Are State Employment Numbers Just Too Good To Be True?
The state received some encouraging news from the state Executive Office of Labor and Workforce Development last month.
Massachusetts created 19,100 new jobs in April as the unemployment rate fell from 9.3 percent in March to 9.2 percent.
The labor department was quick to point out that the job gain was the largest month-to-month increase in 17 years. Seven of the state’s 10 employment sectors added jobs during the month, including long-suffering sectors like manufacturing.
But not everyone was convinced that all the excitement was justified.
Rational Exuberance
On the same day that the labor department’s numbers were published, Andre Mayer, senior vice president of communications and research at the Associated Industries of Massachusetts, wrote:
“If 19,000 new jobs in one month seems too good to be true — well, it probably is.” AIM “found the magnitude of the gain unlikely, suggesting a statistical anomaly in the survey or (that) seasonal adjustment produced an inflated figure.”
Seasonal adjustment is a tool widely used in any statistics-gathering operation. It helps organizations or agencies like the labor department or AIM keep their eyes on larger, overarching trends rather than temporary fluctuations. (Click here to see a graph of the state's unemployment trends.)
But if fluctuation is so great as to undermine an underlying trend, is there really a trend?
Yes, according to Meyer.
“What’s important is not what we see that month, but what we see over time,” Meyer said. “We have had several months of improvement. The trend is there, and clearly it’s good news.”
Still, seasonal adjustments use data from the past — about spikes or drop-offs in business as a result of the end or beginning of the school year, or the number of days between Thanksgiving and Christmas, for example — to compensate for fluctuations in the present.
“Each month, there’s an assumption that what happened in the past is what happens, and if it isn’t, it can make the numbers jump around,” Meyer said.
Noisy Numbers
Without seasonal adjustments, it can be difficult to recognize an overall trend in a statistic like job creation, said Rena Kottcamp, the state labor department’s research director.
Unadjusted, raw data “bounces all over the place. It’s what we call noise,” she said.
Anyone can seasonally adjust data.
The U.S. Census Bureau provides a number of seasonal adjustment calculations for use by agencies like the labor department.
Also, anyone can apply his or her own seasonal adjustment to state labor department data, as does noted Northeastern University economist Alan Clayton-Matthews.
During this time of year, construction hiring typically spikes while school bus drivers are laid off. In the winter, those bus drivers will be back at work and construction workers will be laid off.
While those changes are important within respective industries, they make delineating overall job creation trends across all sectors complicated.
The adjustments also clue analysts, economists and researchers into the drivers of overall trends, making it more likely that they’ll recognize “when something happens every year for 10 years and now it’s different,” Kottcamp said.
For example, the labor department usually sees a drop-off in state and local government employment around this time of year. This year, however, state and local governments, already cut to the bone, have not done so to the extent they have in the past, Kottcamp said.
What agencies like the state labor department do not do is predict the future. In fact, Kottcamp said, “these are sample-based estimates and we don’t know for six or eight months after what really happened.”
Gains And Losses
That’s where Meyer and AIM take issue with the labor department announcing in terms no more uncertain than to note that its numbers are “seasonally adjusted” that Massachusetts created 19,000 jobs in April.
“It looks like a trend, but it’s not like we’re going into overdrive in job creation,” Meyer said.
The survey sample the department uses to generate its numbers “is not very large,” Meyer said, making job creation numbers even less stable.
The fluctuations are the reason why the labor department revises the numbers it puts out each month. This month, the department will report job creation numbers for May and a revised number for April.
The labor department’s statistics are meant to be read in groups of months, not just month-to-month, in order to make sense.
For Meyer, the cycle of gathering statistics, adjusting them to smooth out seasonal volatility and then revising them, while it does offer a monthly glimpse into employment trends, can be a tool of public officials to help paint a portrait of economic development activity in the state.
“It is political,” he said. “There’s some spin in the (labor department’s) press release sometimes.”
WBJ Web Partners
Most Recent
Most Recent
0 Comments