By Leah Schnurr
NEW YORK (Reuters) - Mark Zandi has more riding on this week's U.S. labor data than most other economists.
The chief economist of Moody's Analytics is helping to overhaul the ADP report on hiring by U.S. companies. It is a closely watched report but it often proves to be out of line with the government's jobs numbers reported two days later, probably the most important U.S. economic indicator.
As part of the surprise revamp announced by Automatic Data Processing Inc <ADP.O> in October, the report now has a more explicit mission statement: it will seek to predict not what the government data will look like 48 hours later, but rather what the revised third reading of the official jobs number will be when it is released two months later.
If ADP's change is successful, it will provide a reliable precursor of the Labor Department's official figures and could capture an upswing in hiring earlier than the government numbers, which would mark an important turning point in the economic recovery.
Trying to narrow what has sometimes been a wide gap between the two reports may put Zandi's reputation on the line.
"Now I'm sweating about Wednesday," Zandi said with laugh, referring to the ADP release on Dec 5.
The people behind the numbers say they are trying to achieve two things: their newly stated target, and to come up with a report that can stand on its own as a "true" look at private sector employment.
"I suspect there will be months when we have turning points and ADP is out ahead of the Bureau of Labor Statistics, and we may come under criticism for not being close to the BLS number," said Zandi.
Moody's Analytics was brought on to replace ADP's previous partner Macroeconomic Advisers, which co-developed the report with ADP in 2006. Modifications to the methodology include using a larger number of companies and changing some outside economic data used in the modeling.
Differences between the jobs numbers from ADP and those from the government have varied widely. ADP either overshot or underestimated the first payrolls reading by an average of 57,000 a month from January through September, the last month that ADP used the old methodology, according to IFR Markets, a unit of Thomson Reuters.
On the other hand, the jobs data from the government gets revised on a monthly and annual basis, often significantly. The annual revisions tend to push up payroll readings during times of job growth and lower the figures in times of job losses, according to IFR.
ADP tested its new model on data going back to 2001 and found it had a very high correlation of 96 percent with the private sector component of the revised government figures.
After only one report so far using the new method, it's too soon to see how well the data matches up in real time, and whether it can assuage critics of its past performance.
"If ADP didn't come out before the payroll data, no one would use it at all or ever look at it," said Paul Dales, senior U.S. economist at Capital Economics in London.
LOOKING FOR CRYSTAL BALL
Wall Street continues to look for ADP to be its crystal ball, something that ADP is aware of even with its new focus on the third reading of the government numbers.
To that end, the firm is experimenting with the idea of slicing the data to come up with another figure that would predict the first reading, Zandi said.
"Ultimately, I think we'll be judged by how close the ADP number is to the BLS over time," said Zandi.
Another possibility is to release weekly data, he said, and regional data looking at big metropolitan areas will be implemented soon.
Jan Siegmund, ADP's new chief financial officer who has been involved with the report since its beginnings, said the revised methodology should make the index quicker to spot changes in the pace of hiring.
"At least that's the hope. We haven't seen the change in momentum yet," he said.
The people working on the data say it is a more complete, real-time report because ADP culls its data all the way up until the weekend before its report, unlike the Labor Department which conducts a survey up until a cut-off date that is in the middle of the month it is reporting on.
"I have always thought in the back of my mind that maybe we should be listening more to ADP and less to the government data," said Nicholas Colas, chief market strategist at ConvergEx Group in New York.
"I understand they're not a good forecast of the Friday number, but it's a really big and robust sample and I do believe there's value in looking at it."
It is likely to take months before it is clear how well the new ADP survey is tracking the government payrolls report. Catching a change in momentum is particularly difficult given all the moving parts, said Brian Bethune, professor of economics at Gordon College.
"I think it's a heroic attempt but I think it's unrealistic to think that they can do better than Macroeconomic Advisers did."
(Reporting by Leah Schnurr; Editing by Lisa Shumaker)