Yesterday, out of left field, Goldman hiked its March NFP forecast from +175,000 to match consensus at +200,000. This is rather odd, considering Goldman's recent bearish spin on economic data. As it turns out the justification for this is not only to align with the trendline in ADP and claims data, but because now, suddenly, Goldman thinks that the 100,000 jobs boost due to warm weather, will not be unwound until April. In retrospect this makes sense: Goldman also recently gave up on the Fed announcing the NEW QE in April, as a result the next such opportunity will be June, which in turn means that a rapid deterioration in the economy will have to take place just before the FOMC meeting, rather than a gentle slowing down. Which is why today's NFP has now become a crapshoot, especially since it is still all in the seasonal adjustments. One thing is certain: the quality of jobs, as first suggested here [5], will continue to go down: because in an election year, one dilutes everything, up to and including jobs.
From Goldman Sachs:
Payroll Payback?
According to the National Oceanic and Atmospheric Administration, the 2011-2012 winter was the fourth-warmest ever recorded in the United States, in a series that goes back to 1896. Unseasonably warm (or cold) winters can have significant effects on the economic indicators. The main mechanism is that warm weather keeps activity in weather-sensitive sectors higher than “expected” by the seasonal factors, which are based on the average employment decline in prior years. So seasonally adjusted employment increases in the winter and falls back in the spring, when temperatures return to normal and the weather in any case no longer matters as much for employment.
The weather issue has received widespread attention over the past few months. According to the minutes of the March 13 FOMC meeting, “[s]everal participants noted that the unseasonably warm weather of recent months added one more element of uncertainty to the interpretation of incoming data, and that this factor might account for a portion of the recent improvement in indicators of employment and housing.” Our own research has also looked at the weather issue repeatedly.
Job Gains Skewed to (Normally) Cold States
How large has the boost been, and how much payback should we expect? One argument against a big effect is the observation that construction—which is generally viewed as the single most weather-sensitive industry—has not seen unusually rapid job gains in recent months. As shown in Exhibit 1, construction payrolls logged a decent gain in December and January, but gave back some of the increase in February and have generally not looked too different over the past few months relative to prior months.
However, simply looking at the raw change in national construction employment is not sufficient for gauging the weather impact. This is partly because the data are inherently noisy and subject to revision, and partly because we do not know what would have happened in the absence of the weather deviation from the seasonal norm. After all, it is possible that the failure of construction to add more jobs simply reflects the continuing weakness in home demand or other nationwide factors that have been weighing on the sector.
A much richer perspective on the weather impact is available from state-level employment data, which are published about 2-3 weeks after the national employment report. The state-level data are based on somewhat different samples and seasonal adjustment methods, and are not identical to the national data when aggregated across states. However, they are similar enough and allow us to exploit the enormous regional variation across the United States in estimating the impact of the weather on employment.
Exhibit 2 breaks down the United States into two equally sized sets of warm and cold states, as defined by the average number of heating degree days in the December-March period in a normal year. The entire acceleration in payroll growth in recent months has occurred in the normally cold states. These are the states where temperatures have been furthest above seasonal norms, and also those where the severity of the winter matters most for employment.
The state-level data also provide a somewhat different perspective on construction employment relative to the national survey. Exhibit 3 is identical to Exhibit 2, except that we focus on the construction sector only. It shows that there may have been a bigger pickup in construction-sector payroll growth than shown in the national survey. It also shows that the normally cold states have sharply outperformed the warm states in terms of construction employment gains.
Weather Has Boosted Payrolls Substantially…
So how large has the weather impact been over the past few months, and how much of a payback can we expect? To assess this, we turn to a slightly modified version of the state panel data analysis presented in the March 28 US Daily. We estimate the impact of the weather by regressing the state-level percent change in seasonally adjusted nonfarm payrolls on the change in the state-level deviation of heating degree days from the average level for that particular month over the prior four years. We include a set of state dummy variables (which allow for different average state employment growth rates) as well as a set of time dummies (which remove all aggregate variation in employment growth rates).
The bottom chart on the front page shows the results by summing the predicted impact across all 50 states. According to our analysis, the weather deviation has cumulatively boosted the level of employment since November by 116,000, with the largest month-to-month boost occurring in January and smaller boosts in December and February. This number is slightly above our previous estimates as we now use a more precise measure of state-level weather, as we focus on the number of heating degree days in the four weeks up to the survey week, rather than the monthly average.
…But the Payback Shouldn’t Come Until April
While the boost from the unseasonably warm weather has been substantial, we do not expect a meaningful payback in Friday’s employment report for March. The simple reason is that March was roughly as warm relative to the seasonal norm as February. As a result, the bottom chart on the front page still shows a (very small) positive weather impact for the March report.
Other indicators also suggest that the March employment report should look quite good. The ISM employment indexes for both the manufacturing and nonmanufacturing sectors rose in March and now stand at 56.1 and 56.7, respectively; online help-wanted advertising showed a healthy seasonally adjusted gain in March; and jobless claims have continued to trend down. On Thursday, we therefore lifted our March payroll forecast to +200,000 from +175,000 previously. We continue to expect a small drop in the unemployment rate to 8.2% from 8.3% as well as 0.1% gain in average hourly earnings.
However, we do expect a sizable payback for the weather boost in April and May. Assuming a return to normal temperatures in April, the model says that the drag should amount to 77,000 in April and 34,000 in May. Together with the fading of the inventory boost to GDP growth and the impact from higher oil prices, this is likely to result in a weaker picture of US economic activity than seen over the past few months.