Private payroll firm ADP has a new job: making its monthly jobs report more accurate.
The firm announced today that it is changing how it computes its closely-watched monthly jobs estimate, which comes out ahead of the all-important report from the Department of Labor. The company's estimate of private payroll growth, released each month just before the federal government issues its take on the job market, is a closely watched predictor of what the government figure will be, influencing analysts' expectations and the markets.
Among the changes that ADP is making are an increase to the sample size of firms and employees it uses to compute its estimate, as well as a new partner in producing the number. While the firm once collaborated with St. Louis-based Macroeconomic Advisers, it will now work with Moody's Analytics.
The idea is to make the ADP estimate track more closely to the Labor Department's numbers.
"I think it will be more accurate for a number of reasons," says Mark Zandi, chief economist at Moody's Analytics, pointing to the larger sample size, as well as more frequent data collection. He says that already, the new way looks promising.
"The two months where we've been running this alongside the previous methodology in real time, the accuracy is actually quite good," says Zandi.
The change comes amid criticism that the payroll processing firm has recently too often overshot the government's jobs estimate. The company's September estimate of new private-sector jobs came in at 162,000, for example, while the private-sector count from the Labor Department's establishment survey came in at 104,000. Under the new methodology, however, ADP's count comes in at a much-closer 88,200. Even the government's figures have been questioned of late, amid the fractious political climate and at a time when voters are focused on the state of the job market.
According to Marketwatch, ADP overestimated private-sector job growth by around 53,000 per month over the last six months. The new approach brings that gap down to around 1,000.
Still, Zandi stresses that even the Labor Department doesn't nail the jobs figure on the first try. The department will still issue two further monthly revisions to the September jobs figure, for example, bringing it closer and closer to its "true" value.
Zandi says that the goal of the latest changes is to reduce the discrepancy between ADP's estimate and the Labor Department's figure after that third estimate.
Though ADP's estimates have been off at times, they still overall tend to track closely with the Labor Department's numbers.
The general trend, rather than the specific month-to-month figures, have been the key value of ADP figures, says one economist.
"You always saw that they moved in the trend the same way. Were they the exact same numbers? No. Were they sometimes a lot higher or a lot lower? Yes," says Joel Naroff, president and chief economist at Naroff Economic Advisors, an economic consulting firm based in Holland, Pa. "I never use the numbers as an absolute. If it comes out at 150 [thousand], it doesn't mean it's 150 [thousand]."
He also points out that, even if ADP's new estimates fall perfectly in line with the Labor Department's second revision, the government still further revises its establishment numbers after the fact. Last month, the department released its preliminary annual benchmark revision to the numbers and found that in the 12 months leading up to March 2012, the economy likely added 386,000 more jobs than previously estimated.
So can the ADP figure be a better predictor? The nation is about to get a first look. The next ADP estimate will be released on the morning of November 1, just a day before the Labor Department's estimate--and barely days before voters go to the polls to choose the president and the next Congress.