GDP continues to grow, the stock market flirts with record highs, and workers produce more per hour than ever before. Yet polls show that most Americans disapprove of President Bush's handling of the economy.

Republican political consultant Frank Luntz explains the apparent conundrum thusly: "Some people who aren't partisans say, 'Yes, the economy's pretty good, so why are people so agitated and anxious?' The answer is they don't feel it in their paychecks." [Real Wages Fail to Match a Rise in Productivity, By Steven Greenhouse And David Leonhardt, New York Times, August 28, 2006]

Mr. Luntz is 80 percent right. The richest 20 percent of American households—and only the richest 20 percent—have enjoyed higher real incomes during the Bush expansion. Everyone else has lost ground; the lowest 20 percent has actually lost a full 1.8 percent. (For details, click here: Table 1.)

With the economy now slowing, the current recovery is on course to become the first since World War II in which incomes of most workers declined.

This is new. From the end of World War II until the late 1960s, the rich-poor divide was remarkably stable, even narrowing over long stretches. [Table 2] Things started to come apart around 1970, as can be seen by eyeballing the trend in mean and median family income:

Mean is the average income, calculated by dividing total income by the number of families. Median is the mid-point, the income at which half families are above and half below.

You may recall from Statistics 101 that if all the objects (e.g., family incomes) in a sample grow at the same rate, its mean and the median will move in lockstep. If, however, the top half grows faster (or falls more slowly) than the bottom half, the mean will pull away from the median.

Such pulling away is painfully evident in the graphic, especially—and we think not coincidentally—in the years following the 1986 illegal alien amnesty. In 2005, mean family income was a record 30.4 percent above median family income. In 1986, the gap was just 18.6 percent. [Table 2]

Until recently, economists rarely mentioned the I-word when explaining the income distribution. The consensus among most academics was that the primary cause of increased inequality was "skill-biased technical change" (SBTC)—i.e., increased economic rewards to educated, technically savvy workers.

In a word, SBTC compensation was based on merit. How quaint!

Northwestern University economists Ian Dew-Becker and Robert J. Gordon broke from the group naiveté in a paper published last year:

"If SBTC had been a major source of the rise in inequality, then we should have observed an increase in relative wages of those most directly skilled in the development and use of computers. Yet in the 1989-97 period….total real compensation of CEOs increased by 100 percent, while those in occupations related to math and computer science increased only 4.8 percent and engineers decreased by 1.4 percent." [Where did the Productivity Growth Go? Inflation Dynamics and the Distribution of Income, (PDF) Ian Dew-Becker and Robert J. Gordon, Northwestern University]

In debunking SBTC the authors make a broader historical point regarding immigration:

"To be convincing, a theory must fit the facts, and the basic facts to be explained about income equality are not one but two, that is, not only why inequality rose after the mid-1970s but why it declined from 1929 to the mid-1970s. Three events fit neatly into this U-shaped pattern, all of which influence the effective labor supply curve and the bargaining power of labor: (1) the rise and fall of unionization, (2) the decline and recovery of immigration, and (3) the decline and recovery in the importance of international trade and the share of imports…"

"Partly as a result of restrictive legislation in the 1920s, and also the Great Depression and World War II, the share of immigration per year in the total population declined from 1.3 percent in 1914 to 0.02 percent in 1933, remained very low until a gradual recovery began in the late 1960s, reaching 0.48 percent (legal and illegal) in 2002. Competition for unskilled labor not only arrives in the form of immigration but also in the form of imports, and the decline of the import share from the 1920s to the 1950s and its subsequent recovery is a basic fact of the national accounts."

Of course, immigration is not the only factor skewing the distribution. The enormous income gains in the top 1 percent, for example, are attributed by the authors to a relative handful of sports and entertainment superstars, plus CEOs who enjoy

"…a halo of reputation that leads a board of directors to shower him or her with tens of millions of compensation, often without corresponding performance, when an equally capable but less famous alternative might have been willing to do the job at one-tenth the compensation."

But for ordinary workers, that "alternative" is increasingly immigration—and stagnant incomes.

Edwin S. Rubenstein (email him) is President of ESR Research Economic Consultants in Indianapolis.