Goldman's Andrew Tilton dissects today's NFP number, explaining why if it is weaker than expected (+146k) it is due to snow, and why if it stronger than expected, it is entirely due to the "economic recovery" (and not Bernanke's hyperinflationary mandate). Bottom line: win-win, while North African (and soon Middle East) regimes: lose-lose.
Winter Weather Vs. Improving Fundamentals--A January Job Market Showdown
The January employment report is setting up to be a duel between improving job market fundamentals and nasty winter weather. The fundamentals have been strong, with accelerating GDP growth, signs of a pickup in job openings, and a declining trend of layoffs in recent months. But very cold and snowy winter weather—including a “major” snowstorm in the January payroll survey week—likely kept some people from getting to work, suppressing the payroll count.
- We continue to estimate a gain of 175,000 in nonfarm payrolls for January. The underlying trend is very likely stronger than this—or will be soon—but bad weather will mask some of that improvement. We expect the unemployment rate to tick back up to 9.5% after December’s very sharp decline. Overall, market risks are probably skewed to the side of a “strong employment” reaction—weaker numbers may be downplayed as the result of weather, whereas a strong report is likely to be viewed as evidence of an improving trend.
- Tomorrow’s report will also feature the annual “benchmark revisions” to the employment data; the Labor Department’s preliminary estimate is for a 366,000 downward revision to the level of payrolls in March 2010, implying that job growth over the previous year was overestimated by about 30,000 per month. Typically, revisions for subsequent months are directionally similar but smaller in magnitude.
The January employment report is setting up to be a duel between improving job market fundamentals and nasty winter weather. In general, labor market fundamentals have looked quite strong in recent months. But very cold and snowy winter weather—including a “major” snowstorm in the January payroll survey week—likely kept some people from getting to work, suppressing the payroll count. Below we take a look at each of these forces in turn.
On the fundamental side, good labor market omens continue to accumulate:
1. Accelerating economic growth. Real GDP growth picked up to 3.2% in the fourth quarter, the second consecutive acceleration. The composition of the report was very encouraging as well; one way to see this is that nonfarm output rose 4.5% annualized in Q4 (helping productivity growth to a solid performance in a report earlier today). Generally, employment growth follows GDP growth with a short lag of one to a few months, so accelerating GDP growth should translate into somewhat faster employment growth in fairly short order.
2. A steady trend down in layoffs. The four-week average of new jobless claims reached 413,000 in the January payroll survey week, down from an average of 465,000 through the first three quarters of 2010. Adjusted for the size of the labor force, new claims have fallen to mid-1990s levels, the onset of a period of very strong employment growth.
3. An increase (possibly large) in job openings. The Conference Board’s monthly report on job advertising showed an enormous surge in open positions. The 16% month-on-month gain in new online job advertisements was the largest in the six-year history of the series. However, this must be taken with a grain of salt given that the report included significant revisions to historical data (apparently due to a new process to remove duplicates and inauthentic ads), and given that the underlying non-seasonally adjusted data is not available (despite our requests) for analysis. The Monster employment index was directionally consistent but much less spectacular, posting a 3-point gain after seasonal adjustment (slightly over a 2% increase).
4. A robust ADP report. Though we were rightly suspicious of the large gain in December’s ADP report (see “Growth is Accelerating—When Will Employment?,” US Daily, January 5), its January reading should not suffer from the “purging” issue that concerned us last month. (Through most of 2010, ADP missed to the low side; adjusting for its average error last year, the gain of 187,000 private sector jobs forecasted by the ADP report would imply 246,000 private payrolls.) This month the issue is different: ADP’s employment count is based on the number of people on payrolls, regardless of how many hours they worked, so it does not exhibit a significant snowstorm effect. The Labor Department report only counts people as employed “who received pay for any part of the pay period that includes the 12th day of the month” (for more details, see http://www.bls.gov/ces/cesfaq.htm#scope2), so hourly employees who didn’t work at all during this time due to weather would not count as employed in the official data whereas they do in the ADP report.
5. Better survey data. Both of the monthly surveys conducted by the Institute for Supply Management (ISM) in January featured gains in indexes of employment. In fact, the employment component of the ISM manufacturing survey rose to its highest reading since 1973 (61.7, from 58.9 in December). The nonmanufacturing survey’s employment index rose to 54.5, the highest reading for this survey since May 2006. Less importantly, the Conference Board’s survey of consumers saw an improvement in perceptions of job availability, with the difference between the share of consumers viewing jobs as “plentiful” versus “hard to get” falling to -38.2%--a dismal level, but the best reading in 20 months.
If the fundamentals were the only thing we had to worry about, this month’s payroll call would be fairly simple—better than the last report, probably much better. But as so often happens, there’s a complicating factor. While the fundamentals look encouraging, January’s weather was terrible:
1. Cold weather. Winter temperatures can have a significant impact on employment in seasonally sensitive sectors (for a primer on the effects of weather on the economic data more broadly, see “What’s With the Weather?” US Economics Analyst 07/02, January 12, 2007). January was much colder than the seasonal norm—for the month as a whole, essentially tied for the coldest January since 1994. However, the payroll report measures month-on-month changes, and December was substantially colder than normal as well (the coldest in a decade, in fact). Adjusting for the timing of the employment survey and other factors, we estimate the incremental impact in January at perhaps 10,000-20,000 jobs. This should show up in weaker-than-usual growth of sectors like construction.
2. “Major” snowstorm. The Midwest and Northeast were hit by a large snowstorm late in the payroll survey week. This caused at least a few firms to reduce work hours and likewise kept at least some employees from reaching their workplaces. When we published our original payroll forecast, the National Oceanic and Atmospheric Administration (NOAA) had not yet issued a rating for the storm; it now has a preliminary categorization of “major” (the middle of five rankings), bigger than we expected. If we include a variable for snowstorm intensity in our models of employment, we find that a storm of this magnitude has an average impact in the neighborhood of 40,000-50,000 on reported payroll employment. We emphasize that this is quite uncertain, as we don’t have many data points (large snowstorms during the payroll survey week) to go on. Any weather effect should be fully unwound the subsequent month (i.e. if January were 50,000 below the underlying trend, February should be 50,000 above it), assuming of course that we don’t see another large snowstorm next week. Most of the snowstorm effect occurs in the construction and leisure/hospitality (hotels and restaurants) sectors, so job gains or losses in these two sectors, which have averaged -5,000 and +19,000 respectively over the past three months, will help us gauge the extent of any weather effect on the report.
When all is said and done, we haven’t changed our preliminary forecast for the employment report. We expect an increase in nonfarm payrolls of 175,000, with the unemployment rate at 9.5% (giving back a small part of its large drop in December), and average hourly earnings up 0.1%.
Why haven’t we changed our forecast this week? We’ve gotten better-than-expected news on the fundamentals (particularly online job advertising, the ADP report, and the ISM employment indexes), but worse-than-expected news on weather (the NOAA’s quantitative ranking of the snowstorm suggests a considerably bigger impact than we’d thought at first), and we think these are at least partly offsetting. So there is a lot of uncertainty, but it runs in both directions. In any case, in terms of market impact, we view the risks as skewed to the side of a “strong employment” reaction—weaker numbers are likely to be downplayed as the result of weather distortions, whereas strong numbers are likely to be viewed as evidence of an improving trend. Note that barring a sharp shift in the fundamental data or another big snowstorm, the February employment report is likely to look very strong, as it should benefit from mean reversion on the weather front.
One final consideration in Friday’s report: the Labor Department will indicate the extent of its annual “benchmark revision” to nonfarm payrolls. Each year, the Labor Department adjusts the level of payrolls the previous March to reflect a near-comprehensive count of unemployment insurance records up to that point. Last October, the Labor Department estimated that it would have to revise down the level of March 2010 payrolls by 366,000. A downward benchmark revision would in turn result in a downward revision to monthly changes since the previous benchmark in March 2009 (if the adjustment turned out to be 366,000, then the average monthly change between March 2009 and March 2010 would be revised down by about 30,000). The Labor Department will also adjust monthly changes since March 2010 to reflect its assessment of what the comprehensive data imply for more recent employment growth; typically these revisions are directionally similar but smaller in magnitude.