For years, critics have accused the Census Bureau of substantially overstating the number of poor people in America.

Now, after a comprehensive study, the government's leading statistical agency has finally admitted that the critics were correct.With bureaucracy's tunnel vision, the bureau counted only cash income in its poverty studies. This flew in the face of common sense, which sees real value in non-cash benefits provided by government and private employers.

In 1986, the last year for which complete figures were available, the Census Bureau reported officially that 13.6 percent of Americans were below the poverty level, which it defined as $11,203 in cash income.

However, when the bureau crunched in figures for all the benefits people receive, it found that the actual poverty rate was 10.3 percent, one-fourth lower than it had previously reported.

To arrive at the "real" amount of poverty, the bureau took into account taxes, capital gains, Social Security, welfare, Medicare, Medicaid, food stamps, subsidized housing, unemployment compensation, school lunches, employer-provided health insurance and other benefits.

Now that Washington has a more accurate fix on poverty, it's time to correct another key figure: the unemployment rate. It's also off.

The Bureau of Labor Statistics estimates how many people were jobless the month before by sending out poll-takers. Unfortunately, plenty of people don't wish to jeopardize their welfare payments or food stamps by telling the poll-takers that they indeed are working - in a regular job, or "off the books" or in an illegal activity.

Since much public policy is based on the jobless number, the Labor Department should emulate the Census Bureau and get it right. The Commerce Department already gathers employment data far more reliable than the Labor Department's surveys. Why not use them?