Inequality and institutions in 20th century America

Frank Levy, Peter Temin

15 June 2007



A central feature of post-World War II America was mass upward mobility: individuals seeing sharply rising incomes through much of their careers, and each generation living better than the last. It therefore is problematic that recent productivity gains have not significantly raised incomes for most American workers. In the quarter century between 1980 and 2005, business productivity increased by 71%. Over the same quarter century, median weekly earnings of full-time workers rose from $613 to $705, a gain of only 14% (figures in 2005 dollars), as our recent research shows.1 Detailed analysis of these years shows that college-educated women are the only large labour-force group for whom median compensation grew in line with labour productivity.

Since productivity growth expands total income, slow income growth for the average worker implies faster income growth elsewhere in the distribution. In the US case, growth occurred at the very top.  Piketty and Saez estimated that the share of gross personal income claimed by the top 1% of tax-filing units – about 1.4m returns – rose from 8.2% in 1980 to 17.4% in 2005. Among tax returns that report positive wage and salary income, the share of wages and salaries claimed by the top 1% rose from 6.4% in 1980 to 11.6% in 2005.2

To place these developments in historical perspective, we construct the following ratio: 


The numerator of this ratio is the sum of median annual earnings of full-time workers and the value of estimated fringe benefits. The denominator is Non-Farm Business Productivity – the standard labour productivity measure - expressed as an annual dollar amount. We can think of this ratio as a bargaining power index, BPI for short, since it is the share of total output per worker that the average full-time worker captures in compensation.


Figure 1 displays this Bargaining Power Index from 1950-2005. For purposes of comparison, Figure 1 also displays the Piketty-Saez estimate of the 99.5th income percentile on federal tax returns – the median income of the top 1% of reported incomes – adjusted for fringe benefits and normalised by Non-Farm Business Productivity.

In the “Golden Age” of 1947-73, labour productivity and median family income each roughly doubled. The median compensation of full-time workers (the numerator of the BPI) and labour productivity (the denominator) grew at the same rate from 1950 to the late 1970s. Simultaneously, income equality increased as very high incomes (illustrated by the 99.5th percentile) grew more slowly than labour productivity. 

In the 1970s stagflation, median compensation of full-time workers began to lag behind productivity growth, a trend that accelerated after 1980. In Figure 1, the lag is illustrated by the BPI declining from 0.6 in 1980 to 0.53 in 1990 and to 0.43 in 2005. This declining bargaining power of the average full-time worker is a useful way to describe why significant productivity growth since 1980 has translated into weak growth in earnings and compensation.

Many economists attribute the average worker’s declining bargaining power to skill-biased technical change – technology, augmented by globalisation, which heavily favours better-educated workers. In this explanation, the broad distribution of productivity-gains during the Golden Age is often assumed to be a free-market outcome that can be restored by creating a more educated workforce.

We argue instead that the Golden Age relied on market outcomes strongly moderated by institutional factors. Following the literature on economic growth that emphasises the role of institutions in economic outcomes, we argue that institutions and norms affect the distribution of economic rewards as well as their aggregate size. Our argument leads to an explanation of earnings levels and inequality in which skill-biased technical change, globalisation and related factors function within an institutional framework. In our interpretation, the recent impacts of technology and trade have been amplified by the collapse of these institutions, a collapse which arose because economic forces led to a shift in the political environment over the 1970s and 1980s. If our interpretation is correct, no rebalancing of the labour force can restore a more equal distribution of productivity gains without government intervention and changes in private sector behavior. 

We do not challenge the existence of technology’s and trade’s effects on labour demand.3 Rather, we argue that technology and trade’s impacts are embedded in a larger institutional story. We argue that the current trend toward greater inequality in America is primarily the result of a change in economic policy that took place in the late 1970s and early 1980s. The stability in income equality where wages rose with national productivity for a generation after the Second World War was the result of policies that began in the Great Depression with the New Deal and were amplified by both public and private actions after the war. This stability was not the result of a natural economy; it was the result of policies designed to promote it. We have termed this set of policies the Treaty of Detroit, after the most famous labour–management agreement of the postwar years.

This agreement was replaced in the 1980s (and surrounding years) by another set of institutional arrangements which we call the Washinton Concensus.4 These new policies also originated in a time of economic distress, albeit nowhere near the distress of the 1930s. In a process similar to the experience of the Great Depression, policy-makers – unable to comprehend the macroeconomic causes of distress – instituted microeconomic changes in an attempt to ameliorate the macroeconomic problems. In both cases, the measures taken were only partially successful, and recovery came from diverse influences. The microeconomic changes, however, had durable impacts on the distribution of economic production.

These microeconomic changes were not inevitable. Different labour-market institutions within Western Europe are compatible with similar rates of unemployment, and different labour-market institutions in Western Europe and America appear to be compatible with similar rates of economic growth. Rapidly rising incomes among the very rich appear in the U.S., England and Canada (largely in response to U.S. competition.) but do not appear in most continental European countries or Japan. 

Globalisation clearly does not determine institutions. Some economists and commentators have asserted that globalisation has made more than one set of institutions not viable, yet the variety of institutions that are found in Western Europe shows only very limited signs of disappearing. Finally, economic shocks do not determine institutions. The Vietnam War and the oil shocks deranged the international economy, yet countries responded to these shocks in idiosyncratic ways. The contrast between the US and Japan in the 1970s is only one example of the great diversity.

Deregulation, floating exchange rates, international capital mobility, low minimum wages and taxes, and the destruction of labour unions, were not unique responses to the oil crisis or the productivity collapse. The effects of these policies have been amplified by skill-biased technical change and, in the extreme, winner-take-all markets. But the technology did not fully determine who received the rents produced any more than technology fully determined who got the rents from the great postwar expansion; African-Americans were largely excluded from the GI Bill and other public policies by a series of political and bureaucratic actions.

The elements of the Washinton Concensus were adopted in the name of improving economic efficiency. But there is growing recognition that the current free-market income distribution – the combination of large inequalities and stagnant wages for many workers – creates its own “soft” inefficiencies as people become disenchanted with existing economic arrangements.  People suffering from stagnant incomes —both here and in some similar countries—have begun to protest. Our analysis suggests that the trends in the distribution derive in part from the shift from one complex set of policies to another—from the Treaty of Detroit to the Washinton Concensus. There is no single determinant, whether education, minimum wage, capital or labour mobility, that determines the path of income distribution. Any specific measure therefore can alleviate the distress of some people, but it cannot change the overall distributional trends shown in our graphs.

The last six years of U.S. federal tax history have involved an inhospitable politics in which winners have used their political power to expand their winnings. But political sentiment does shift. Economic distress like that of the 1930s can induce such a shift. Even the smaller economic distress of the 1970s was enough to redirect American economic policy. Only time will tell if more economic distress is needed to change policy yet again.




1 Levy, Frank, and Peter Temin, “Inequality and Institutions in 20th Century America,” NBER Working Paper 13106, May 2007.

2 See Thomas Piketty and Emmanuel Saez. 2003. “Income Inequality in the United States,” Quarterly Journal of Economics, and the updating of their figures to 2005 on Emmanuel Saez’ website. Their calculations are based on pre-tax market income (wages, partnership income, interest, dividends, rents, etc.), excluding transfer payments. A tax-filing unit represents a tax return (which may be single or joint). Piketty and Saez estimate the total number of tax-filing units that would occur if all U.S. households filed federal income taxes and figures like the “top 1% of tax-filing units” refer to the top 1% of that estimated number rather than the top 1% of those who actually file.

3 Card, David, and John E. DiNardo 2002. “Skill-Biased Technical Change and Rising Wage Inequality: Some Problems and Puzzles.” Journal of Labor Economics, 20 no 4 (October), pp. 733-783.

4 This term normally is used for LDCs, but the spirit of this concept applies well to the changing institutions within the United States. We use the term here to refer to the microeconomic policies of deregulation and privatisation of the consensus, not the macroeconomic policies of fiscal discipline and stable exchange rates.



Topics:  Economic history Poverty and income inequality

Tags:  Labour Markets, labour market policy

Professor of Urban Economics at MIT

Elisha Gray II Professor Emeritus of Economics, Massachusetts Institute of Technology