Flows From Unemployment to Employment & Extended Unemployment Insurance

I was not surprised to learn that conservatives assert that the recent increase in the growth of employment and reduction in unemployment are due to the failure to extend extended unemployment insurance. The headline labor force statistics have changed in the direction they predicted.

Steve Benen noted the argument and linked to counter arguments by Danny Vinik and Ben Casselman. While their posts are very worth reading, I want to add more. Vinik mostly and correctly says we can’t yet determine the effect of briefer unemployment insurance on flows from unemployment to employment. Casselman focused on another issue and showed 12 month rolling averages which aren’t ideal for detecting a change January 1st.

update: In comments Casselman noted that the recent post is a brief reminder of a … well I’ll just hand him the mike

Thanks for noting my piece on long-term unemployment. I actually did a much deeper dive on the question of benefits, using CPS microdata to identify likely EUC cutoff victims, in May. Same basic conclusion: the end of emergency benefits hasn’t forced people back to work. You can see that piece here.

I think the most nearly useful thing I can do with FRED is to look at the matching function which relates flows from unemployment to employment to the number of unemployed workers and vacant jobs. The conservative’s claim must be that the flow increased January 1st as workers who were no longer receiving UI accepted jobs which they would previously have turned down.

I have long thought that the best guess of the matching function is that hires of the unemployed are proportional to Vancancies to the 0.7 times number unemployed to the 0.3. To be kind to the Conservatives, I calculate the ratio of monthly flows from unemployed to employed to the square root of the product of vacancies and number unemployed.


There isn’t any sign of a shift in January. One might claim that the increase in December was due to anticipation of the end of extended unemployment but I think that is nonsense (the failure to extend was a surprise).

By the way this shows I was wrong to be skeptical of claims that matching had worsened (based on the Beveridge curve) and that Krugman was right (I know you are shocked).

Here I focus in on the past 12 months

update: Originally I posted a graph highlighting November (not January) which is now deleted. Now I post a very simple graph of the flow from unemployed to employed. The green line is seasonally adjusted, the blue line isn’t. Neither shows an increase in the flow when extended UI ended.


There is little support for the conservative’s story. The data are noisy and I am looking at all of the unemployed not specifically those affected by the change, so Vinik is right it is hard to tell. But really, the conservatives don’t seem to have a case.