The worst, for the US economy, is over.
That's the conclusion from Goldman's chief economist Jan Hatzius, who writes that "the US economy now appears to be through the trough." Hatzius points to the following chart to show that "a range of timely new big data measures indicate that activity has started to pick up in recent weeks as some parts of the US have begun to ease lockdowns."
Hatzius explains that his optimistic take is based on "the release of the Q1 GDP and April employment data, insights from new big data sources, and the first glimpses of the reopening process, we now have a clearer view of the impact of the virus shock on the US economy and the path forward."
Here are some more details on how Goldman estimates the hit to real-time GDP using real data:
We begin by estimating the first-round virus hit to GDP through its impact on consumer services, manufacturing, construction, intellectual property products, government spending, and other minor GDP components. Throughout, we make use of new alternative data sources. Big data is especially useful in the current cycle, both because far more of it is available today and because the huge swings in the economy make the noise in these datasets more tolerable by comparison.
To estimate the decline in consumer services spending, we use the official consumer spending data from the Bureau of Economic Analysis (BEA) and two alternative data sources, consumer transaction data and location data. Our primary source of consumer transaction data is Second Measure, and we also consult similar spending data referenced by BEA economists. Our primary source of location data is FourSquare, and we also consult other mobility data provided by Google and SafeGraph.
The official BEA data are only available through March. To use them to forecast April, we make the rough assumption that the economy adjusted to a new lockdown level of spending in mid-March and remained at that level in April, which implies that the April month-over-month growth rate should be roughly the same as the March growth rate. The two alternative data sources already cover all or most of April.
For each category, we use the estimates from the three data sources, shown as blue bars in Exhibit 2, to form a judgmental estimate of the decline in spending in each category by April relative to the pre-virus level. Our judgmental estimates, shown as red diamonds, also take into account the timing and specific provisions of lockdown orders in the US as well as some special features of the BEA’s methodology.
Despite its newly-found optimism, Goldman further cut its economic outlook, and after "revisiting our GDP and labor market forecasts", the bank now estimates that "the level of US GDP declined by nearly 18% in April (vs. 13.4% previously)", and forecasts an even slower rebound, namely "quarterly annualized GDP growth rate of -8.2%/-39%/+29%/+11% for 2020Q1-Q4 (vs. -8.2%/-34%/+19%/+12% previously)."
In year-on-year terms, Goldman expects -12.6% in Q2, -7.3% in Q3, -5.4% in Q4, and -6.5% for 2020 as a whole, "all slightly below our prior forecasts."
Exhibit 8 shows our estimate of the virus impact on the level of GDP by component. The solid red line shows the total impact of the virus hit, while the dotted red line shows our previous estimate of the total impact for comparison. The orange line shows our GDP forecast in levels by adding potential growth to our estimate of the virus impact. Our forecast implies that real output will return to the pre-virus level by late 2021, but a roughly 3pp output gap relative to potential will remain.
Of course, what's a forecast without some version of a V-shaped recovery:
As Exhibit 8 shows, we strongly expect the level of output to recover from here, for several reasons. First, authorities are likely to continue relaxing lockdown orders and social distancing is likely to diminish. Second, businesses, individuals, and government officials will find ways to adapt to minimize the economic costs of social distancing while keeping virus spread under control. Third, improvements in treatment later this year should reduce fear and increase willingness to be around others. Fourth, wider antibody testing should allow those who are hopefully now immune to resume normal activity
Worse, Goldman now expects the headline U3 unemployment rate to peak at 25% (vs. 15% previously) and U6 to peak at 35% (vs. 29% previously).
We have made three key changes to our previous forecast. First, we now expect more job losses in total. Our best guess of job losses in the month through the April survey week is 30mn, the decline in employment plus the increase in the employed-but-not-at-work category in the household survey. Second, consistent with the somewhat more V-shaped growth path, we now expect job losses to be more front-loaded and job gains to be somewhat quicker under our baseline assumption that reopening goes smoothly enough to permit the rapid rehiring that usually follows temporary layoffs. Third, we now expect more job losers to be classified as unemployed, as was the case in April, rather than as not in the labor force, where a larger share appeared in March.
Under our new forecast path, the U3 unemployment rate peaks at 25% (vs. 15% previously), the broader U6 underemployment rate peaks at 35% (vs. 29% previously), and total labor market slack peaks at just under 30% (vs. 23% previously), as shown in Exhibit 11. We expect the unemployment rate to stand at 10% by end-2020 (only slightly higher than our previous estimate), still as high as at the peak of the Great Recession, and at over 8% by end-2021.
Goldman's conclusion on the unemployment rate: "A return to the pre-virus rate is likely years away." We wonder if this means that Hatzius is one of those "rich guys" that Trump was railing against this morning.