Buckle Up: This Recovery Could Be Bumpyadmin
“Until the public is confident that the disease is contained, a full recovery is unlikely.”
- Federal Reserve Chairman Jerome Powell
Federal Reserve Chairman Jerome Powell got a lot of blame for throwing a wet blanket on the rally in the stock market a few weeks ago with the Fed’s downbeat assessment of the economy and it’s recovery. He may end up being right after all.
During his testimony before the Senate Banking Committee last week he gave us a rough framework of the phases of the economy during the lockdown crisis and recovery. It looks something like this:
1) Shutdown: “First was the shutdown and we’ve seen what that would produce … we may be reaching a bottom on that now.”
2) Bounce-back: “We would expect to see large numbers of people … coming back to work during this second period, call it the bounce back or the beginning of the recovery.”
3) Regaining confidence: For sectors that demand close human-to-human interaction, see quote above.
The trouble is, confidence that the coronavirus is contained seems to be fading. And for good reason.
One of your authors received a version of the following chart from an astute client last week (hat tip MA), showing the correlation between restaurant traffic in Arizona and the spike in cases two weeks later. Here we lagged by two weeks the seven-day moving average of new cases per day in Arizona (in red) versus OpenTable’s measure of restaurant activity in the state (in blue).
Coincidence? Who is to say? But certainly worth tracking. Also note the stall in restaurant activity as the virus has ramped. We learned from watching this OpenTable data back in early March that people do a pretty good job of policing themselves, as most metropolitan areas saw massive drops in activity even before state-mandated lockdowns were put in place.
Looking at new cases per day can be problematic because if states increase testing, which they should be and are doing, the number of new cases per day should rise. Looking instead at the percentage of positive results from testing might give us a better indication of whether the rise in new cases is being driven by community spread.
We look at just that in the analysis below. We started by grouping states based on when they reopened restaurant activity. Instead of looking at when guidelines were changed for each state, which would have been tedious and not much fun at all, we instead categorized each state as “open” when its OpenTable reservations were better than -75% compared to a year prior on that date. It’s a rough proxy, but we like rough proxies. The chart below looks at positive test rates by state, broken out by when states reopened restaurants.
If you go back to mid-May, when the first states started seeing increases in restaurant activity, the positive test rates were very similar across all states. That is not the case today. The first thing to note is that we have two groups of states, both among the earliest openers, which are seeing increases in COVID-19 spread. These states are largely located in the South, from the Carolinas down to Florida, across to Texas and into the aforementioned Arizona. As we see clearly in the graph above, the increase in community spread looks like it starts roughlytwo-to-three weeks after reopening.
Many are calling this the “second wave” of COVID infections, but we hesitate to use that term. Most of these states are actually experiencing their first wave, which is starting later than the densely populated areas of the northeast which became the epicenter for the virus in the U.S. and which are opening more gradually. We’ve spent a lot of time looking at how the less-dense, car-focused areas of the country have outperformed the dense, mass transit-dependent areas. Will this analysis continue to be true going forward, or will we start to see new lockdowns and social distance enforcement in the coming months in these states? News of Florida and Arizona intensive care units reaching capacity is worrisome.
Echoing Powell’s sentiments, a lack of virus containment creates a ceiling for how quickly the economy can recover. Right on cue, on Friday Apple announced that it would be re-closing some stores due to COVID spikes. After the blockbuster May jobs report and more good news last week on retail sales, it has been looking like we were firmly in step two of the Fed chair’s recovery framework (i.e. the bounceback). But the new COVID spikes and a deeper dive into the data suggests that the path from lockdown to full recovery may not be linear.
Data on small business hiring from human resources software provider Homebase has been very useful for tracking the labor market and was certainly a good indicator that May would be a good month. But progress has been limited since then. Amazingly, fewer employees worked at the end of the third week of June than the second. If we break out the Homebase data using our reopened states framework we find that states that opened on May 15 saw growth in employee hours peak on May 15. States that opened on June 1 saw growth in employee hours peak on … June 1. Those that opened on June 15? You guessed it.
The slowdown in small business hiring isn’t a great sign. Perhaps as more states move out of the “Not Reopened” category we will see those areas take over the lead in the recovery. That bucket still includes places like New York and Massachusetts that have largely gotten the virus under control and are now beginning to reopen. As usual, we watch and wait.
Retail sales data provided another reason to be optimistic last week. Up 18% in May! Huge!
A month ago , as we were getting a clearer picture of how deep of a hole the economy is in, we wrote “sometimes, when the monthly changes are so big, it helps to think in levels.” We were speaking of big negative monthly changes back then. That framework is useful during the rebound as well. The chart below looks at the level of retail sales (blue), non-farm payrolls (red) and industrial production (green) all indexed to 100 in 2012 so they can all be on the same axis.
That big increase in retail sales is still impressive on this chart, as it looks like the start of a V-shaped recovery. It makes sense that the consumer is leading the way. Fiscal stimulus appeared to keep consumer expectations intact relative to current conditions, which bodes well for spending over the short-run. This is why an extension of enhanced unemployment benefits after they run out in July could be very important.
But even after that big jump, the level of retail sales remains only at the third quarter of 2017 levels. Even after better-than-expected increases in payrolls and industrial production (which rose 1.4% in May), those remain stuck at 2010-11 levels.
Let’s break down that big bounce in retail sales. Here are retail sales by category, showing the percentage change by month and the total change from pre-pandemic levels in yellow.
While every category saw gains in May, the levels relative to February remain drastically different. Only five categories are up since the pandemic began: nonstore retailers (online shopping), grocery stores (of course), building materials, auto part stores and hobby stores (sports, books, etc.). Meanwhile, some categories like clothing, restaurants, electronics and gas stations are still down 30% or more! Clothing store spending was only $8 billion in May, compared to a five-year average of $21 billion per month. The success of the five categories that are up since February shows households realizing they can still read, do yard and car work at home, but that does nothing for the restaurants and clothing stores still closed or severely limited.
This pattern in the retail sales data, where we see a good headline number with huge dispersion under the hood, is showing up in many places. The Industrial Production data we mentioned early, for one. The biggest increases in May were in those industries that have been the hardest hit and remain the lowest even after the rebound: transit equipment, clothing, auto products, durable consumer parts (things used to make computers, cars). Not much analysis is needed here – manufacturing remains in a rut.
And if chairman Powell is right, these rolling waves of new COVID infections will make the journey to a full recovery tempestuous at best.
The Week Ahead …
Data releases for May continue next week, highlighted by the publication of personal income and spending data on Friday. April’s report was one of the most interesting of 2020, showing a large spike in income due to one-time stimulus payments as spending fell off a cliff. May will likely show the opposite, as spending returned as seen in the retail sales jump, while the one-time income support payments wore off. The marriage of the two will be the personal saving rate, which set a record at 33% in April and is likely to settle lower in May.