U.S. Underestimated Job Losses, State Finds : Economy: California has suffered its worst plunge in more than half a century, records show.

The federal government may have seriously underestimated the effect of the recession on California and some other major states and undercounted job losses nationwide by more than 2 million, according to an analysis by the California Department of Finance.

Payroll tax filings in California, New York and other states indicate an extraordinary free fall in jobs in late 1990 and early 1991. California has suffered its worst job losses in more than half a century, according to these records. The U.S. government may have underestimated this plunge in California by more than 500,000 jobs, according to state officials.

“To me it’s very obvious that the recession is far, far deeper than the (Labor Department) figures are indicating,” said Ted Gibson, principal economist with the Department of Finance in Sacramento.

The surprising state reports--while far from definitive--are gaining attention because they are based on statements filed quarterly by almost all private employers. By contrast, the Labor Department’s widely quoted jobs data comes from smaller surveys that may not fully detect sudden, sharp swings in the economy for many months.

The payroll tax figures not only raise questions about the accuracy of information relied on by U.S. policy-makers as the recession gathered steam in 1990 and 1991, but also may provide a clue as to why consumer confidence last year fell to the lowest levels in a decade, according to economists in government and the private sector.

“There must be something rotten in Denmark with those national numbers, because we haven’t gone through a mild recession,” said Allen Sinai, chief economist with the Boston Co. “I think what the jobs data in states like California is finding is much more telling.”

The nation’s jobs picture is drawn from findings by the Labor Department’s Bureau of Labor Statistics, in cooperation with state agencies, such as the California Employment Development Department. At issue are the payroll employment statistics, which describe how many jobs are gained or lost each month.

Analysts in and outside government watch the payroll figures closely; many economists believe that they offer more insight into the economy’s course than the unemployment rate. A puzzling and vast gulf seems to have opened between the official payroll jobs totals--meant to reflect the number of jobs in the economy--and what employers reported to the government early last year.

Consider the experience of California. Since the recession began in mid-1990, the state appears to have lost 660,000 jobs, many at the beginning of last year, according to a study of payroll tax records by the Department of Finance. There has not been such a staggering drop--5% of the entire job total--since 1938.

By comparison, the official job totals computed by the state Employment Development Department and the federal Bureau of Labor Statistics tell a dramatically different story, recording a drop of just 160,000 jobs in California.

Nationally, the disparity appears to exceed 2 million, Gibson said, led by apparent underestimates of the recession last year in California, New York, Massachusetts, New Jersey and Michigan. According to an internal Bureau of Labor Statistics document, payroll tax records from New York last March suggested a loss of more than 400,000 jobs that had not been counted in official labor statistics.

“The federal figures are too optimistic and they come too late,” said Claudia Hutton, a spokeswoman for the governor’s budget office in New York, where budget planners “have gotten burned” by federal revenue projections that proved too optimistic. “In New York, I don’t think it’s an exaggeration to say that just about everyone knows someone who has lost a job.”

When asked about the gap, Bureau of Labor Statistics officials said they annually review their statistical assumptions about job totals, in addition to making routine revisions, and are expected to adjust last year’s findings on the number of jobs for California and the nation in the coming months. Bureau officials in Washington have said that they may revise the nation’s job count downward by 650,000.

Some economists also point out that the payroll tax records, for all the interest they are sparking, are not perfect and should be considered in context. It is possible that some of the tax data is inaccurate, reflecting cheating by hard-pressed employers. And discrepancies between official job counts and state payroll records are not universal. They appear to be focused in a handful of major industrial states in the East and West.

Still, the gaping disparity has perplexed the bureau’s statisticians and prompted unusual reviews of data that in non-recessionary times cause more in the way of glazed eyes than public controversy.

“We’ve been rechecking the figures daily for five months,” said Merlin E. Meyer, a Bureau of Labor Statistics official in the San Francisco regional office.

Gibson said: “The EDD (Employment Development Department) office here has been crawling with feds ever since this thing arose.”

The official jobs figures for California are collected by the state Employment Development Department based on a sample survey of 37,000 employers out of the state’s 838,000 employers.

“We feel they’re very representative,” Meyer said of the employer sample. “All industries are covered.”

However, critics say that the government statisticians fail to adequately weigh changes affecting small firms, ranging from building contractors to business services, that may quickly emerge and vanish from the economy. They also argue that new firms, an important part of the jobs picture, are not adequately represented in the sample.

The disparity first became apparent last year when the payroll tax records for the first quarter suggested that 390,000 jobs vanished in California. The loss seemed so drastic as to test credulity, although it has stood up to review so far.

A 390,000-job plunge seemed “preposterous,” said David G. Hensley, director of the UCLA Business Forecasting Project. But later in the year, when payroll tax records filed in the second quarter showed little change, “I decided at that point that we had better go with those numbers,” Hensley said.

Withholding taxes paid by California employers became another piece of evidence that something was amiss with the official labor statistics. They came in $125 million below what would have been expected if the official labor figures were accurate, Gibson said.

Donald Ratajczak, a Georgia State University economist, said the gap is evident in statistics from Florida and Virginia, but does not emerge in most of the Midwest. “Conditions are dramatically mixed around the country,” he said. “Some people are having real bad times and for others, it’s like ‘big deal.’ ”

Miscalculations in the size of the job rolls do more than raise questions about the intensity of the slump. They are an important part of revenue projections for hard-pressed state governments, so mistakes can aggravate budget deficits and create troubling side effects for the economy.

“We’ve usually trusted those (Bureau of Labor Statistics) numbers in the past and this data suggests there are problems with them,” David Wyss, an economist at DRI--McGraw Hill in Lexington, Mass.

There are fewer jobs out there than the official statistics say there are, he added, noting that “it’s something that makes us very nervous right now.”