Our economic outlook remains unchanged from last week. This is technically a neutral development, but can be argued as a positive to the extent that forecasts had previously deteriorated nearly every week. Presenting a similar perspective, a poll of our investment teams finds that the majority have also maintained a steady perspective on COVID-19 over the past week, with a significant minority becoming less concerned. Financial markets must feel similarly, with risk assets having staged a substantial recovery.
Beneath the surface, there have been a mix of positive and negative developments. Most positives revolve around the spread of the disease itself, whereas the negatives are more varied.
- Evidence is mounting that quarantines are working as the new daily cases of COVID-19 diminish in a wide range of countries.
- Exit strategies are being discussed more and more, with Spain even lifting restrictions on certain sectors. Whether significant easing will prove realistic any time soon is an open question, but it is heartening that we can now ponder the timing of the economic upswing.
- The recent OPEC+ agreement is certainly good news for the oil market, though prices should remain very low.
- The handful of countries that are now enjoying a falling number of new daily cases have not managed as rapid a decline as did China and South Korea.
- Testing numbers have increased massively – the U.S. is now conducting roughly 140K tests per day and Canada is managing around 27K. But that is still wildly shy of the ideal of being able to test nearly everyone every week to ensure a confident restart of economic activity.
- We are becoming more nervous about the speed of the economic recovery once quarantines are lifted. It is far from certain that this occurs quickly or smoothly.
Key virus figures
COVID-19 now registers 1.8 million infections globally, but the daily new case count has managed to decline in each of the past two days, to a level of around 72K. The single-day peak was 93K nine days ago. As such, it is possible that the global outbreak is now past its peak, but we hold off the celebration until firmer evidence is at hand, particularly given that poorer parts of the world have made only limited attempts to constrain COVID-19.
Indeed, whereas the epicenter of the disease started in China, then shifted to Europe and has now landed in the U.S., it does not take a great leap of the imagination to think that it will eventually move into the developing world. Whether testing proves sufficient to officially capture this or not is another matter.
Promisingly, the number of global deaths has also declined somewhat in recent days, peaking five days ago. This may also ultimately prove premature, in that the case count itself has not yet peaked with any certainty, and the number of deaths should in theory continue to rise for around two weeks after that event.
One thing is reasonably certain: the global transmission rate is in decline. In mid-March, each sick person could expect to infect three others, whereas this is now down to just a hair above the critical level of one today (see chart). Once it has fallen below one, the virus can be said to be in global decline. Several countries have already achieved this.
Global transmission rate declining, but still above key threshold of one
Note: As of 04/13/2020. Transmission rate calculated as 7-day % change of underlying 5-day moving average of new daily cases. Source: ECDC, Macrobond, RBC GAM
China and South Korea managed to gain control over the virus roughly two months ago, creating a roadmap for other nations. Now, others are following their lead (as demonstrated by transmission rates below 1 in the next chart).
Transmission rate above one suggests continued growth (based on new cases)
Note: As of 04/13/2020. Transmission rate calculated as 7-day % change of underlying 5-day moving average of new daily cases. Source: ECDC, Macrobond, RBC GAM
- Italy’s new cases peaked on March 22.
- Germany peaked on March 28.
- Spain and France peaked on April 1.
- Switzerland has been declining since April 3.
Beyond Europe, Canada has also tentatively staked out a peak on April 3, while the U.S. and U.K. may have peaked on April 11. In the case of these last two, the peak date is still too recent to report with much confidence.
Still, all of this adds to our understanding of the parameters of COVID-19. Based on the Chinese experience, our hypothesis that other countries should enjoy a declining rate of new infections roughly two weeks later has held up quite nicely. Each of those countries peaked between 11 days (Canada) and 20 days (U.S.) after their hardest form of quarantine was installed.
However, the number of new cases post-peak is still not declining as readily outside of Asia as it did in China and South Korea. To illustrate, by three weeks after their peak, China and South Korea had reduced their daily case count to just 10% to 20% of the peak figure. In contrast, Italy is still trundling along at 50% to 70% of its peak number. This is a bit concerning. France has had better success, it should be noted, but still not quite to the Asian standard. While one might pose questions about the accuracy of the Chinese figures given revelations discussed last week, South Korea’s numbers are not considered similarly suspect.
While it is certainly possible to note that not every quarantine imposes identical constraints on a country’s citizens, that observation does not actually – at least based on our initial analysis – help to explain the different peak timings and rates of decline. For instance, a variety of real-time metrics argue that Italy’s quarantine has been much more extreme than Canada’s, and yet Canada enjoyed a faster transition from quarantine to peak. Similarly, France has managed a faster subsequent decline in its rate of new infections relative to Italy, and yet once again Italy has maintained the more aggressive quarantine.
We know that every country has more infected people than the official tallies suggest simply by virtue of the facts that:
We can even guess which countries are undercounting to the greatest extent. These are likely the countries with the fewest tests per million people and/or the highest positive rate among those tested (see chart).
COVID-19 testing varies significantly by country
Note: Most recent available data as of 04/11/2020. Source: Our World in Data, ECDC, RBC GAM
By these measures, it is likely that India, Mexico, Brazil and Japan have an especially large number of uncounted cases as proxied by their low testing rate; while Iran, the U.K., France and the U.S. likely have many uncounted cases given the large proportion of their tests that are coming back positive. We exclude China from this list because it has shared little about the number of tests it has conducted.
Despite all of this, we do not believe the COVID-19 peak dates have been seriously distorted by these imprecisions. So long as a country is not testing a declining fraction of its population over time, the recorded peaks should either be accurate or even slightly lagged. It is simply that the breadth of the illness is likely underreported and the overall fatality rate is likely over-reported.
While our rough quarantine-to-peak math has proven enormously useful, it is worth following what the true experts say on such matters.
The University of Washington has put together a particularly good set of estimates (with the one glaring exception that Canada is excluded from the large list of countries). The forecasts focus on the date of peak deaths as opposed to the date of peak new cases. The results are fairly similar, though logically lagged to reflect that mortality normally occurs a few weeks after the onset of the virus.
- Italian daily deaths should have peaked on March 27 and Spain on April 1, with the U.S. following later with a peak between April 14 and April 18.
- The U.K. is expected to peak even later, between April 17 and 22.
- Canada would presumably be slightly sooner than the U.S. and U.K. given its more favourable peak daily case date.
A glance at the U.S. regional predictions are also informative, logically predicting that the states that imposed quarantines first should similarly enjoy a declining death toll first. New York State should have peaked on April 10 and Washington State on April 12, whereas Texas and Alabama may not peak until April 21 and April 26, respectively.
But let us not pretend that a single set of forecasts has a monopoly on the truth. These numbers make the most sense to us, but several epidemiological models anticipate later peaks and have far higher mortality figures. And, of course, much depends on how enthusiastically governments re-normalize economic conditions and relatedly whether a second peak eventually forms.
While most developed regions have enacted a fairly similar set of quarantine policies, a handful have not. It will be worth watching these countries to understand how the disease and economic factors develop under other protocols.
Sweden has avoided mandatory quarantining, instead encouraging a less extreme form of social distancing. This can be seen in the country’s Google Mobility data, which shows that trips to workplaces and retail outlets are down just 18% and 25%, respectively. In comparison, the U.S. is down 40% and 49%, and Italy’s equivalent figures have fallen 62% and 95%.
Hong Kong is another interesting location. Although it was among the first places to be affected by the virus and has not attracted a great deal of commentary, it hasn’t actually closed down as much as most other markets. Its Google Mobility figures are -27% and -37% for trips to workplaces and retail outlets, respectively.
Japan is another important country to watch, with Google Mobility numbers that are merely down 13% and 25%. It was initially proclaimed as an example to the world as the country managed to fend off its initial outbreak. However, a second outbreak has occurred and Japan is now scrambling to respond. It is particularly vulnerable given its large elderly population.
The University of Washington expects Sweden to be among the worst hit from a death-per-capita perspective. That seems reasonably likely, though Hong Kong’s virus figures haven’t been too bad despite grappling with the disease for several months now.
The disease itself
We have already discussed the transmission rate, which is extremely high under ordinary conditions but declining nicely in countries that have imposed significant social distancing measures.
While the case fatality rate outside of China stands at a worrying 6-7%, recent analysis of data from Gangelt, Germany find that the overall population fatality rate in that region is just 0.37%. Helping to explain this, the researchers estimate that 14% of the region’s population has already been infected. An earlier Icelandic study arrived at a similar conclusion, though its methodology has been criticized. For the moment, we continue to use a 0.5%-1.0% fatality rate assumption in our own work. This leaves COVID-19 notably worse than the flu (0.1%) but less fatal than most major pandemics.
A smattering of research continues to argue that warmer weather may reduce the spread of COVID-19. Of course, that simultaneously increases the risk of a second wave next fall.
Efforts to quash the virus via medical innovations continue. There are now at least 43 vaccines under development, with three already in Stage 3 trials. Computational chemists have also uncovered some 80 molecules that have enjoyed tentative success limiting the spread of COVID-19, and are evaluating 3,500 more promising candidates. A host of existing medications intended to address other problems are also being tested on the disease. The rapid development of effective therapeutics or a vaccine remains the best-case scenario. While some prospective therapeutics are already being prescribed, experts remain firm that a true vaccine is unlikely in less than 12 to 18 months.
Less death from other things
It is important to highlight silver linings where they can be found. One is surely that there should be fewer deaths from other ailments during this quarantine.
Contrary to popular belief, the death rate actually tends to decline during recessions. This occurred during the Great Depression relative to the 1920s boom, during the Global Financial Crisis and even during the recent period of ultra-high unemployment in Spain and Greece.
Why do deaths decline during such trying times? For a variety of reasons, including lower pollution, fewer on-the-job accidents, and because car accidents diminish due to less traffic.
One can imagine those fortuitous forces proving even more potent during this quarantine. Emissions are declining massively, commuting has all but disappeared and even the crime rate has reportedly fallen. Simultaneously, the human distancing component of this recession means that deaths from communicable diseases should also decline. Recall that tens of thousands of Americans normally die from the flu each year.
There are clearly partial counterpoints to all of this good news in that doctors are delaying some medical procedures, stress levels are surely higher, depression and loneliness are inevitably up, and poverty is certainly rising (and a predictor of earlier death).
Each week brings new insight into how the economy is being damaged by COVID-19.
We have continued to enhance our suite of real-time indicators, now numbering more than 30 measures. Among the prominent findings:
- Google Mobility trends reveal U.S. retail and recreation trips down 49% versus normal, visits to transit stations down 54% and workplace visits down 40%. In comparison, Canada’s numbers have fallen somewhat more sharply (-63% / -67% / -46%), while the Italian decline is even more severe.
- Measures of U.S. online restaurant bookings, movie ticket sales and air travel are down 100%, 100% and 96%, respectively.
- U.S. steel production is down nearly 20%.
- U.S. mortgage applications have declined by around 30%.
- Somewhat ominously – given that other measures point to an enthusiastic rebound – the Financial Times’ Chinese economic activity tracker argues that economic output there is still around 25% below normal. We suspect it undersells the Chinese recovery.
- U.S. initial jobless claims again reported another six million newly unemployed in the latest week – worth a nearly 4 percentage point increase in the unemployment rate and arguing that the true unemployment rate may now be in the realm of 14%. This is beginning to challenge our medium depth/medium elasticity forecast of a 15.5% unemployment rate peak (though let the record show we have other, more adverse scenarios). Still, there was also a sprinkling of good news in that the number of new jobless claims was actually slightly lower than the prior week – this makes sense as the economic shock is extremely front-loaded. Future weeks of jobless claims should continue to decline from here.
- A Peterson poll finds that 73% of Americans say that COVID-19 has reduced their family income, with 24% hit very significantly. We are inclined to imagine the smaller figure as representing those who have lost their job or suffered a significant diminishment of hours (and keeping in mind that a family would presumably have responded in this way if just one source of income had been affected in a two-income family). The remaining part of the larger figure might include those whose retirement savings are affected.
- As of early April, only around 70% of rent had been paid for the month in Canada and the U.S. But this is not as bad as it sounds, as in the U.S. normally only 81% of rent would have been paid by that point. Thus, rent collection is running around 15% below normal.
- U.S. bank loans to commercial and industrial enterprises are now up 17% since the beginning of March – a happy trend and one that confirms the credit channel is indeed conveying government stimulus.
Turning to more conventional economic figures, Canada had its first run-in with (nearly) the full brunt of the COVID-19 shock. The country’s March job numbers were abysmal, with 1.01 million jobs shed in a single month. That is easily the all-time record for a month, and unwinds 40 months of net employment gains. What’s more, the April job numbers should be worse given what has been reported about jobless claims in Canada. Statistics Canada further estimated that were one to include absences from work due to business conditions and those who suffered a reduction in working hours, the true figure could be closer to 3.1 million Canadian affected.
The burning question is how did Canada lose significantly more jobs than the U.S. in March (1.01 million versus 701K) despite roughly one-tenth the population? The answer comes in three parts:
- We suspect the Canadian economy will indeed be hit somewhat worse than the U.S. This suspicion is rooted in Canada’s large and beleaguered energy sector, in the fact that Canada’s sector mix leaves it more exposed than the U.S. to the ravages of quarantining, and in the fact that the Google Mobility data argues Canada’s quarantine is more severe than the U.S. But this hardly justifies a 13-to-1 job loss mismatch on a per capita basis.
- Crucially, the U.S. reference week for its jobs report was March 8-14, whereas for Canada it was March 16-22. It just so happens that the middle of March was when COVID-19 morphed from a source of anxiety to a job-destroying monster. Canada’s numbers caught a significant part of the blow, whereas the U.S. figures did not.
- Also important is that Canada uses a household survey which asks if people were working that very week. Conversely, the U.S. focuses on its payrolls survey that simply captures how many workers received their pay packet in the latest week. For someone paid bi-weekly, they might not have worked since the beginning of March and yet would have registered as employed in the U.S. To illustrate the magnitude of the payroll versus household survey effect, whereas U.S. payroll employment fell 701K, the less-examined U.S. household employment number was -3.0M (and this was still captured a week earlier than in Canada).
A final thought on the economic data: as we flagged last week, the numbers are becoming not just more extreme, but ever-less trustworthy. Businesses and households have more important things to attend to than answering surveys. Demonstrating this, the U.S. household employment survey response rate was just 73%. That is 10 percentage points lower than normal, and the next month will probably see a significant further drop.
The March consensus growth outlook has now been published, and forecasts have unsurprisingly been slashed. The median forecaster now looks for a 4% decline in both U.S. and Canadian 2020 GDP. Let the record show that our medium depth/medium trough forecast remains somewhat worse than this, at -7.7% for the U.S. and moderately worse for Canada.
Of course, there is still substantial uncertainty around these numbers, which is why we maintain a full set of nine scenarios that range from the optimistic (-2.0% U.S. growth in 2020) to the extremely pessimistic (-30.0%). Refer to the following table.
There are arguably five key assumptions in any post-COVID-19 growth forecast, as depicted in this diagram and discussed next.
Five key economic questions
Source: RBC GAM
- Initial bounce:
- GDP to prior peak:
- GDP to prior trajectory:
Versus other crises
Having just cited the Great Depression and the Global Financial Crisis, it is important to reiterate that COVID-19 is neither of those things. The following table highlights the differences. The COVID-19 shock is likely to be milder in every way than the Great Depression. While COVID-19 should suffer a sharper quarterly and annual decline in output than the Global Financial Crisis, we still suspect the cumulative foregone output from the Global Financial Crisis may prove worse due to the lingering nature of the event.
We maintain unchanged views on the peak-to-trough decline of different economic sectors (see chart). Such sectors as entertainment & tourism, real estate and retail should experience the largest hits.
Medium scenario: U.S. output levels during and post-COVID-19 outbreak
Note: As of April 2020. Bars show the peak and trough of fraction of normal output for each U.S. sector in 2020 and 2021 in a medium depth and medium duration scenario. Pre-COVID output level = 1. Source: Haver Analytics, RBC GAM
Interestingly, U.S. job losses in March were skewed roughly 10-to-1 toward entertainment & tourism as opposed to retail. However, we suspect retail may make up for lost time in future months, as it was already clear that people were cancelling vacations over the first half of March, whereas the shutdown of retail establishments did not happen until later.
While many sectors are constrained by demand – people can’t go to stores or restaurants, which limits the demand for such products – others are being hit on the supply side. For instance, in Canada the agricultural sector relies on around 60,000 temporary foreign workers to harvest crops. After considerable debate, these workers will still be allowed into the country, but it is unlikely to be seamless: some may be unwilling to travel during a pandemic and all will be forced into two weeks of self-isolation upon arriving.
The oil market and the oil industry has been among the most adversely affected in recent weeks, not just by plummeting demand for oil as transportation needs dry up, but by an oil supply glut induced by a battle between Saudi Arabia and Russia.
The good news is that the price war is now over. Saudi Arabia, Russia and indeed OPEC+ have agreed to an unprecedented quota reduction of 9.7M barrels per day. The U.S. helped to broker the deal. There is furthermore hope that other non-OPEC+ nations could contribute a further 5M barrel reduction over time.
However, caution is appropriate on several fronts:
- The 10-15M barrel reduction in supply doesn’t even go half way toward addressing what is thought to be a 30M+ barrel reduction in demand. The oil market is still in incredible imbalance.
- It is unclear whether the rest of the world will voluntarily cut oil production. Possibly more likely is that prices will have to remain sufficiently low to drive excess production from the market via natural forces. Those could arise sooner rather than later, as there is now very little remaining space to store the excess oil that has already been pumped.
- For the moment, the most stringent part of the cut only extends through June, though one imagines there is scope to extend the quota cuts should economies not recover by then.
In the end, the OPEC+ announcement is a positive development. However, oil prices have not budged from earlier lows and the sector is likely still in store for an extremely challenging multi-year period given the remaining supply-demand mismatch and the oil inventory glut that is building.
A word on inflation
A quick word on inflation: real-time daily measures of inflation are indeed going down in major markets. There had been some initial debate given supply chain issues and worries about quantitative easing, but there are far more fundamental forces pushing downward than upward right now, including lower oil prices and declining economic demand.
We continue to work with the assumption that total CPI (Consumer Price Index) in North America will experience negative annual prints while core CPI will settle down into the 0.5% to 1.0% range.
Further to our observation that economic data will become less trustworthy, recall that the CPI examines a fixed basket of goods and services. Its inflexibility may cause it to underestimate the true rate of inflation by picking up very low prices on products that no one can buy during a quarantine and failing to allocate a large enough weight to the things that everyone is now purchasing (and that are accordingly becoming more expensive).
The latest stimulus
Fiscal and monetary stimulus remains astonishingly large, well-targeted and rapidly delivered.
Committed fiscal stimulus is already reaching 10% of GDP in some major markets, and it would not be a surprise if it ended up closer to 20% over the next several years. Keep in mind that the fiscal boost arises not merely from sparkling new government promises but also from automatic stabilizers that naturally kick in when people lose their jobs and gain access to pre-existing government programs.
Furthermore, we continue to believe – by virtue of the speed and reasonably good aim of the new fiscal programs – that the fiscal multipliers will prove large, likely well in excess of one. In non-technical language, the money is being well allocated, and likely to benefit the economy by more than the actual amount of money being spent. Of course, we will never be able to directly observe the effect because it is difficult to say how many businesses and households would have gone bankrupt without the assistance.
To be sure, the stimulus is far from perfect. Application processes can be challenging, the delivery of aid is not as rapid as would be ideal, some people and businesses fall through the cracks (such as the self-employed in many countries) and others receive money who would have been fine without it (such as via universal transfers in the U.S.).
On the monetary side, the pace of quantitative easing (QE) in the U.S. is already blowing away prior records. Now, the Fed has supersized its actions yet again, committing to go beyond the relative safety of sovereign debt, government-backed mortgages and investment-grade corporate debt. It will now purchase AAA-rated commercial mortgage-backed securities, AAA collateralized loan obligations and fallen-angel investment grade debt, among others. The Fed has also clarified its commitment to smaller businesses, providing $600B in loans to small and medium-sized businesses. It will also buy up to $500B of short-term debt from states, counties and cities.
Two concerns immediately spring to mind when one hears that the Fed is moving so aggressively:
- Might all of this money printing and bond buying stir inflation? It is not impossible, but note that even if the Fed balance sheet doubles over the next year to $9 trillion – not an unreasonable expectation given all that it has pledged – that still only puts the U.S. monetary base in line with the Eurozone, where inflation remains hard to see. Canada’s QE is comparatively much punier.
- Has the story fundamentally changed now that the Fed is buying risky products on which it might suffer a significant loss?
- The answer is “yes” in that one can criticize the Fed for bailing out investors who knowingly invested in a risky corner of the market. However, the Fed is dead-set on ensuring that financial markets remain liquid and that corporate bankruptcies are kept to a minimum, and so is willing to unintentionally reward some investors if that is unavoidable.
- But the risk of inflation is arguably no higher than before. While any investment losses by the Fed can never be unwound (whereas liquidity injections reverse automatically and standard QE can be unwound manually), this need not mean that the seeds of unavoidable inflation have been sowed in the U.S. This is because – in something akin to the very opposite of fears about monetary financing – the U.S. Treasury has committed $75B to cover any losses incurred by the Fed in its acquisition of riskier securities. The losses need not be added to the monetary base.
Financial markets have now significantly reversed course over the past few weeks, nearly halving the extent of the stock market’s global decline. We remain underweight fixed income and overweight equities in our recommended asset mix for global, balanced investors.
Our primary worry remains the magnitude of the economic and earnings hit. While this should ultimately only be temporary, some damage should linger for multiple years as discussed earlier.
Nevertheless, positive signs abound. Not only is the virus itself beginning to come under control, but many investment-grade companies have enjoyed success issuing corporate debt at rates not unfamiliar before COVID-19 arrived.
Furthermore, corporate executives in March bought the most of their own company’s stock in a single month in seven years. It is a positive sign that “insiders” feel their own companies are on sale. Technical signals are also fairly positive.
Stepping beyond the short-term perspective, the risk premium between stocks and government bonds remains unusually large, if less extreme than several weeks ago. Still, stocks can reasonably be expected to beat bonds to an unusual extent over the long run from this starting point.
More long-term musings
Each of the past several weeks, we have tacked on yet more musings about the long-term repercussions of COVID-19. Refer to prior #MacroMemos for discussions about the prospect of more debt, ultra-low interest rates, working from home, and more.
We begin this week’s edition with a familiar reminder that it is far from certain that the world will actually change in a permanent way. It’s human nature to revert back to familiar habits. Historical records from the Spanish Flu increasingly emphasize just how similar the initial fears and economic hit was, and yet people ultimately returned to restaurants and theatres.
Nevertheless, we present three further long-term issues this week.
- Does the enormous expansion of government influence persist after COVID-19 is done? While small elements could endure, we are dubious. The programs introduced are inherently temporary and will naturally expire; furthermore, eligibility for such programs will diminish as people regain their jobs and businesses cease to desire additional loans. Even if the programs were tweaked in a way that made them permanent, keep in mind that society has only acceded to half of what would need to be done to keep government services at an elevated level. The other half is to find the necessary revenue to pay for such services. That would require massive tax hikes that are unlikely to prove popular. Recall also that there is precedent for a temporarily large government shrinking back down to normal – during World War II, governments were involved in every aspect of economic activity and yet shrank back to size thereafter.
- Inequality seems likely to rise as a result of COVID-19. Job losses nearly by definition widen the gap between haves and have nots, and these particular job losses are not random. To the contrary, the sectors most affected are those such as tourism and retail sales that tend to employ lower-wage workers. Thus, it is the lowest-paid workers who are more likely to lose their jobs and the highest-paid that are more likely to keep them. Furthermore, to the extent that this experiment with remote working encourages a further structural push toward automation and online shopping, some of those lower-skilled jobs could be lost forever. Additionally, poor children are less likely to have access to the necessary online resources to further their education while in quarantine, and perhaps also less likely to have a parent qualified to assist them in that pursuit.
- Some have expressed concern about Canadian national unity during this time of great economic stress. This is not an unreasonable worry, but ultimately it seems more likely that the country will be pulled together by the experience as opposed to torn apart. The bulk of fiscal stimulus is coming at the federal level, and those with the greatest need are arguably those that have historically had the greatest grievances – Alberta (which has been hit hard by the oil shock) and Quebec (which has experienced the greatest number of Canadian COVID-19 cases). For that matter, there is nothing like a common enemy to unite people, and reflecting that, incumbent politicians almost everywhere are becoming more popular.
-With contributions from Vivien Lee and Graeme Saunders