THE NOTION that America is losing its middle-class jobs has been so widely popularized that it is rapidly becoming conventional wisdom. Stories frequently appear in national newspapers and magazines with headlines like "The Disappearing Middle Class" or "A Low-Wage Explosion," while local news stories tell of the family breadwinner who lost his factory job and now finds himself responding to "help wanted" signs in fast-food restaurants.

The stories about individual workers are unfortunately true. Many workers have lost what they regarded as stable jobs and have experienced considerable distress when they were unable to find jobs that paid comparable wages. Fortunately, however, the generalizations that the stories seem to point to are wrong. For the economy as a whole, the average worker's pay has gradually increased and the share of the work force with low earnings has gradually declined.

Why has the view that jobs have increasingly slipped from middle-class to low-wage status become so popular? One reason is the intuitive appeal of explanations linked to the growth of jobs in services. Average hourly wages in services are, in fact, considerably lower than average wages in manufacturing. Moreover, the huge rise in employment -- over 50 percent during the past 20 years -- is almost entirely accounted for by jobs outside manufacturing. And it is also true that average wages, after adjusting for inflation, stopped growing after 1972 whereas real wage growth had averaged more than 2 percent a year over the '50s and '60s.

Putting these facts together it's easy to conclude that the fast growth in service jobs has caused average real wages to stagnate and the share of workers with low wages to expand at the expense of the middle class. Easy, but wrong. The slowdown in wage growth since the early '70s is associated with lower productivity gains dating from the same period -- a legitimate area of concern. But our calculations show that virtually none of that slowdown is linked to the shift to services that has been going on for 40 years.

Moreover, part of the stagnation in wages is more apparent than real. For example, much recent growth in compensation has come in the form of better "fringe" benefits such as health and pension plans. Counting non-wage benefits would jack up real wage growth by a substantial seven-tenths of a percent a year. Overcounting inflation (specifically housing costs in the Consumer Price Index) also makes real wage gains seem smaller than they actually were.

Correcting for these factors still leads to the conclusion that overall wage increases have slowed down considerably. But that slowdown has not come from an increase in poorly paid jobs as alleged in such widely publicized studies as that released by the Joint Economic Committee in December 1986.

That study, entitled "The Great American Job Machine: The Proliferation of Low Wage Employment in the U.S. Economy," assigned wage and salary workers to high-, medium- and low-earnings categories based on their annual earnings in 1973, 1979 and 1984, adjusted for inflation. The central conclusion of the study is that the share of employment in the "low-wage" category increased disproportionately between 1979 and 1984, compared with the change between 1973 and 1979. According to the authors, "nearly three fifths of the net new employment generated between 1979 and 1984 was low-wage, compared with less than one fifth during the preceding period." {See chart.}

Our re-examination of data from the same basic source (the Current Population Survey) points to radically different conclusions. Instead of a substantial increase, our analysis shows a decline in the share of new jobs with low annual earnings. The share of new jobs that paid less than 50 percent of median earnings declined from 24 percent between 1973 and 1979 to 13 percent between 1979 and 1986.

It is also important to note that most of the workers with low annual earnings worked only part-time or part of the year -- 92.5 percent in 1985. In other words, the main reason for low annual earnings is not low wages but low working hours.

The middle group of jobs did have proportionately smaller gains than in the earlier period but only because the share of higher paying jobs increased substantially.

A substantial part of the difference in our results comes from updating our analysis from 1984 to 1986 (and also using a better inflation measure). But adjusting for these technical differences does not affect our central finding: The share of jobs with low earnings has generally been declining since 1967. The changing skills and work experience of the labor force, especially the massive influx of young, entry-level workers during the past 20 years, have changed earnings distributions for some broad groups of workers. Earnings relationships have adjusted to accommodate the changing mix of workers as jobs were tailored to their qualifications and goals. Despite these compositional changes, however, the distribution of earnings for the work force as a whole has remained essentially unchanged.

Proponents of the view that job quality has very seriously deteriorated have argued that various policies should be considered to counter these adverse trends. Policies suggested include increased public works spending, raising the federal minimum wage, enacting plant-closing legislation and imposing protectionist trade barriers. The case for such policies, however, receives no support from careful analysis of trends in the distribution of earnings.

Marvin Kosters is director of economic policy studies and Murray Ross is a research associate at the American Enterprise Institute.