It feels as though the COVID-19 narrative has now slowed to a more normal clip after the chaotic sprint that characterized the middle of March through the middle of April.
Positive developments include:
- The number of new infections continues to slow.
- Further evidence argues for a low fatality rate.
- Four-plus weeks of quarantine are now complete, meaning significant progress is being made toward the finish line.
- Government efforts to control the virus and limit the economic damage remain enormous and well-targeted.
- Focus rightly continues to shift away from how to control the virus toward how to restart the economy.
- Financial markets have functioned well and have lately proven buoyant.
- Net-Present-Value calculations argue that risk assets aren’t necessarily too optimistic (though there are many other ways of gauging the appropriate market response, with differing conclusions).
There are also a handful of negative developments:
- The short-term economic damage is set to be enormous. Our forecasts already recognize this, but many 3rd-party forecasts continue to be revised downward.
- The genie isn’t going back into the bottle. Even China’s extreme quarantine has failed to snuff out the virus altogether. Instead, the goal must be to keep the rate of infection fairly low until herd immunity is achieved or a vaccine is developed.
- In turn, the economic recovery is likely to be sluggish for a mix of both artificial reasons (governments will only be able to partially restart economies) and natural reasons (skittish post-quarantine demand).
On the aggregate, we feel slightly better about the COVID-19 situation than a week ago, but the prospect of a sluggish recovery is both serious and unwelcome.
COVID-19 has now officially infected nearly 2.4 million people globally, with almost 165,000 deaths. Fortunately, the rate of new infection and the incidence of additional deaths appear to be stabilizing or even slowing (see next two charts). The virus is now infecting just over 70,000 additional people worldwide per day.
Spread of COVID-19 globally
Note: As of 04/20/2020. Spike on 02/13/2020 due to methodology change. Source: ECDC, Macrobond, RBC GAM
Note: As of 04/20/2020. Spike on 02/13/2020 due to methodology change. Source: ECDC, Macrobond, RBC GAM
Naturally, poorer countries are still not testing in any great numbers and so certainly undercount the disease. Still, the global figures are not a total fiction: the rate of transmission has now fallen below one in the bulk of major developed nations where testing is more extensive (next chart). Figures below one mean that the number of new cases is declining over time.
Transmission rate above one suggests continued growth (based on new cases)
Note: As of 04/20/2020. Transmission rate calculated as 7-day % change of underlying 5-day moving average of new daily cases. Source: ECDC, Macrobond, RBC GAM
The U.S. peak was tentatively on April 11, with an undulating but ultimately declining trend since (see next chart). Less testing was reportedly conducted over Easter, resulting in an artificial subsequent lull and then bounce.
Spread of COVID-19 in the U.S.
Note: As of 04/20/2020. Source: ECDC, Macrobond, RBC GAM
The Canadian situation is a bit messier, as April 17 and 18 officially registered new highs (next chart). We are still inclined to call April 3 the true peak as the 17th and 18th were known to have captured the results of an unusually large number of new tests (around 26K per day relative to the 15K per day norm beforehand). Delayed Easter testing may also have bled across the week. For the moment, Canada’s transmission rate is now slightly back above the critical one threshold. Additional information is needed, but even if the April 18 high is genuine, the subsequent two days have been lower.
Spread of COVID-19 in Canada
Note: As of 04/20/2020. Source: ECDC, Macrobond, RBC GAM
In general, while many countries have now tentatively experienced a peak in their daily cases, the rate of subsequent improvement is still not as impressive as that of China. After international criticism, China recently revised its estimated number of COVID-19-related Wuhan deaths upward by roughly 50% after an official review. But the total number of infections barely rose and accusations of undercounting remain.
Regardless of whether the Chinese downward slope represents a realistic aspiration for others, it is important to highlight that not all quarantines are alike (see next two charts). Google mobility data shows that Spain and Italy have been more aggressive than others. The U.K. has been more extreme than Canada, and Canada, in turn, has had a more aggressive quarantine than the U.S. Sweden trails behind with its experiment in social distancing without explicit quarantining. Fascinatingly, South Korea shows the smallest decline in visits to stores and workplaces despite having managed among the best COVID-19 outcomes. We ascribe this disparity to its early response to the virus, paired with aggressive testing, tracing and mask-wearing.
Google mobility: Retail and recreation
Note: As of 04/11/2020. Source: Google, RBC GAM
Google mobility: Workplace
Note: As of 04/11/2020. Source: Google, RBC GAM
A growing number of emerging-market countries are now imposing quarantines of their own, with India’s particularly notable (see the country in the charts, above). However, as a general rule, it will be harder for these countries to halt the progression of the disease given their more limited room for fiscal support, lesser testing and medical care capacity, and larger fraction of subsistence workers. Then again, their populations tend to be younger and their climates tend to be warmer – both helpful things.
Although the raw global case fatality rate is a distressing 17%, it is likely that the true fatality rate will ultimately be substantially less.
We continue to assume a much lower 0.5—1.0% rate given the belief that the true number of infected is being substantially undercounted. Indeed, a variety of one-off studies have lent support to this view.
On the heels of earlier research from Germany, Italy and Iceland, there is a new study from the Netherlands that finds that 3% of Dutch blood donors already have COVID-19 antibodies. Extrapolated to the national population, this suggests 520K total cases – 17 times more than the official count of 30.5K at the time of study. Furthermore, it argues that the true Dutch fatality rate may be just 0.67%.
For context, 21% of Netherlands COVID-19 tests have been positive, suggesting a similar amount of undercounting in other countries with a similar positive rate, such as the U.S. and U.K. While the numbers still leave countries a significant distance from herd immunity, this might be achieved more quickly than conventional estimates suggest.
Although the debate remains furious around the precise extent of asymptomatic and pre-symptomatic cases, our read is that there appear to be very many. Additionally, a recent Nature study estimates that 44% of infections happen during the pre-symptomatic stage. No wonder COVID-19 is so hard to contain.
The focus for policymakers has shifted over the past few weeks, away from how to control the spread of the virus, and toward how to re-open economies. Naturally, the two issues are intertwined, with re-opening only possible once the virus has been substantially tamed, and when enhanced virus control measures can be introduced.
It no longer seems realistic to aim for the complete elimination of the virus via quarantine. No country has achieved that goal, including China. Instead, the goal of restarting the economy is to minimize the economic damage in a way that tolerates an ongoing steady trickle of new cases, with an end goal of a) achieving herd immunity or b) waiting until a vaccine is developed and mass-produced.
If new infections are to be kept to a small number, it is either the case that the economy can only be partially reopened or that aggressive tracking and tracing of an intensity that is not yet logistically possible will be necessary to limit the disease. While South Korea has managed to keep its economy largely open via incredible testing and tracing efforts, it never had as many cases to track as much of the developed world. As such, initial efforts to restart the economy cannot hope for a full return to normality.
Fortunately, Google and Apple are now working together on a Bluetooth technology that would allow for real-time tracking of contact between people, though one worries about the precision of the technology given the crucial importance of differentiating between strangers passing one another at a distance of 2 metres versus 1 metre. Imprecision here would potentially result in hundreds of phantom contacts, each requiring two weeks of quarantine. Another proposal involves testing 7% of the population each day – a sufficient fraction to control the spread of the disease. Some propose testing people for temperatures before they are permitted into stores, though this would fail to pick up many with COVID-19.
Ironically, the countries with the most aggressive quarantines will take the longest to achieve herd immunity, meaning that their economies might be limited for the longest period of time. On the other hand, they should suffer the fewest deaths and should ultimate resolution come via the early arrival of a vaccine rather than herd immunity, this approach may be vindicated. There is now apparently the possibility of a vaccine by this fall, rather than the usual 12 to 18 months.
In the meantime, successful management of the virus could be akin to surfing a wave, repeatedly tweaking the amount of social distancing to keep the economy functioning and progressing toward herd immunity, without overloading the health care system.
Many jurisdictions and entities have now presented loose exit strategy frameworks. Some are even beginning to reopen. Spain has allowed its construction and manufacturing sectors to restart. Denmark is re-opening schools. Germany is permitting smaller shops to open this week, with a further step on May 3. France is looking toward May 11. Austria, the Czech Republic, Norway, Australia and New Zealand are also tentatively reducing the severity of their quarantines.
Among the common components of these plans is that the most vulnerable groups should remain isolated for longer. Further, some forms of social distancing should remain, including limits on large crowds, the use of facemasks and the continued practice of quarantining infected or exposed people. Illustrating that China continues with social distancing a full 2.5 months after the virus’ peak there, Beijing subway ridership is still 65% below normal.
Re-opening international borders does not appear to be a priority. On the other hand, it is obviously key that spare medical capacity exist before restarting economies.
From an economic standpoint, the plans use some combination of the following criteria to determine which segments of the economy to re-open first:
- Work that can be done from home (though one would imagine most of this is already happening)
- Jobs with low interpersonal contact
- Sectors with a high economic importance
- Sectors that are essential to allowing a broader restart of the economy
- Schools, both to ensure the acquisition of human capital and permit parents to resume work
- Social activities, including going to restaurants and presumably recreational sports (must wait until later)
- Mass gatherings (must wait until herd immunity or a vaccine are achieved).
Emphasizing just how incremental this process is likely to be, the Eurozone has even talked about waiting a month between each major step to see what the effect is on the spread of the virus.
A further question mark that could inform whether social distancing and economic constraints persist for weeks, months, or even years is the extent to which exposure to the virus imparts immunity, and if so, whether this is temporary or enduring. One Chinese study found that 70% of those that had been infected had achieved immunity to a further bout, whereas the other 30% had not. That means a vaccine could prove necessary to truly quell the disease. Fortunately, there are now over 100 potential vaccines under testing.
The government stimulus delivered so far is enormous, far higher than that delivered in response to any other modern crisis. This is true both of fiscal and monetary stimulus. Here are several fresh observations.
New stimulus continues to be delivered:
The Eurogroup finally approved 540 billion euros of fiscal stimulus in recent weeks, with this sum coming on top of national-level fiscal programs.
Japan had already announced a large fiscal package, but has now pivoted toward a less targeted approach involving lump sum cheques to all Japanese citizens. While more targeted measures would be theoretically superior, they are proving difficult to deliver in some countries.
In Canada, the Bank of Canada has now upgraded its quantitative easing efforts to purchase Canadian provincial and corporate debt.
Existing programs will grow:
Prominently, the U.S. small business fund has already committed all of its $349 billion allotment, with expectations now that further funding will be secured.
We suspect many existing programs will be expanded or extended over time.
As stimulus starts to roll out the door, gaps in support are being identified. For instance, in Canada, contract workers and the self-employed whose incomes have plummeted but not quite fallen to zero had previously been ineligible for unemployment insurance payments and other special government support. The federal government will now permit up to $1,000 per month in income while still qualifying for such funds.
Pivoting toward the medium term:
Some countries have now mused about infrastructure investment as a means of filling economic holes. Economists have mixed feelings on the subject. To be sure, infrastructure spending tends to generate an unusually high return on investment, making it highly attractive under normal circumstances.
However, infrastructure spending is arguably not the right solution for COVID-19 in 2020. Infrastructure funds are rarely deployed in real time – it takes quarters or even years to get projects properly started. Furthermore, it is often the case that national infrastructure spending simply displaces pre-existing state or provincial level plans, rendering the return on investment nil. Finally, the COVID-19 shock has not physically damaged infrastructure, unlike such examples as the Japanese earthquake/tsunami of 2011, when infrastructure spending was put to excellent use.
This is not to say that infrastructure spending has no role. To the extent the fiscal stimulus is front-loaded but the economic recovery may prove sluggish, additional policy support will likely be needed over the latter half of 2021 and beyond. Perhaps there is room for infrastructure spending over that timeframe, but it isn’t a “today” issue.
The economic data for March and even April is now flowing, giving insight into COVID-19.
New York State’s Empire Manufacturing Index has plummeted to -78 in April, much worse than the -21 reading in March and the 0 level that indicates flat output. It is now more than twice as negative as the worst month of the global financial crisis, which shouldn’t come as a surprise given that we forecast the peak-to-trough decline in economic output should be five times greater.
The qualitative U.S. Beige Book has now been released, covering March and early April, and unsurprisingly there are big drops reported everywhere. From a sector perspective, leisure and hospitality is worst, with non-essential retail activity also radically reduced. Loan demand is high – we see this reflected in bank lending figures. Alas, businesses report plans to cut further jobs, so the fact that job losses are front-loaded does not mean there won’t be some losses later.
On the subject of employment, U.S. weekly jobless claims have again been released, this time recording a further 5.2 million Americans applying for benefits. While this now makes a startling 22 million newly unemployed Americans over a mere four weeks, we take some solace that the number of new claims has declined in each of the past two weeks, from 6.9M to 6.6M to 5.2M. This ameliorating trend should continue, but not to the point of halting the rising unemployment rate. A further silver lining about the U.S. job market is that a recent survey argues that around 70% of layoffs are currently viewed as temporary by employers, whereas the equivalent figure was less than 1% during the global financial crisis. The employment rebound should therefore be much faster.
U.S. consumer confidence numbers have now fallen significantly, from roughly 100 before COVID-19 to just 71 in April. Happily, this remains better than during the global financial crisis. Conversely, retail sales – theoretically, the manifestation of that confidence – fell by 8.7% in March. This was twice as bad as the worst monthly decline during the global financial crisis, and it significantly understates the true peak-to-trough decline given that the quarantine only began toward the end of that month. In fact, weekly Redbook sales are now down around 27%. How to reconcile confidence with spending? Consumer confidence may reflect desired spending whereas retail sales represent actual spending. The gap between the two results from quarantine limitations. This argues for a rebound in spending post-quarantine.
China’s early encounter with COVID-19 continues to provide useful insights into how economic conditions may evolve in other markets.
To the extent China’s economic decline was entirely contained within the first quarter, the recent release of the country’s Q1 GDP report is enormously informative. The economy shrank by a large 9.8% from the prior quarter on a non-annualized basis. Keeping in mind that China would normally have grown by around 1.5%, that represents a big 11.3ppt miss.
We are dubious these figures capture the full extent of the peak-to-trough economic decline given that the Chinese economy was functioning fairly normally over the first few weeks of January and was then rebounding enthusiastically over the final month of the quarter. Inserting a variety of reasonable assumptions into our week-by-week economic model yields a possible peak-to-trough decline of 25% in China. This is actually a fairly happy conclusion, as we had feared that the Chinese economy might have declined by 30% or worse given the severity of its quarantining and tentative real-time data. Of course, the usual caveats about Chinese data quality apply.
Recent Chinese data also says something about the country’s nascent rebound. March industrial production – the country’s rebound month – was only down 1.1% on a year-over-year basis. Granted, it would normally be up around 5%, but this still represents considerable progress relative to the 13.5% decline recorded in January-February.
On a more cautious note, Chinese retail sales were down 16% YoY in March, only a slight improvement from the -20% in January-February. It would appear that China’s economic supply is reviving nicely, but demand is proving somewhat more hesitant.
For a third consecutive week – a new record! – we maintain the same base economic forecast of a 7.7% decline in U.S. 2020 real GDP. Beneath the surface, this represents our medium/medium scenario, constituting a 20% peak-to-trough economic decline that then lasts 12 weeks.
Our international forecasting has now been updated, based on four key inputs:
- Sector weights: Taking pre-existing forecasts for each economic sector, we use the varying sector weights for each country to construct different economic forecasts. Some countries, such as Italy, are unusually reliant on tourism-related sectors and so suffer a worse hit. Others, such as the U.S., are more tech oriented and so can continue unimpeded.
- Quarantine strength: We then use Google Mobility data to estimate which countries have imposed the most aggressive quarantines, with the implication that their economies will be depressed to a greater extent. Italy is among the most aggressive, whereas the U.S. and Japan have lighter quarantines that will do less damage.
- Economic scorecard: We use a scoring system to evaluate a number of other relevant national attributes, including:
- Extent of the virus’ spread in each country
- Testing efforts
- Strength of the health care system
- Elderly share of the population
- Public debt profile
- Reliance on trade and immigration
- Flexibility of the labour market (a negative thing since it permits more job losses).
The combination of these three factors are conveyed in the following chart.
COVID-19 impact on GDP is substantial
As of 04/17/2020. Note: Decline in output level estimated based on factors including transmission rate, testing, containment, health care system, education, demographics, globalization, labour market, public debt, monetary and fiscal response, sector-level impact, and Google Community Mobility data. Source: CIA, Google COVID-10 Community Mobility Report, kita.org, Our World in Data, national government websites, Knoema, Haver Analytics, Macrobond, RBC GAM
- Economic stimulus: Fiscal and monetary stimulus is unlikely to substantially limit the peak-to-trough decline in output since no amount of money will get people spending if the stores are closed. However, it is entirely relevant to the subsequent economic trajectory, and so interacts significantly with overall 2020 and 2021 growth. We have manually tempered by half the expected stimulus differences by country on the view that policy measures will likely converge with time. All of this work yields the forecasts conveyed in the following table.
Global growth forecast 2020
Medium depth / medium length scenario
Global inflation forecast 2020
As of April 2020. Source: RBC GAM
To summarize, the U.S. economy does the best of the developed world by virtue of its superior sector mix, lighter quarantine and aggressive stimulus. Japan also does well, mainly because of its extremely light quarantine. Conversely, the Eurozone and U.K. are hit worst of all due to their much stronger quarantining. Altogether, global growth declines by 4.9% in 2020.
From an inflation standpoint, our rough modelling continues to argue that headline inflation should undershoot the prior trend by around 2ppt (yielding the figures in the second table, above).
Canada’s forecasted peak-to-trough decline of 23% and a 2020 GDP forecast of -10.3% growth are substantially worse than the U.S. This is for good reason, as Canada has enacted more aggressive quarantining than the U.S. and has a less helpful sector mix. Embedded within this sector mix is Canada’s outsized oil sector pain.
We can further support this forecast via four additional pieces of evidence that confirm the weighty hit to the Canadian economy.
- The Bank of Canada’s latest rate decision lacked a forecast but acknowledged the potential for a 15—30% peak-to-trough decline in Canadian economic output. Our assumption is precisely in the middle of this range.
- Statistics Canada has just published a flash estimate for March GDP. This is something new that they have cobbled together far more quickly than normal. The agency estimates that Canadian GDP declined by 9% in March from February. But given that the sharp economic decline only began in the middle of the month, we can extrapolate that this could well mean more like an 18% decline in output over the second half of the month.
- Canada’s monthly job numbers have now been out for several weeks, with a stark 5% decline in employment. But what we had initially missed was the simultaneous massive 15% drop in hours worked. This can be interpreted in different ways. Optimists will note that this means companies found a way to handle two-thirds of the 15% decline in required labour without resorting to layoffs. However, pessimists will note that this means that a much larger fraction of Canadians are suffering reduced incomes than the raw job numbers suggested. More to the point, the very fact that hours worked declined by 15% in March despite the shock only arriving in the middle of the month (plus given that some companies are almost certainly hoarding unneeded labour) argues that the true decline in economic output could well be on the order of 30% or worse. Furthermore, the tentative jobless claims figures in Canada have been reliably worse than in the U.S. on a population-adjusted basis.
- Canadian oil production is already estimated to be down 325K barrels per day, with an eventual 1M barrel decline considered likely – around 25% of production. The hit to cap ex will almost certainly be greater. Given the country’s relatively high cost of oil extraction, it stands to reason that Canada’s oil sector will be hurt worse than the average country, and yet our sector-by-sector analysis merely inserted the same mining sector assumption as everywhere else in the world.
Naturally, the economic outlook is subject to massive uncertainty – a reason why we have generated nine different scenarios for the U.S. We have not fully replicated that effort elsewhere, but the sentiment still applies – these numbers could be substantially too optimistic or too pessimistic, but they constitute a best guess for highly uncertain times.
We have also generated approximate provincial GDP forecasts in much the same manner as the international forecasts were constructed, acknowledging different sector weights by province, different severity of quarantining by province, and other idiosyncratic differences relating to such matters as the spread of COVID-19, reliance on migration and provincial demographics. The peak-to-trough results are in the following chart.
2020 GDP peak-to-trough decline by province
Note: As of 04/15/2020. Decline estimates based on sector exposure/impact & quarantine effectiveness. Medium scenario. Source: Statistics Canada, RBC GAM
By way of brief discussion, it comes as an initial surprise that our methodology reveals Quebec to be the most negatively affected province. Most money would presumably have been bet on Alberta or Newfoundland given their immense oil sector challenges. These numbers hardly constitute the final word on the matter, and let us not forget that the Alberta economy was already unusually weak when COVID-19 struck, making any further decline particularly uncomfortable. In addition, do not forget that real GDP only cares about the number of things being produced – the fact that the price of oil is also down is not a directly relevant factor when forecasting GDP, even though it matters enormously for government, household and corporate income.
In the end, the Quebec hit was larger than the rest primarily because the Google Trends data reveals that the province has engineered a much more aggressive quarantine than the rest of the country.
Financial markets have surprised many with their resilience in recent weeks despite the worst peak-to-trough decline in economic output in 90 years. The S&P 500 has reclaimed half of its initial losses to sit just 17% below its prior peak.
One institutional survey reveals that over three quarters of market participants think it is more likely that equities will re-test earlier lows than that they will next claim new highs. Of course, the very fact that so many investors feel that way and are presumably positioned thusly could actually limit the ability for markets to decline from here since all of that pessimism is already embedded in the price.
Other important reasons for the relative resilience of risk assets include:
- The artificial and temporary nature of the recession
- Its roots in human health as opposed to financial market excesses
- The herculean effort by governments to minimize bankruptcies and defaults
- The substantial assistance being offered by central banks to keep financial markets liquid.
Net-present-value analysis actually argues that financial market levels may be entirely rational. Even if a 7.7% decline in U.S. GDP translates into a 40% to 50% decline in earnings that takes several years to fully unwind, the decline in the net present value of the future earnings and thus the fair value of the stock market is surprisingly slight. The precise figure depends on the discount rate deployed and how many years of future earnings one actually cares about. But suffice it to say that even unfavourable assumptions struggle to justify a 15% decline in the stock market, and perfectly reasonable figures even argue that the fair-value decline in stocks should be less than 5% as a result of COVID-19.
Thus, while darkening sentiment, a failed exit strategy or mounting corporate damage could well pull equities downward, it needs to be said that stocks are not only already cheap on a relative basis to sovereign bonds, but also arguably cheap on a net-present-value basis.
We continue to accrue musings about the long-term implications of COVID-19 in our weekly #MacroMemos. Older thoughts can be found here.
We add three small items this week.
- Although standard economic figures fail to reveal a calamitous economic decline during the Spanish Flu of 1918, historical newspaper articles give the impression that the story was more similar to COVID-19 than commonly imagined. Quarantining and social distancing were the norm, and a Wall Street Journal article from October 24, 1918 notes that production had fallen by 50% in parts of the country, and was down nearly everywhere. The retail sector was cited as particularly affected. We mention all of this to highlight that the world nevertheless eventually returned to normal, both in the sense of resumed economic growth and also more fundamentally in that society was not permanently altered: handshaking was not abandoned forever and movie theatres and mass sporting events didn’t cease to exist.
- How much damage is happening to human capital as children miss out on their schooling? You might imagine that those aged 5—19 are perhaps learning at half the usual rate through the final 30% of their school year. North Americans have an average of 14 years of schooling. Thus, those of a school age might theoretically be losing 1% of their school-generated human capital. Of course, learning also happens outside of the school (call it one-third), schools will make an effort to catch students up in future years (perhaps reclaiming a third of what was lost). Furthermore, only around 15% of the overall population is aged 5-19. In conclusion, only 0.07% of the economy’s human capital has been lost as a direct result of closed schools. This seems tolerable (though it neglects skill loss by unemployed people).
- There has been some speculation that a baby boom might materialize as a result of all the time people are spending at home. There are undeniably scattered claims to this effect from past stay-at-home events such as blackouts, ice storms and other disasters. However, few hold up to close scrutiny – there just isn’t much evidence that this actually happens, and for that matter one might even imagine that the high level of uncertainty about the future could actually discourage parenthood in the near term. In China, a more visible consequence of forced quarantining has actually been a rising rate of couples separating.
-With contributions from Vivien Lee and Graeme Saunders