The March unemployment rate will be released Friday, and unlike previous releases this report is likely to be outdated as soon as it is printed. According to the Bureau of Labor Statistics, February had 9,035 unemployed and a labor force of 404,494 in North Dakota. That was an unemployment rate of 2.2%. The report only covers events from a single week, the one that includes the 12th of the month which is again why it is horribly outdated. North Dakota did not report a case of COVID-19 until March 11 and so the bulk of business changes occurred after the survey date.
Add to this that the labor situation in North Dakota has always been difficult, mostly due to a labor constrained economy. This typically meant more jobs than people to fill them. With distancing guidance in effect and closures of businesses there are not the same number of openings and we have seen increased layoffs in the state. Add to that a real question about the ability to even apply for openings that may still exist and you have a recipe for a really high number.
As an exercise for my class I made them forecast the unemployment rate for a state of their choice. I looked at North Dakota and the situation is not pretty. Let’s start with a few assumptions and some discussion of the data.
I looked at five unemployment claims reports from March 7 to April 4 to gather initial unemployment claims. That total by the way is 35,772, but it clearly includes days in April that would not count in March unemployment. Realistically then I excluded about 9,000 claims from the last report. I looked at a daily average and just excluded the four days in April. Clearly an assumption and not much to say it is right or wrong, and so we proceed forward. That adjustment gives me 26,533 claims for unemployment in the month. That is quite a bit.
Let’s add another assumption: the labor force is not changing. The lack of mobility due to distancing and closures gives people few options to move to other jobs at this time. I suspect the number cannot actually change much right now. Historically, which is a risky reference when there is no precedent in history, we have not seen enormous labor force fluctuations in North Dakota even in response to economic crisis.
Let’s further assume that nobody who was unemployed from the February data was able to get a job. This is likely the worst assumption of the lot, but if true we get a theoretical maximum for North Dakota unemployment. We would take the February number and add the total for March to get 35,568 unemployed. With the labor force from February that translates to an unemployment rate of 8.79%, or nearly quadruple the February value.
Now it seems unlikely that nobody gets rehired, especially for those in the February unemployment data. With that in mind I looked at reducing the early unemployed in the data sample in 10% increments. What I found was that for about a 10% reduction in the early unemployment, assuming the later really have no option to find work, we see about a .22-.24% reduction in the unemployment rate.
If we reduce our sample to the start of the month with the reporting restrictions the unemployment rate should only climb to around 2.5%. While accurate given the reporting structure I worry that this such a horribly outdated number will give the wrong signal about the impacts of the virus and related policies.