As the election nears, a fair assessment is that the economy is actually doing well. Gross domestic product has grown at more than 4% per year since early 2003, and investment and productivity have been strong. Inflation seemed to tick up for a while, but the consumer price index for the last two months has been tame. And after lagging, the labor market has seen a solid expansion of jobs in 2004. Since mid-2003 the unemployment rate has fallen by nearly a full percentage point, to 5.4%, below the average rate of 5.6% since 1948. This favorable picture is unlikely to change much before the election, because only two major reports remain -- the September employment report on Oct. 8 and the third-quarter GDP report on Oct. 29.
Information about the labor market comes from two main sources, the payroll and household surveys. The payroll survey is good for assessing changes in employment. This survey indicates that 1.7 million net jobs have been created since the low point in August, 2003, but 0.9 million net jobs have been lost since the start of the Bush term in January, 2001. In the early part of the term, weak employment growth reflected the 2001 recession. The weak recovery in 2002 can be explained partly by the terrorist attacks of September 11.THE LATEST PAYROLL REPORT FOR AUGUST was more favorable than generally recognized. Employment rose by over 200,000 when one factors in the upward revision for July. Some say that 150,000 to 200,000 new jobs are needed each month to keep up with labor-force growth. This estimate seems wrong, because the great expansion of labor-force participation from the 1960s through the 1990s, which was driven especially by women, has probably ended. It's reasonable to expect that the overall participation rate will now expand with population at 1% per year. Thus job creation of about 100,000 per month is sufficient to keep up with labor-force growth.
The other measure of labor market performance, the household survey, shows that for more than a year employment growth has surpassed labor-force growth. This is why the unemployment rate fell from 6.3% in June, 2003, to 5.4% today. The problem with the household survey is not that the survey covers only 60,000 households, a number bigger than those used to measure electoral sentiment. The real difficulty is that we do not know the underlying population of households. This information arrives only every 10 years from the Census. Therefore, the household survey is unreliable for gauging month-to-month changes in employment and the labor force. Still, it is good for measuring the unemployment rate, which is the fraction of persons who want jobs but do not have them.
A one-month drop in the unemployment rate signals a stronger labor market even if the measured labor force declines in the month. In August, 2004, the unemployment rate fell from 5.5% to 5.4%, while the measured labor force fell by 152,000. By comparison, in May, 2004, the unemployment rate stayed at 5.6%, while the measured labor force expanded by 233,000. It is reasonable to infer that August was stronger than May.
The payroll survey also has deficiencies -- failing to count the self-employed and workers at home and on farms. The household survey is better in these respects, and the month-to-month errors in measuring labor-force changes may not be so serious over longer periods. Thus, it is significant that household employment increased by 1.9 million since January, 2001 -- 2.8 million better than the payroll survey.
Traditionally, the unemployment rate has been accepted by Democrats and Republicans as one of the most important macroeconomic indicators. Thus it is disturbing that during this Presidential campaign, Democrats are trying hard to ignore the strength implied in the falling unemployment rate. If the rate were 11%, as in the worst of the 1982-83 recession, Democrats would be appropriately screaming. Robert J. Barro is a professor of economics at Harvard University and a senior fellow of the Hoover Institution (email@example.com)