Thanks to Harvey and Irma, payrolls fell last month, but underlying job market remains strong

October 6th, 2017 at 9:29 am

Payrolls contracted by 33,000 last month due to the impacts of hurricanes Harvey and Irma. The unemployment rate, which BLS tells us was not affected by the storms, fell to 4.2 percent, its lowest rate in over 16 years, and it fell for “good reasons” last month, i.e., not because discouraged workers left the labor force. In fact, the closely watched labor force participation rate rose to 63.1 percent, its highest level since March of 2014.

Thus, to evaluate the strength of the current US job market, look at the unemployment rate, not the negative payroll number. The former is on trend; the latter is a weather-induced outlier. Another important and strong indicator from September, also one that was unlikely to be influenced by the storms, was the healthy bump to employment rates of prime-age (25-54 year-old) workers, a closely watched indicator in this recovery. Overall, it climbed from 78.4 to 78.9 percent, the highest since July of 2008. For men, it went from 84.9 to 85.5, the highest since August 2008. For women, the employment rate went from 72.1 to 72.4.

Wage growth was above trend last month, as average hourly wages rose 0.5 percent over the month and 2.9 percent over the past year. This spike is also likely hurricane related, reflecting the fact that lower-paid workers tend to be the ones not paid when they can’t get to work, and thus they dropped out of the average wage calculations last month. As I’ll show below, the trend in wage growth remains slightly north of 2.5 percent.

More evidence of the storms’ impacts can be seen in restaurant employment. BLS points out that while jobs in food services and drinking places have been rising at a decent clip of around 25,000, last month they declined by 105,000. The Bureau reported that, “In this industry, a large majority of workers are not paid when they are absent from work. Hence, if these employees were unable to work during the September survey reference pay period because they had evacuated, or because their establishments were not open for business due to power failures or other effects of the hurricanes, they were not included on September payrolls.” This dynamic dampened the job numbers and boosted the wage results.

Thus, the impact of the storms are the most important message from this month’s report. The Texas and Florida hurricanes were clearly responsible for the negative low topline employment number, a decline that is not indicative of a sharply worsening trend in job growth. Prior to this morning’s report, the BLS pointed out that “about 11.2 million workers were employed in March 2017 in the FEMA-designated disaster counties and represented about 7.7 percent of national employment” (about 70 percent in FL and the rest in TX). Estimates suggest that employment growth last month might have been 100,000-150,000 higher had the hurricane not so severely disrupted commerce directly in Texas and Florida and indirectly in other parts of the country (note: Puerto Rico is not included in the monthly national employment report; the island is included in the state job report out later in the month).

Another way to show this negative impact is to look at the number of people whose absence from work was weather-related. The spike shown in the figure below is the largest in this series in 20 years, according to BLS.

Historically, severe weather disruptions akin to those in September are short-term, temporary events, and we expect the job numbers to bounce back up to their underlying trend of about 150,000 per month within another month or two (e.g., hurricane Katrina’s impact lasted about two months in the national data).

To get a better feel for the underlying trend, our monthly smoother takes averages over the last 3, 6, and 12 months. While the hurricane impact is part of the average in each bar, it is only 1/6th or 1/12th of the second two bars, which thus give a better impression of the trend.

Given the outlier data for September’s payrolls, this is a good moment to take a closer look at the statistical noise in the monthly job numbers.  This month’s smoother shows a new twist – the horizontal black lines depict the upper and lower bounds of the 90 percent confidence interval around the 1-month, 3-month, 6-month, and 12-month estimates.  Given how large the potential range is for 1-month estimates – BLS estimates that we had something in between 153,000 jobs lost and 87,000 jobs created in September – relying on longer-range estimates makes more sense.  As noted above, we’re still averaging about 150,000 jobs per month over the past year.

Wage growth remains more subdued than expected given the historically low unemployment rate. The trend lines below (six-month moving averages) show that wage growth clearly accelerated from 2 to 2.5 percent between 2015 and 2016, and since then has stalled out at around that level. Part of this is reasonably attributable to low productivity growth, but based on my analysis, even with low productivity growth, nominal wages should be growing at least half-a-percent faster right now.

In sum, the low unemployment rate reveals a solid job market that, despite large storms that were devastating for many, is still closing in on full employment. The trend in payroll employment, about 150,000 or so, is strong enough to continue tightening the job market, which, if it proceeds apace, should generate more wage pressures. That’s an essential dynamic to watch for, as many workers are only now beginning to reap the benefits of an economic expansion that’s been underway for nine years.

Print Friendly

16 comments in reply to "Thanks to Harvey and Irma, payrolls fell last month, but underlying job market remains strong"

  1. Tom in MN says:

    What role do you think Fed tightening has played in the slow down in hourly earnings growth? They seem determined to not let the economy do any catch up back to pre-recession trends.

    • Brett Showalter says:

      There has been no slowdown. The 2.5% trend has been established since late 2015 when the economy emerged from the recession back to “normality”. A 3.0%+ rate would be a sign of a overheating economy and best prepare for a future recession(probably driven by a correction in the secular bull market).

      Looks like NFP growth was about 250,000 without the Hurricane, which makes sense since the BLS “steals” from August to pay September. Wage growth was going to bounce as well, probably back up to 2.7%(which was the underlying trend).

  2. Smith says:

    It is fundamentally wrong to use the term “closing in on full employment” as if it were imminent. Here below is my argument that trends and historical indicate we are two years away from full employment.

    Following the recessions of 1990 and 2001, participation of prime age fell a years and then climbed up, taking six years before reaching previous level, or never in the case of 2001 since the 2007 recession began while still 1.5 percentage points short of 2000 peak. Previous recessions didn’t have this effect as participation either stagnated or continued to rise. Women’s continued expansion into the workplace could be a reason, and also explain that during a pre 1990 recessions, women may have had even more incentive to enter job market during the downturn offset effect of increased male unemployment. To show the gloomy side of things (which explains wage stagnation) if you take yearly averages, we’re at the 2008 level, and at the bottom of the post 2000 recession level, above the bottom of the 1990 recession dip (by .6 percent), but in a non recession affected year, you still have to go back to 1988 to find similar levels. From 1995 to 2001, prime age participation was about 1 percent higher. 2006 and 2007 also achieved 79.8 and 79.9 rates, while still falling short of 81 percent average of 1997 – 2001. This means it’s not unreasonable to assume 1 percent still left ready to be lured back, and possibly 2 percent. 1 percent is 1.2 million people. 100,000 new jobs a month are needed to meet population growth, so the 150,000 expected projection means 50,000 are left over each month to absorb part of the 1.2 million. That would take two years, as in 50,000 x 24 months.

    The mistake being made is not giving prime age workforce participation it’s proper due. Once you have, stagnant wages are easily explained by slack in labor pool hidden by categorization 1 percent prime age population as non participants. One can even argue that the participation rate is the more important indicator, since employment prospects have to be especially bad to not even consider looking for work.

    Also, it’s silly to say wages stagnate because of small productivity growth if productivity growth is mostly dependent on higher wages driven in part by a higher growth economy.

    Absent too, is any consideration that wages are predominately affected by labor supply, and employers are willing to give up revenue to maintain profits.

    In gathering data, I’ve noticed Fred workforce prime age participation data does not match BLS, not sure how significant this is, but would not affect my analysis either way (I used BLS numbers and FRED figures are even more favorable to my point)

    • Smith says:

      Crap, not enough proof reading again. Most importantly, the one place where a statement’s meaning is changed:

      Absent too, is any consideration that wages are predominately not affected by labor supply, and employers are willing to give up revenue to maintain profits.

      Left out all important ‘not’ as in time to stop thinking full employment has to lead to higher wages (not saying it can’t be an important factor, and sometimes predominate)

      First sentence, left out word ‘data’ from
      It is fundamentally wrong to use the term “closing in on full employment” as if it were imminent. Here below is my argument that trends and historical data indicate we are two years away from full employment.

      • Smith says:

        “fell a few years” in first sentence second paragraph, missing “few”
        “during a pre 1990 recessions” doesn’t the article “a”
        “it’s proper due” should be “its proper due”

    • Brett Showalter says:

      We are close to full employment. 4% is the full employment rate. adjust cohort size and it represents 5% in the 90’s expansion or 4.7% in the 80’s and 00’s expansion………….notice what didn’t happen in the 80’s. You keep on whining about prime age population, but that had to fall when the Boomers started leaving the index in 2001. Now with the millies coming it, it is pushing it back up, slowly, but surely. Each first quarter it will burst higher as they come in. Think about that for a second. Add 3 more year of millie influxes into that index and it will probably be over 80% again. Patience grasshopper.

      Matter of fact, look at FRB data for labor market tightness in August, we were as tight as the peak of the 1980’s supposed “boom” and a better economy than what peaked in 1979. Just further proof how bad the 1974-1995 era was. A bleak era of dollar devalued post-gold triggered inflation, a country that lost a war to a pip squeak and was emasculated, mass sexual perversion(that hit two peaks in 1979 and 89) and stimulant drug use that blows the so called opium shit out of the water.

      Since 1996, the US has had steady real wage growth and a boom in consumption like no other. Obviously there are problems in any era. The lack of big ideas and projects has been downer. Nobody wants to spend on risk anymore, including the voters own government, they elected. Health Care costs got out of control since 2002, but that is just rent seeking. Up to voters to put people in charge that will take care of that. They don’t. Their fault. But compared to the burned out shit hole of the 74-95? Things are all right man, just fine.

      • Smith says:

        1974 to1995 real wages went up, but from 1980 t0 2000 it was accompanied by sharply rising inequality.
        Since prime age population participation removes demographic effects, I fail to see any relevance to boomer or millennial generations. Prime age participation is just a percentage of the group total, 25 to 54 years of age, and matters not whether that group of 120 million is a a few million larger or smaller. Wage measures might be affected to small degree by age distribution, younger workforce, lower wage, but even that effect is minor when you do the math.
        Rust belt factory workers who still had jobs 1974 to 1995 would heartily disagree with your assessment of that pre-NAFTA, pre WTO China era, which is why you have Trump.
        From 1996 to 2000 there was real wage growth, but also tremendous growth in inequality. Since 2000 virtually no wage growth, including for college graduates. Even the top ten percent have gained little as most income gains accrued to the top 1 percent and top .1 percent.
        All these statements are backed up by solid data. Check State of Working America, BLS, EPI, FRED, Saez.

        Minor point, how come this blog lacks a automatic filter for cuss words?
        Also, the term you mean to use is opioid, not opium, though related (opioid being the synthetically produced substance produced by big pharma vs. our friends in Afghanistan)

        What is true about the 1974 to 1995 era is a level of urban crime now unimaginable in the now gentrified thriving revived cities. While some freakonomics minded economists attribute the turnaround to abortion, others give credit to changed policies, luck, changing economics, and the natural burnout of several associated plagues and movements.

  3. Wondering says:

    Has anyone ever tried to incorporate consumer debt levels into this idea of wage pressures through mathematical modeling? If so, are they using the right definition of consumer debt?

    One of the primary problems I see with the division between consumer debt measurements and capital investment is the student loan issue. I don’t know the standard accounting methods on this. It seems to me that some might want to model educational investment as capital investment, but that reality dictates that if that education is not leading to higher productivity jobs, then it should be recategorized as consumer spending (consuming information) for no economic benefit.

    • Smith says:

      No. I don’t think public school education of about $5,000 per pupil per year for 50 million children is in the same category of what we consider capital investment. That represents $250 billion a year. But then what about the food, clothing and shelter of children, an investment in future workers that also doesn’t pay dividends for 18 to 22 years later?
      Also college enrollment is 20 million, mostly 3/4 public, 2/3 4 year programs, 2/3 full time, average tuition $5,000 to $10,000 (not including room and board, but including $50,000 a year Harvard, Yale, etc averaged in). 20 million x $7,000 is another $140 million on education.

      • Smith says:

        $140 billion, not million. Hence in the neighborhood of $400 billion a year.

        • Wondering says:

          Thank you. Clearly students loans for college generate a lot of consumer spending, but on the loan side I believe this is probably categorized as investment rather than savings. What I’m trying to get at is that the portion that isn’t returning dividend to the students in higher wages is not investment but rather savings. This would cause a miscalculation and misunderstanding of the SI curve.

          • Smith says:

            Bad investments (college expenses not yielding expected dividends) are not the same as savings. Direct investment is usually spent, whereas savings (unless you are putting cash under a mattress) often yields to more demand (unless business borrows to pay off higher interest debt). The greater demand is because savings are leveraged by banks to loan more money than deposits.
            The problem with college loan debt is very much more straight forward. People in debt have less money to spend. It matters little whether the debt was incurred buying a car and clothes or paying for necessities, or paying for exorbitant salaries of college presidents and bloated college admin staff and building programs.
            Meanwhile, the oversupply of college graduates has a disproportionate effect on wages of everyone. Why? Big theory here with simple mechanism:
            The same level of total unemployment has different effects depending on whether higher income jobs have a surplus of job seekers. The history of most labor markets was the opposite until a surge in college graduates during the 1970s (spurred partly by Vietnam War college deferments, previous college affordability, and prosperous baby boomer parents. As the cohort of boomer graduates matured and became an ever more significant factor, and as the percent of graduates was sustained, 1/3 of the entering labor force since 1970s vs less than half ten years earlier (not checking these stats right now, but should), the effect became more pronounced. Labor surplus at higher incomes pushes everyone’s wages down, higher income due to surplus, lower incomes as higher displace the less credentialed. Same level of unemployment produced different wage effect because shortage of higher income labor would create upward pressure there, and also pull up wages of lower income workers as some found employment in higher income market.

            Form New York Fed:
            Are Recent College Graduates Finding Good Jobs?
            2014 Jaison R. Abel, Richard Deitz, and Yaqin Su
            “While many of these graduates will eventually find employment or transition into higher-skilled jobs as they gain experience and as the labor market normalizes, recent research suggests that those who begin their careers during such a weak labor market recovery may see permanent negative effects on their wages.”

          • Smith says:

            See also
            Here’s Exactly How Much the Government Would Have to Spend to Make Public College Tuition-FreeAnd the grand total is…Jordan Weissmann Jan 3, 2014

      • Smith says:

        Oops, this says public school costs $640 billion a year, and averages $12,000 a student.
        That is nearly 5% of GDP and financed mostly by flat wealth tax on the middle class (property tax which picks up about 3% of GDP).

Leave a Reply

Your email address will not be published.