They conclude that changes in labor's share of income play no role in rising inequality of labor income: by one measure, labor's income share was almost the same in 2007 as in 1950.

However, there are gender differences in income inequality: between 1979 and 2005, for example, the income gap between women working for the median wage (the 50th percentile) and low-earning women (at the 10th percentile) grew much more than it did for men at those income levels during the same period. That suggests that the decline in the real value of the minimum wage over that time played a causal role, the authors argue. That's not surprising, in one sense, since women are, roughly speaking, twice as likely to work for the minimum wage as men are.

If women were more affected by the minimum wage, men bore the brunt of the decline in unionization over the least three decades, the survey finds. One study the authors cite suggests that the fall in organized labor's share of the workforce can explain 14 percent of the rise in the variance among male wages between 1973 and 2001 (but it had no apparent effect on the variance of female wages).

There is little evidence on the effects of imports. And, the ambiguous literature on immigration implies a small overall impact on the wages of the average native-born American, a significant downward effect on the wages of high-school dropouts, and a potentially large impact on previous immigrants who work in occupations in which immigrants specialize.

The authors introduce two new issues, disparities in the growth of price indexes and in life expectancy between the rich and the poor. "While the poor may do better when price indexes are corrected, they do much worse when their health outcomes are considered," the authors write. They cite evidence that between 1980 and 2000 the life expectancy of the bottom 10 percent of earners increased at only half the rate of the top 10 percent. "This may be the most important single source of the increase in inequality in the United States, and it combines not only unequal access to medical care services and insurance, but also to differences in personal habits and environment related to education and income," the authors conclude.

The most controversial section of the survey looks at the question of why the rich have gotten so much richer. In a 2005 study, the authors found that the top 10 percent of earners saw their share of overall income rise from 27 percent in 1966 to 45 percent in 2001. But that study also documented that fully half of that increase came from the relative gains made at the very top of that spectrum – those at the 95th percentile and above. That study also distinguished between "superstars," whose incomes were market-driven, and CEOs, whose incomes were "chosen by their peers." In their new survey, the authors carve out a third group – high-income professionals, especially lawyers and investment bankers, whose pay is market-driven but who don't enjoy the benefits of "audience magnification," whereby the superstars can fill entire arenas or sell recordings to millions of people. Their point: income inequality is growing even among the top 10 percent of earners as the superstars an! d CEOs increase their pay faster than lawyers and investment bankers. But at least the pay of the superstars, lawyers, and investment bankers is market-driven. The pay of CEOs is not.

Their review of the CEO debate places equal emphasis on the market, in showering capital gains through stock options, and an arbitrary management-power hypothesis based on numerous non-market aspects of executive pay. "CEOs, through compensation committees and inbreeding of boards of directors, have a unique ability to control their own compensation," the authors write. "Furthermore, if a director approves a higher compensation package, that may subsequently lead her to receive more compensation at her own firm."

They cite one study of 1,500 firms that found that the compensation earned by the top five corporate officers in 1993-5 equaled 5 percent of their firms' total profits during that period; by 2000-2, that ratio had more than doubled to 12.8 percent. The trend was caused in equal parts by arbitrary pay decisions by corporate boards and by the showering of stock options on CEOs, they conclude.

Furthermore, the survey cites a study showing "ample evidence that firms work to disguise the magnitude of CEO pay," such as lifetime healthcare, below-market-rate loans, and above-market-rate loans when CEOs defer their compensation, to lessen shareholder outrage. Such research "is important because it tells shareholders what to expect and where their outrage constraint should be set," the authors write.

This skewing of pay at the very top in the United States contrasts with other countries, especially Japan. There, the income share of the top 0.1 percent peaked at 9.2 percent in 1938, reached stability of close to 2.0 percent after 1947, and ended up at 1.7 percent in 1998. Initially, America also saw an initial peak (8.2 percent in 1928) fall to a low (1.9 percent in 1973). But then the income share of America's top 0.1 percent rebounded (7.3 percent in 2000). Canada and the United Kingdom mimicked the U.S. pattern, though their most recent upturns were less dramatic. France, like Japan, has seen the income share of its very top earners stay quite stable since the mid-1940s.

Why the disparity? America's CEOs have had their pay inflated by generous stock options and having their pay set by peers on corporate boards, the survey finds, as well as institutional differences between the United States and other nations, including such things as regulations, unions, and social norms.