Column: Statistics can be both fuzzy and correct
Lies, bigger lies, and statistics.
That, or a variation on it, is among the first replies we hear at the News-Topic each time we have a story about the local unemployment rate dropping, as we did in Wednesday’s paper. The N.C. Department of Commerce reported that Caldwell County's unemployment rate in November dropped to 7.5 percent, the lowest it has been since June 2008, mirroring a statewide trend.
Some of those who say the rate is a lie mean that in the plainest sense: that it's a fiction by politicians to make themselves look good. Some mean it in a more technical sense: that the way the figure is calculated leaves out many people.
The former is extremely unlikely because of the complex, multilayered way that the numbers are gathered. The latter is undoubtedly true – but the question is to what extent it is true.
The unemployment rate derives not from precise counts of every job and every working-age resident but from national monthly surveys involving businesses and more than 100,000 individuals. Samples are carefully chosen, and mathematicians use them to calculate state and local results.
So there likely were not exactly 36,915 people in Caldwell County in November who were either employed or actively seeking work, as appeared in the state's report. That number is a statistical projection made from a smaller sample.
As with all surveys, while the projection may be only close to the true number, the trend over time is generally accurate.
One trend that is true when looking at North Carolina’s overall unemployment figures, according to Harvard economist Lawrence Katz, is that 95 percent of the reduction in the unemployment rate has been due to people "leaving the labor force," which may mean such things as giving up or drawing early retirement.
The percentage would vary in each county, and but we know from Federal Reserve economic data that Caldwell County’s civilian labor force (those employed or actively seeking work) in November 2012 was 38,279, more than 1,300 higher than the preliminary data show for November 2013. In February 2010, when the local unemployment rate peaked at 17 percent, the labor force was 39,959, about 3,000 more than November 2013.
But it’s hard to know what that reduction means. Are all of those 3,000 still here but relying on food stamps and welfare? Did some move away? Or retire?
What we know is that during that time Caldwell County has gained employers, who have more workers now than they did when they arrived. Skeptics scoff that the companies are hiring from outside the county. To some extent, that has to be true – there are always people crossing county lines – but it would make no sense to put your business here if all your workers had to commute or you had to pay relocation costs for a majority of your workers.
The month-to-month data in the surveys used to calculate the unemployment rate also sometimes provide puzzles. For instance, John Quinterno, a principal with South by North Strategies Ltd., a research firm specializing in economic and social policy, pointed out a quirk in the November surveys: “On the one hand, employers reported having fewer positions on their payrolls than in October, but on the other hand, more people reported being employed.”
People often think that math is simple, black and white. But it’s not when you are dealing with estimates and projections, which is what the unemployment rate is.
That’s why those who track the rates say that more important than any single monthly number are the long-term trends. Over time, statistical anomalies wash out.
Over time, the number of people in Caldwell County who say they have a job has gone from the low point of fewer than 32,000 in December 2009 and barely more than 33,000 in February 2010 to more than 34,000 in November 2013. Having a job is not the same thing as having the kind of job you want or the pay level you want, but employment is all we can measure.
Employment is not back to pre-recession numbers – in June 2008, when the unemployment rate was the same as last November, the number reporting they had jobs was more than 35,000 – but it is going up. The true amount is somewhere in the fog of statistics, but it’s a fog, not a lie.