The May employment report was a shocker. Jobs were added instead of being lost, while the unemployment rate came in well below expectations. Were the numbers correct, wrong or manipulated? The answer is none of the above. The economic data mills are operating in an environment that makes collection of the numbers difficult and the interpretation of the results problematic.
First, let’s get out of the way the idea that just because a number doesn’t meet expectations or is beneficial or harmful to a political candidate, it had to have been manipulated. As far as the government’s statistics go, that can be answered with an unequivocal no; politics played — and will continue to play — absolutely no role in the creation of the numbers.
The Bureau of Labor Statistics (BLS) is responsible for the employment numbers as well as a variety of other data. Hundreds of seasoned professionals who take pride in their work are involved in the process.
Political appointees are walled off from the data collection and analyses and this wall actually works. And what is true for BLS is true for the other government agencies that collect and disseminate data. Please remember this when a report comes out that doesn’t help your candidate.
The idea that political interference alters the data, especially a government number, is disturbing and can be damaging. In 2012, when the September employment report was released a month before the election, it showed that the economy did better than expected. Former GE CEO Jack Welch (since deceased) basically called it phony, a complaint picked up by many commentators.
The fact of the matter was that the drop in the unemployment rate was simply part of a longer-term, saw-tooth decline that was going on for three years and would continue until this February. Meanwhile, the payroll increase was actually underestimated by about 50 percent.
Given that BLS put out a number that understated the strength of the economy, it is clear it didn’t manipulate the data to help a president get reelected then and it will not now. Nevertheless, its reputation was tarnished.
As for the private-sector numbers, whether they are created by firms looking to raise their visibility or trade associations providing insights into their sectors, if it turns out that politics affected the numbers, trust in the data would disappear and they would be out of business.
So, why am I warning you to be wary of the data? Because we are in extraordinary times that play havoc with the ability to accurately collect and determine the meaning of the numbers.
Take the May employment report. It turns out that BLS misclassified some of the responses to the surveys. This happened not just in May but in March and April, as well.
Without getting too technical, many workers who were furloughed and didn’t get paid should have been classified as unemployed. However, the questionnaire allowed for a response that grouped those workers with those on vacation, on jury duty, or taking care of a relative (i.e., not unemployed).
The misclassification led to a huge undercount of the number out of work, greatly understating the unemployment rate. The May rate should have been about 16.3%, not the 13.3% reported. Similarly, the April rate should have been nearly 20%, not 14.7%, while the March 4.5% level should have been about 5.5%.
Those are huge differences that significantly change the image of the economic landscape. Since the peak unemployment rate would represent the size of the hole the downturn has dug, it is likely a lot deeper than the data imply.
Because BLS doesn’t reclassify data so that the responses are kept free of interpretation (or political interference), we are stuck with these numbers.
As for the May job growth gain, that may have been somewhat overstated. It could even have had the wrong sign, meaning that there were job losses, not increases.
The adjustment factors that account for seasonal economic changes may be largely meaningless. The recent downturn and upturn in activity have occurred more because of states’ allowing activities to reopen than normal seasonal issues. That departure from the norm could lead to really odd numbers.
For example, dentists’ offices accounted for 10 percent of the May jobs increase. In normal times, this sector employs only about 0.6% of total payrolls. The wildly outsized impact makes little sense.
The public- and private-sector number crunchers are doing the best they can. Unfortunately, the magnitude and timing of the changes that were created by government decisions to close and then reopen the economy, the randomness of those decisions, and the variations in the size, speed and extent of the reopenings, make it difficult to produce numbers that reflect accurately that state of the economy.
So, what should the average person do? Stop, look and listen, as it is not clear the direction from which the train is coming.
Don’t react immediately. Yes, the markets may move on the releases, but a knee-jerk response is as likely to be wrong as right. Remember that although the unemployment numbers are not revised on an ongoing basis, as more information comes in, a lot of the other data are. What we think is happening today may not be the case next month.
Read a variety of comments on the numbers. We know the data spinners are always out there. It is an election year. But even economists are uncertain, so don’t depend on one view.