Figures behind our “targeting the Black rate” essay

June 14th, 2020 at 9:18 am

Janelle Jones and I have a new piece coming out wherein we explain why and how the Federal Reserve should target the Black unemployment rate in setting monetary policy.

The first figure to which we refer is the share of quarters since 1972 (when the Black jobless rate data start) that the unemployment rate for different racial groups has been below CBOs estimate of the “natural rate.” Whites enjoyed full employment labor markets almost 60% of that time. The Black rate, conversely, has never fallen below the estimated full employment rate (which I’ve, for the record, long argued is biased up, meaning these figures are optimistic).

Source: BLS, CBO

The next figure relates to our discussion of who benefits most from tight labor markets. It shows that pre-crisis, the pace of nominal Black median weekly earnings surpassed that of Whites.

Source: BLS

Surprise! One report does not a new trend make but reopening may be occurring sooner than expected.

June 5th, 2020 at 11:23 am

Before even getting to the facts of the case on today’s very surprising jobs report, let me share a few insights.

–One jobs report does not a new trend make.

–Keynes was right.

–Economists are terrible at catching turning points.

–Don’t ignore levels for trends.

In one of the more surprising jobs reports I’ve seen, payrolls rose–as in, went up!–last month by 2.5 million and the unemployment rate fell from just below 15 percent to 13.3 percent (though, as I’ll show, racial disparities may be resurfacing). These monthly reports are always noisy, so we don’t want to completely rethink our priors, but expectations were for unemployment to shoot up to about 20 percent and jobs to tank by another 5-10 million.

So, what happened?!

Well, first of all, let’s start with some context. The payroll figure below shows that job losses in March and April summed to an unprecedented 22 million. Thus, even if this uptake pace continues, it would take a year to just make up lost ground.

Second, we should also recognize that 13.3 percent is recessionary-level unemployment, far above the 10 percent jobless rate peak of the last recession. Again, it will be a long time–I’d guess well more than a year–until we’re back to full employment. Also, if we include a group of workers who are absent from work and may well be unemployed, the jobless rate jumps to about 16 percent (note: with that same adjustment, April’s rate was 19.7 percent, meaning the rate fell even faster last month if you include these absentees).

Third, in my last month’s jobs report, I noted that the magnitude of the shock dampened the usual disparities between changes in white and non-white labor market indicators. Also, the fact that Black and Brown workers are disproportionately “essentials” is also a factor limiting the usual disparities we see in downturns, where non-white’s indicators deteriorate faster than whites. As the table below shows, today’s report may reveal those disparities resurfacing.

Fourth, we knew this was coming. It just may be appearing sooner than we, or at least I, expected. All forecasts show a bounceback in the second half of this year of GDP, jobs, etc, though the strength of the bounce is still contested. Again, if this sticks, it means that bounce started before we thought it would.

Fifth, as noted above, Keynes was right! There should be no question in anyone’s mind that the fiscal and monetary actions taken to ameliorate the downturn and hasten the recovery have been effective.

Turning back to today’s numbers, the table below suggests some gaps are developing in the Black and Hispanic unemployment and employment rate changes. Since February, for example, the Hispanic jobless rate is up 13.2 percentage points compared to 9.3 points for whites. To be clear, those are both huge spikes in unemployment over just a few months, but historical patterns predict that such diversion will continue.

Source: BLS

The next figure is designed to provide some context for understand where we are and the depth of labor market weakness that still prevails. It first shows how payrolls fell off a cliff as we shut the economy down in the pursuit of control the spread of the virus. The uptick, while very welcome, clearly doesn’t significantly change the picture. It would take almost of year of gains of May’s magnitude to get back to where we were.

Source: my analysis of BLS data.

But as the forecast line shows, just getting back to where we were–i.e., gaining back 22 million lost jobs–would be under-performing relative to the pre-pandemic trend. To get back to that trend would require an extra 3.5 million jobs.

Other revealing numbers from the report:

–The problem facing state and local governments is still as real as ever. Since February, their combined job losses are 1.6 million, 8 percent of their total jobs. These sectors desperately need fiscal relief, like by yesterday, or they’ll have no choice but to continue aggressive layoffs of essential workers.

–As Martha Gimbel pointed out on Twitter, the payroll jobs diffusion index, which tracks the percent of industries adding jobs, went from 4 percent in April and 70 percent in May, the biggest monthly jump on record. If that holds, it means a much larger set of both industries and firms is reopening sooner than expected.

–After growing almost 5 percent in April over March, nominal wages fell a percent last month. These weirdly large monthly changes are being driven by “composition effects:” many low wage workers left the labor market in April, kicking up the change, and returned in May, pushing it back down.

–Speaking of that, one number I’m tracking is the share of unemployed workers on layoff, meaning those who could be quickly called back to work as firms reopen or ramp back up. That share was a hugely elevated 78 percent in April, but was expected to perhaps fall sharply in May, as temporary layoffs became permanent. That didn’t happen, and the share remains elevated at 73 percent, implying a large share of the unemployed are still at least somewhat attached to work.

So, does this mean those of us who expected a much worse jobs report should sharply update our priors? Surely not based on one jobs report. That said, my sense is that there is real information in today’s report telling us that our aggressive fiscal and monetary responses to the sharp downturn have probably helped to pull the recovery forward. Even so, the return to full employment is many months, if not years away, unemployment is still extremely high, and millions of Americans are still experiencing deep and lasting economic shortfalls.

In other words, our countercyclical repair work is helping, but it’s far from complete.

The US job market catches the virus and crashes.

May 8th, 2020 at 10:24 am

Due to the shutting down of the American economy to control the spread of the coronavirus, the bottom fell out of the job market last month. I’ve been writing up monthly jobs reports for decades, and I’ve never seen anything remotely close to this. Employment gains that were made over almost a decade vaporized in two months.

Payrolls collapsed by 20.5 million in April, by far the worst month on record for a data series that begins in 1939. Combining March’s losses of 870,000, revised from -701K, payrolls are down by 21.4 million over the two months. This takes the level in April—131.1 million total, nonfarm jobs—down to the lowest level since February of 2011, meaning in two months, the job market shed almost a decade worth of employment gains (see figure).

In the last recession, the worst month for payroll job losses was 800,000, and the totality of jobs lost was around 9 million. In other words, we just recorded two months of jobs losses that were more than twice that of the whole of the recession formerly known as “great.”

The unemployment rate also hit an historical peak of 14.7 percent, but even that elevated value was biased down by 6 million labor force exits. If all the exiters were actively unemployed, the rate would have been 18 percent. Underemployment, a broader measure of labor market slack that includes millions of part-time workers who want full-time work, rose to a record high 22.8 percent last month, as the number of involuntary part-timers almost doubled to almost 11 million.

The reliable pattern in downturns is for African American and Hispanic conditions to deteriorate faster than that of whites. But over the past two months, the changes by race were roughly comparable (though worse for Hispanics) as shown in the table below showing how much unemployment rates have risen and employment rates have fallen since the crisis began by for workers. There are at least two explanations for this unusual, relatively equal pattern. It may be that because Black and brown workers are disproportionately in essential services, their jobless rate was slightly insulated. For example, one of the only sectors that added jobs last months was jobs in “warehouse clubs and supercenters,” which disproportionately employ non-whites, at least in non-managerial positions.

But more likely is the fact that force of the massive demand contraction is, for now, overwhelming the historical racial patterns. My strong prediction is that this will change in coming months and the usual racial disparities will resurface.

BLS does not break out underemployment by race. In a forthcoming paper with Janelle Jones, we estimate this rate, based on historical relationships. Based on that estimate, we predict black underemployment was 33 percent last month, the Hispanic rate was 32 percent and the white rate with 19 percent.

Jones and I argue that these quickly deteriorating conditions must be considered in light of the devastating, disproportionate impact of the virus on the black community, a function of the structural racism that is putting African African’s at uniquely high risk of illness and death. Brookings fellow Rashawn Ray summarizes this tragedy: “There is a saying — ‘When America catches a cold, Black people get the flu.’ Well, in 2020, when America catches coronavirus, Black people die.” Correlates of this outcome — what epidemiologists call comorbidities—include neighborhood density and safety, diminished health-care access, exposure to pollutants and more.

Even what looks like a piece of good news in today’s report is deceptive. Wages rose 4.7 percent over the month and 7.9 percent since last April, a notably higher growth rate than their previous trend of around 3 percent. But the spike occurred not because those who stayed on the job got big wage increases but because lower wage workers dropped out of the job market. Imagine a job market with three workers who earn $6, $10, and $20 per hour. The average wage is 36/3 or $12 per hour. If the $6/hour worker loses her job, the average rises to $15 per hour (30/2), a 25 percent increase. This is analogous to what happened in April.

One question today’s report begs is: what’s next? Is this the bottom of the labor market trough?

It almost surely is not. Forthcoming reports are unlikely to be this negative, but, even as some commerce is slowing coming back online, payroll numbers will likely carry a negative handle for months to come as layoffs exceed hires. Most forecasts, for what they’re worth in this unique and thus hard-to-model moment, have unemployment peaking in the third quarter of this year.

That said, one relatively hopeful number in the report is that out of 23 million unemployed, 18 million are on temporary layoff, meaning they still retain some connection to work. However, that connection will fray the longer the downturn last, yet another reason to implement for more rigorous testing and tracing than has heretofore occurred.

But for now, we’ve yet to see the bottom of this cycle. If anything positive comes out of the undeniable disaster that is this jobs report, it will be that it forces Congress to revisit the urgency they appeared to understand earlier in the crisis and quickly get back to administering the necessary economic relief. Just as critical, we need the Trump administration to finally set forth a coherent virus-control plan to get to the other side of the crisis, but I recognize that the likelihood of that occurring is diminishingly small.

At some point, a lot of economically vulnerable people will need to get back to work.

April 30th, 2020 at 2:17 pm

Yesterday, we learn the economy contracted in the last quarter at the fastest rate since 2008, when we were in what used to be called the Great Recession. As this decline captured only a tiny share of the time we’ve been in shutdown, it’s the tip of the iceberg that’s sitting atop an economy still in deep freeze. This morning, we learned that another 3.8 million people filed claims for Unemployment Insurance. That’s 30 million claims, which are a fair proxy for layoffs, in six weeks. In a month-and-a-half, we’ve experienced more than three times the layoffs we had in the whole of the recession formerly known as “great.”

The unemployment rate implied by these numbers is 18 percent, much higher than any previous peak.

Such numbers complement the info from a new NPR/PBS NewsHour/Marist poll that gives one a sense of pervasive hardship. As expected, the poll reveals a familiar partisan divide on Trump’s handling of the virus and the economy, but when asked about layoffs or reduced hours at work, about half of the respondents of all political stripes said they or someone they know has been hit by the shutdown. That’s up sharply from 18 percent in April. Lower-income persons, non-whites, and non-college graduates have been hit hardest, but the pain is widespread.

The poll also shows that despite the economic hardship, large majorities do not want to open up the economy too soon. Still, there’s a contrast that caught my eye. Over 90 percent said they thought it was too soon for large groups to attend sporting events, but a significantly lower 65 percent said it was too soon to get back to work.

Given that GDP is falling much faster in this quarter than last, that the jobless rate is probably heading for 20 percent, and that the bottom half of households in America have virtually no savings to fall back on, I suspect that this totally understandable reticence to mess with an invisible, potentially fatal virus could quickly flip. This is especially likely as some governors and White House officials decide that it’s safe to go back in the water, even while epidemiologists are shouting “not so fast!”

This impulse to get back to something resembling normality is also completely understandable. Neither the broad economy nor the people who comprise it–at least, those not deemed “essential”–are designed to freeze in place. Moreover, all the deep racial and economic imbalances that predated the crisis are playing out in ways that are as predictable as they are fatal, especially to African Americans. I know that many of those “liberate Michigan” protesters are motivated by shady, political forces. But there are many others who don’t just want to go back to work. Their livelihoods and that of their families depend on it.

For now, some of the fiscal measures we’ve taken, including expanded UI compensation and checks to households are helping. But the UI expansion runs out at the end of July, and as partisan politics appears to be reawakening from its very brief slumber, it’s no slam dunk that it either checks or plussed-up-UI benefits will get repeated (though they definitely should).

If I’m right that the share of respondents saying it’s too soon to get back to work falls precipitously with each new UI claims report, what must happen to accommodate their shifting views?

That’s easy: what’s now invisible must be made visible through testing at least one million persons per day as opposed to the ~200,000 or so we’re currently testing.

And that requires the Trump White House to stop talking about drinking breach, preening about their great work, and kicking everything important to the states. Simply put, they need to for once get ahead of this thing by taking charge on widespread testing and tracing. And they need to do so before majorities of people decide they can’t afford to shelter in place much longer.

The tip of the wave: Jobs report shows large losses, but predates the worst of it

April 3rd, 2020 at 9:34 am

Payrolls fell by 701,000 in March, their first monthly decline in almost 10 years, and the jobless rate ticked up to 4.4 percent (from 3.5) as the coronavirus and efforts to contain it pounded the U.S. labor market last month. Because of the timing in the surveys in this report, it only picks up the front end of tsunami of layoffs that occurred in the second half of March, when initial claims for Unemployment Insurance rose by almost 10 million, an increase most economists would have considered inconceivable before this crisis. But the report clearly identifies the tip of the wave.

The surveys were fielded in the middle of March, and thus better reflect conditions in the first half of the month, when containment measures were just taking hold. Commerce, travel, and broad consumer activity was slowing but hadn’t shut down as they would in March’s second half. Even so, the report is far more negative than recent vintages, and shows how remarkably quickly conditions have reversed in the job market.

For example, the 0.9 point increase in the jobless rate was the largest one-month increase since 1975; the 1.7 point increase in the underemployment rate, to 8.7 percent, is the largest increase on record since this measure was introduced in 1994. The Bureau reports that the “bulk of the increase in unemployment occurred among people on temporary layoff, which increased 1.0 million in March to 1.8 million.” Measures of labor market participation also fell sharply, a clear reflection of the reversal in labor demand. This shift is especially disheartening as prior to the virus, the tight job market was pulling typically left-behind workers into the job market. Such gains are quickly unraveling, a point I return to below.

As readers know, we typically feature our jobs smoother which averages monthly payroll changes over 3, 6, and 12 month windows in order to pull out the underlying signal. We print the smoother below, but it too is less informative this month, since a backwards looking average by definition downplays the sharp shift in trend that occurred in the past two weeks (to emphasize this point, we include a bar for the 701K loss in March alone).

A better, simpler approach this month is to plot monthly payroll levels, which show the sharp trend reversal in March.

Source: BLS

Hourly pay stayed on track last month, up 3.1 percent and beating inflation that’s been running just north of 2 percent, though price growth will likely slow (boosting real wage growth) due to very low energy prices. However, wage trends can be deceptive at times like this because of “composition effects.” For example, as more low-wage workers face layoffs relative to high-wage workers, this can show up as accelerating wage growth. I’ll try to parse out this potential bias in forthcoming reports.

Different sectors have different degrees of exposure to the jobs impact of the virus, of course. One way to think of this difference is that if you can draw a paycheck by clicking into Zoom meetings, you’re in a less exposed sector. So, restaurant, hotel workers, flight attendants and anyone in a face-to-face services (and their suppliers) has a much higher chance of a layoff relative to many in professional services like legal, accounting, or research. The food vendor who works at a professional sports venue is directly exposed; the team’s lawyer is not.

For example, employment in restaurants and bars fell by over 400,000 in March, a one-month loss of over 3 percent, by far the sector’s worse month on record. Conversely, employment in professional and technical services was up slightly in March, by about 6,000. True, that’s a weak month for the sector, and most sectors (outside of those that are directly responsive to containment efforts) are being hit by the sharp downturn. But magnitudes of losses will differ by exposure.

It is far from incidental, of course, that there’s an inequality divide implicit in that distinction. A useful analysis from the St. Louis Federal Reserve split workers by occupations into high and lower risk of unemployment. About half of the workforce fell into each category (to be clear, that doesn’t mean unemployment will hit 50 percent; not every exposed worker will get laid off). The analysis found that “the highest risk of unemployment also tend to be lower-paid occupations. The average annual earnings of the low-risk occupations is $64,600, about 75% higher than earnings in the high-risk occupations, at $36,600. This indicates the economic burden from this health crisis will most directly affect those workers who are likely in the most vulnerable financial situation.”

Source: Charles Gascon, St. Louis Fed.

We’ve never shut the U.S. economy down as we have done in response to the virus. This was a wholly necessary response to its threat, but it came at point when the labor market was persistently closing in on full employment, providing meaningful employment and wage gains for workers who are often left behind under more slack conditions. Much as full employment provides out-sized benefits for the most vulnerable workers, the reversal we’re now witnessing metes out the most pain on those same groups of workers. Many of these laid off workers lack paid leave and their savings rate is zero or negative. That is, they are the least insulated among us against this sort of sudden shock.

That is why our relief efforts must scale to the unprecedented size of the problem. Recent stimulus measures, with their emphasis on checks to household and expanded Unemployment Insurance, are a good start but we a) must ensure these measures are quickly implemented, and b) we will need further trips back to this well.

The risk at times like these–risks we’ve seen borne out in both the health and economic responses–is doing too little. As Dr. Fauci said the other day (paraphrasing): If you think I’m overreacting, I’m probably getting it right.