Flawed Federal Statistics Paint Confusing Economic Pictures

MOST people have no independent means of assessing how a national economy is doing. If they don't get a reliable picture from summary data, they can't make sound judgments. A news release this week noted that federal analysts have begun reviewing deficiencies in measures of United States economic performance. It's high time. The deficiencies are many.

Some actions are directly tied to economic findings. Upward adjustments in Social Security benefits are by law linked to the Consumer Price Index (CPI), even though experts agree the index is flawed. Testifying before Congress last week, Federal Reserve Chairman Alan Greenspan said that it overstates the inflation rate by as much as 1.5 percentage points. Politics obviously plays a part in this measurement problem. Billions of dollars are at issue in entitlement spending.

Technical issues are involved as well. It's hard to measure inflation in an economy with the amount of choice ours offers and the extent of discretionary spending. If one item in the market basket of goods used to calculate the CPI has a big price spike, millions may not buy it. Given the huge array of shopping alternatives, it's exceedingly difficult to determine what the operative price actually is.

The deficiencies in data on family income are even more disturbing. Few people are aware that the standard income data are derived from surveys, as opposed to calculations of actual wages and other income. The information comes from the Current Population Survey, conducted monthly by the US Census Bureau. Large national samples are interviewed, much as in the familiar opinion poll.

The Census Bureau does the sampling so well that sampling error probably isn't much of a problem. Nonetheless, these census surveys are susceptible to all the other errors that afflict normal polls. People may genuinely not know what their total family income is, a situation that seems to be growing given the number of multiple-wage-earner households. Other respondents consciously misrepresent their income. Some may believe that if they have underreported to the IRS, they shouldn't give the accurate (higher) figure to another agency. In any case, we know that the survey-derived family income data are wrong: They are contradicted by other information that we have reason to conclude is more reliable.

The census survey data show Americans making practically no income gains over the last quarter-century. The median family income line is depressingly flat. If this picture is accurate, it should be of great concern. It would mean that an important part of the American dream has been shattered - the promise of growing material prosperity.

Other data present a different picture. Using sources that do not involve asking people their incomes, the Bureau of Economic Analysis of the US Department of Commerce publishes compilations of the output of goods and services, known as the Gross Domestic Product (GDP). I have calculated changes in real per capita GDP from 1890 to the present on a decade-by-decade basis. The gain between 1980 and 1990 was 21 percent. This recent experience is about the same as that of the 1970s, better than the 1950s, 1930s, or 1920s, not as good as the 1940s and 1960s.

The GDP story from 1980 on tells of middle-of-the-pack national economic performance - average for a country whose long-term growth record is exceptionally strong. Real per capita GDP data include components that do not show up in personal income. Still, it's inconceivable that the US could have experienced a 22 percent real per-person gain in the '70s and 21 percent in the '80s, while median family income remained flat.

What's more, data show huge consumption increases over this span - far more than increased borrowing could explain. Taking per-person spending for all forms of recreation and controlling for inflation, we find that spending more than doubled between 1970 and 1991! Spending for video and audio products grew by 500 percent. Spending for boats and other sporting equipment more than doubled. Rates of foreign travel surged. Real GDP data show impressive growth over the last 20 to 25 years. Consumer spending is consistent with this picture.

The family income, survey-derived information is the odd man out. As noted, some people misrepresent their family income. Apart from this, the size of the average household has declined significantly - meaning that whatever income is reported is, on average, now divided among fewer people. In 1960, the average size of a US household was 3.33 persons; in 1993, it was just 2.63 - a decline of more than 20 percent. Also, the CPI data used to deflate current family income data to a constant-dollar level almost certainly greatly exaggerate inflation - thus understating income growth.

The American economy has many problems. Those who want to find areas where constructive change is needed will not lack for good candidates. It's essential that our economic measures be as accurate as possible, and that they not be politicized - distorted to advance various policy interests.