US Watch – This time it’s different

by: Bill Diviney

  • The covid-19 recession is unfolding like no other – in its cause, speed, and depth
  • The unique nature of the downturn means the recovery is likely to be different, too
  • We expect unemployment to rise to levels not seen since the Great Depression…
  • …but it is likely that many of the newly unemployed will quickly return to work
  • There will, nonetheless, be significant lasting damage done to the economy
  • Forecast uncertainty is unprecedented; we present three plausible scenarios
  • Outcomes will depend significantly on pandemic control, and fiscal policy success (or failure)
US-Watch-covid-19.pdf (354 KB)

A policy-induced downturn

The covid-19 pandemic has led to a ‘sudden stop’ in economic activity of the kind unseen in modern times. As of writing, 316 million people (96.7% of the US population) in at least 42 states, three counties and 10 cities are under ‘stay at home’ orders. This is already leading to sharp declines in measured activity, with jobless claims rising at an unprecedented pace, and the New York Fed’s new Weekly Economic Index tracker pointing to an almost vertical drop in activity. As we will explain, the pace of contraction is likely to be followed by a similarly sharp rebound, although the net effect will still see unemployment well above pre-pandemic levels, and activity some 5.2pp below trend by the end of 2021. Indeed, while a significant part of the economy will recover as quickly as it declined, there will be parts of the economy that see longer-lasting dislocations due to bankruptcies and reduced aggregate demand from higher unemployment. As such, following the initial dramatic bounce, we do not foresee sustained above-trend growth until the latter part of 2021.

Our lockdown assumptions: A phased re-opening from mid-May

While stay-at-home orders are decided at the state and local level, president Trump has extended federal social distancing guidelines to at least 30 April, and many local orders are likely to extend beyond this date. By mid-May, we expect some phased re-opening of the economy. This will likely start with the lowest risk sectors in parts of the country where the pandemic has peaked, and where social distancing is more feasible and extra hygiene measures (such as mask-wearing) can lower the chance of transmission. For certain sectors in the services industry – notably eating and drinking establishments, and hosting of events – activity is likely to resume, but be depressed for longer, potentially beyond the summer. For instance, there might be rules on the number of people allowed in an establishment at a given time. Even if there are not, many individuals (particularly those more susceptible to severe disease) are likely to stay at home more.

Consumption the main driver of the downturn

Our starting point for analysing the impact is from consumption, which makes up 70% of US GDP, and is the part of the economy most affected by lockdowns. As a base case, we assume a 20% peak-to-trough fall in consumption, with the bottom occurring in April, and a gradual normalisation from May-September. This estimate takes full account of essential consumption that should remain unaffected by lockdowns (housing, energy, food, healthcare, etc. – around 60% of consumption), as well as consumption that is seeing substitution effects (more online shopping, higher grocery spending, restaurant takeaways). We also expect a significant fall in investment, although not on quite the scale of the Great Recession of ’08-‘09, as there will be some substitution effects (for instance increased healthcare spending, and investment by businesses to adapt, eg. to enable people to work from home).

Economy will recover, but still be 5pp below trend by end-2021

What is the net effect of all this on GDP? A deep recession is already underway, as has been made clear by the unprecedented surge in jobless claims in recent weeks. With lockdowns starting in parts of the US in mid-March, and activity already declining significantly in the run-up to then, we expect Q1 GDP to already show a contraction of 7.7% qoq annualised (this may not be apparent in the first estimate, on 29 April, but is likely to be revised lower later). In Q2 we expect the deepest contraction, by a dramatic 31.9%. Thereafter, we expect significant snapbacks in growth of 24.7% in Q3, and 11.8% in Q4. While these numbers might look optimistic, note that they imply GDP will still end the year 3.2pp below Q4 19 levels, and closer to 5pp below where GDP would have been were growth to have continued at trend in 2020 (assuming the Fed’s estimate of trend growth at 1.9% pa). Indeed, when quarterly growth numbers become this large, we think it is more instructive to look at GDP forecasts in levels, and in the recovery phase at how much of the ‘lost ground’ from pre-pandemic trend the economy has made up.

Our base case could be too optimistic, or pessimistic

Naturally, the longer lockdowns or other restrictions remain in place, the bigger the impact will be on GDP. We assume 6-8 weeks of lockdowns for most parts of the US with a phased easing of restrictions thereafter. We make no assumptions about possible treatments, vaccines, or renewed outbreaks. Our base case could, therefore, prove to be too optimistic, or pessimistic. In a more positive scenario, widespread testing might show significant parts of the population are already immune to covid-19, or new treatment options could become available enabling a more abrupt, less phased resumption of activity. In a more pessimistic scenario, lockdowns might need to last longer, or the re-opening might need to be more gradual.

Then there are the second-round effects, from increased unemployment, tighter financial conditions, and weaker confidence. These effects are hard to predict with precision, and depend significantly on the success or failure of fiscal and central bank policy in cushioning the impact and enabling a more rapid restart in activity. Unemployment has already risen by 17 million as of the time of publication, and is likely to see further rises in the coming weeks. Our base case is that unemployment peaks at a staggering 18% in May, before declining almost as rapidly as it rose for a time, but settling at a level well above the pre-pandemic low – we expect it to end the year at 9.4%, almost 5pp higher than February’s 3.5%. Such a dramatically higher rate of unemployment will inevitably weigh on consumption and slow the pace of recovery.

Most newly unemployed so far are furloughs

As with GDP, there are extremely wide uncertainty bands around this forecast. Aside from uncertainty around how controlled the pandemic becomes and the phasing of reopening the economy, there is significant uncertainty over how many of the layoffs that do happen prove to be temporary, and can quickly and easily return to their old jobs. Almost all of the 1 million layoffs in the March nonfarm payrolls data were ‘furloughed’ – or temporarily laid off. Indeed, the proportion of such furloughs in total unemployed is by far the highest on record, which goes back to 1967. This suggests employers are – at least for now – seeing this downturn as very much temporary in nature, and that workers can quickly return to their previous jobs once restrictions are lifted.

Fiscal and monetary measures a vital support, but how much they will help is unclear

We also do not know how (un)successful the $2.1trn CARES Act will be in incentivising companies to keep significant numbers of employees on payroll that would otherwise have been laid off. The longer restrictions remain in place, and the longer it takes to get fiscal support to companies in distress, the greater the permanent ‘hysteresis’ type effects will be on the economy. Meanwhile, although financial conditions have eased significantly in the past few weeks, conditions remain tight. The Fed has made a ‘whatever it takes’ type commitment to purchase as many government bonds as necessary, and could yet implement 1940s-style yield caps to keep rates low. These factors – combined with expectations for a recovery – should ease financial conditions over time. For the now, however, conditions remain drag on activity.

High frequency data will give a timely signal on the downturn and the recovery

Given the unique uncertainty we face in this crisis, high frequency data has become much more valuable – including jobless claims, transportation, and energy production – the type of data that is typically overlooked by economists because it is too ‘noisy’. Now, this data will be crucial in tracking the progress of the economy relative to forecasts, as it will give an early indication not only of the extent of the fall in activity we are seeing, but also the pace of the eventual recovery. The New York Fed has begun publishing a weekly index that combines a range of such data, and this looks to have been a good early signal of historical 4-quarter GDP growth (see chart on front page of this document).