October 24, 2017
Two major hurricanes recently hit the U.S. mainland, leaving considerable destruction in their paths. Hurricane Harvey made landfall in Texas on August 25 resulting in severe damage along the Texas coast, deluging the Houston area with heavy rains and extensive flooding. Houston is the largest metropolitan area in Texas and the fourth largest in the country, with a population of 6.5 million people. Shortly thereafter, Hurricane Irma reached the Florida Keys on September 10th and quickly enveloped most of the state by the following afternoon. Ahead of the storm, Florida officials ordered 7 million of its residents, about a third of the state’s population, to evacuate their homes and by Tuesday, two days after the storm hit the state, nearly 3 million people were without power. The storm eventually dissipated after dumping heavy rain on Georgia and other neighboring states. While it is too early to assess the damage, recent estimates could place the number near or even higher than the damage caused by the record-setting Hurricane Katrina in 2005.
These severe weather events, coupled with the negative national employment numbers for September (the first since September 2010), raise the question of how costly weather may be to the U.S. economy. Rebuilding storm-ravaged areas can boost an economy, but the shutdown of businesses for extended periods of time could have devastating effects on the economy, particularly when critical commodities like gasoline and petroleum products are in short supply.
Even if we are patient enough to wait for the GDP numbers for the quarter encompassing September to be released, they will not tell us the net impact of the storms, since we somehow need to isolate the effects of the hurricanes from other events that occurred during those two months that may have affected the economy.
The Bureau of Labor Statistics, through its monthly Current Population Survey—the same survey used to estimate the nation’s unemployment rates—offers another option. They ask the 60,000 households surveyed each month two questions related to weather. The first reflects the number of workers (both full-time and part-time) who were absent from work for the entire reference period because of bad weather. The second reflects the number of full-time workers who worked fewer than 35 hours during the reference week because of bad weather. The bad weather could either prevent workers from getting to a business that continues to be open or cause businesses to close and workers would have no work to go to that week. The reference week refers to the week each month that includes the 12th of that month.
The timing of the weather event relative to the reference week is critical for measuring the effect of weather on employment. For example, in August, the week including the 12th began on Sunday the 6th and ended on Saturday the 12th. Hurricane Harvey did not make landfall until nearly two weeks later, and its effect would not be recorded in the August survey. Hurricane Irma, on the other hand, hit the Florida coast on September 10th, which was the beginning of the week that included the 12th, so that much of the missed work days would be captured by the September survey. It may also capture those individuals whose employers were still closed or their circumstances prevented them from returning to work due to Hurricane Harvey.
Combining the responses to the two weather-related questions shows a spike in September, more than likely due to at least Hurricane Irma, if not the aftermath of Hurricane Harvey. Figure 1 shows an abrupt increase from 471,000 workers in August to 4.4 million in September. This accounts for 2.9 percent of household employment. Roughly a third of the responders reported missed work and the other two-thirds reported reduced hours for full time workers.
September 2017 recorded the largest number of workers affected by weather during the four-month period from July through October during which hurricanes are most likely to hit the U.S. mainland, since the BLS began asking the question in 1976. As shown in Figure 2, the next largest number of workers affected during hurricane season was September 1999, when Hurricane Floyd hit North Carolina and continued up the eastern seaboard. Hurricane Floyd hit on Thursday September 16th, which was included in the reference week that month. The hurricane was one of the largest on record and prompted massive evacuations from Florida to Long Island. Nearly 3.5 million workers were affected, mostly due to a reduction in work hours for full-time workers, accounting for 2.6 percent of employment that month.
Unfortunately for tracking purposes, not all hurricanes cooperate in hitting landfall during the reference week. A case in point was Hurricane Katrina, which holds the record as the costliest hurricane in U.S. history, with a damage assessment exceeding $108 billion in 2005 dollars and a tragic death toll of nearly 2,000. It first hit landfall on August 25, 2005 and then quickly skipped across southern Florida to restrengthen in the Gulf before moving onto the Louisiana shore. It finally dissipated on September 1, after moving all the way to Lake Ontario and beyond. Regardless of the devastation, the hurricane hit after the August reference period and while cleanup and rebuilding took years in various places along its path, it was nearly another two weeks before the reference period in September. Therefore, only a minimum number of workers were recorded as affected by the storm—only 421,000 in September. Two additional powerful hurricanes followed Hurricane Katrina that month, hurricanes Rita and Wilma, but neither one hit during the reference week. Nonetheless, their aftermaths probably accounted for the 467,000 workers who reported that weather affected their hours in October. The numbers for September and October were both well above the monthly averages for those two months.
While hurricanes are often tragic events and capture the public’s attention, the largest number of workers affected by weather are recorded in the winter months. Figure 3 shows the average number of workers and the percentage of employed workers who missed work because of bad weather, as recorded by the Current Population Survey. Note that the average number of workers recorded in January is more than four times the average for September. The largest number affected by weather, according to this measure, occurred in January 1996. During that month, 9.8 percent of all employed workers were affected by what became known as the “Blizzard of 1996.” The storm started on January 6 and by January 8, much of the region was under up to 4 feet of snow with strong winds creating whiteout conditions. The reference period for January captured the brunt of the storm, beginning on Sunday, January 7, at the height of the blizzard. A similar weather event happened in January 1982, when extreme cold enveloped much of the eastern half of the country, affecting nearly 9 percent of employed workers. Even with our current focus on hurricanes, the effect of Hurricane Irma was rivaled by a late-season winter storm in March of this year. Winter Storm Stella hit the upper northeastern part of the country from March 9 to March 17, dumping up to 3 feet of snow in the hardest hit areas. As shown in Figure 1, nearly 3.3 million people missed some work, which accounted for more than 2 percent of employment.
While winter storms still occur frequently, their effect on days missed of work is not nearly as great in recent years. Between 1976 and 1999 the number of times in January and February in which more than 2 percent of employment was affected by weather was 12; since 2000, the number of times is 5.
Should the loss of 33,000 jobs in September be blamed on hurricanes? Obviously, the severe weather had some effect, but does it account for the swing in employment from a gain of 169,000 in August to a loss in September? Probably not, or at least not all of it. In fact, the correlation between monthly employment change and the number of workers who missed work because of weather is only -0.02, which suggests virtually no simple relationship. Therefore, while weather events can have devastating effects on property and lives, they seem to have much less of a lasting effect on employment.
 This is the same method used in “Work absences due to bad weather: analysis of data from 1977 to 2010,” Issues in Labor Statistics, Bureau of Labor Statistics, Summary 12-1 / February 2012, by Mary Bowler. All data used in this note are from the Bureau of Labor Statistics, including the two series from the household survey on weather-related work absences. Information on weather events is taken from National Oceanic and Atmospheric Administration (NOAA).
 In addition, total nonfarm monthly employment change fell from 196,000 in August 2005 to 67,000 in September, 84,000 in October and then 341,000 in November. See “CPS Program Response: The Current Population Survey Response to Hurricane Katrina,” Monthly Labor Review, August 2008, pp.40–51.