Given the debate about extending unemployment insurance benefits, North Carolina has been in the news recently for cutting its benefits in July 2013. Last month, the American Enterprise Institute’s
Did reducing the number of North Carolina residents eligible for federal extended unemployment benefits boost employment? These data suggest it did not, a reality Washington policymakers might want to consider.
And just today, Paul Krugman commented
North Carolina is an interesting place these days, and I mean that in the worst possible way…[I]f there were anything to the theory that cutting unemployment benefits encourages job search and somehow translates into higher employment even in a slump, harsh policies should work better at the state than at the national level. But there is no sign at all that North Carolina’s harshness has done anything except make the lives of the unemployed even more miserable.
Pethokoukis relies on just labor market figures from North Carolina to make his point. Krugman does a bit better in comparing North Carolina to the rest of the country.
I thought I could go a little farther in ensuring that differences in employment and labor force are due to policy changes, not other factors affecting labor markets. My approach is to only look at those counties in North Carolina and neighboring states that are located along the state border.
Neighboring counties are likely sharing overall economic experiences, so there is a stronger case that differences in employment and labor force trends are linked to North Carolina’s “policy experiment.”
So what does the unemployment rate look like in these border counties?
North Carolina’s border counties have a higher unemployment rate than their neighbors, but that difference narrowed considerably after July. Specifically, the aggregate unemployment rate fell from 9.4% to 7.9% (1.5 percentage points) while the comparison counties saw their rate fall from 7.8% to 7.0% (only 0.8 percentage points).
The change is more noticeable if you index the unemployment rate by its July 2013 level.
Of course, Pethokoukis and Krugman note that the fall in the unemployment rate may be misleading. The unemployment rate is a ratio of the number of unemployed and the size of the labor force. As workers become discouraged and stop looking for work, the Bureau of Labor Statistics no longer counts them as unemployed or in the labor force. Consequently, the unemployment rate can fall by workers giving up— not exactly the sign of a robust labor market.
Indeed, that does seem to be the case.
The decline in the size of the labor force is small (-0.67% in North Carolina counties compared to -0.42% in comparison counties) and the trend appear to precede the end of extended benefits in July, but North Carolina border counties have seen their labor force shrink more than neighboring counties.
But that is not to say the conservative argument is bunk (side note: the term “bunk" comes from Buncombe County in North Carolina). North Carolina border counties have seen a noticeable bump in the number of employed residents relative to their neighborhoods. Like the change in the labor force, the change here is small (0.97% compared to 0.43%), but should be noted.
We can actually quantify which effect —the decrease in labor force or the increase in employment— is most responsible for the decline in the unemployment rate. If you were to hold the size of the labor force constant at its July 2013 level but allow employment to increase as it actually did, the unemployment rate would have fallen 0.9 percentage points in North Carolina border counties at 0.4 percentage points in neighboring counties. If you hold the level of employment constant and allow the labor force to shrink, the unemployment rate falls 0.6 percentage points in North Carolina and again 0.4 percentage points in neighboring counties. So in the comparison group the decline in the unemployment rate is equally accounted for by both effects, but in North Carolina the decline is somewhat more due to the increase in employment than the decline in labor force.
I should note that the small differences here and the margin of error in the household survey used by the Bureau of Labor Statistics likely means none of this is statistically significant and that a longer time series would prove more useful, but none of that will stop pundits from making claims about the North Carolina experience and its relevance to the national debate