The reality of the American Dream

American workers have been battered by a bewildering array of powerful and sometimes contradictory forces over the past fifty years: globalisation; technological change; immigration; declining trades unions. Each of these trends, all important in their own right, have been carefully studied by economists to identify the winners and losers in society and the resulting shifts in the gap between rich and poor. But in an ambitious new paper, CEPR Research Fellow Robert Gordon and his co-author, Ian Dew-Becker, attempt to disentangle these and other factors, and to understand the key causes of the rise in inequality in the US.

By reviewing a multitude of separate pieces of contemporary research into the extent and causes of inequality, and adding a number of new insights of their own, the authors are able to corroborate some explanations for rising inequality, and dismiss others.

They begin by examining the empirical facts about various measures of inequality since the 1950s. One common piece of evidence often advanced for the declining fortunes of the median worker in the US is that the share of labour income in GDP has declined; but Gordon and Dew-Becker reject this idea showing that once the labour earned by those who are business-owners is also included, the share has barely changed since 1950. Instead, they argue, it is the 'skewness' of the labour income towards those at the very top which has changed.

To get a better picture of this skewness they chart how the ratio between earners at different points in the income scale has shifted since 1979. In general, workers at the 90th percentile - the boundary for the top tenth of earners - have taken home a larger multiple of the earnings than those at the 50th and the 10th percentile over the years. There are, however, strikingly different patterns for men and women.

For men, there was a large jump in the gap between 50th and 10th percentile between 1979 and 1986, but a recovery by 1998. Similarly, the largest jump in the gap between the 90th and 10th percentiles had taken place by 1987. This was the period in which unionisation of the workforce declined particularly sharply, and Gordon and Dew-Becker suggest it is therefore consistent with the explanation that falling trade union power has been one factor in widening inequalities.

For women, however, the gap between income levels at the 90th percentile and the 10th percentile increased much more sharply and the effect persists into the most recent evidence. This different pattern supports the idea of a role for the statutory Minimum Wage in determining women's income. Around twice as many women as men are subject to the wage, which declined in real terms through the 1980s and into the early 1990s. It was worth 45% of average hourly earnings in 1979 and 31% by 2005; but the sharpest decline in real terms was between 1980 and 1986, coinciding neatly with the period when the gap in women's earnings between top and bottom widened most rapidly.

The decline in unionisation has been one explanation advanced for the reversal of what Goldin and Mayo, in a 1992 paper, referred to as the 'Great Compression' in the income distribution between World War II and about 1970. They suggested trade and immigration as other factors and Gordon and Dew-Becker examine these in turn.

Rising imports have been advanced as an explanation for widening income inequality because trade tends to be concentrated in labour-intensive industries such as manufacturing; falling prices as a result of rising imports have depressed the marginal product of the average unskilled worker and hence tended to reduce wages; and globalisation has redirected surplus funds towards foreign investment opportunities instead of potentially job-creating projects at home.

There is plenty of economic evidence about the impact of imports on inequality, some of it ambiguous or contradictory. The authors cite an influential paper by Feenstra and Hanson suggesting that outsourcing, and the resulting decline in relative demand for unskilled workers, accounts for 15-25% of the wage gap between the top and bottom of the scale. More recent work by Lawrence, however, in a book published earlier this year, suggested that the influence of trade on wage distribution has declined in more recent years.

For immigration, the authors find there is evidence of some downward effect on the wages of lower-skilled workers as they compete with new arrivals from overseas; but recent research reveals the largest impact has tended to be on previous waves of immigrants who specialise in the types of occupations, such as in the hospitality industry for example, which new migrants are also likely to work in.

Another common explanation among economists for shifts in the income distribution is what has become known as 'skill-biased technological change' - the idea that as new innovations, such as computers, have become more widespread, the mixture of types of worker needed has altered. Specifically, more highly-skilled workers are required in this new, hi-tech world; so wages at the top end of the distribution tend to rise faster than those at the bottom.

Gordon and Dew-Becker have questioned this theory in the past pointing out that the wages of several key occupational groups, including engineers and computer technicians, have not risen rapidly in real terms. However, they welcome more recent research by Katz and Kearney, which splits the workforce into three distinct groups. At the top are those carrying out 'non-routine, cognitive' tasks, such as investment bankers, lawyers and so on. These workers are highly skilled, but their jobs are also very difficult to outsource to lower cost economies because they require direct interaction with clients, defendants and so on.

In the middle of the distribution are a very large group of people carrying out routine, repetitive work, such as accountancy and certain kinds of engineering work, which are more prone to being outsourced to cheaper locations (such as IT helpdesks and other 'back office' functions being relocated to India, for example).

Thirdly, at the bottom, there is a group of workers whose jobs are manual but require direct interaction, such as nurses, truck drivers and waiters, whose work it is extremely difficult to outsource.

Gordon and Dew-Becker argue that this more subtle three-way split helps to show how the forces of technological change have hit different groups in different ways - and in particular how the median worker, in that middle, more vulnerable group, has seen their income decline relative to those in the top 10% of the earnings scale.

Finally, the authors also examine one of the most contentious issues in the debate about American inequality - the pay of those at the very top of the pile, in particular CEOs, who have often walked away with billions of dollars in stock options on top of their salaries.

Gordon and Dew-Becker separate CEOs from two other groups who are also in the very top of the distribution, but whose compensation is more closely tied to the market. 'Superstars,' such as actors, music artists and top sports stars have benefited from technological changes - such as video, cable TV, mass distribution of CDs, and so on - which have hugely amplified the size of the audience they can reach with the same amount of effort, and hence increased the rewards enormously.

A long way below them, but still tied to the vicissitudes of the market, are professionals such as investment bankers. This group has also benefited from technological change - electronic trading means more deals can be done, faster, giving them a slice of a larger pie- but they must still meet and deal with clients face to face so cannot be catapulted into the superstar bracket.

But CEOs, the authors argue, are in a category all of their own. A large number of pieces of research suggest that they have been able to secure disproportionately large rewards in recent years. Bebchuck and Fried show in one recent paper, for example, that among 1500 firms reporting to the Securities and Exchange Commission, CEOs took home 5% of company profits in 1993-5; but that had risen to an extraordinary 12.8% by 2000-02.

Bebchuck and Fried argue that the classic principal-agent model, with board members acting as the agents of shareholders in setting executive pay, simply cannot explain the sharp rise in compensation over the past decade.

Instead, they suggest CEOs can in effect control their own pay, subject only to the 'outrage constraint' - that if shareholders are sufficiently appalled by the size of a particular package they may vote it down. Directors are in theory independent but they are often paid large salaries, the fear of the loss of which is likely to considerably outweigh any potential benefits from insisting on better governance.

The authors argue that the lengths boards often go to in disguising the true size of executive compensation, by providing other benefits such as housing, pensions, low interest loans and so on, supports the idea that there is some concern about the 'outrage constraint' - otherwise why not just pay it all in cash? In general, Gordon and Dew-Becker argue, 'CEOs, through compensation committees and inbreeding of board directors, have a unique ability to control their own compensation.'

Assessing the various causes of rising inequality in the US over the past thirty years is a complex task, but Gordon and Dew-Becker try to bring together a range of different explanations. They find that declining union power, the falling real value of the minimum wage, rising immigration and rapid growth in world trade have all had at least some part to play - and it is informative to look separately at the patterns of inequality for men and women. Technological change has had an impact - both on separating out 'superstars' from their less talented peers, and exposing certain occupational groups to the threat of outsourcing and hence lower wages.

But Dew-Becker and Gordon's most striking contribution is in exposing the fact that where the immense rewards falling to America's CEOs are concerned, finding a rational, market-based explanation is impossible.

DP 6817 Controversies About the Rise in American Inequality - a Survey

Robert J Gordon and Ian Dew-Becker