Real-Time Data Shows That After Peaking In Late June, Consumer Spending Is Now Declining

When we last looked at real-time consumer spending data one month ago, we saw a stunning rebound in Bank of America credit and debit card spending trends, with total card spending ex-autos essentially recovering pre-covid levels by early June.

No doubt, a big part of this was due to the surge in Personal Income since the start of the current recession, which as we explained earlier was a function of the extremely generous fiscal stimulus which meant that on a per capita basis, claimants received roughly $788/week ($41k annualized) on average, well above the usual amount of roughly $300 in a normal labor environment ($15-$16k annualized).

Unfortunately, both this massive government handout and this impressive spending spree are now is coming to an end, and as JPM writes, in data through Sunday, July 5, the bank's tracker of spending by a panel of 30 million Chase credit and debit cardholders remains below its recent peak on June 22, and it appears to have flattened out at this lower level as COVID-19 spreads rapidly in some parts of the country.

What is notable, is that contrary to widespread anecdotes that sunbelt states have led the slowdown in spending, JPM notes that this modest pullback in nationwide spending has not been driven by a sharp pullback concentrated in states where the virus has spread rapidly, but instead by modest pullbacks that are widespread across states. On the other hand, this pattern raises the concern that behavior has not changed enough to stem the spread of the virus in the hardest-hit states.

To be sure, holidays like July 4 can make it difficult to distinguish signal from noise in high-frequency data, and some methods of calculating over-year-ago spending comparisons have naturally swung dramatically in recent days given the timing of the holiday. But even when looking through these swings, spending has begun to flatten out at a somewhat lower level than the peak seen on June 22.

JPM also points out that the pullback in spending since late June has been widespread across states. There is some correlation between the spread of the virus over the last two weeks and the pullback in spending, but the correlation has been modest so far. This pattern suggests that there are some cautious consumers in all states who have pulled back on spending as the virus has resumed its spread, but that the rapid spread of the virus in certain states has not produced a significant change in behavior by the entire population in these states.

To be fair, there are somewhat larger recent declines in spending in buckets likely to be most affected by the spread of the virus. For example, “card-present” (essentially in-person) spending has fallen more in Texas than in New York, and spending by Texas millennials—which had returned to year-ago levels by mid-June— has fallen more than by New York millennials. Still, the bank's economists are most struck by the relatively small differences in changes in behavior across these groups in recent weeks.

Meanwhile, and as we first pointed out last week, JPM has observed a correlation between spending levels and the subsequent spread of the virus three weeks ago, and it has remained strong over the last three weeks. Indeed, to date the bank finds that states with higher levels of spending—especially card-present restaurant spending—have seen more rapid growth of the virus in subsequent weeks.

To summarize, JPM has documented two important facts about the interaction between spending and the spread of the virus—higher spending levels have predicted the spread of the virus, but spending levels have not fallen much in the states where the virus has spread most rapidly recently.  These two facts raise the concern that behavior in the hardest-hit states has not changed enough to stem the spread of the virus going forward.

Still, it is possible that other behavioral changes that we cannot measure could stop the spread. As one example, it is possible that the relationship between restaurant spending and virus spread is just a correlation, and that it could possibly be driven by other behaviors like going to bars or parties which are correlated with restaurant spending. With bars now closed again in Texas, Florida, and southern California, it is possible that the spread of the virus will slow even as restaurant spending remains high. But the facts shown in JPM data do not present much reason for optimism.