Employment & Productivity

“The boss told me, I’d get paid weekly…

…and that’s exactly how I’m paid.” Johnny Paycheck

If it’s so hard to find workers…

The Bureau of Labor Statistics announced in August that real hourly earnings were down 0.1% for all employers in July (+0.3% in earnings +0.3% increase in CPI for urban consumers, and rounding—more on that in a bit), and real weekly earnings were down 0.3%, adding in the 0.3% decrease in the average workweek.

Over the year, real hourly earnings were up 1.3%, but the workweek was down 0.3%, and average weekly earnings were up just 0.8%.

For production and non-supervisory workers, real hourly earnings were down 0.2% in July (+0.2% in earnings and +0.4% in the price index for Urban Wage Earners and Clerical Workers), and weekly earnings were down 0.5% in the month, adding in the decline in the workweek. Over the year, real earnings were up 1.6%, but a 0.9% decline in the workweek spiked that down to +0.7% in weekly wages.

It’s tough getting used to a falling workweek—it was so easy just to leave 34.5 hours circled on the forecast spreadsheet—but the declines seem to be real and persistent, and not just coming from the sectors one would expect, like retail, which has been falling for decades. The Bureau of Labor Statistics has, of course, added in specific sectoral detail to their releases over time and, as one would expect, the goods-producing sectors were the first to come online.

We started the graphs below in 1965 for consistency’s sake, though manufacturing goes back all the way back to 1941 (and some sectors don’t begin until 1972). World War II really shows up in the early years of the series, when the workweek exceeded 45 hours, but it fell back towards 40 with demobilization. It’s too bad we can see what other sectors looked like back then.


Please note the difference in hours worked in the graphs: the goods sector shows no long-term decline, leisure & hospitality is low and weakening, and the split between wholesale and retail trade widening, for obvious reasons.

The all-workers series starts in 2006, so we didn’t use it for graphing, but we did compare recent performance of the all workers and production work weeks. For private service-providing workers, the production side lost 0.3 hour in 2014 but all workers did not, meaning supervisory hours were up, but recently they have moved together, although production workers’ week is 6 minutes longer.

Production workers hours are more jagged in the logging and mining sector and have seen some declines that don’t show up in all hours. (Logging and mining isn’t graphed, because things were getting crowded.) Also, production workers’ hours, at 47.2, are below their 2014 peak of 47.9, while all workers’ weeks are longer than they have been since 2006. In manufacturing, production hours’ slide from 42.4 to 41.5 since April 2018 is a bit steeper than all worker’s decline, but in retail trade, oddly, production workers’ hours have risen 8/10s to 30.2 while all workers fell from 31.8 hours to 30.7.

In leisure & hospitality, all workers are half an hour above their prior trough while production workers are below, and in bars & restaurants production workers are now slipping in a modest jagged way, but still up about half an hour from the 2010 trough, while all workers are slightly down.

The two series travel together in construction, and in education & health, although, you guessed it, health care hours are stronger than education hours. One could make a “we don’t need no education” crack about that, but since there are a lot of people working for those “under federal investigation” for-profit outfits on that line, we’ll raise a glass instead.

Using a long line graph obscures recent movements, so we’ll note that over the year total all-worker private workweeks are down from 34.5 to 34.3 hours, and within that goods production is down 0.4 tenths of an hour, construction down 0.3 over the year and down 0.2 over the month, manufacturing down 0.6 over the year and 0.3 over the month. Retail was steady over the month, at 30.7 hours, which is down 0.4 from last year, and education and health are flat for the year and month, but down 0.1 from earlier in 2019. Leisure & hospitality is down 0.3 over the year, and transportation and warehousing down a full hour to 38.1 hours. Manufacturing overtime for both durable and non-durable goods continues to fall.

If it is so hard to find workers, and firms are holding on to their current employees, many of whom want to work longer hours, why are workweeks falling?

by admin· · 0 comments · Employment & Productivity

Benchmark Blues: BLS to Cut 0.3%, or 501,000 jobs, from 2019 levels

The extrapolation methods used by the Bureau of Labor Statistics in producing their monthly estimates (their word) of NonFarm Payroll (NFP) growth can obscure the magnitude of cycle turns, which is why it is important to pay attention to the annual benchmark, derived from the Unemployment Insurance filings mandated by federal law that cover 97% of the NFP universe.

The rule of thumb at the BLS is that if the benchmark falls between +/-0.2%, the average of the last 10 years, everything is copacetic, but if it exceeds that there is real information there. This morning the BLS released the preliminary benchmark for 2019 and, unfortunately, there is information there. The overall employment level is slated to be taken down by -0.3% or 501,000 jobs when it is made formal in January 2020, and in the private sector -0.4%, or -514,000 of the jobs previously estimated will be benchmarked away.

Largest losses are in logging & mining, -2.2%, or a scant -16,000 jobs, leisure & hospitality -1.1%, a not-so scant -175,000 jobs, retail trade -0.9%, or -146,400 jobs, professional/business services, -0.8%, or -163,000 jobs, and wholesale trade, -0.6%. Transportation & warehousing will be revised up 1.4%, or about 80,000 jobs, information 1.2% or 33,000 jobs, and government 0.1%, or 13,000 jobs. That’s it for the plus signs.

As you can see on the table, this is both out of trend, and the largest negative benchmark since the 2010 decline in the aftermath of the great recession.

Dwindling Labor Share

Here we revisit a familiar topic: the decline in the labor share of national income. We were prompted to revisit by a recent post to the St. Louis Fed’s website with the provocative title “Capital’s gain is lately labour’s loss” (Anglo spelling in original). It draws on the work of Loukas Karabarbounis and Brent Neiman (KN), which we’ve also discussed, though mostly in passing. KN’s data runs only through 2012; the St. Louis Fed post draws on their in-house FRED database to update it through 2017 for five major economies.

The declining labor share is interesting for several reasons. For decades, most economists assumed the share to be constant, which makes it far easier to develop economic models of production functions and economic dynamics. But the labor share is not constant. In their examination of 59 countries with at least 15 years of data between 1975 and 2012, BN found 42 with a declining labor share, measured against corporate value-added. A major reason is the declining cost of investment goods; the St Louis researchers present a series showing a near-relentless decline in the price of capital goods vs. consumer goods since 1948, with an acceleration in the 1980s. (The average decline from 1948 through 1979 was 1.6%; from 1980 to 2016, 2.6%.) That makes it easy to substitute capital for labor, to the detriment of labor’s share.

The extension of the data for the five years doesn’t change the fundamental story. The labor share in the five economies shown in the graph on the top of p. 7 was little changed between 2012 and 2017, despite sharp declines in the unemployment rates in most. You might think that a tightening labor market might boost labor’s share, but that hasn’t happened.

As the graphs above show, the declines happened to different degrees and over different intervals for the five countries shown. The declines range from 3 percentage points in the US to 8 in Canada; expressed as percentage (not point) declines, they range from 5% in the US to 11% in Canada (with the other four countries not far behind). Remember, it had been a well-established axiom in economics that this was not supposed to happen.

BN find that most of the decline in the labor share has happened within industries, so it can’t be explained by compositional changes such as the shift from manufacturing to services, which, among other things suggests things other than globalization are at work (given the varying exposure of different industries to international competition). And they also find a decline in the labor share within China, which makes it an unlikely culprit for the decline in the labor share in the richer countries.

They conclude that the decline in the price of investment goods accounts for about half the decline in labor’s share. An interesting question is what accounts for the other half. They don’t examine the effect of labor market deregulation and the declining power of unions, but those seem like worthy avenues of investigation, as they have been in other research pieces.

Longview: ECI

You might think a sub-4% unemployment rate would be nudging the employment cost index (ECI) higher, but it’s not.

In the first quarter, private sector compensation costs were 2.7% at an annual rate and were also up 2.7% for the year. That yearly gain was the lowest since the end of 2017. The softening was led by benefits, up 2.1% in the quarter and 2.4% for the year. Wage growth accelerated some, but still isn’t much above where it was at previous lows on the graph. And growth in overall compensation costs is also around the level of earlier lows. Wages remain the dog that didn’t bark.

by admin· · 0 comments · Employment & Productivity

NIPA review

There were some huzzahs over the first quarter GDP number. The 3.2% quarterly rate, while still below two of 2018’s quarters, was at the upper end of its recent range. And the yearly rate, also 3.2%, was the best in almost five years. But the quarter’s numbers do nothing to close the gap with the long-term trend, and the composition of growth was unimpressive.

As the graph below shows, actual GDP remains well below its 1970–2007 trendline, which is based on an average of 3.2% growth. Since 2019Q1 just matched that average, the trendline remained just as distant. GDP is now almost 17% lower than it would be had the 3.2% average prevailed after the Great Recession. That gap works out to a deficiency of almost $5,000 in per capita terms, which, given long-term relationships, works out to about $3,800 in after-tax personal income. Consumption is 16% below trend; nonresidential fixed investment, 15% below; residential investment, 40% below; and government, 20% below.

And important components of GDP sagged badly in the quarter.

Consumption contributed just 0.8 point to real GDP, less than half what it did the previous quarter, and well below the average for both this expansion and the average of all expansions between 1949 and 2007. The contribution of goods consumption was negative. Nonresidential investment was weak, with both equipment and structures contributing nothing, and only intellectual property making any contribution. Residential investment was a drag on growth. Imports were weak, which is actually a positive contribution to growth, since they’re counted as a subtraction from GDP, but it’s not a sign of strong domestic demand. Inventory investment—which was 0.6% of GDP in nominal terms, twice the average since 1980—contributed a hefty 0.7 point to the headline growth number.

But the headline number did look pretty decent.

by admin· · 0 comments · Employment & Productivity