Economy

Special Report

What Can 1918/1919 Teach Us About COVID-19?    “Those who cannot remember the past are condemned to repeat it” George Santayana – 1905   Chart 1Coronavirus: As Contagious But Not As Deadly As Spanish Flu Today’s economy is very different to that of 100 years ago. Many countries then were in the middle of World War I (which ended in November 1918). The characteristics of the Spanish Flu which struck the world in 1918 and 1919 were also different to this year’s pandemic. COVID-19 is almost as contagious as the Spanish Flu, but it is much less deadly (Chart 1). Healthcare systems and treatments today are far more advanced than those of a century ago: many people who caught Spanish flu died of complications caused by bacterial pneumonia, given the absence of antibiotics. Influenza viruses tend to mutate rapidly: the influenza virus in 1918 first mutated to become far more virulent in its second wave, and then to become much milder. Coronaviruses have a “proofreading” capacity and mutate less easily.1  Nevertheless, an analysis of the pandemic of 100 years ago provides a number of insights into the current crisis, particularly now that policymakers are easing social-distancing rules to help the economy, even at the risk of more cases and deaths. Among the lessons of 1918-1919: Non-pharmaceutical interventions (NPIs) do lower mortality rates. The speed at which NPIs are implemented and the period of implementation are as important as the number of measures taken. Removing or relaxing measures too early can lead to a renewed rise in mortality rates. It is hard to compare current fiscal and monetary policies to those taken during the 1918 pandemic, since policy in both areas was already easy before the pandemic as a result of the world war. However, a severe pandemic would certainly call for a wartime-like fiscal and monetary response. The economy was negatively impacted by the pandemic in 1918-19 but, despite the shock to industrial activity and employment, the economy subsequently rebounded quickly, in a V-shaped recovery. Introduction Predicting how the economy will react to the COVID-19 pandemic is hard. Governments and policymakers face multiple uncertainties: How effective are different containment measures? Will cases and deaths rebound quickly if lockdown measures are eased? When will the coronavirus disappear? When will a vaccine be ready? With an event unprecedented in the experience of anyone alive today, perhaps there are some lessons to be learned from history. For this Special Report, we attempt to draw some parallels between the current situation and the 1918-19 Spanish flu. We focus on the different containment efforts implemented, the role that fiscal and monetary policies played, the impact on markets and the economy, and whether history can throw any light on how the COVID-19 crisis might pan out. The 1918 Spanish Flu Chart 2The Spanish Flu Hit The World In Three Waves The 1918 influenza pandemic was the most lethal in modern history. Soldiers returning from World War I helped spread the pandemic across the globe. The first recorded case is believed to have been in an army camp in Kansas. While there is no official count, researchers estimate that about 500 million people contracted the virus globally, with a mortality rate of between 5% and 10%. The pandemic occurred over three waves in 1918 and 1919 – the first in the spring of 1918, the second (and most deadly) in the fall of 1918, and the third in spring 1919 (Chart 2). In the US alone, official data estimate that around 500,000 deaths (or over 25% of all deaths) in 1918 and 1919 were caused by pneumonia and influenza.2 The pandemic moved swiftly to Europe and reached Asia by mid-1918, but became more lethal only towards the end of the year (Map 1).3 Map 1The Spread Of Influenza Through Europe Initially, scientists were puzzled by the origin of the influenza and its biology. It was not until a decade later, in the early 1930s, that Richard Shope isolated the particular influenza virus from infected pigs, confirming that a virus caused the Spanish Flu, not a bacterium as most had thought. Many of those who caught this strain of influenza died as a result of their lungs filling with fluid in a severe form of pneumonia. In reporting death rates, then, it is considered best practice to include deaths from both influenza and pneumonia. The first wave had almost all the hallmarks of a seasonal flu, albeit of a highly contagious strain. Symptoms were similar and mortality rates were only slightly higher than a normal influenza. The first wave went largely unnoticed given that deaths from pneumonia were common then. US public health reports show that the disease received little attention until it reappeared in a more severe form in Boston in September 1918.4  Most countries did not begin investigating and reporting cases until the second wave was underway (Chart 3). Chart 3Most Countries Began Reporting Only When The Second Wave Hit This second wave – which was more lethal because the virus had mutated – had a unique characteristic. Unlike the typical influenza mortality curve – which is usually “U” shaped, affecting mainly the very young and elderly – the 1918 influenza strain had a “W”-shaped mortality curve – impacting young adults as well as old people (Chart 4). This pattern was evident in all three waves, but most pronounced during the second wave. The reason for this was that the infection caused by the influenza became hyperactive, producing a “cytokine storm” – when mediators secreted from the immune system result in severe inflammation.5 Simply put, as the virus became virulent, the body’s immune system overworked to fight it. Younger people, with strong immune systems, suffered most from this phenomenon. Chart 4A Unique Characteristic: Impacting Younger Adults By the summer of 1919, the pandemic was over, since those who had been infected had either died or recovered, therefore developing immunity. The lack of records makes it difficult to assess if “herd immunity” was achieved. However, some historical accounts and research – particularly for army groups in the US and the UK – suggest that those exposed to the disease in the first mild wave were not affected during the second more severe wave.6 The failure to define the causative pathogen at the time made development of a vaccine impossible. Nevertheless, some treatments and remedies showed modest success. These varied from using a serum – obtained from people who had recovered, who therefore had antibodies against the disease – to simple symptomatic drugs and various oils and herbs. The Effectiveness Of Non-Pharmaceutical Interventions (NPIs) What we today call “social distancing” showed positive effects during the 1918-19 pandemic. These included measures very similar to those applied today: school closures, isolation and quarantines, bans on some sorts of public gatherings, and more. However, there were few travel bans.  The number of passengers carried during the months of the pandemic did noticeably decline though (Chart 5). Chart 5Travel Slowed...Just Not Enough Table 1, based on research by Hatchett, Mecher and Lipsitch, breaks down NPIs by type for 17 major US cities. Most cities implemented a wide range of interventions. But it was not only the type of NPIs implemented that made a difference, but also the speed and length of implementation. Table 1Measures Applied Then Are Very Similar To Those Applied Today Further research by Markel, Lipman and Navarro based on 43 US cities shows that the median number of days between the first reported influenza case and the first NPI implementation was over two weeks. The median period during which various NPIs were implemented was about six weeks (Table 2). Table 2NPIs Were Implemented Only For Short Periods Markel, Lipman and Navarro's findings show that a rapid public-health response was an important factor in reducing the mortality rate by slowing the rate of infection, what we now refer to as “flattening the curve.” There were major differences in cities’ policies: both the speed at which they implement NPIs, and the length of the implementation period. Chart 6 shows that: Cities that acted quickly to implement NPIs slowed the rate of infections and deaths (Chart 6, panel 1) Cities that acted quickly had lower mortality rates from influenza and pneumonia (Chart 6, panel 2) Cities that implemented NPIs for longer periods had fewer deaths (Chart 6, panel 3) Chart 7 quantifies the number of NPIs taken, the time it took to implement the measures, and the length of NPIs to gauge policy strictness. Cities with stricter enforcement had lower death rates than those with laxer measures. Chart 6Fast Response And Longer Implementation Led To Fewer Deaths... Chart 7...So Did Policy Strictness For example, Kansas City, less than a week after its first reported case, had implemented quarantine and isolation measures. By the second week, schools, churches, and other entertainment facilities closed. Schools reopened a month later (in early November) but quickly shut again until early January 1919.  While we do not have definitive dates on when each NPI was lifted, some sort of protective measures in Kansas City were in place for almost 170 days. By contrast, Philadelphia, one of the cities hardest hit by Spanish Flu, took more than a month to implement any measures. Its tardiness meant that it reached a peak mortality rate much more quickly: in 13 days compared to 31 days for Kansas City. Even after the first reported case, the Liberty Loans Parade was still held on September 28, 1918 – with the knowledge that hundreds of thousands of spectators might be vulnerable to infection.7,8 It was not until a few days later that institutions were closed and a ban on public gatherings was imposed. Many other cities also held a Liberty Loans Parade, including Pittsburgh and Washington DC, but Philadelphia’s was the deadliest. Studies also show that relaxing interventions too early could be as damaging as implementing them too late. St. Louis, for example, was quick to lift restrictions and suffered particularly badly in the second wave as a result. It later reinstated NPIs up until end of February 1919. Other cities that eased restrictions too early (San Francisco and Minneapolis, for example) also suffered from a second swift, albeit milder, increase in weekly excess death rates from pneumonia and influenza (Chart 8). Chart 8Relaxing Lockdown Measures Too Early Can Lead To A Second Rise In Deaths... Chart 9...And So Can Highly Effective Measures Of course, NPIs cannot be implemented indefinitely. A recent research paper by Bootsma and Ferguson raises the point that suppressing a pandemic may not be the best strategy because it just leaves some people susceptible to infection later. They argue that highly effective social distancing measures, which allow a susceptible pool of people to reintegrate into society when the measures are lifted, are likely to lead to a resurgence in infections and fatalities in a second peak (Chart 9).9 They suggest an optimal level of control measures to reduce R (the infection rate) to a value that makes a significant portion of the population immune once measures are lifted.   The Impact Of The Spanish Flu On The Economy And Markets How did the Spanish Flu pandemic affect the economy? Many pandemic researchers ignore the official recession identified by the NBER during the months of the pandemic (between August 1918 and March 1919).10 The reason is that most of the evidence indicates that the economic effects of the 1918-19 pandemic were short-term and relatively mild.11 Disentangling drivers of the economy is, indeed, tricky given that WW1 ended in November 1918. However, it is easy to underestimate the negative impact of the pandemic since the war had such a big impact on the economy, as well as investor and public sentiment. Various research papers support the fact that, while the pandemic did indeed have an adverse effect on the economy, NPIs did not just depress mortality rates, but also sped the post-pandemic economic recovery.12 Research by Correia, Sergio, and Luck showed that the areas most severely affected by the pandemic saw a sharp and persistent decline in real economic activity, whereas cities that intervened earlier and more aggressively, experienced a relative increase in economic activity post the pandemic.13 Their findings are based on the increase in manufacturing employment after the pandemic compared to before it (1919 versus 1914). However, note that the rise of manufacturing payrolls in 1919 was high everywhere given the return of soldiers post-WWI. The researchers also note that those cities hardest hit by the pandemic also saw a negative impact on manufacturing activity, the stock of durable goods, and bank assets. Because Spanish flu disproportionately killed younger adults, many families lost their breadwinner. In economic terms, this implies both a negative supply shock and negative demand shock. If fewer employees are available to produce a certain good, supply will fall. The same reduction in employment also implies reduced income and therefore lower purchasing power. Both cases will result in a decrease in output. However, the change in prices depends on the decline of supply relative to demand. In 1918-19, the impact was disinflationary: demand declined by more than supply, and both spending and consumer prices fell during the pandemic (Chart 10). Chart 10Short-Term Price Impact Was Disinflationary US factory employment fell by over 8% between March 1918 and March 1919 – the period from the beginning of the first wave until the end of the second wave. It is important to note, however, that few businesses went bankrupt during the pandemic years (Chart 11). Additionally, the November 1918 Federal Reserve Bulletin highlighted that many cities, including New York, Kansas City, and Richmond, experienced a shortage of labor due to the influenza.14  Factory employment in New York fell by over 10% during this period.  Chart 11Loss Of Middle-Aged Adults = Loss Of Breadwinners The link between the labor shortages and the decline in industrial production is unclear. Industrial activity in the US peaked just before the second wave, contracting by over 20% during the second wave (Chart 12). Various industries reported disruptions: automobile production fell by 67%, anthracite coal production and shipments fell by around 45%, and railroad freight revenues declined by over seven billion ton-miles (Chart 12, panels 2, 3 & 4). However, some of this decline is attributed to falling defense production after the war. Chart 12Activity Slowed, But Rebounded Quickly Chart 13The War Had A Bigger Impact On The Stock Market Than The Pandemic The equity market moved in a broad range in 1915-1919 and fell sharply only ahead of the 1920 recession (Chart 13). Seemingly, stock market participants were more focused on the war than the pandemic. The lack of reporting of the pandemic could have contributed to this: newspapers were encouraged to avoid carrying bad news for reasons of patriotism and did not widely cover the pandemic until late 1918.15   Chart 14Monetary Policy Was Easy...Even Before The Pandemic Started The Federal Reserve played an active role in funding the government’s spending on the war, and so monetary policy was very easy during the pandemic – but for other reasons. The Fed used its position as a lender to the banking system to facilitate war bond sales.16 Interest rates were cut in 1914 and 1915 even before the US entered the war. The US economy had been in recession between January 1913 and December 1914. Policy rates remained low throughout 1916 and 1917 and slightly rose in 1918 and 1919. It was not until 1920 that Federal Reserve Bank System tightened policy rapidly to choke off inflation, which accelerated to over 20% in mid-1920 – rising inflation being a common post-war phenomenon (Chart 14). The Lessons Of 1918-19 For The Coronavirus Pandemic Non-pharmaceutical interventions should continue to be implemented until a vaccine, effective therapeutic drugs, or mass testing is available. Relaxing measures prematurely is as damaging as a tardy reaction to the pandemic. Reacting quickly and imposing multiple measures for longer periods not only reduces mortality rates, but also improves economic outcomes post-crisis. The economy suffers in the short-term: supply and demand shocks lead to lower output. The demand shock however is larger leading to lower prices and disinflationary pressures, at least during and immediately after the pandemic.   Amr Hanafy Senior Analyst amrh@bcaresearch.com Footnotes 1  Please see the Q&A with immunologist and Nobel laureate Professor Peter Doherty, published by BCA Research April 1st 2020: BCA Research Special Report, “Questions On The Coronavirus: An Expert Answers,” available at bcaresearch.com 2 Please see “Leading Cause of Death, 1990-1998,” CDC Centers for Disease Control and Prevention. 3 Please see Ansart S, Pelat C, Boelle PY, Carrat F, Flahault A, Valleron AJ, “Mortality burden of the 1918-1919 influenza pandemic in Europe,” NCBI. 4 Please see Public Health Report, vol. 34, No. 38, Sept. 19, 1919. 5 Please see Qiang Liu, Yuan-hong Zhou, Zhan-qiu Yang Cell Mol Immunol. 2016 Jan; 13(1): 3–10. 6 Please see Shope, R. (1958) Public Health Rep. 73, 165–178. 7 The Liberty Loans Parade was intended to promote the sale of government bonds to pay for World War One. 8 Please see Hatchett RJ, Mecher CE, Lipsitch M (2007) "Public health interventions and epidemic intensity during the 1918 influenza pandemic,"PNAS 104: 7582–7587. 9 Please see Bootsma M, Ferguson N, “The Effect Of Public Health Measures On The 1918 Influenza Pandemic In U.S. Cities,” PNAS (2007). 10Please see https://www.nber.org/cycles.html 11Please see https://www.stlouisfed.org/~/media/files/pdfs/community-development/research-reports/pandemic_flu_report.pdf 12Please see https://libertystreeteconomics.newyorkfed.org/2020/03/fight-the-pandemic-save-the-economy-lessons-from-the-1918-flu.html. 13Please see Correia, Sergio and Luck, Stephan and Verner, Emil, Pandemics Depress the Economy, Public Health Interventions Do Not: Evidence from the 1918 Flu (March 30, 2020). Available at SSRN: https://ssrn.com/abstract=3561560 or http://dx.doi.org/10.2139/ssrn.3561560. 14Please see Board of Governors of the Federal Reserve System (U.S.), 1935- and Federal Reserve Board, 1914-1935. "November 1918," Federal Reserve Bulletin (November 1918). 15Please see https://newrepublic.com/article/157094/americas-newspapers-covered-pandemic. 16Please see https://www.federalreservehistory.org/essays/feds_role_during_wwi.

Special Report

Highlights Fed/BoE NIRP: It is too soon for either the Fed or Bank of England to consider a move to a negative interest rate policy (NIRP), even with US and UK money markets flirting with pricing in that outcome. Lessons from “NIRP 1.0”: In the countries that did go to negative rates in 2014-16 (Japan, Switzerland, the euro area, Sweden and Denmark), there existed some combination of weak economies, near-0% inflation, anemic credit growth or unwanted currency appreciation. Negative rates were needed to help fight those trends by driving down longer-term bond yields. NIRP 2.0?: Among the major countries without negative rate policies in effect (the US, UK, Canada and Australia), there is no evidence that longer-term borrowing rates need to fall further to boost credit growth, even in the midst of deep recessions. However, additional strength of the stubbornly resilient US dollar could be the deflationary shock that eventually forces the Fed into NIRP. Feature Chart 1NIRP 2.0 Would Trigger A Surge Of Negative Yielding Bonds Within a 20-month window in 2014-16, the central banks of Japan, Sweden, the euro area, Switzerland and Denmark all cut policy interest rates to below 0% - where they remain to this day. Fast forward to 2020, in the midst of a global pandemic and deep worldwide recession that has already forced major developed market central banks to cut rates close to 0%, there is now increased speculation that the negative interest rate policy (NIRP) club might soon get a few new members. The Federal Reserve has been front and center in that group. Fed funds futures contracts had recently priced in slightly negative rates in 2021, despite Fed Chair Jerome Powell repeatedly saying that a sub-0% funds rate was not in the Fed’s plans. The Bank of England (BoE) has also seen markets inch toward pricing in negative rates, although BoE officials have been more open to the idea of negative rates as a viable policy choice. Even the Reserve Bank of New Zealand has suggested that negative rates may be needed there soon. An expansion of the list of countries that have moved to negative rates, beyond the “NIRP 1.0” group of 2014-16, has the potential to drive down global bond yields even further. Already, there is $11 trillion of negative yielding debt within the Bloomberg Barclays Global Aggregate index, representing 20% of the total (Chart 1) If there is a shift to negative rates in the potential “NIRP 2.0” group of major developed economies with policy rates now near 0% – a list that includes the US, the UK, Canada and Australia – then the amount of negative yielding debt worldwide will soar to new highs. An expansion of the list of countries that have moved to negative rates, beyond the “NIRP 1.0” group of 2014-16, has the potential to drive down global bond yields even further. In this report, we take a look at the conditions that led the NIRP 1.0 countries to shift to negative rates in the middle of the last decade, to see if any similarities exist in non-NIRP countries today. We conclude that the conditions are not yet in place for a shift to sub-0% policy rates in the US, the UK, Canada or Australia – all countries where central banks still have other policy tools available to provide stimulus before resorting to negative rates. How Negative Interest Rates Can “Work” To Revive Growth Broadly speaking, central banks around the world have had difficulty meeting their inflation targets since the 2008 Global Financial Crisis. The main reason for this has been sub-par economic growth, much of which is structural due to aging demographics and weak productivity. Since central bankers must stick to their legislated inflation targeting mandates, they are forced to cut rates when economic growth and inflation are too low. If real economic growth remains weak for structural reasons, then central banks can enter into a cycle of continually cutting rates all the way to zero, or even below zero, in order to try and prevent low inflation from becoming entrenched into longer-term inflation expectations. If growth and inflation continue to languish even after policy rates have reached 0%, then other tools must be used to ease monetary conditions to try and stimulate economies. These typically involve driving down longer-term borrowing rates (bond yields) through dovish forward guidance on future monetary policy, bond purchases through quantitative easing (QE) and, if those don’t work, moving to negative policy interest rates. A nice summary indicator to identify this intertwined dynamic of real economic growth and inflation is to look at the trend growth rate of nominal GDP. Chart 2 shows the policy interest rates three-year annualized trend of nominal GDP growth for the NIRP 1.0 countries, dating back to before the 2008 crisis. Japan stands out as the weakest of the group, with trend nominal growth contracting during and after the 2009 recession, while struggling to reach even +2% since then. The euro area, Sweden and Switzerland all enjoyed +5% nominal growth prior to 2008, before a plunge to the 1-2% range during and after the recession. After that, the three countries had varying degrees of economic success. Between 2016 and 2019, Sweden saw trend nominal growth between 4-5%, while the euro area struggled to achieve even +3% nominal growth and Switzerland maintained a Japan-like pace. Chart 2Fewer Tools Left For NIRP 1.0 Countries To Boost Growth Chart 3NIRP 2.0 Candidates Can Still Expand QE First The European Central Bank (ECB), Swiss National Bank (SNB), the Bank of Japan (BoJ) and Sweden’s Riksbank all cut policy rates aggressively in 2008/09, helping spur a recovery in nominal growth. The central banks had to keep rates lower for longer because of structurally weak growth, leaving far less capacity to ease aggressively in response to the growth downturn a few years later. Eventually, the ECB, SNB, BoJ and Riksbank all went to negative rates between June 2014 and February 2016. The BoJ and SNB, facing persistent headwinds from strengthening currencies, also resorted to aggressive balance sheet expansion to provide additional monetary stimulus – trends that have continued to this day, with both central banks having balance sheets equal to around 120% of GDP. The experience of these four NIRP 1.0 countries showed that the move to negative rates was a process that began in the 2008 financial crisis. Central banks there were unable to raise rates much, if at all, after the recession, leaving little ammunition to fight the varying growth slowdowns suffered between 2012 and 2016. Eventually, rates had to be cut below 0% which, combined with QE, helped generate lower bond yields, weaker currencies and, eventually, a pickup in growth and inflation. Looking at the NIRP 2.0 candidate countries, nominal GDP growth has also struggled since the financial crisis, unable to stay much above 3-4% in the US, Canada and the UK. Only Australia has seen trend growth reach peaks closer to 5-6% (Chart 3). The Fed, BoE, Reserve Bank of Australia (RBA) and Bank of Canada (BoC) all also cut rates aggressively in 2008/09, with the Fed and BoE doing QE buying of domestic bonds. Rates were left at low levels after the crisis in the US and UK, with only the RBA and, to a lesser extent, the BoC hiking rates after the recession ended. When growth slowed again in these countries during the 2014-16 period, the RBA and BoC did lower policy rates, but negative rates were avoided by all four central banks. Today, nominal growth rates have collapsed because of the COVID-19 lockdowns that have shuttered much of the world economy. Central banks that have had any remaining capacity to cut policy rates back to 0% have done so, yet this recession has already become so deep that additional declines in rates may be necessary to stabilize unemployment and inflation. The experience of the NIRP 1.0 countries shows that negative rates can also be effective in boosting growth – especially in countries suffering unwanted currency strength. One way to see the problem that monetary policymakers are now facing is by looking at Taylor Rule estimates of appropriate interest rate levels (Charts 4 and 5). Given the rapid surge in global unemployment rates to levels that, in some cases, have not been seen since the Great Depression (Chart 6), alongside decelerating inflation, Taylor Rule implied policy rates are now deeply negative in the US (-5.6%), Canada (-2.9%) and euro area (-1.7%).1 Taylor Rules show that moderately negative rates are also needed in Sweden (-0.5%), Switzerland (-0.2%) and Japan (-0.2%). Only in Australia (+1.3%) and the UK (+0.3%) is the Taylor Rule indicating that negative rates are not currently required. Chart 4Taylor Rule Says More Rate Cuts Needed Here … Chart 5… But Rates Are Appropriate Here Chart 6The Main Reason Why Taylor Rule Implied Policy Rates Have Plunged Among the potential NIRP 2.0 candidates, the negative rate option has been avoided and aggressive QE balance sheet expansion has been pursued by all of them – including the BoC and RBA who avoided asset purchase programs in 2008/09. Balance sheet expansion can be an adequate substitute for policy interest rate cuts by helping drive down longer-term bond yields and borrowing rates, which helps spur credit demand and, eventually, economic growth. Yet the experience of the NIRP 1.0 countries shows that negative rates can also be effective in boosting growth – especially in countries suffering unwanted currency strength. How negative rates worked for the NIRP 1.0 countries For the ECB (Chart 7), BoJ (Chart 8), Riksbank (Chart 9) and SNB, the path from negative policy rates in 2014-16 to, eventually, faster economic growth and inflation followed a similar process: Chart 7The Euro Area's Negative Rates Experience Chart 8Japan's Negative Rates Experience Chart 9Sweden's Negative Rates Experience Moving to negative policy rates resulted in a sharp decline in nominal government bond yields The fall in yields helped trigger currency depreciation Nominal yields fell faster than inflation expectations, allowing real bond yields to turn negative Credit growth eventually began to pick up in response to the decline in real borrowing costs Inflation bottomed out and started to move higher. In Japan, the euro area and Sweden, this process played out fairly rapidly with credit growth and inflation bottoming within 6-12 months of the move to negative rates. Only in Switzerland (Chart 10), where the SNB gave up on currency intervention in January 2015, was the process delayed, as the surge in the currency triggered a move into deeper deflation and higher real bond yields. It took a little more than a year for the deflationary impact of the franc’s surge to fade, allowing real bond yields to decline and credit growth and inflation to bottom out and recover. The implication is clear – negative rates are good for real assets, but troublesome for banks.  Of course, we are talking about the pure economic effect of negative rates as a monetary policy tool. There are side effects of having negative nominal interest rates and deeply negative real bond yields, like surging asset values (especially for real assets like housing). Bank profitability is also negatively impacted by the sharp fall in longer-term bond yields that hurts net interest margins, even with higher lending volumes and reduced non-performing loans. Chart 10Switzerland's Negative Rates Experience Chart 11Negative Rates Are Good For Real Assets This can be seen in Charts 11 & 12, which compare the performance of real house prices and bank equities (relative to the domestic equity market) in the years leading up to, and following, the move to negative rates in 2014-16 for the NIRP 1.0 countries. The implication is clear – negative rates are good for real assets, but troublesome for banks. Chart 12Negative Rates Are Bad For Bank Stocks Nonetheless, the experience of the NIRP 1.0 countries suggests that the potential NIRP 2.0 countries could see similar benefits on growth and inflation – but not before other policy options are exhausted first. Bottom Line: In the countries that did go to negative rates in 2014-16 (Japan, Switzerland, the euro area, Sweden and Denmark), there existed some combination of weak economies, near-0% inflation, anemic credit growth or unwanted currency appreciation. Negative rates were needed to help fight those trends by driving down longer-term bond yields and helping spur credit growth and, eventually, some inflation. Depreciating currencies had a big role to play in generating those outcomes. Negative Rates Are Not Necessary (Yet) In The NIRP 2.0 Countries As discussed earlier, the sharp surge in unemployment because of the COVID-19 global recession means that negative interest rates may now be “appropriate” in the US and Canada, based on Taylor Rules. Negative rates are not needed in the UK and Australia, however, although policy rates need to stay very low in both countries. A similar divergence can be seen in inflation. Headline CPI inflation rates were already under severe downward pressure from the recent collapse in oil prices. The surge in spare economic capacity opened up by the current recession can only exacerbate the disinflation trend. However, the drop in inflation has been more acute in the US and Canada relative to the UK and Australia, suggesting a greater need for the Fed and BoC to be even more stimulative than the BoE or RBA (Chart 13). A renewed breakout of the currency to new cyclical highs could be the deflationary signal that triggers the Fed into an even more aggressive policy response. There is one area where the Fed stands alone in this group. The relentless strength of the US dollar, even as the Fed’s rate cuts have taken much of the attractive carry out of the greenback, hurts US export competitiveness in a demand-deficient recessionary global economy. The strong dollar also acts as a dampening influence on US inflation. A renewed breakout of the currency to new cyclical highs could be the deflationary signal that triggers the Fed into an even more aggressive policy response (Chart 14). This would mirror the experience of the NIRP 1.0 countries prior to the move to negative rates, where unwanted currency strength crippled both economic growth and inflation. Chart 13The Threat Of Deflation Could Trigger NIRP Chart 14Could More USD Strength Drag The Fed Into NIRP? For now, the Fed has many other policy options open before negative rates would be seriously considered. The reach of its QE programs could be expanded even further, even including equity purchases. The existing bond QE could be combined with a specific yield target (i.e. yield curve control) for shorter-maturity US Treasuries, helping anchor US yields at low levels for longer. Summing it all up, we do not see the need for any of the NIRP 2.0 candidates to move to negative rates anytime soon. The need for such extreme policies is not yet necessary, though, both in the US and the other NIRP 2.0 candidate countries. Bank lending is expanding at a double-digit pace in the US, and still at a decent 5-7% pace in the UK, Canada and Australia, even in the midst of a sharp recession (Chart 15). This may only be due to the numerous loan guarantees provided by governments as part of fiscal stimulus responses, or it may be related to companies running down credit lines to maintain liquidity. The experience of the NIRP 1.0 countries, though, suggests that credit growth must be far weaker than this to require negative policy rates to push down longer-term borrowing costs. Chart 15These Already Look Very "NIRP-ish" Chart 16Too Soon For Global NIRP, Maintain Neutral Global Duration Exposure Summing it all up, we do not see the need for any of the NIRP 2.0 candidates to move to negative rates anytime soon. In terms of investment implications, we continue to recommend an overall neutral stance on global duration exposure, as we see little immediate impetus for yields to move lower because of reduced expectations of future interest rates or inflation (Chart 16). We will continue to watch currency levels and credit growth as a sign that policymakers may need to shift their tone in the coming months. Bottom Line: Among the major countries without negative rate policies in effect (the US, UK, Canada and Australia), there is no evidence that longer-term borrowing rates need to fall further to boost credit growth, even in the midst of deep recessions. However, additional strength of the stubbornly resilient US dollar could be the deflationary shock that eventually forces the Fed into NIRP.   Robert Robis, CFA Chief Fixed Income Strategist rrobis@bcaresearch.com The GFIS Recommended Portfolio Vs. The Custom Benchmark Index Footnotes 1 Our specification of the Taylor Rule uses unemployment rates relative to full employment (NAIRU) levels as the measure of spare capacity in the economies. For the neutral real interest rate, we use the New York Fed’s estimate of r-star for the US, Canada, the euro area and the UK; while using the OECD’s estimate of potential GDP growth as the neutral real rate measure for countries where we have no r-star estimate (Japan, Sweden, Switzerland and Australia).

Special Report

Feature The SPX suffered its third 5.3-7.3% pullback since early April last week, which we deem a healthy development as markets cannot go up in a straight line. While there is a chance this latest pullback may morph into a correction, our sense is that equities will remain range bound in the near-term consolidating the vast gains made since the March 23 lows. Now that earnings season is practically over and macro data will remain backward looking, a large void signals that technicals will dominate trading. On that front, this looming lateral move will likely confine the SPX between the critical 50-day and 200-day moving averages – a roughly 10% range between 2,712 and 3,000 – until a catalyst breaks the stalemate (top panel, Chart 1A). With regard to the cyclical outlook, ultra-accommodative fiscal and monetary policies remain the dominant macro themes, and underpin our sanguine equity market view for the next year. Chart 1AConsolidating Gains Dollar The Reflator Importantly, King Dollar is a key macro variable that we are closely monitoring and as we highlighted last week, the Fed is indirectly aiming at jawboning the greenback.1 US dollar based liquidity is one of the most important determinants/drivers of global growth. The longer US dollar liquidity gets replenished, the more upward pressure it will put on SPX momentum and SPX EPS (Chart 1B). Sloshing US dollar based liquidity will serve as a much needed catalyst for a global growth recovery. Chart 1BHeed The Message From US Dollar Liquidity: Chart Of The Year Candidate The Yield Curve, Interests Rates And Profits Meanwhile, the yield curve, in fact a number of different yield curve slopes, troughed prior to the SPX in March, preserving its leading properties both near equity market tops and bottoms (middle & bottom panels, Chart 1A). The Fed orchestrated the steepening of the yield curve – which is typical during recessions – with the two preemptive cuts in March. Crucially, the yield curve is signaling that in the back half of the year SPX profits will also trough. True, a profit shortfall is upon us in Q2, and the steeper the fall, the higher the chance of a V-shaped recovery, owing to base effects (yield curve shown advanced, Chart 2). Chart 2Steep Yield Curve Slope Will Reflate Profits Encouragingly, the Fed reiterated last week that it will remain ultra-accommodative. While it will refrain from delving into NIRP, QE5 can expand anew and sustain the perching of the 2-year and even the 5-year and 7-year Treasury yields near zero. In fact, the shadow fed funds rate is already below zero as we highlighted last week.2 This monetary backdrop coupled with rising fiscal deficits as far as the eye can see – which will put upward pressure on long-term Treasury yields – will ensure a steep yield curve, and thus engineer a profit recovery (Chart 2). With regard to the interplay of interest rates and profit growth, the two are tightly inversely correlated (Chart 3). Empirical evidence suggests that since the mid-1980s profit growth is the mirror image of the year-over-year change in 7-year Treasury yields, albeit with a significant lag. Chart 3Interest Rate Pummeling Is A Boon For EPS What EPS Growth Is Discounted? Currently, if the relationship between profits and yields were to hold, then SPX EPS growth would stage a sizable come back in 2021. Chart 4 depicts the sell side’s quarterly EPS forecasts all the way to end 2021. Indeed, following a steep contraction, a brisk V-shaped profit recovery is looming in 2021 as we first argued three weeks ago that “historical precedents show an explosive year-over-year growth increase in EPS from recessionary troughs”.3 In more detail, Chart 5 breaks down 12-month forward EPS growth per sector. Tech comes out on top and by a wide margin with a near double-digit profit growth rate in absolute terms. This gulf is even more pronounced relative to the contracting SPX EPS growth rate. In fact, tech relative profit growth just reached the highest level since 2004 and explains the broad market’s tech dependence. As a reminder, tech market cap is back to the 2018 peak despite the fact the GOOGL and FB have now moved to the newly formed S&P communication services index. If one were to add the pair and AMZN back to the tech sector’s weight, it would comprise over 36% of the SPX, higher even than the dotcom bubble era (Chart 6)! Chart 4V-Shaped Profit Recovery Chart 5Tech… Chart 6…Reigns Supreme Tech Titans Digression A brief digression is in order as it pertains to the tech titans. We have been inundated with requests recently on the subject of valuations and the concentration of returns in the top five SPX stocks. We first commented on this in January, and reiterate today that the current tech sector’s supposed overvaluation is nowhere near the dotcom excesses .4 Back then, the top five SPX stocks commanded a forward P/E over 60, but today’s valuation pales in comparison with the late-1990s, as the equivalent P/E is roughly half that multiple (please refer to Chart 2 of the January 27, 2020 Weekly Report). Why? Because at the turn of the millennium, tech stocks had very little earnings to show for, but now the tech sector has the largest profit weight among its GICS1 peers. Thus, tech stocks trade at a modest 9% premium to the broad market whereas in 1999 they were changing hands at more than twice the SPX multiple (Chart 7). Chart 8 attempts to shed more light on the subject. The top panel shows the overall SPX market cap and also excluding the top five stocks. Then we subtract the top five stocks’ forward P/E from the broad market and show where the S&P 500 ex-top five stocks P/E trades (second panel, Chart 8). Since the FB IPO, these stocks have indeed increased their influence on the broad market’s valuation (third panel, Chart 8). Chart 7What Relative Overvaluation? Chart 8Top Five Are Pricey, But For Good Reason Sectorial Profit Growth Breakdown Circling back to the breakdown of 12-month forward EPS growth per sector, traditional defensive sectors (utilities, staples and health care) all enjoy positive 12-month forward profit growth in absolute terms, and so do communication services that just kissed off the zero line. All other sectors are contracting at differing degrees (Chart 5). On a longer-term basis, as expected no GICS1 sector is slated to contract, but their five-year growth rates are widely dispersed. Consumer discretionary, real estate, materials and tech occupy the top ranks with double digit growth rates, while utilities, consumer staples, energy, industrials and financials are in mid-single digits and at the bottom of the pit. Communication services and health care hover in the middle, on a par with the broad market (Chart 9). Chart 9Long-Term Growth Has Reset Lower Higher Profits Are Synonymous With Higher Returns Intuitively, the higher the forward profit growth rate, the higher each sector’s trailing return. Chart 10 depicts this positive correlation on the GICS1 sectors and corroborates that the laggard energy sector has the lowest year-to-date return, whereas tech stocks lead the pack. Importantly, SPX sector profit weights are extremely important. Chart 11 ranks the GICS1 sectors 12-month forward profit weights. Tech, health care and financials comprise roughly 60% of total S&P 500 earnings for the coming year. Whereas the drubbing in the energy sector (83% projected EPS contraction) has drifted into oblivion within the SPX context and has a mere 0.5% profit weight (Chart 11). Chart 10Higher Growth = Higher Returns Chart 11Top three Comprise 60% Of Profit Weight Bottom Line: While the top three sectors inherently carry the bulk of the risk on the SPX earnings front courtesy of the high concentration, our sense is that both tech (neutral) and health care (overweight) will deliver according to the messages from our macro EPS growth models (Chart 12). Financials (overweight) profits are a question mark, and therefore pose the greatest risk to our still constructive 9-12 month broad equity market view.     Anastasios Avgeriou US Equity Strategist anastasios@bcaresearch.com Chart 12EPS Growth Models Emit Positive Signals   Footnotes 1     Please see BCA US Equity Strategy Weekly Report, “The Bottomless Punchbowl” dated May 11, 2020, available at uses.bcaresearch.com. 2     Ibid. 3    Please see BCA US Equity Strategy Weekly Report, “Gauging Fair Value” dated April 27, 2020, available at uses.bcaresearch.com.          4    Please see BCA US Equity Strategy Weekly Reports, “Three EPS Scenarios” dated January 13, 2020 and “When The Music Stops...” dated January 27, 2020, available at uses.bcaresearch.com.  

Special Report

Highlights The duration of this crisis and the details of the plan to reopen the economy will determine whether the initial uptick in median home prices will prove to be transitory. Phase I provides room for construction to resume at least partially, while demand for homes is likely to recover more gradually. This temporary supply/demand imbalance is unlikely to result in a meaningful price contraction as significant mitigating factors are at play. Government actions to support households and the availability of credit as well as low mortgage rates should prop up the homeownership rate. Housing’s wealth effect has decreased and is unlikely to drive consumption in a pandemic-related recession. Construction employment was highly affected though resuming work in this sector is more likely to boost steel demand than have a significant impact on the unemployment rate. Feature A recession typically occurs amidst imbalances in the economy and the 2008 sub-prime episode arguably embodied the epitome of housing excesses. Housing’s contribution to GDP has significantly decreased over the past seventy years, and today’s well-balanced housing market is unlikely to be the center of attention, but home prices are cyclical and large fluctuations can have repercussions in other areas of the economy. Social distancing is leading supply to contract first, altering the typical recessionary chain of events. We examine the COVID-induced shock to housing and its potential ramifications. Under the working assumption that a vaccine and/or effective treatment will allow economic activity to fully resume within the next twelve months, we conclude that home prices are unlikely to contract meaningfully. The homeownership rate should remain well supported and consumption is more likely to be impacted by unemployment than housing wealth effects. Meanwhile, a tightening in lending standards at the margin should not get in the way of credit availability. A Sellers’ Market Chart 1COVID-19 Is Destroying More Supply Than Demand The latest housing data releases strikingly contrast with the employment data and GDP growth estimates. The median home price actually increased by 8% on a year-on-year basis while new home sales contracted by 10%, suggesting that supply has been decreasing at a faster pace than demand (Chart 1). In the typical recession sequence of events, home prices slip as falling employment dents demand which in turn leads homebuilders to defer new starts and reduce prices on the existing supply of new homes. This is not a typical recession, however, and the supply shock preceded the demand shock. Confinement measures prevented construction professionals from going to work, thereby immediately halting the production of new homes. Meanwhile, the fact that most job losses have been temporary thus far has led to a relatively slower pace of demand destruction. Moreover, real estate transactions take a couple of months to close and the latest data may simply reflect purchase decisions that were made before the US became an epicenter of the pandemic. Median home prices may also be holding firm because sellers are not compromising on their asking price when social distancing prevents in-person visits (Chart 2), or because sellers are waiting things out before re-listing their property for sale. The housing market is effectively in a time-out where a reduced number of transactions is preventing prices from adjusting in a timely fashion. Chart 2Prospective Buyers Taking Social Distancing To Heart Prices Subject To Mitigating Forces Chart 3A Well-Balanced Housing Market The duration of the COVID-19 crisis and the details of the phases of economic reopening will ultimately determine whether this initial uptick in median home prices proves to be transitory. The housing market can remain a sellers’ market for as long as the mortgage forbearance allowed under CARES Act protects mortgage owners from defaults. It currently allows applicants to postpone their mortgage payments for up to a year amid COVID-related economic hardships. These payments are then tacked on to the end of the forbearance period and paid back over time in a mortgage modification. The winds will change if a vaccine is not mass-produced by then and Congress does not provide new aid. A wave of defaults would lead to mass property listings by desperate sellers, exerting significant downward pressure on home prices. Local homebuilders’ associations are making their case to Washington to be considered essential. The current plans to reopen the economy would provide room for residential construction to resume at least partly under Phase I, as mandated social distancing measures can be implemented on an open-air construction site. The US Census Bureau estimates that the average length of time from start to completion ranges between 7 and 15 months depending on the type of construction. Even if no new project begins until the end of the recession, currently pending constructions will resume and add another 730,0001 new homes on the market by fall - a conservative estimate that excludes any potential existing homes that might go up for sale. Existing homes account for the lion’s share of total inventory. Meanwhile, it would take much longer for demand to recover even in the unlikely event that the virus miraculously disappeared and life returned to normal in a fortnight. It generally takes time for the unemployment rate to recover to pre-recession levels, as matching available workers with employers is time-consuming and feedback loops are at play whereby unemployment leads to less spending which in turn reduces the incentive for firms to hire. The temporary nature of the layoffs and the government financial support to households will be mitigating factors, but precautionary savings tend to rise after a recession and unemployed workers might have drawn down their bank accounts. All these factors should contribute to a slower pace of housing demand recovery. Even though demand might take longer to recover, a generally well-balanced market will support prices. This temporary supply/demand imbalance scenario is bearish for home prices. However, it is worth remembering2 that unlike the previous downturn, the housing market was well balanced before this crisis began, another important factor that should mitigate the magnitude of any potential price decline (Chart 3). Bottom Line: Under the working assumption that a vaccine will be available and mass-produced within twelve months, this atypical recession is unlikely to result in a severe home price contraction. Support For Credit Chart 4Loan Deferrals Exert Pressure On Banks... An increasing share of banks have tightened residential mortgage lending standards at the margin (Chart 4), an unsurprising outcome given that a recession has arrived and payment deferrals reduce the net present value of any given mortgage. Securitization may also become more difficult or costly as mortgage servicers’ resources are strained by delayed reimbursement from Fannie Mae and Freddie Mac for the interest payments they have to advance to holders of agency mortgage-backed securities. Three-quarters of residential mortgages are backed by federal agencies, and banks presumably have little appetite to tie up limited capital with new loans at the onset of what might be a brutal recession. They will presumably be eager to get loans off their balance sheets by selling them into securitization pools, but if servicers are wary in an environment when 7.5% of all mortgages are already in forbearance, they would be well-advised to underwrite them as if they were going to have to hold them. However, banks have exerted significant restraint since their pre-Great Financial Crisis frenzy.3 Their loan books - across all core lending categories, but most prominently in the real estate segment - have grown at a markedly slower pace in the past decade than they did in any other postwar expansion4 (Chart 5). Banks are also better capitalized than they used to be, strengthening their ability to sustain losses (Chart 6). Chart 5...But Their Restrained Behavior In The Late Expansion...   Chart 6...And Increased Equity Capital Strengthen Their Ability To Sustain Losses Bottom Line: Financing should remain available to prospective home buyers. There are no excesses in the overall banking system and regulators will not allow the mortgage securitization machinery to break down. Resilient Homeownership Rate Just as the pandemic is unlikely to result in a drastic decline in home prices, the homeownership rate is unlikely to deteriorate meaningfully. Chart 7Better Situated Households Taking Advantage Of Competitive Rates COVID-19 may have claimed a staggering 33 million jobs and counting, but CARES Act forbearance will shield the most vulnerable households for the next twelve months, propping up their current rate of homeownership. Meanwhile, low mortgage rates create opportunities for better-situated households. Data from Corelogic suggest that millennials have driven the bulk of the uptick in mortgage applications (Chart 7). They are also the cohort most inclined to transition from renting to owning and their increasing access to homeownership these past few years suggests that their financial situation is not as dire as widely believed (Chart 8). Chart 8Millennials' Transition From Renting To Owning Low mortgage rates have also increased homeownership’s competitiveness relative to renting (Chart 9). This trend is unlikely to reverse in the near term. Eviction protection programs and rent forbearance under the CARES Act will only temporarily cap rent growth. Meanwhile, mortgage rates are set to remain competitive beyond the timeframe of this recession. Chart 9Owning Is More Attractive Than Renting... Low mortgage rates and relatively easy lending standards have prevailed since 2013 but home price appreciation has outpaced wage growth, denting housing affordability (Chart 10). While the tendency to build smaller housing units would contribute to decreasing median home prices at the margin (Chart 11), income growth will take a while to catch up. The labor market will have to tighten anew before income growth can revive. Chart 10...Even Though Homes Have Become Less Affordable Chart 11Is Smaller Becoming Better? Still, declining affordability has not prevented the homeownership rate from recovering to its long-run average. It may stand at a lower level today than it did in 2007 when it reached 69%, but it reflects sounder lending behaviors. Bottom Line: The COVID-19 crisis does not pose an immediate risk to the currently healthy level of homeownership. Better-situated households can take advantage of low mortgage rates but decreasing housing affordability will prevent homeownership from grinding meaningfully higher. Fading Wealth Effect Amid COVID-19 Consumers tend to spend more when the value or perceived value of their assets rises. Housing accounts for a sizable portion of homeowners’ equity, but the wealth effect of housing may have become less significant than most investors believe. The contribution to spending from housing wealth mirrors the decrease in housing as a share of households’ aggregate net worth (Chart 12). The latter now stands at 15%, way off its 1980s and 2006 peaks, while pension entitlements and equity and mutual fund holdings have filled the void, each accounting for a quarter of homeowners’ net worth. Chart 12The Wealth Effect Of Housing Is Decreasing... The wealth effect of housing remains positive. However, fluctuations in home prices are not evident to consumers in real time (Chart 13) and COVID-19 has precipitated the swiftest recession on record. The immediate or perceived future loss of employment and income are much more likely to drive consumption than home prices. Chart 13...And Is Unlikely To Influence Spending In A Pandemic Bottom Line: In a pandemic-induced downturn, home prices alone are unlikely to have a meaningful effect on consumption patterns. A Marginal Impact On Employment Overall housing-related sectors of the economy account for a marginal share of total employment. Construction activity makes up a mere 5% while related sectors including the sale and manufacturing of furniture, appliances and wood products, amongst others, chip in another 4.5% (Chart 14). On a rate of change basis, however, housing has been at the forefront. While the airline and leisure and hospitality sectors have been the center of attention in the past couple of months, construction has also suffered markedly. Total construction employment contracted by a third in April alone, behind only leisure and hospitality (Chart 15). Chart 14Housing's Marginal Impact On Overall Employment Chart 15Construction Was Highly Affected By COVID-19 A Phase I economic reopening will make room for activity in housing and many other sectors to resume and restore at least a portion of the jobs temporarily destroyed. The leisure and hospitality sector, however, is most likely to be the real game changer. 40% of the job losses so far have been in this single sector. While restaurants will be able to resume partial activity under Phase I, traveling is unlikely to return to normal for some time, even after a vaccine is mass-produced. It took several years after 9/11 for individuals to feel safe traveling again and for air traffic to reach its pre-crisis levels. Bottom Line: Although housing employment has been highly affected by COVID-19, it accounts for a small share of nonfarm payrolls and a pickup in this sector is unlikely to have a meaningful impact on the overall unemployment rate. A Significant Source Of Global Steel Demand A revival in housing activity is more likely to significantly impact commodity prices than the overall unemployment rate. Homebuilders are a key driver of lumber demand and construction activity accounts for half of the demand for steel and copper (Chart 16). The US is the largest net importer, making it a heavy player in the steel market, but its influence on copper prices is dwarfed by the demand stemming from Asia. Chart 16A Revival In Construction Would Boost Demand For Commodities Putting It All Together Over the past seventy years, housing has accounted for a steadily decreasing share of the economy and homeowners’ net wealth. In the absence of excessive lending and overbuilding, its ramifications for employment, consumption and the rest of the economy should remain muted in this crisis. BCA researchers tend to leave the thorough bottom-up analysis to professional stock pickers and instead focus their attention on the fundamental 30,000-feet top-down macroeconomic perspective. Although we do not expect overall home prices to contract drastically, “location, location, location” has always been real estate’s modus operandi. We would note that home prices in cities like Las Vegas or Orlando with economic activity tied to tourism, arts and entertainment, restaurants and recreation might be disproportionally affected by COVID-related externalities. It is too early to assess whether the widespread social distancing measures will result in lasting structural changes on society, housing preferences and the way we conduct business. There is sound basis, however, to hypothesize that cooped-up city dwellers might find suburbs and satellite cities to be more attractive going forward, and that lasting work-from-home arrangements will enable them to make that life-style change.   Jennifer Lacombe Associate Editor jenniferl@bcaresearch.com Footnotes 1 The housing start data is seasonally adjusted. Starts averaged 1,466 million in 1Q20 and 1,443 million in 4Q19 meaning that a quarter of these projects actually started in 1Q20 and 4Q19 (367K and 361K starts, respectively). 2 Please see US Investment Strategy Special Report titled "Housing: Past, Present And (Near) Future", published November 19, 2018. Available at usis.bcaresearch.com. 3 Please see US Investment Strategy Special Report titled "How Vulnerable Are US Banks? Part 2: It’s Complicated", published April 6, 2020. Available at usis.bcaresearch.com. 4 Until the NBER makes the official designation, our working assumption is that the current recession began in March.