Evan Soltas
Aug 23, 2013

Unemployment Among the Tarheels

North Carolina cut the duration of unemployment benefits to 19 weeks in July, a reduction of 80 weeks since the start of 2012 and a full month less than the standard 26 weeks to which the federal government is willing to contribute in good economic times and bad. The state also cut the weekly payment from $535 to $350.

You might have not been watching this closely, but I have been -- in part because I've spent quite a lot of time in the state, but mostly because the action means the state has volunteered for an epic public-policy experiment. It's one that we should not be having, but given that we're having it, macroeconomists should be watching closely: the magnitude of the policy change is enormous and it is in a single state, so it's pretty close to a controlled experiment.

I'll be writing about this more soon. But let's share the basic data, because we're beginning to see some results that, by my analysis, appear significant. I'll also point out a few areas where results are not yet significant but I anticipate to see substantial change over the next quarter.

The most important: Initial claims for unemployment benefits are way, way down.

The number of people that applied during the week of Aug. 10 is 35 percent less than the number in the corresponding week a year ago: 7,323 claims now versus 10,977 then. And no, the drop isn't macroeconomic. At this time of the year, North Carolina usually represents about 3 percent of total initial claims nationwide -- and it's just over 2 percent right now.

Now for some disclaimers: Initial claims data is noisy week-to-week and highly seasonal. Sometimes the seasonal trend shifts a little and the year-over-year comparison messes up. But as a check, I applied the seasonal decomposition tool in R.

For those who haven't used it before: It's highly conservative in that the underlying trend it reports is very smoothed and slow to react to changes. It detects a significant and abrupt change in trend as of the first week in July -- a 10-percent drop so far, though the trend is still adjusting -- which for me is the most convincing evidence I've presented. An quick-and-dirty graph is available here. Someone needs to run the de-seasonalized trend through X12, which is the official Bureau of Labor Statistics tool that I don't know how to use.

There's been no drop in continued claims for unemployment benefits yet, but there was some interesting behavior in the immediate run-up to July 1. It looks like there was some sort of scramble or administrative change. I expect continued claims to drop significantly for obvious reasons. 19 weeks is 19 weeks.

It's strange that there's been a significant drop in initial claims but nothing yet in continuing. My expectation coming into this was that it would be the other way. Trying to explain the result after the fact, could it be that the cuts to duration and weekly benefit dissuade new applications, but if you're already on unemployment compensation, you're just going to use them up until you hit 19 weeks? I am also uncertain as to whether the 19-week standard applies retroactively to people who signed up expecting many more weeks of benefit or if it only applies going forward. If it's the latter, the results make more sense.

I am also watching the North Carolina labor-force count. When people lose their unemployment benefits, they have three options available to them: (1) they can find a job, (2) they can actively search for work without benefits, and (3) they can stop searching for work.

I don't expect to see a significant change in payroll employment as a result of the cut to initial claims, as employment is pretty clearly demand-constrained in the U.S. and making the already-suffering a little more desperate won't reap any supply-side benefits.

I do expect a large fraction of those who lose benefits to keep looking for work, but it's quite likely that a significant fraction will choose to stop looking for work, given the loss of unemployment benefits that require an active work search. The result is that we should expect to see a minor drop in the unemployment rate -- if only because they are no longer counted as unemployed, though still jobless -- and a drop in the civilian labor force. The trend will be clearer in the labor-force count, as macroeconomic changes and noise tend to obscure state-level unemployment rates.

There has been a small drop in the N.C. labor force, but it is not significant and nothing to hang an argument upon at this point.

At the end of the day, the state of North Carolina is doing this to save state revenue. That I understand. They gave $594 million in unemployment benefits in the first quarter of 2013, though federal money covered some of that. For comparison, the state appears levied $22 billion in taxes in 2011, so I figure unemployment benefits is somewhere around a tenth of all state spending. A question I'd have for an N.C. specialist is: When the state loses all of this federal money as a result of their cuts, how much do they actually end up saving? It can't be that much.

I expect that we're about to see a very significant percentage drop in N.C. unemployment-benefits spending next quarter. A rough estimate would be a 60 percent drop, the first 35 percentage points from the cut to weekly benefits and the other 25 from the cut to duration. That calculation assumes that per-beneficiary spending scales linearly with the weekly benefit and duration because I really don't know what else to assume. (These appear to be pretty close with CBPP's numbers, also.)