UI Claims Versus Job Loss

The release of the Bureau of Labor Statistics state level jobs data for April provides an opportunity to see how reliable an indicator Unemployment Insurance initial claims are of employment loss for particular states. UI data for this purpose has the virtue of being timely, but is confounded by the varying UI laws, rules and practices of states which can affect the initiation of claims and the awarding of benefits.

The first chart plots, for each state, its percent payroll employment loss from February to April (the payroll week that includes April 12), against its UI initial claims made from March 8 (when they started to climb) through the week ending April 18, as a percentage of the February labor force. Each dot is a state (or the District of Columbia or Puerto Rico). There is an obvious relationship between the extent of job loss and the extent of UI initial claims, but there is a great deal of variation.

One UI initial claims number that has caught my eye has been Kentucky. It’s been running first or second for a few weeks. The reasons have not been obvious. It hasn’t been particularly hard hit by COVID-19. It doesn’t disproportionately rely on the industries most affected by the crisis. In fact, what we see with today’s BLS data is that its job losses are above average (15.9% job loss versus 14% average for the country) but it only ranks 14th among the states. Yet as of April it was second in UI initial claims.

It is evident that we should not consider UI initial claims data as a very reliable indicator of the relative level of disemployment occuring in states. As aluded to above, different states have different eligibility rules and application processes vary in how hard they are to navigate.

The second graph compares employment loss to the number who have actually received UI benefits, “the insured unemployed,” instead of inital claims, as a percentage of the February labor force (in this case for the week ending April 25). Given how much states vary in eligibility requirements it isn’t surprising that the relationship between job loss and UI receipt is an imperfect one. The placement of Kentucky, however, makes more sense here — right in the heart of the scatter at a 16% job loss and and insured unemployed at 12 percent. It suggests that people have been applying for benefits in Kentucky out of proportion with job loss but have only received benefits at a level generally consistent with other states.

Note that interactive versions of these graphs are available here and here. On those you can point at the dots and see which state they are.