GOLDMAN: A Little More Than Half Of The Economic Slowdown Is Due To Bad Weather
Key economic data points like employment, housing, and retail sales have been deteriorating in the U.S. over the past two or three months.
The big question on the minds of market participants is how much of the weakness is due to unseasonally harsh winter weather, and how much of a boost in economic activity the U.S. will see once the weather abates.
In a report discussing the effects of recent weather conditions on economic activity, Goldman Sachs economist David Mericle estimates that a bit more than half of the slowdown has been caused by transitory weather-related factors. He expects GDP to take a 0.5 percentage point hit in the first quarter, but expects a subsequent 0.5-0.75 percentage point boost in Q2.
A measure of population-weighted heating degree days was well above average in January and February.
So too has the National Oceanic and Atmospheric Administration's NESIS Snow Score, which combines meteorological measurements with population information to give an indication of a storm's societal impacts.
"Snowstorms have been more frequent than usual: of the 52 snowstorms ranked as high-impact on the NESIS scale since 1956, five have occurred this winter," writes Mericle in the report. "And as with temperatures, snow too has been most severe in February."
In the report, Mericle explains how he quantifies the weather's economic effect this winter in terms of Goldman's proprietary Current Activity Indicator (a proxy for economic growth) and discusses his findings:
We attempt to quantify the impact of recent adverse weather using a bottom-up estimate of the hit to our CAI. We start by building simple models for each of the CAI's 25 components. Each model explains the growth rate of a given indicator using the recent trend of the dependent variable, the deviation of heating degree days from seasonal norms during the current and last three months, and the deviation of the cumulative NESIS score from seasonal norms during the current and last three months. We drop snowfall from models in which it does not have a statistically significant effect. Finally, we impose the restriction on each model that the long-run impact of weather deviations on the level of activity must eventually be zero.
We confirm our finding from earlier research that colder weather has a negative and statistically significant impact on most major indicators, and we also find a statistically significant snowstorm effect on some. We find that weather tends to have the largest effect on housing indicators such as starts and construction; employment indicators such as payrolls, hours worked, and jobless claims; and spending measures such as retail sales. We find a somewhat more modest impact on business surveys, and only a small impact on measures of consumer confidence.
Exhibit 4 shows Goldman's estimates of the effects of the harsh winter in December and January on various economic indicators, as well as forward assumptions for the weather-related effect on February and March data.
"We assume, conservatively, that the heating degree days deviation in the second half of February will be only ¼ that of the first half of the month, and will then return gradually to 0 by June," says Mericle. "While subject to considerable uncertainty, our models anticipate a significant bounce-back effect beginning in March and continuing in the following months as weather conditions normalize."
"The chart suggests that a bit more than half of the ¾-1pp decline from the pre-December trend is a result of weather effects," he writes.
"The January aggregate weather effect is smaller because the weather effects for the employment indicators are positive, reflecting that month's unusually warm reference week. The chart also points to bounce-back from the weather hit in March and Q2. While the exact timing of this weather bounce-back effect is highly uncertain, it is likely to provide a substantial boost."