as headline inflation accelerated. For now we would advise investors to focus on the other factors in the model: the lack of overvaluation, and our view that policy will remain easy on a measured basis, supports an overweight stance towards domestic tech over the coming year. Telecom Services Chart II-10 Table II-10 Our domestic telecom services relative performance model highlights that the sector is defensive like its investable peer, but the factors driving performance are somewhat different. The only similarity between the two models is that periods of outperformance are negatively related to rising broad market stocks prices for both investable and domestic telecom services, with domestic telecom stocks responding with a lag. Among the macro factors included in the model, periods of domestic telecom services outperformance are negatively and coincidently related to our LKI leading indicator, and positively related to tight monetary policy (with a slight lead). Oversold technical conditions have also proven to help predict future episodes of outperformance. The model failed to predict a brief period of outperformance in mid-2018, but has generally accurately predicted underperformance of domestic telecom stocks since early-2017. Barring a collapse in the US/China trade talks or considerably weaker near-term economic conditions than we expect, domestic telecom services will likely continue to underperform until the specter of tighter monetary policy emerges. This is unlikely to occur until the middle of 2020, at the earliest. Utilities Chart II-11 Table II-11 Overall, our domestic utilities model has considerably worse predictive power than our model for investable utilities. The model shows that the performance of domestic utilities is negatively related to rising core inflation (with a lag) and rising broad market EPS, but these relationships are not particularly strong. We noted in our June 19 Special Report that domestic utilities ranked highly on the impact that relative EPS had on predicting relative stock prices , yet relative sector earnings did not register as a significant predictor in our model. This apparent discrepancy is resolved by differences in the time horizon between these two approaches. The analysis that we presented in our June 19 Special Report examined the relationship between earnings and stock prices over the entire sample period (2011-2018), meaning that it examined the predictive power of earnings over the long-term. The models built in this report have focused strongly on explaining periods of outperformance over a 6-12 month time horizon, there have been enough deviations in the trend between the relative performance of utilities and relative utilities earnings that the relationship between the two was not sufficiently strong to show up in the model. In other words, the long-term link between utilities relative earnings and stock prices is strong, but the short-term link is fairly weak. Real Estate Chart II-12 Table II-12 Similar to investable real estate, our model shows that domestic real estate is a counter-cyclical sector in that it is negatively related to periods of rising economic activity, a rising LKI leading indicator, tight monetary policy, and rising core inflation. Overbought technical conditions have also aided in predicting future episodes of domestic real estate underperformance. Our model for domestic real estate stocks has performed quite well on average, but its predictive success since late-2017 has been mixed. This period of atypical underperformance has coincided with a considerably weaker rebound in residential floor space sold than has occurred in previous recoveries in the real estate market. This suggests that domestic real estate stocks are more susceptible to trends in housing sales than their investable peers (which appear to be mostly sensitive to rising house prices). We noted in our November 6 Weekly Report that floor space sold is picking up , but it still remains weak when compared with history. This, in combination with our view that the Chinese economy will improve over the coming year, suggests that investors should avoid domestic real estate exposure relative to the overall domestic equity market. Footnotes 1  Please see China Investment Strategy Special Report "A Guide To Chinese Investable Equity Sector Performance," dated October 30, 2019, available at cis.bcaresearch.com 2  Please see China Investment Strategy "Six Questions About Chinese Stocks," dated January 16, 2019, available at cis.bcaresearch.com 3  Please see China Investment Strategy Special Report "Chinese Equity Sector Earnings: Predictability, Cyclicality, And Relevance," dated June 19, 2019, available at cis.bcaresearch.com 4  Please see China Investment Strategy Weekly Report "China Macro And Market Review," dated November 6, 2019, available at uses.bcaresearch.com Cyclical Investment Stance Equity Sector Recommendations

Dear Client, In addition to this short weekly report, you will also receive our 2020 outlook, published by the Bank Credit Analyst. Next week, I will be on the road visiting clients in South Africa. I hope to report my discussions and findings the following week. Best regards, Chester Ntonifor Highlights According to a simple attractiveness framework, the most desirable currencies are the Norwegian krone, the Swedish krona, and the Japanese yen. The least attractive are the New Zealand dollar and the British pound. Take profits soon on our long GBP/JPY position. Feature In this report, we use a simple framework for ranking G10 currencies. First, we consider the macroeconomic environment using as proxies a country’s basic balance and external vulnerability. Next, we look at valuation metrics,  surveying a variety of both short-term and longer-term models. Finally, we consider positioning, to gauge if our view is mainstream or out of consensus. Below are our results. Basic Balance Chart I-1Basic Balance We consider the basic balance to be one of the most important concepts in determining the attractiveness of a currency. In a nutshell, it captures the ebb and flow of demand for a country’s domestic assets. Persistent basic balance surpluses are usually associated with an appreciating currency and vice versa. The euro area sports the best basic balance surplus in the G10 universe, followed by Norway and then Australia (Chart I-1). In simple terms, this means there is constant strong underlying demand for these currencies - either for domestic goods and services, or for investment into portfolio assets. The UK and the US rank the worst in terms of basic balances, driven by Brexit uncertainty and the ebbing of tax reform benefits in the US. We will explore balance of payments dynamics within all of the G10 countries in detail next week. External Debt A currency is sometimes only as vulnerable as its external liabilities. In an absolute sense, external debt as a share of GDP is highest in the UK, euro area, and Switzerland (Chart I-2). However, what matters most times for vulnerability are net external assets rather than gross liabilities. On this measure, Japan, Switzerland, and Norway are the most attractive countries, while the US and Australia rank the worst (Chart I-3). Chart I-2External Vulnerability Chart I-3US Is Least Attractive Purchasing Power Parity (PPP) Chart I-4PPP Model Various models have shown PPP to be a very poor tool for managing currencies, but an excellent one at extremes. However, there is a roadblock that comes from measurement issues, since consumer price baskets tend to differ in composition from one country to the next. In order to get closer to an apples-to-apples comparison across countries, two adjustments are necessary. First, categorizing the consumer price index (CPI) into five major groups. In most cases, this breakdown captures 90% of the national CPI basket. This includes food, restaurants and hotels (1), shelter (2), health care (3), culture and recreation (4), and energy and transportation (5). The second adjustment is to run two regressions with the exchange rate as the dependent variable. The first regression (call it REG1) uses the relative price ratios of the five groups as independent variables. This allows us to observe the most influential price ratios that help explain variations in the exchange rate. The second regression (call it REG2) uses a weighted average combination of the five groups to form a synthetic relative price ratio. If, for example, shelter is 33% in the US CPI basket, but 19% in the Swedish CPI basket, relative shelter prices will represent 26% of the combined price ratio. This allows for a uniform cross-sectional comparison, as opposed to using the national CPI weights. The US dollar is overvalued, especially versus the Swedish krona, Japanese yen, and Norwegian krone.  The results show the US dollar as overvalued, especially versus the Swedish krona, Japanese yen, and Norwegian krone. Commodity currencies are closer to fair value, and within the safe-haven complex, the Japanese yen is more attractive than the Swiss franc. The euro is less undervalued than implied by the overvaluation in the DXY index (Chart I-4). Intermediate-Term Timing Model (ITTM) Back in 2016, we developed a set of currency indicators to help global portfolio managers increase their Sharpe ratio in managing currency exposure. The idea was quite simple: For every developed world country, there were three key variables that influenced the near-term path of its exchange rate versus the US dollar. Our intermediate-term timing models are not sending any strong signals at the moment.  Interest Rate Differentials: Under the lens of interest rate parity, if one country is expected to have lower interest rates versus another, the incumbent’s currency will fall today so as to gradually appreciate in the future and nullify the interest rate advantage. Chart I-5Intermediate-Term Model Inflation Differentials: Assuming no transactional costs, the price of sandals cannot be relatively high and rising in Mumbai versus Auckland. Either the Indian rupee needs to fall, the kiwi rise, or a combination of the two has to occur to equalize prices across borders. Risk Factor: Exchange rates are not government bonds in that few treasury departments and central banks can guarantee a par value on them. Ergo, the ebb and flow of risk aversion will have an impact on the Norwegian krone as well as the yen. For the most part, our models have worked like a charm. On a risk-adjusted return basis, a dynamic hedging strategy based on our ITTMs has outperformed all static hedging strategies for all investors with six different home currencies since 2001. These results give us confidence to continue running these models as a sanity check for our ever-shifting currency biases. That said, our intermediate-term timing models are not sending any strong signals at the moment. The Swedish krona, Norwegian krone, and New Zealand dollar are the most attractive currencies, while the British pound and Swiss franc are the least attractive (Chart I-5). Long-Term Fair Value Model Chart I-6Long-Term Model Our long-term FX models are also part of a set of technical tools we use to help us navigate FX markets. Included in these models are variables such as productivity differentials, terms-of-trade shocks, net international investment positions, real rate differentials, and proxies for global risk aversion. These models cover 22 currencies, incorporating both G10 and emerging market FX markets. The models are not designed to generate short- or intermediate-term forecasts. Instead, they reflect the economic drivers of a currency's equilibrium. Their main purpose is to provide information on the longevity of a currency cycle, depending on where we are in the economic cycle. Our long-term FX models are not sending any strong signals right now, with the US dollar at fair value. The cheapest currencies are the yen, the Norwegian krone, and Swedish krona (Chart I-6). The priciest currencies are the South African rand and the Saudi riyal. Real Interest Rates One defining feature of the currency landscape is that pretty much across the G10 countries, we have negative real rates (Chart I-7). Within  the G10 universe, the US and New Zealand dollars are the highest-yielding currencies, while the British pound and Swedish krona are the least attractive. Chart I-7Real Rates Speculative Positioning Being long Treasurys and the dollar has been a consensus trade for many years now (Chart I-8). According to CFTC data, this has been expressed mostly through the aussie and kiwi, although our bias is that the Swedish krona and Norwegian krone have been the real victims. Chart I-8Positioning That said, flow data highlights just how precarious being long US dollars is right now. Net foreign purchases by private investors are still positive, but the momentum of these flows is clearly rolling over. This is being more than offset by official net outflows. As interest rate differentials have started moving against the US, so has foreign investor appetite for Treasury bonds. Concluding Thoughts Should the nascent pickup in global growth morph into a synchronized recovery, it will go a long way in further eroding the US’ yield advantage. More specifically, the currencies that have borne the brunt of the manufacturing slowdown should also experience the quickest reversals. For example, yields in Norway, Sweden, Switzerland, and Japan have risen by much more than those in the US since the bottom. The most attractive currencies are the Swedish krona, the Norwegian krone, and the Japanese yen. The least attractive are the British pound and New Zealand dollar. This is the message being sent by an aggregate of our ranking model. The most attractive currencies are the Swedish krona, the Norwegian krone, and the Japanese yen. The least attractive are the British pound and New Zealand dollar (Chart I-9). Take profits soon on our long GBP/JPY position. Chart I-9Favor Norway, Japan and Sweden   Chester Ntonifor Foreign Exchange Strategist chestern@bcaresearch.com Currencies U.S. Dollar Chart II-1USD Technicals 1 Chart II-2USD Technicals 2 Recent data in the US have been mixed: Retail sales grew by 0.3% year-on-year in October. Industrial production contracted by 0.8% month-on-month in October. On the housing market front, building permits and housing starts both increased by 5% and 3.8% month-on-month in October. However, MBA mortgage applications contracted by 2.2% for the week ended November 15th. The NY Empire State Manufacturing index fell to 2.9 from 4 in November. The Philly Fed manufacturing index, on the other hand, soared to 10.4 from 5.6 in November. The DXY index depreciated by 0.3% this week. The FOMC minutes released this Wednesday showed that the Fed now sees little need to further reduce rates. Last week, we did a reassessment of global growth and the USD, and entered a limit sell for the DXY index at 100. Report Links: Place A Limit Sell On DXY At 100 - November 15, 2019 Signposts For A Reversal In The Dollar Bull Market - November 1, 2019 On Money Velocity, EUR/USD And Silver - October 11, 2019 The Euro Chart II-3EUR Technicals 1 Chart II-4UR Technicals 2 Recent data in the euro area have been mostly positive: The seasonally-adjusted trade balance fell to €18.3 billion in September. The current account surplus slightly narrowed by €0.3 billion to €28.2 billion. Headline and core inflation were both unchanged at 1.1% and 0.7% year-on-year respectively in October. Consumer confidence improved from -7.6 in October to -7.2 in November. EUR/USD increased by 0.5% this week. The improvement in soft data confirms that the economy is in a bottoming process in the euro area. The fact that the largest economy, Germany, skirted a recession last week also boosted investor confidence. We continue to remain overweight the euro. Report Links: On Money Velocity, EUR/USD And Silver - October 11, 2019 A Few Trade Ideas - Sept. 27, 2019 Battle Of The Central Banks - June 21, 2019 Japanese Yen Chart II-5JPY Technicals 1 Chart II-6JPY Technicals 2 Recent data in Japan have been positive: Exports decreased by 9.2% year-on-year in October. Imports slumped by 14.8% year-on-year. The total trade balance shifted to a surplus of ¥17.3 billion.  The industry activity index increased by 1.5% month-on-month in September. USD/JPY fell by 0.2% this week. While global growth is set to improve given a possible trade détente and easy monetary policy worldwide, uncertainties continue to loom. The US Senate unanimously passed legislation on the "Hong Kong Human Rights and Democracy Act," adding more difficulties to finalize the Phase I trade deal. Global trade uncertainty is positive for safe-haven demand. Report Links: Signposts For A Reversal In The Dollar Bull Market - November 1, 2019 A Few Trade Ideas - Sept. 27, 2019 Has The Currency Landscape Shifted? - August 16, 2019 British Pound Chart II-7GBP Technicals 1 Chart II-8GBP Technicals 2 Recent data in the UK have been positive: The Rightmove house price index increased by 0.3% year-on-year in November. Public sector net borrowing increased by £3 billion to £10.5 billion in October. The British pound continues to appreciate by 0.7% against the US dollar this week. With Brexit being less of a threat, the pound is poised to rise through next year. We are long GBP/JPY in our portfolio and it is in the money at 6.1%. Report Links: A Few Trade Ideas - Sept. 27, 2019 United Kingdon: Cyclical Slowdown Or Structural Malaise? - Sept. 20, 2019 Battle Of The Central Banks - June 21, 2019 Australian Dollar Chart II-9AUD Technicals 1 Chart II-10AUD Technicals 2 Recent data in Australia have been soft: The Westpac leading index fell by 0.1% month-on-month in October, following a slight decline the previous month. AUD/USD has been more or less flat this week. In the monetary policy minutes released this week, the RBA expressed their expectations for stronger growth at 2.75% in 2020 and around 3% in 2021, supported by accommodative monetary policy, infrastructure spending, stabilizing house prices, and strong steel-intensive activities in China. The minutes also presented an argument against lower interest rates: while lower interest rates can support the economy through the usual transmission channels, they could be negative for savers and confidence. That said, the RBA is "prepared to ease monetary policy further if needed." Report Links: A Contrarian View On The Australian Dollar - May 24, 2019 Beware Of Diminishing Marginal Returns - April 19, 2019 Not Out Of The Woods Yet - April 5, 2019 New Zealand Dollar Chart II-11NZD Technicals 1 Chart II-12NZD Technicals 2 Recent data in New Zealand have been positive: Both output and input components of the producer price index have increased in Q3: the output component grew by 1% quarter-on-quarter and input component by 0.9% quarter-on-quarter. NZD/USD increased by 0.7% this week. Both growth and inflation in New Zealand are showing signs that the economy is in a bottoming process. We are positive on the kiwi against the US dollar while we remain short against the Australian dollar and Swedish Krona. Report Links: Place A Limit Sell On DXY At 100 - November 15, 2019 USD/CNY And Market Turbulence - August 9, 2019 Where To Next For The US Dollar? - June 7, 2019 Canadian Dollar Chart II-13CAD Technicals 1 Chart II-14CAD Technicals 2 Recent data in Canada have been negative: Manufacturing shipments fell by 0.2% month-on-month in September. Both headline and core inflation were unchanged at 1.9% year-on-year in October. ADP employment showed a loss of 22.6K jobs in October. The Canadian dollar fell by 0.6% against the US dollar this week. While a possible trade détente between US and China and rising oil prices could put a floor under the loonie, the pipeline constraints in Canada have dampened the correlation between the oil prices and the loonie.  This will limit the upside potential for the Canadian dollar. Report Links: Making Money With Petrocurrencies - November 8, 2019 Signposts For A Reversal In The Dollar Bull Market - November 1, 2019 Preserving Capital During Riot Points - September 6, 2019 Swiss Franc Chart II-15CHF Technicals 1 Chart II-16CHF Technicals 2 Recent data in Switzerland have been positive: The trade surplus narrowed to CHF 3.5 billion in October from CHF 4.1 billion the previous month, due primarily to growth in imports, which grew by 1.9 billion month-on-month. Exports also increased by 1.3 billion month-on-month. Import demand remains firm for chemical products. Industrial production grew by 8% year-on-year in Q3. USD/CHF increased by 0.2% this week. The trade balance still remains at a high level in Switzerland, which is bullish for the franc. Moreover, global uncertainties could underpin the safe-haven franc. Report Links: Notes On The SNB - October 4, 2019 What To Do About The Swiss Franc? - May 17, 2019 Beware Of Diminishing Marginal Returns - April 19, 2019 Norwegian Krone Chart II-17NOK Technicals 1 Chart II-18NOK Technicals 2 Recent data in Norway have been positive: The trade balance shifted to a surplus of NOK 5.9 billion in October, after a deficit of NOK 1.4 billion in September. However, this is compared to a surplus of NOK 32.6 billion in the same month last year. On a year-on-year basis, exports slumped by 27%, caused by a decrease in exports of mineral fuels and chemical products. The Norwegian krone appreciated by 0.3% against the US dollar this week, supported by the oil price recovery. On Wednesday, the EIA posted an increase of crude oil inventories by 1.4 million barrels from the previous week, lower than expectations. WTI crude oil prices thus surged by 4% on the news. Going forward, we remain overweight energy prices and the Norwegian krone. Report Links: Making Money With Petrocurrencies - November 8, 2019 A Few Trade Ideas - Sept. 27, 2019 Portfolio Tweaks Into Thin Summer Trading - July 5, 2019 Swedish Krona Chart II-19SEK Technicals 1 Chart II-20SEK Technicals 2 Recent data in Sweden have been positive: Capacity utilization increased to 0.5% in Q3, up from 0.1% in the previous quarter. The Swedish krona increased by 0.7% against the US dollar this week. The Swedish krona has depreciated by 23% against the USD since its 2018 peak. A global growth revival is likely to give a boost to the krona from a valuation perspective. Report Links: Where To Next For The US Dollar? - June 7, 2019 Balance Of Payments Across The G10 - February 15, 2019 A Simple Attractiveness Ranking For Currencies - February 8, 2019 Trades & Forecasts Forecast Summary Core Portfolio Tactical Trades Limit Orders Closed Trades

Special Report

Highlights US politics are the chief source of global geopolitical risk over the coming year – and likely beyond. President Trump’s reelection remains our base case – the sitting president rarely loses if the economy is expanding. Yet the risk of a Democratic victory is high – Trump’s low approval rating, impending impeachment trial, and various policy troubles threaten his reelection bid. Trump’s tactics and the Democrats’ turn to the progressive left pose threats to BCA Research’s cyclically bullish house equity view. Feature If a time-traveler had accosted you in the fall of 2014 and told you that Donald Trump, the host of the reality TV show The Apprentice, would be the next American president, would you have believed him? What if the time-traveler had gone on to say that President Trump’s unconventional behavior would get him into hot water and that in 2020 he would become the first president in US history to be impeached and removed from office? Granting the premise, the second proposition is easier to imagine. And yet Trump is highly unlikely to be removed from office. He is in fact favored to be reelected. Just as his victory in 2016 proved more likely than the consensus held at the time, so his reelection in 2020 is more likely than the consensus holds today. The reason comes down to political constraints. First, the bar for removal in the Senate is very high. Second, it is easier for a sitting president to get reelected than it is for the opposition to convince voters to start over with something entirely different. Especially if the economy is in decent shape. In what follows we present our quantitative 2020 election model and our qualitative, constraints-based analysis of the election and likely market responses. Trump's fate is only one factor. But US politics is the chief source of market-relevant global political risk over the next 12-24 months. Not A Lame Duck (Yet) After a harrowing year in which global manufacturing slumped due to China’s tight credit policy and Trump’s trade war, the probability of a US recession is now – tentatively – subsiding (Chart 1). This is good news for Trump, whose presidency is hanging by a thread. Chart 1Recession Averted? Or Trump's Death Knell? Chart 2Bookies Expect A Democrat Victory Betting markets like PredictIt.org suggest that Democrats are slightly more likely than Republicans to win the White House next November (Chart 2). The narrow spread is appropriate given that the balance of evidence is fairly even. However, if there is to be a tilt, it should go the opposite way, i.e. toward Republicans as the incumbent party. The history of US elections since 1860 shows a strong tendency for the incumbent party to hold the White House when the sitting president is running at the head of the ticket. This is especially true when there has not been a recession during the president’s four-year term. It is even true when the ruling party has lost seats in preceding congressional elections, as occurred in 2018 and as is often the case (Chart 3). Other than recession, the biggest exception to the sitting president’s victory – especially in modern times – is when a major scandal has occurred, as with Gerald Ford in 1976. This is clearly relevant to today. In these rare cases the incumbent president’s and incumbent party’s historic reelection rates are both 50/50. The implication of Chart 3 is that Trump’s odds, from a historical point of view, are slightly above 50%. Of course, history does not afford an example of a first-term president being impeached, acquitted, and running for election again.1 Yet this is the most likely outcome today, as there is not an overwhelming popular demand to remove Trump from office. Despite the revelations and public hearings in the impeachment inquiry so far, support for removal stands at 47%, while opposition to removal stands at 45% (Chart 4). In other words, there is no majority in favor of removal, but only a narrow plurality. Removal – nullifying an election result – requires more. Chart 3History Says Trump More Likely To Win Than Not Chart 4No Consensus On Removal From Office The spread is conspicuously close to the 46%-to-48% popular vote spread for Trump and Hillary Clinton, respectively, in 2016. The impeachment is not a tsunami of public opposition to the administration. It is a bare-knuckle power struggle: Trump tried to have his top rival investigated and tarred with corruption allegations, the Democrats are retaliating by trying to remove Trump prior to the election. Support for removal will fluctuate, but it will take more than 47% of the population to generate a 67-vote supermajority against Trump in a Republican-held Senate. Republican senators would be taking a grave risk in voting against their base when they have the option of deferring to voters in just 11 months’ time. Both Richard Nixon and Bill Clinton were in their second terms when Congress began moving articles of impeachment: the public had no other recourse in the event that they committed “high crimes and misdemeanors.” Trump is in his first term and is due for the public’s verdict shortly. Nixon resigned when it became clear that grassroots Republicans had lost faith in him and the Senate would not acquit. Trump’s political base has not yet lost faith – his approval among Republicans is still 90%, higher than the average of Republican presidents and at the high end of his term in office (Chart 5). When it comes to the final vote, some Republican senators may defect, but it would take 20 to remove Trump from office. This will require a Nixon-like hemorrhage of support. Remarkably Trump’s general approval rating has not been affected by the impeachment inquiry (Chart 6). His approval rating is still comparable to President Barack Obama’s rating at this stage in his first term (as well as Ronald Reagan’s). While Trump is highly unlikely to break above 50%, he is emphatically not a lame duck … at least not yet. Presidential approval tends to rise as the opposition nomination is settled and the election approaches. If Trump’s approval revives to the 46% of the popular vote he won in 2016, then he remains competitive in the swing states where the election will be fought and won. Chart 5Trump’s Political Base Geared Up For Battle Chart 6A Precarious Approval Rating What about the Republicans’ heavy losses in the midterm elections and special elections since 2016? Haven’t national voting trends already condemned Trump and the Republicans to a loss in 2020? Not necessarily. Democrats lost elections more dramatically in 2009-11 than Republicans lost in 2017-19 – both in voter support and turnout (Table 1) – and yet President Obama secured the victory in 2012. Presidential elections are a different beast. Table 1Democrats Suffered More Post-2008 Than Republicans Post-2016 … Yet Obama Won Reelection Chart 7GOP Governorships At Low End Of Rising Trend The same goes for Republican losses in recent gubernatorial races. In Kentucky the incumbent governor was a Republican and lost; in Louisiana the incumbent governor was a Democrat and won. The catch is that the number of Republican governors was extremely elevated prior to 2018. Recent losses have merely brought the Republicans back to the bottom of their upward channel as a share of the nation’s 50 governors (Chart 7). Thus while the interim elections are a warning sign to Trump and the GOP, they are not a death knell – as long as the economy rebounds and President Trump’s approval rises as the election approaches. Bottom Line: Trump is not a lame duck yet. His administration is embattled and the impeachment process could permanently damage his standing. But so far his general approval rating and the specific impeachment polling suggest that he will stay in office and remain competitive in the 2020 race. If the election were today he would almost surely lose, but a lot can change in 12 months. If the economy avoids recession, then investors should take reelection as their base case. Cyclical Constraints Will Prevail A recession is the surest way to render a president a lame duck. It does not have to be a technical recession. The contraction in the manufacturing sector – and corresponding cutbacks in lending in the manufacturing-heavy and electorally vital Midwest – are extremely threatening to a president who promised to revive manufacturing and trade (Chart 8). Incumbency, economic growth, failed impeachment, and partial policy victory are enough to win the key swing states. Having declared that “trade wars are good and easy to win,” President Trump will not be able to hide from a deeper slowdown in the industrial heartland. State-level wage growth is positive, but swing states, particularly Trump swing states, are seeing a sharp drop-off from the highs prior to the trade war (Chart 9). The solution is the trade ceasefire being pursued with China. Trump is now in the position of the Federal Reserve Chairman: he can no longer afford to hike (tariff) rates, and the equity market may force him to cut, as long as he can reasonably hope to improve the economy. If the economy is lost, the trade war is back on. Chart 8An Urgent Need For A Trade Ceasefire Chart 9Trump Swing States Took A Hit From The Trade War Chart 10Buttigieg And Warren More Favorable Than Others Are incumbency, economic growth, failed impeachment, and partial policy victories enough to get Trump over the line in the key swing states?2 Subjectively, we think so. The Democrats have to win all of the states they won in 2016 plus Michigan and Florida (or two other states in place of Florida, such as Wisconsin and Pennsylvania). President Trump can afford to lose Michigan and one other state (but not Florida). This assessment has little to do with the Democratic presidential nominee – as yet unknown – and everything to do with whether the incumbent president or party has been fundamentally discredited. Democratic candidates like Senator Elizabeth Warren and Mayor Pete Buttigieg are generally more competitive than consensus holds. Warren, for instance, is one of the few candidates in recent elections who has a net positive favorability rating (Chart 10). But her favorability is not enough to overturn a sitting president – that will most likely require a shock that renders the status quo intolerable. The cyclical constraints on Trump and his opponents are thus clear. What of the structural constraints? Trump’s 2016 victory is often attributed to long-running structural trends in the US such as deindustrialization, immigration, and racial attitudes. The Democrats’ “blue wall” in the Rust Belt crumbled because Trump courted the working-class voter there and/or stoked racial anxieties. The implication, however, is that Trump still has an advantage in these swing states. Older voters and especially white voters have drifted toward Republicans for several years – the trend was interrupted only by the Great Recession, which saw a surge in Democratic support that has now subsided (Chart 11). Chart 11Old And White People Drifting To GOP Over Time ... Excepting The Great Recession While the white share of the swing states is falling over time, that trend is not sufficient to prevent Trump from winning the Electoral College in the year 2020. Instead the rapidly changing racial and ethnic composition of society should be seen as motivating the attitudes that Trump exploits. Trump’s electoral strategy of maximizing white turnout and support for the Republican Party, which we dubbed “White Hype” in 2016, is still the only way for him to achieve a popular vote victory in 2020, and hence the clearest pathway for him to achieve an Electoral College victory (Chart 12). Needless to say, tensions and controversies over race and immigration will swell in the coming year. Chart 12Electoral College Scenarios Show Trump Win Still Possible Chart 13Swing State Turnout Follows Unemployment By the same token, demographic change means that the Democrats can theoretically win by performing no better than they did in 2016 in terms of voter turnout and support rates (see the “Status Quo” scenario in Chart 12). This is a low hurdle for Democrats – suggesting once again that the election will be extremely close, that Trump can win only through the Electoral College (not the popular vote), and that the election outcome will ultimately swing on the cyclical factors outlined above, particularly the state of the economy. A final word about voter turnout. The greatest electoral risk to President Trump is an increase in voter turnout among traditionally low turnout groups that heavily favor the Democratic Party, such as young people and minorities. Given the surge in turnout for the 2018 midterm elections, and the extremely controversial and heated environment surrounding Trump’s presidency, there is considerable reason to suspect that 2020 will be a high-turnout election. Other things being equal, this would likely penalize Trump’s reelection prospects. However, it is important to recognize that voter turnout in swing states is fairly well correlated with the unemployment rate (Chart 13). Depending on the state, surges in turnout occurred in 1992, in the wake of recession; 2004, in the wake of recession, terrorism and war; and 2008, in the wake of the great financial crisis. The exception is Pennsylvania, where a surge in white voter turnout helped Trump pull off a surprise win in the state. Turnout is the hardest political variable to predict, so it is not clear whether Trump’s scandals and impeachment will do the trick. But an increase in the unemployment rate would virtually destroy Trump’s bid, being negatively correlated with presidential approval and positively correlated with voter turnout. Bottom Line: Trump’s executive powers give him the potential to achieve some additional policy victories that could boost his approval rating – namely a trade ceasefire with China that simultaneously improves the economic outlook. Meanwhile structural factors such as demographics do not forbid Trump from winning the Electoral College – on the contrary, aging and the decline in the white share of the population mean that Trump’s electoral strategy could succeed again in 2020, but will be much harder to pull off after 2020. Introducing … BCA’s Geopolitical Strategy 2020 US Presidential Election Model The BCA Geopolitical Strategy Presidential Election Model is a state-by-state model that uses political and economic variables to predict the Electoral College vote. What differentiates our model from that of others is that it attempts to predict the probability of the incumbent party winning the Electoral College votes in each of the 50 states. The model would have predicted the past five elections correctly on an out-of-sample basis, even the controversial win of George W. Bush over Al Gore in 2000. Why do we predict the electoral vote rather than the popular vote? First, the winner of the presidential election is determined by the Electoral College, not the popular vote. Second, in recent history, two candidates who lost the popular vote (George W. Bush in 2000 and Donald Trump in 2016) won the election. It is possible that we will see a similar result in 2020, given President Trump’s low national popularity yet distinctive policy pitch for the Midwestern states (e.g. economic patriotism, hardline on immigration). With only minor exceptions, electoral votes are allocated based on a winner-take-all process, as opposed to proportionately to the popular vote. Hence the best way to forecast the presidential election winner is to predict the probability of winning each state, i.e. receiving all the electoral votes assigned to each state.3 Due to the data availability of our input variables, our sample size includes nine elections (1984 to 2016) across 50 states, making for a total of 450 observations. We designed the model to be as succinct as possible. It includes four explanatory variables: A weighted average of the Federal Reserve Bank of Philadelphia State Leading Index, from the beginning of the previous presidential term until September of the election year. The state leading indexes predict the 6-month growth rate of the state coincident indexes, which include nonfarm payroll employment, average hours worked in manufacturing by production workers, the unemployment rate, and wage and salary disbursements deflated by the consumer price index (U.S. city average).4 Chart 14Voters Make Up Their Minds Ahead Of Time We use a weighted average of all the monthly forecasts in the presidential term preceding an election, where later months are weighted more heavily than earlier months. Our sample includes 6-month growth rates up to and including September of the election year, which means it includes a rough forecast of the direction of the state’s economy in Q1 of the new president’s term. Since we weigh recent months more heavily, our model assigns more importance to forward-looking factors. It is sufficient to end our calculations of the average state leading indexes in September of the election year. First, the October data comes out in early November, just days before the election, which would be an insufficient lead-time for our final forecast. Second, most voters make their decision at least one month in advance of the election and last-minute changes in economic forecasts will likely not influence their decision (Chart 14). The incumbent party’s margin of victory in the previous presidential election in each state. This is measured as the incumbent party vote share minus the non-incumbent party vote share. Simply put, if the incumbent party failed to secure a solid win in a given state in the previous election, the probability of securing a solid win in the current election is much smaller. Average national approval level of the incumbent president in July of the election year. We tested the correlation between presidential approval in every month leading up to the election versus the election outcome and found that July approval levels have the second-highest correlation with the popular vote and Electoral College vote (Chart 15). Average October approval levels have slightly higher correlation with election outcomes, but not sufficiently so to sacrifice three months of lead-time. A “time for change” variable. This is a categorical variable indicating whether the incumbent party has been in the White House for one or more terms. Academic literature shows that a party that has occupied the White House for two terms or more is much less likely to win an election than a party that is running for a second term.5 Chart 15Voters Mostly Decided By July The output of our model is the probability of an incumbent win in each state. There are two ways of aggregating these probabilities to produce a national-level outcome: Allocate the number of Electoral College votes won by the incumbent proportionally to their probability of victory in each state, and then sum them up across all states. This method would smooth out potential errors in our forecast. The Republican Party is expected to win with 279 Electoral College votes in 2020. Assume a probability threshold of 50%: any state with an incumbent win that is at least 50% likely is fully assigned to the incumbent. While this method could significantly sway our forecast towards one of the parties because of small changes in probability, it is closer to the political reality. Even the smallest majority in a given state will (usually) result in the winning candidate getting all of the state’s Electoral College votes. We therefore adopt this method in our aggregation.6 Our model performs well in back tests: it correctly predicted every election in in-sample tests and every election from 2000 to 2016 in out-of-sample tests (Chart 16). Chart 16BCA Research Geopolitical Strategy Election Model: Back Tests Accurate Chart 17 shows our initial 2020 prediction. Overall, the Republican Party is expected to win 279 Electoral College votes, a 25-vote decrease from its 2016 result. Chart 17Trump Narrowly Slated To Win 2020 With 279 Electoral College Votes As of the latest available data, our model predicts that the Republicans will lose Michigan and Wisconsin (critical victories in 2016). Wisconsin, Pennsylvania, and New Hampshire become borderline or “toss-up” states: the probability of a Republican win in these states is 48.77%, 50.17%, and 46.90%, respectively. Even the smallest change in our inputs can shift these states to either party. The two inputs that can affect our forecast are the state leading index and President Trump’s approval level, since the other two inputs – the time for change variable and last election’s margin of victory – are fixed. Table 2 shows the predicted Electoral College votes for the Republican Party for various scenarios of these two variables. According to the model, President Trump is currently at the lowest level of approval and weakest state-by-state economy that he can afford. If one of these factors stabilizes below today’s level, Trump will lose his reelection bid. Table 2Small Decline In State Economies Could Ruin Trump’s 2020 Bid In the worst-case scenario for Trump – if his approval and the state leading indexes drop to the lowest levels they have touched in Trump’s presidency – the Republican Party will only manage to secure 230 Electoral College votes. The opposite, optimistic scenario would see them winning with 329 votes. An interesting takeaway from our model is that it captures the increase in American political polarization that has been widely observed by scholars. The 2020 forecast shows that many states will be won or lost by the incumbent party with extreme certainty (0% or 100%). Results of in-sample predictions show that this trend has been increasing since 1992 (Chart 18, top panel), which is also in line with our own measure of polarization (Chart 18, bottom panel). Since the results are based on in-sample estimations, the coefficients remain constant, so the differences in the results can be attributed to the underlying data. The impression of ever-intensifying polarization in the US is correct. What does this mean for Trump? He cannot be written off simply because he has a relatively low approval rating. Structural political factors that propelled him to the White House are still in place. His approval and the economy must deteriorate to change this base case. The chief risk to our model is the accuracy and interpretation of presidential approval polling. While polling data always has a margin of error, it is possible that approval polling is underestimating Trump’s support, particularly on the state level, as was witnessed in 2016 (Chart 19). Chart 18Rising Polarization – It’s Empirical Chart 19State-Level Polling Still A Risk We have a high degree of confidence in professional pollsters, who have also made improvements since 2016.   But asking Americans whether they “approve” of the unorthodox Trump may be a different proposition than in the past, disguising voting intentions to some degree. By choosing the level of Trump’s approval in our model (see Appendix), we are guarding against overstating his support and not allowing much room for any dampening effects or self-censorship, which is thus a risk to our model. Bottom Line: Quantitative modeling, entirely independent of our qualitative assessment, suggests that Trump is favored to win the 2020 election. However, he is skating on very thin ice with regard to key cyclical variables such as state-level economic performance and popular approval rating. If his approval level suffers from a slowing economy, or scandal and impeachment, then he will lose the critical toss-up states and the White House. Investment Conclusions In this report we have outlined a case where President Trump, despite his extreme unorthodoxy in general, and acute vulnerability at this moment in time, is still the most likely winner of the 2020 election. Elections are a Bayesian process in which investors should establish a clear prior, or starting place, and update their probabilities according to reliable data streams. This report establishes our prior and our key data streams. So what? Does it matter if Trump is reelected? Is it relevant to investors? From a bird’s eye view, Trump has made a few decisions that clearly distinguish his term in office from that of previous presidents. First, by replacing Janet Yellen with Jerome Powell at the Federal Reserve, Trump arguably accelerated the normalization of monetary policy, which contributed to a rise in bond yields, an increase in market volatility, a strong dollar, and a global slowdown. Second, by embracing sweeping Republican tax reform, Trump initiated pro-cyclical fiscal stimulus that widened the US’s monetary and economic divergence from the rest of the world, while exacerbating the US’s long-term fiscal woes. Third, by adopting protectionist trade policy to confront China’s mercantilism, Trump rattled global sentiment and contributed to a manufacturing recession. As long as our view remains correct, investors will have a base case that is cyclically bullish. Of these three macro developments, the only one that the election could substantially change is trade policy – and yet the Democrats are also taking a more hawkish approach to China. On the fiscal front, the Democrats will raise taxes, but they will not impose austerity – instead they propose large expansions of entitlements that the populace increasingly demands. Populist social spending combined with geopolitical struggle with China ensures that the deficit/GDP ratio will go up regardless of the party in power. From a market point of view, the historical record suggests that presidential elections – specifically elections that lead to gridlock between the White House and Congress, since we do not expect the Democrats to lose the House of Representatives – usually see a rising US stock market beforehand and a higher degree of volatility afterwards (Chart 20). Relative to developed market equities, US stocks typically underperform, and only resume their rise in the second half of the following year (i.e. 2021). Comparing Trump to other first-term presidents, it is clear that his “pluto-populism” (populism plus tax cuts for the rich) has exerted a reflationary effect on the equity market (Chart 21). As long as the data show that he has a fair chance of reelection, investors will have a base case that is cyclically bullish, despite the volatility to come from the Democrats’ taxation and regulation proposals. Chart 20Equity Outcomes Surrounding US Presidential Votes Chart 21Trump A Reason To Be Bullish What is most striking about Trump’s presidency is the low real total return on US Treasuries. This is despite his aggressive foreign and trade policy, which has motivated safe-haven flows into Treasuries this year (Chart 22). The bottom line is that the output gap is closed, the labor market is tight, and fiscal policy is expansive, putting upward pressure on yields. Given that Trump needs to cultivate a China ceasefire and economic improvement for reelection, this trend should continue until the next recession looms. Chart 22Trump Marks End Of Bull Market In Bonds The risk, however, is that Trump’s precarious China negotiations fall through, or that his scandals cause a permanent downshift in his approval rating, rendering him a lame duck. Not only would this free him of the election constraint that currently forces him to pursue pro-market policies, but it would also make a Democratic victory more likely. The Democratic nomination, meanwhile, could easily produce a progressive populist in the figure of Elizabeth Warren, who is still a frontrunner in the Democratic nomination. A bear market could develop quite easily if a normal equity market correction, which improves the odds of a Democratic victory becomes entangled in expectations that Warren is set to win the nomination. If the opposition can summon enough votes to unseat an incumbent president, chances are that the circumstances will include a “blue wave” that also sees the Democrats take the Senate. This would institute another sweeping change to American policy, this time in a direction that is unfriendly to corporate profits. As the probability of such a scenario rises, the equity market will have to discount it. Expectations of a Trump victory will spur the market upward – but investors should be wary. If this very long bull market has continued all the way to November 3, 2020, and President Trump is confirmed in office, the positive stock market reaction will likely provide an excellent time for booking profits and reducing risk. In a second term, Trump will be unshackled from his electoral constraints – very much unlike a first-term Democrat. This would free him to pursue his trade wars with fewer inhibitions – against China but also likely against Europe. A continuation of the trade war has important impacts across the full slate of global assets, as outlined in Chart 23, which depicts the movement of assets on days in which US equities reacted negatively to trade war developments. Chart 23A Trump Second Term Means Trade War With Fewer Constraints With 11 months to go, we are a world away from the election. The party nomination process, or third-party candidates, could overturn all expectations. But if there is one certainty, it is that polarization and political risk will rise in the coming 12-24 months. The losing side of the population will have deep heartburn. A crisis of legitimacy could easily haunt the next administration. There could be hanging chads, vote recounts, faithless electors, or contested results. The outcome of the election could turn upon unprecedented developments in the Electoral College, Supreme Court, or even in cyberspace. If the Democrats win, redistribution will amplify partisanship. If Trump wins, inequality will rise. There is no easy way forward for the United States.   Matt Gertken Vice President Geopolitical Strategist mattg@bcaresearch.com   Ekaterina Shtrevensky Research Analyst ekaterinas@bcaresearch.com Appendix 1: The Approval Question: Level Or Change? Chart 24Trump’s Historically Low Approval Rating The chief risk to our model is the interpretation of the presidential approval rating and its impact on the election. President Trump’s approval rating is notoriously low compared to the average president (Chart 24). While many authors use approval rating (or popularity) in their models, some argue that it is not the approval level, but the change in approval leading up to the election that matters.7 Consider the following: if President Trump’s approval increases from today’s level of 43% by 5%, he would be at the same level of approval as the average president if their approval were to drop by 5%. A model based on approval level would place these two presidents equally, while a model based on the change in approval would favor Trump. So which one is correct? We compare the incumbent’s popular vote in post-WWII elections with four different “variations” of incumbent president approval: the average level in July of the election year (as in our model); the deviation of the average October level from the election-year average, the change during the last two years of the term; and the range throughout the entire term. Directionally, the results are as expected. Level and change in approval are positively correlated with the popular vote, while a less stable approval (higher range) is negatively correlated (Chart 25A). We also find that approval level has the best fit with the election outcome, followed by the change in approval in the two years leading up to the election. However, if we restrict the sample size to the range of elections used in our model, 1984 to 2016, we find that the change in approval has a much better fit than the level (Chart 25B). In other words, in modern elections the presidential candidate’s momentum matters more in the final outcome. Chart 25AHigh, Rising, And Stable Approval Ratings … Chart 25B… Help Presidents Win Elections We tested each variation of approval as an input in our model instead of the July approval level. Table 3 summarizes the results. Trump wins in all four versions. Table 3All Measures Of Approval Favor Trump In 2020 Our current model penalizes Trump the most, while the model based on approval range favors him. This makes sense, given that President Trump’s approval is relatively low but very stable (Chart 26). Chart 26Trump Approval Very Low … And Very Stable We will continue to use approval level in our model to generate updated predictions, given that this measure has the best long-term historical fit with the election outcome. However, given that President Trump is performing relatively well on these other measures of approval, there is upside risk to his 2020 performance. Appendix 2: A Word About The Probit Model Table 4 presents the regression coefficients of our model. Since this is a probit model, the coefficients cannot be directly interpreted as they would in an ordinary regression. The coefficients in a probit regression model measure the change in the Z-score associated to each independent variable for a one-unit change in that variable. Table 4BCA 2020 US Presidential Election Model Statistics The sign of the coefficient corresponds to the direction of change in probability. So increases in the state leading index, presidential approval, or the incumbent’s margin of victory in the last election increase the probability of the incumbent winning a state. Of course, the latter variable is fixed and will not change until the election. At the same time, having occupied the White House for two terms or more decreases the probability of an incumbent win. But this is not the case in the current election. Footnotes 1 Andrew Johnson, the first to be impeached, did not run in 1868; Ulysses Grant bowed out after two terms in 1876, amid the “Great Barbecue” scandal; Warren Harding died before the election of 1924, amid the infamous “Teapot Dome” scandal; Harry Truman stepped down amid scandal after two terms in 1952; Richard Nixon resigned before the election of 1976; Bill Clinton was impeached and hit the two-term limit before the election of 2000. For these examples, and the electoral impact of great scandals in general, please see Allan J. Lichtman, Predicting The Next Presidency: The Keys To The White House 2016 (Rowman and Littlefield, 2016). 2 Trump’s policy record contains one major legislative victory, the Tax Cut and Jobs Act of 2017, along with a number of works in progress. The Republicans’ failed attempt to repeal and replace the Affordable Care Act (Obamacare) exacted an opportunity cost: it deprived Trump and the GOP Congress of time needed to legislate a southern border wall, while mobilizing the opposition for all subsequent elections. As for other policies, the renegotiation of NAFTA is only a partial success as the USMCA has not been ratified. The promised infrastructure package will become a campaign pledge for the second term. We expect some kind of North Korea deal. 3 To this end, we use a probit model, where the dependent variable is stated as 1 = incumbent party won all Electoral College votes in this state, or 0 = incumbent party did not win any Electoral College votes in this state. This model allows us to measure the probability that a state with certain characteristics will fall into one of these two categories. 4 “The leading index for each state predicts the six-month growth rate of the state’s coincident index. In addition to the coincident index, the models include other variables that lead the economy: state-level housing permits (1 to 4 units), state initial unemployment insurance claims, delivery times from the Institute for Supply Management (ISM) manufacturing survey, and the interest rate spread between the 10-year Treasury bond and the 3-month Treasury bill.” See the Federal Reserve Bank of Philadelphia, www.philadelphiafed.org. 5 Alan I. Abramowitz, “Forecasting the 2008 Presidential Election with the Time-for-Change Model,” Political Science and Politics, Vol. 41, No. 4 (Oct., 2008), pp. 691-695. 6 We also assume that the Democrats always win the District of Columbia. 7 Please see Michael S. Lewis-Beck, Charles Tien, “Forecasting presidential elections: When to change the model,” International Journal of Forecasting, Volume 24, Issue 2, April–June 2008, Pages 227-236, and Mark Zandi, Dan White, Bernard Yaros, “2020 Presidential Election Model,” Moody’s Analytics, September 2019.