Tuesday, April 28, 2015

The Real Impact of Weather on Economic Growth

When the rate of economic growth slows down, particularly in the first quarter of the year, the mainstream media seems to focus on the "fact" that bad winter weather has had a negative impact on consumer spending which is responsible for about 70 percent of GDP.  A recent study by Justin Bloesch and Francois Gourio at the Federal Reserve Bank of Chicago looks at the connection between the fairly bad weather in the first quarter of 2014 and the slowing of the economy in the same quarter.

In case you've already forgotten, here is an example of the type of story that circulated in the mainstream media when Q1 2014 GDP growth stumbled:

Among the issues that connected the weather to the stalling economy were stories about employees who couldn't get to work, construction sector supply chains that were interrupted and households that delayed making retail purchases because they were spending more on heating costs and repairing storm damaged homes.

Here is a figure showing six economic indicators and how they were impacted with each of the three dots representing December 2013, January 2014 and February 2014:

While there was clearly a somewhat of a slowdown, it was unclear whether the slowdown was due to winter weather or another unrelated underlying trend.  At the time, the cause of the slowdown was particularly important because the Federal Reserve Open Market Committee had decided in December 2013 that it would start to taper its purchases of assets because the economy was showing some strength.  The challenge facing the FOMC was to disentangle how much of the economic weakness was weather related and how much was caused by outside forces.  This is more difficult than one might expect given that aggregate snowfall and temperature data are area weighted (i.e. they represent a geographic area) rather than being population weighted.  In other words, if winter temperatures are much colder than normal and snowfall amounts are much higher over less populated areas, the potential impact of harsh weather on the economy will be much lower than if the colder temperatures and increased snowfall amounts take place in a more populated region.  The goal of the research in this paper was to provide us with more robust statistical evidence regarding the effects of the weather on economic activity.

Let's start by looking at the weather side of the equation to see whether the winter of 2013 - 2014 was actually harsher than normal.  The authors' source of weather measurements data was taken from the U.S. Historical Climatology Network.  This data set has daily measures of temperature, snowfall and total precipitation from 1200 weather stations throughout the United States over the past 65 years.  The authors then calculated the monthly deviation or the difference between the monthly average temperature and the long-run normal temperature.  The temperature data is then averaged across all stations in a state and then aggregated the state data (weighted to their employment) to regional (i.e. Midwest, West, Northeast and South) and national weather indices.  The authors then did the same thing to the snowfall data.

These two figures show the entire national record of both temperature and snowfall deviation  from the late 1940s to 2014:

It is quite clear that on a weighted national basis, temperatures were lower and snowfall amounts were higher during the winter of 2013 - 2014.

Now, let's look at how much of an impact that the weather had on economic activity using the authors' state-level model as shown on this table:

1.) At its peak, the weather contributed negatively to the growth of non-farm employment over the period from November to March to the tune of -0.04 percent in the month of February or about 50,000 to 60,000 jobs.  By March, there was no impact of weather on non-farm employment and by April and May (plus 0.03 percent in each month), the weather effects on non-farm employment were completely reversed.

2.) The weather effect on the unemployment rate was negative in both November and December (-0.02 and -0.03 percent respectively and was positive in February, pushing the unemployment rate up by 0.06 percent.  By March, the impact of weather on the unemployment rate was back to zero percent.

3.) The peak impact of weather on the number of new unemployment insurance claims took place in February when new claims were up by 0.98 percent.  New claims dropped by 0.07 percent in January and 0.31 percent in March, again, showing that the weather effect on new U.I. claims was short-lived at best.

4.) At its peak, weather contributed negatively to housing permits during the month of February when the number of new permits declined by 1.04 percent.  Once again, by March, new housing permits had risen by 1.11 percent, completely reversing the losses of the previous month.

5.) The weather effect on housing starts was negative in February, dropping by 0.67 percent.  By March and April, this weather-related decline had been reversed when housing starts rose by 0.22 percent and 3.35 percent respectively.  It is interesting to note that the weather impact on housing starts was actually more significant in November 2013 (-1.71 percent) than it was during the depths of winter.

In addition, when using the national model, the authors found that the results were less clear.  The number of hours worked fell by 0.60 percent in February of which 0.19 percent is attributable to the weather.  Utility production grew by 3.3 percent in January of which 0.52 percent is attributable to the weather.  

The authors' conclusions show that weather does have an impact on economic activity, however, in the case of the winter of 2013 - 2014, the effect was very short-lived, generally negatively impacting economic growth for little more than a month.   In addition, the effect of weather is not very large, even though the winter weather in 2013 - 2014 was relatively severe when compared to historical records.  The authors' also note that the timing of the decline varied for many of the economic indicators with some indicators declining in December, others in January, February or March or a combination of months.  This tells us that the bad weather during the winter does not account entirely for the weak economy during that period and that other factors must have been at play, factors which include a correction in inventory levels and the impact of foreign trade. 

1 comment:

  1. Years ago a Seinfeld episode was centered around the idea of producing a television show about nothing! Sadly, in many ways this is the direction America has moved towards when it comes to measuring our economic growth. We have allowed numbers that mean "nothing" to seep into how the gross domestic product (GDP) is calculated all in an effort to create the illusion of growth.

    The first comprehensive set of measures of national income was developed by economist Simon Kuznets who in 1934 told the US Congress the formula was problematic. Bottom-line in the words of its creator, "The GDP framework is more or less an empty abstraction devoid of any link to the real world." More on what many see as an important number in the article below.