It pays to look before you statistically leap
Published March 23, 2014
By John Hood
by John Hood, John Locke Foundation and NC SPIN panelist, published in Rocky Mount Telegram, March 22, 2014.
In late February, some 44,000 discouraged North Carolina workers suddenly disappeared.
Relax. It’s not as mysterious as it sounds. What happened was the U.S. Bureau of Labor Statistics released its long-awaited revision of five years of household employment data for North Carolina and the rest of the country.
For months, critics of North Carolina’s decision to exit the unemployment insurance extended benefits program have cited household survey data to pummel Gov. Pat McCrory and the Republican-led legislature.
Ignoring warnings about the preliminary nature of the data, and the statistical insignificance of many findings because of the survey’s small sample size, these critics claimed that some 111,000 people had left North Carolina’s labor force in 2013, dwarfing the scant 13,000 unemployed workers who had obtained jobs during the year.
The first egregious problem with this claim was that extended benefits ceased in North Carolina in July, not in January. As it happens, the first six months of 2013 were the roughest for North Carolina’s labor market, with very little net job growth.
This could not have been caused by a policy change that hadn’t happened yet.
The second egregious problem with the claim was that not everyone exiting the labor force deserves the label “discouraged worker.” Those who have been unemployed for a long time and are so frustrated that they’ve given up form only a small minority – albeit a sympathetic minority – of those leaving the labor force. Others are in the midst of relocating or retraining for a new job. Still others are simply retiring, as the Baby Boom bulge is now affecting labor force participation across the country.
Now there is a third egregious problem with the claim: the revised Bureau of Labor Statistics numbers show something very different. North Carolina’s labor force declined last year by 67,000, not 111,000. For the last six months of 2013, the period after extended benefits ended, the number of employed North Carolinians rose by about 27,000 while the labor force declined by 39,000. Furthermore, because labor-force participation is trending downward for the nation as a whole, and particularly among Southeastern states, it would be silly to attribute North Carolina’s entire 39,000 decline to the end of extended unemployment insurance benefits.
As a matter of fact, even these revised household-survey findings should still be treated with great caution because of the small sample size of the survey. For the last six months of 2013, all that can be said with a high degree of confidence from the household-survey data is that the number of unemployed North Carolinians fell by about 65,000.
During the same six-month period, North Carolina employers added about 60,000 net new jobs, according to the broader survey of employer establishments – the data from which have also just been revised.
Economists and policy analysts typically put more stock in the employer survey because of its far larger sample size. The survey data also are periodically adjusted to comport with actual payrolls reported to the federal government each quarter.
According to this dataset, North Carolina added 86,000 net new jobs in 2013 – more than two-thirds of them after July 2013 – which was a faster rate of growth than the national average and the largest annual job gain the state has experienced since the Great Recession.
Do these revised household and employer surveys depict North Carolina’s economic performance as spectacular? No, not by historical standards. The nation as a whole is experiencing a comparatively weak recovery from a severe downturn. Since 2011, North Carolina has done better than the national average in job creation and income growth, but these were not exactly high bars to clear. Hundreds of thousands of North Carolinians remain unemployed or underemployed.
All sides of the political debate agree that significant economic challenges remain, even as we disagree about the best policies to respond to them. One way to foster meaningful dialogue, however, would be for everyone to stop hyping preliminary data and drawing statistically invalid conclusions from them. Resist the temptation.
