Last year North Carolina's conservative Republican legislature got tough. It sharply reduced the duration of unemployment benefits and made them much more difficult to collect. The changes took effect at the start of July, 2013. Their story was that unemployment insurance and other benefits discourage workers from seriously looking for jobs. If we take away this crutch of unemployment benefits, then workers will figure out how to find jobs. This both saves the government money and is better for the workers themselves since they will actually be making a living on their own.
We now have data for 10 months into the experiment (through May) and John Hood, the chairman and president of the John Locke Foundation, a North Carolina think tank, has a piece in the Wall Street Journal telling us that it is a resounding success. Hood tells readers:
"According to the U.S. Bureau of Labor Statistics, the number of payroll jobs in North Carolina rose by 1.5% in the second half of 2013, compared with a 0.8% rise for the nation as a whole. Total unemployment in the state dropped by 17%, compared with the national average drop of 12%. The state's official unemployment rate fell to 6.9% in December 2013 from 8.3% in June, while the nationwide rate fell by eight-tenths of a point to 6.7%."
Okay, let's take these in turn. North Carolina did have more rapid job growth than the rest of the nation in the period since it cut benefits, but it also has had more rapid job growth than the rest of the nation for the last four decades, before many of the benefit cutters were even born. This because it is in the South, which has been growing more rapidly than the Northeast and Midwest for quite some time. (My explanation is air-conditioning, but you're welcome to throw in other items.)
If we look at North Carolina's labor market over the last year (May 2013 to May 2014) we find that the number of jobs, as measured by the Labor Department's establishment survey, grew at 1.92 percent rate. This beats the 1.86 percent rate for the rest of the South Atlantic region, but the difference certainly is not enough to employ all the people who were cut off from the unemployment rolls. (The South Atlantic region is a grouping of states from Florida to Maryland. It has been used by government agencies for many decades.) If the argument is that the ending of benefits put the fear of God in the unemployed and made them finally get serious about working, these numbers don't do much to support the case.
The situation gets even worse if we pull out the Charlotte-Gastonia-Rock Hill area. The reason for pulling out this relatively fast growing region is that it straddles the border with South Carolina. Many of the workers who have gotten jobs with companies in North Carolina actually live in South Carolina. If unemployed workers' past employment experience had been in South Carolina, they will not have any additional motivation to find work as a result of North Carolina cutting benefits.
We can't know how many of the new workers the Charlotte metropolitan area are from South Carolina, but it is striking that if we pull out this area, North Carolina's job growth slightly lags the rest of the South Atlantic region. Excluding the Charlotte area, job growth in the state was 1.76 percent over the last year, roughly a tenth of a percentage point less than the average for the rest of the region. This means that outside of the Charlotte area, it doesn't seem that the cut in benefits did anything to increase incentives to work. As a practical matter, the differences in both directions are small, but the point is that there is no evidence that cutting benefits did anything to increase employment growth in North Carolina compared with comparable states.
As far as the great news on unemployment that Hood cites, this is entirely a story of North Carolina workers giving up looking for work and leaving the labor market. (in order to collect unemployment benefits, workers must be looking for work.) While the size of the labor force in the rest of the region grew by 1.0 percent over the last year, the labor market shrank by 0.2 percent in North Carolina. Employment growth in North Carolina, as measured by the Labor Department's survey of households, was 1.9 percent over the last year. That compares to 2.2 percent growth for the rest of the region. Again, no evidence that the ending of benefits got people to be serious about looking for work. The evidence is that they gave up looking for work and dropped out of the labor force.
Not content to rest his case that cutting unemployment benefits increases employment on the North Carolina alone, Hood moves on the national picture:
"Still not convinced that leaving the extended-benefits program encouraged both job creation and job acceptance? As of Jan. 1, 2014, the extended-benefits program expired nationwide. Yet there has been no sudden exodus of discouraged workers to the fringes of the national economy. Both job creation and household employment are up. The nation's employment-population ratio was 58.9% in May, up from 58.6% in December. The labor-market effects of reforming unemployment insurance may not be massive. But they certainly don't appear to be negative."
Actually what is most striking in the data for 2014 is the sharp drop in labor force participation. The labor force participation rate is down by an average of 0.5 percentage points from the first six months of last year to the first six months of this year. This corresponds to roughly 1.3 million people leaving the labor force. Usually labor force participation rises during an upturn, so this is certainly consistent with a story of many unemployed workers giving up looking for work when their benefits expire.
In short, if we look at the data instead of playing games with it, the story is pretty clear. There is zero evidence that cutting unemployment benefits in North Carolina or the rest of the country did anything to spur job growth. There is much evidence that it led those who saw their benefits to end to give up looking for work and to drop out of the labor force.
I see John Hood has replied to my post. Apparently he thinks that if we play games with the start and end dates we can say cutting benefits worked.
Okay, I don't know what games they play in North Carolina, but let's just remember what is at issue. The argument for cutting benefits was that if the state pushed people off unemployment insurance (UI) they would be motivated to get a job. We know that North Carolina pushed lots of people off UI. The question is whether there is any evidence this led more people to get jobs.
That gives us two numbers to focus on, how many people were thrown off UI, and how many extra jobs North Carolina has compared to other states. As noted above, the story on the latter is pretty clear. There is no evidence of better job growth in North Carolina than in comparable states over the last year. Let's look more closely on the numbers of people thrown off the UI rolls in North Carolina.
We can get the numbers of people thrown off UI directly from the North Carolina Department of Commerce. They tell us that in May of 2014, 42,382 workers were receiving UI benefits. In May of 2013 the number was 90,858, a decline of 53.4 percent. By comparison, the number of people receiving benefits in the country as a whole fell by 10.4 percent over the same period.
This means that North Carolina removed roughly 32,000 more people off UI than if it had just followed the same pattern as the rest of the country. This is equal to roughly 0.8 percent of the jobs in the state as of May 2013, which is how much more rapid job growth we would have to see in North Carolina than in comparable states if we are to believe that the additional workers thrown off UI got jobs. As noted in my initial post, there is no evidence of more rapid job growth in North Carolina than in comparable states over this period, which means we conclude that the vast majority of the people who lost UI benefits did not find jobs.
Hood is apparently unhappy that I took May to May comparisons. I confess, I took these months because it is the easiest data to grab because every publication gives the prior year's numbers. Therefore the May 2013 numbers are right next to the May 2014 numbers.
But the year over year numbers also have the nice advantage that we don't have to worry about the seasonal adjustments. We know that more people are working in summer months than the winter months and fewer are unemployed. Most of our data are adjusted for this seasonal effect, but the adjustments are far from perfect, especially at the state level. If we use the same month, then it is much less of a problem.
Hood argues the start point of May is unfair because the state didn't put its tighter rules in place until July. This shows a poor understanding of the policy he supports. Unemployed workers read newspapers and listen to news. Most people receiving benefits undoubtedly knew that they would face tougher rules starting in July. If taking away benefits was going to motivate people to get jobs, then the new rules already should have been having this effect by May and certainly June of 2013.
Hood also wants us to end the comparison in December of 2013. That makes no sense since the differential impact of the UI policy in North Carolina and the rest of the country continues to the present. In other words, the state has far fewer people collecting UI benefits than if it had followed the same policy as the rest of the country.
In short, the story is very simple. North Carolina’s UI policy led to a sharp reduction in the number of people collecting unemployment insurance relative to the rest of the country. There is no evidence that it has had more rapid job growth as a result of this policy. We can therefore conclude that most of the people who lost benefits did not get jobs.
Due to quirkiness of the data, North Carolina shows a fall in employment from May, 2013 to July, 2013 of 5,448, the equivalent of a drop in employment of roughly 200,000 nationally. This is almost certainly an error in the data, just like the gain in employment of 31,316 shown for October to December of 2011 (the equivalent of 1.2 million newly employed workers nationally) was almost certainly an error in the data.