At CAHS we delve with data quite often (maybe less than our friends at CT Voices, but we do love numbers nonetheless) so the sudden twists and turns of Connecticut´s jobless rate obviously picked our interest. It turns out that the initial report from the department of labor was "wrong", and that the state did not stagnate with 100 jobs lost in 2012, but actually had some decent growth, with 8,600 jobs created instead. The revisions go all the way to 2011, with the state adding almost twice the amount of jobs initially reported.
Why these changes? Are not the unemployment figures fixed in time and readily available? Well, to tell the truth, they really are not. These figures come from a (very) large survey that is sent every month to tens of thousands businesses across the country. This means that the initial number that we get every month it is likely wrong, and usually is revised several times in the following reports, as the Department of Labor get more data and can update their estimates.
The sample for Connecticut is, obviously, a subset of the larger Federal one, and as a result it is much noisier. If say 10% of businesses survey fail to report on time, close or move to a new address in the sample for the whole country, the effect is there, but it is small. If that happens in Connecticut, the drop might be enough to move the numbers pretty wildly. To this we have to add the fact that new companies are never surveyed, as the department of labor really only updates their panel once a year, and only adds businesses that are more than one year old, and you have the recipe for occasional underestimates and large revisions down the road. Something similar happens during downturns, as companies that close sometimes are mistaken as simply failing to fill their surveys on time. Don´t get me started trying to explain how job creation and unemployment rates are estimated using two different surveys (the former to business, the later to individuals) or we could be here all day.
So no, the governor is not making up the numbers, and the Department of Labor is not run but people that can´t do math. Compiling good data for a country as large as the United States is extremely difficult, and Federal agencies, believe it or not, are exceptionally good at their jobs in this regard.