Friday’s Jobs Report: What Can It Actually Tell Us?
Friday’s jobs report will provide the last snapshot of the labor market before the election on Tuesday. Everyone will want to rush to create a narrative about where the economy is heading into a momentous election off of the jobs numbers. Unfortunately, there is a nonzero chance that the numbers are impacted by the horrific hurricanes that hit the United States at the end of September and beginning of October (Helene and Milton). Workers who weren’t paid during the survey period due to work disruptions won’t be counted as employed, and workers and businesses may be too busy dealing with the aftermath of the storms to respond to surveys. So where should we look in the data to try to see if the hurricanes had an effect – and how low could the numbers go? Spoiler alert: the payroll growth number could be extremely low and there is a nonzero chance of a negative read.
First, Jed Kolko has previously found that three industries (construction, mining and logging, and leisure and hospitality) account for most of the variance in weather-affected job growth. Those industries have accounted for about 15 percent of job growth over the last three months, for an average of 27,000 jobs per month. If job growth in those industries is noticeably below this recent trend, this could indicate that any potential slowdown in job growth may have been due to Helene and Milton. But those industries may understate the impact given the widespread destruction.
Second, we could try looking at which industries are more concentrated in Florida and North Carolina. Note that this gives us a slightly different look than the first approach – while some people in these industries under normal circumstances might have been able to work from home or otherwise accommodate the weather, given the extent of the storms, it is likely that normal working conditions were hurt for a wider swathe of people than would be expected during “usual” weather disruptions. Overall, the distribution of industries looks similar to the national numbers – with Floridians and North Carolinians slightly less likely to be employed in manufacturing, education and health services, and government than workers nationally, and slightly more likely to work in leisure and hospitality, professional and business services, and trade, transportation, and utilities. That being said – the distributions are similar enough that, apart from weather-affected industries above, it is unlikely specific industries will be able to tell us much on Friday. A slower reading in leisure and hospitality might be due to the hurricanes…or just due to luck of the draw. (We do get state specific numbers in a few weeks, although caution is always advised when using the state-by-state employment estimates).
Finally, we can look to historical patterns to help us out. In September and October 2005, job growth was only around 75,000, compared to around 200,000 over the previous few months. However, the initial estimate for September was even lower – at a loss of 35,000 jobs – and in October – adding 56,000 jobs. Assuming that the impact of Helene and Milton was similar to Katrina, that means we could assume the job growth could be negative on Friday even if it is later revised upward. (Note that BLS has historically attempted to account for such disruptions, see here for their explanation of how they handled Katrina). On the other hand, unemployment insurance claims have not moved up as much around the timing of Helene and Milton as they did around Katrina (unemployment insurance claims rose by about a third in the aftermath of Katrina, compared to only about 13 percent in the aftermath of the recent hurricanes), which would imply a low, but not negative number.
In addition to the hurricanes, the Boeing strike is likely to affect the jobs numbers (although not the focus of this piece). While that will hit manufacturing hardest, there could be spillover effects into other industries as well.
In conclusion, it is highly likely that the jobs numbers on Friday give us relatively little information about the state of the economy and much more information about how the hurricanes impacted workers. If voters want to think about the state of the economy, they would do much better to look at the job market over the previous twelve months than at one-month of hurricane-affected data.