If a Greek default is not enough for the compulsive speculators out there, as a reminder today we have that all important February NFP number release, which on one hand we have ADP as indicating in line with expectations of a +210,000 print, on the other we saw both Gallup, Initial claims and the ISM as well as various diffusion indices as pointing to a weaker print. Here is Goldman, which has come in slightly below expectations, with a forecast of 200,000 offset by a further reduction in the unemployment rate to 8.2%. Of course, as we noted last month, once the US participation rate hits 58%, the unemployment rate will actually mathematically go negative. And strangers years have happened in an election year... From Goldman: "We expect tomorrow's employment report to show solid nonfarm payroll growth of 200,000 in February after 243,000 in January. Although unseasonably warm weather should again boost payroll growth in February, we expect a moderation in the rate of job creation due to (1) a likely payback in manufacturing employment; and (2) mixed labor-market news since the last report. Uncertainty around the extent and timing of the weather effect and manufacturing payback suggest risks are probably tilted to the downside of our forecast. We expect the gain in employment to push down the unemployment rate by 0.1 point to 8.2% in February."
Full note: February Payroll Preview
The labor market recovery has gained momentum in recent months, as nonfarm payrolls rose by 243,000 in January and by an average of 200,000 over the last three months. We expect another strong employment report for February (released tomorrow). Specifically, we expect a payroll gain of 200,000 and a dip in the unemployment rate to 8.2%.
The labor-market indicators received since the January employment report point to solid job creation, but at rates below those in January. These include:
1. Steady claims. Initial jobless claims were essentially flat in between survey weeks in January and February (down from 355k in January to 352k in February) suggesting that layoffs have remained low.
2. Weaker ISM employment indexes. Both the ISM manufacturing and nonmanufacturing employment indexes weakened in February, down from 54.3 to 53.2 and 57.4 to 55.7, respectively. The nonmanufacturing index is still at a very healthy level in historical terms.
3. Flat job advertising. Total online advertised job vacancies--as reported by the Conference Board's Help Wanted OnLine Index--rose slightly in February, while the number of new ads has basically been flat since September.
4. Solid ADP. The ADP measure of private employment increased by 216,000 in February--in line with consensus expectations. The ADP report has been a noisy forecast of nonfarm payroll employment gains, but the February print is broadly consistent with a solid yet slightly weaker February payroll number.
Beyond these key indicators, we see two special factors at play in the February employment report:
1. Manufacturing payback (negative). Manufacturing employment rose unexpectedly by 50,000 in January--one of the strongest prints in years. The size of the gain, however, appears out of balance with its trend of around 25,000 in late 2011 and other manufacturing indicators in January, including the ISM manufacturing survey (the employment index of which weakened) and the manufacturing employment gain reported by ADP (16,000 in the revised February report). Some payback, or at least deceleration, in manufacturing employment in February thus seems likely.
2. Unusually warm weather (positive). We have shown in previous research that warm weather can have significant short-run effects on economic indicators, including job creation (see Andrew Tilton and Seamus Smyth, “What’s With the Weather?” US Economics Analyst, January 12, 2007). In particular, we found that the main driver of weather effects for employment data--absent snowstorms or hurricanes--is the deviation in heating degree days (HDD) from the seasonal norm. (As a reminder, the HDD index is a population-weighted measure of how far temperatures are below a benchmark level. For each day, the HDD index measures the number of degrees by which average temperatures fall below 65 degrees, and then totals them for each week or month. Therefore, colder temperatures imply higher values for the HDD index.) The exhibit below shows that we had warmer (and increasingly warmer) weather than the seasonal norm over the past few months. As we find that it is the change in the deviation that matters most for payroll growth, we believe that warm weather has boosted payroll growth over the last two months and will boost it again in February. Although there is a lot of uncertainty around the precise magnitude of the effect, our estimates point to a cumulative impact in the neighborhood of 50-70k as of January, and an incremental boost of 20-30k in February. These estimates are broadly consistent with the details of recent employment reports.
A Boost from Warm Weather
Taken together we regard the labor-market indicators as consistent with another solid employment gain in February. The weakening of some labor market indicators and the uncertainty about the timing and extent of the weather effect, however, probably mean that the balance of risks is tilted slightly to the downside of our forecast.
The household survey should also be solid and, in particular, we expect a decline in the unemployment rate by 0.1 point to 8.2%. We note two points here. First, we have received a few questions about Gallup's weekly survey of the unemployment rate, which reported a sharp increase from 8.3% in mid-January to 9.0% in mid-February. The issues with this survey are that (1) it is only available since January 2010; and (2) it is not seasonally adjusted. Even after seasonally adjusting the survey (with the use of simple monthly "dummy" variables) we found no statistically significant predictive power for forecasting the unemployment rate. [ZH: Well of course not, the "unemployment rate" is that determined by the BLS following Arima models, seasonal adjustments and Birth-Deathing... and yes, it goes as far back as US government propaganda existed - brilliant observation Goldman!]
Second, rounding of the January unemployment rate figure--which was a "low" 8.3% at 8.263%--lowers the bar for a decline of the unemployment rate to 8.2% in February. The intuition is that, all other things equal, a "low" unemployment rate read in one month is more likely to result in a decline in the rounded rate in the next month, and vice versa. In a simple test of this intuition we ran a regression that explains monthly changes in the unrounded unemployment rate with the extent of rounding in the previous month (that is, the difference between the rounded and unrounded unemployment rate). As one would expect, we found a highly statistically significant and negative coefficient (of around one) which provides support to the rule that "low" unemployment rate prints make a decline more likely in the following month--as we expect for February.