Union decline directly caused rising income inequality

The claim

The collapse of US union density — from roughly 35% of the workforce in the mid-1950s to 10% today, and just 6% in the private sector — is not merely a symptom of deindustrialization or technological change. It is a primary structural driver of rising wage inequality and the stagnation of middle-income earnings. The claim holds that this decline was substantially produced by deliberate policy choices — from the Taft-Hartley Act of 1947 through Reagan’s dismissal of PATCO strikers in 1981, the proliferation of right-to-work laws, aggressive employer campaign tactics enabled by NLRB rules, and the Supreme Court’s 2018 Janus decision — and that it could therefore be reversed by different policy choices.

The mechanism

Unions compress wages in two ways. First, they raise wages directly for their members through collective bargaining. Second — and this is the underappreciated mechanism — they establish wage norms across non-union firms in the same industries and regions. Employers competing for workers in heavily unionized labor markets must approximate union wages to attract and retain workers. When union density collapses, this normative pressure evaporates. Non-union workers in formerly unionized sectors lose wage floors they never formally held.

Western and Rosenfeld (2011) call this the “sword effect”: unions do not merely redistribute income within unionized firms — they alter the wage-setting institutions of the entire labor market. As density falls, the sword effect weakens, and wage-setting reverts to pure market power — which, in a concentrated employer landscape, systematically favors capital over labor.

The mechanism breaks down somewhat at the upper tail. Union decline does not easily explain the explosion of incomes among the top 1% — that requires a separate account involving capital income, financialization, and executive pay norms. What union decline explains with greater precision is the widening of the gap between the bottom 90% and the top 10% — the inequality most directly relevant to workers’ material circumstances.

The evidence

Western and Rosenfeld (2011): the foundational decomposition

The most rigorous direct test is Western and Rosenfeld’s 2011 paper in the American Sociological Review. Using Current Population Survey data from 1973 to 2007, they decompose the rise in male wage variance into components attributable to observable factors. Union decline accounts for 20–33% of the rise in male wage inequality over this period, and 8–13% of female wage inequality. These are large shares of variance for any single institutional variable. The authors explicitly model the sword effect, estimating what non-union wages would have been in the counterfactual where union density remained at 1973 levels.

The Canada quasi-experiment

Card, Lemieux, and Riddell (2004) exploit the fact that Canada and the United States share a continent, a language, trade exposure, and major industries, but diverged sharply in union policy after the 1940s. Canadian law — particularly at the provincial level — imposed fewer procedural barriers to organizing, prohibited permanent striker replacement in most provinces, and used card-check certification in several jurisdictions rather than NLRB-style delayed elections. The result: Canadian union density held near 30% through the 1990s while US density halved. The study finds that roughly one-third of the US–Canada gap in male wage inequality is directly attributable to the union density differential. The industries are the same; the technology is the same; the outcomes are different because the institutional rules are different.

Cross-national union density and Gini coefficients

The relationship between union density and wage inequality across OECD countries is one of the most robust patterns in comparative political economy. The 2022 OECD data:

Sources: OECD Trade Union Dataset (2023); OECD Income Distribution Database (2024).

Germany’s case is instructive: its nominal union density has also declined, but sectoral bargaining agreements continue to set wages for roughly 54% of workers regardless of formal membership. The wage floor effect — the sword effect operating through collective agreements — remains in place. The US dismantled this mechanism at both levels.

Policy causation: Taft-Hartley and the legal architecture of decline

The National Labor Relations Act of 1935 (Wagner Act) established a relatively permissive organizing environment. The Taft-Hartley Act of 1947 rewrote the rules in several ways that directly impeded organizing: it prohibited closed shops, authorized state right-to-work laws (allowing workers to free-ride on union contracts without paying dues), restricted secondary boycotts, and required union officials to sign anti-Communist affidavits. The AFL-CIO called it a “slave labor act.” Its passage was followed by the first sustained reversal of union density growth.

Subsequent decades layered additional barriers. NLRB election procedures require a petition, then an election campaign that can extend 3–6 months — a window in which employers can legally hold mandatory “captive audience” meetings and deploy consultants. EPI’s analysis of NLRB data finds that employers challenge certification through unfair labor practice filings in roughly 25% of elections, extending the time to first contract by a median of over a year. By this point, worker turnover has frequently replaced the organizing cohort.

NLRB election win rates and certification delays

Union win rates in NLRB elections have fluctuated between 60–68% in recent years — meaning unions win the majority of elections they contest. The organizing bottleneck is not vote outcomes but the pre-election environment. Workers who sign authorization cards report employer coercion at high rates: a 2009 Hart Research survey found 52% of non-union workers said they would vote union “if an election were held today,” but employer opposition systematically converts or suppresses votes during the campaign window.

Janus and public-sector density

Janus v. AFSCME (2018) overruled Abood v. Detroit Board of Education (1977) and held that requiring public-sector workers to pay agency fees to unions violated the First Amendment. The decision converted all 22 states that had mandatory agency-fee rules to de facto right-to-work status for government employees. Biasi and Sarsons (2022) estimate the decision reduced union revenue by approximately $110 million annually within two years, with downstream effects on political capacity and bargaining leverage. The decision was the product of a coordinated multi-decade legal campaign funded by the National Right to Work Legal Defense Foundation and allied conservative foundations.

Who benefits

The financial beneficiaries of low union density are identifiable and have funded the policy campaign directly:

• The National Right to Work Legal Defense Foundation has received substantial support from the Koch network (the Charles Koch Foundation, Americans for Prosperity) and the Bradley Foundation. It litigated Janus and continues to file decertification suits.
• The American Legislative Exchange Council (ALEC), primarily funded by Koch Industries, the Walmart Walton Family Foundation, and major pharmaceutical companies, has drafted and promoted right-to-work model legislation adopted in 28 states. Right-to-work passage is directly associated with subsequent union density declines and wage compression.
• Major retail and logistics employers — Walmart and Amazon in particular — have invested extensively in union avoidance consulting. Walmart’s reported expenditure on union avoidance training ran to tens of millions of dollars annually in the 2000s. Amazon’s successful campaign against the RWDSU in Bessemer, Alabama in 2021 (later overturned by the NLRB) drew on a well-funded professional opposition infrastructure.
• The Business Roundtable has consistently lobbied against the Employee Free Choice Act (which would have allowed card-check certification and mandatory arbitration of first contracts) and against NLRB rule changes that would shorten the election period.

These organizations have concrete wage-bill interests in suppressed unionization. A 10-percentage-point increase in union density in the retail sector would, based on existing union wage-premium estimates (10–15%), add billions in annual labor costs for large-format retailers.

The counter

The strongest counterargument is that union decline is substantially caused by economic forces — deindustrialization, trade exposure, the shift toward services — that would have occurred regardless of legal framework. Manufacturing employment would have declined even if Taft-Hartley had not passed. Under this view, union law changes are largely epiphenomenal: they ratified and accelerated a shift driven by technology and globalization, but did not cause it.

This position has genuine empirical support. DiNardo, Fortin, and Lemieux (1996) find that deindustrialization and the fall of the real minimum wage together explain a substantial share of wage inequality growth independent of density. Autor, Levy, and Murnane (2003) document skill-biased technological change as a persistent driver of wage dispersion that operates regardless of union status. Trade exposure from China import competition (Autor, Dorn, and Hanson, 2013) contributed to manufacturing employment collapse in ways that no organizing regime would have fully reversed.

The steelman version of the counter says: the residual policy effect is real but smaller than structural accounts claim, and the counterfactual of sustained high union density in a globalized economy is implausible given competitive pressures on tradable-sector employers.

The response from the structural side is that the Canadian counterfactual already embodies this test — Canada faced the same trade exposure and deindustrialization, retained higher union density through different policy choices, and has consistently lower wage inequality as a result. Policy is not the only variable, but it is a causally independent one.

References

Autor, D. H., Dorn, D., & Hanson, G. H. (2013). The China syndrome: Local labor market effects of import competition in the United States. American Economic Review, 103(6), 2121–2168. https://doi.org/10.1257/aer.103.6.2121

Biasi, B., & Sarsons, H. (2022). Flexible wages, bargaining, and the gender gap. Review of Economic Studies, 89(6), 2941–2977. https://doi.org/10.1093/restud/rdab081

Card, D., Lemieux, T., & Riddell, W. C. (2004). Unions and wage inequality. Journal of Labor Research, 25(4), 519–559. https://doi.org/10.1007/s12122-004-1011-z

DiNardo, J., Fortin, N. M., & Lemieux, T. (1996). Labor market institutions and the distribution of wages, 1973–1992: A semiparametric approach. Econometrica, 64(5), 1001–1044. https://doi.org/10.2307/2171954

Economic Policy Institute. (2021). Unlevel playing field: How the broken NLRB election process undermines workers’ freedom to organize. https://www.epi.org/publication/unlevel-playing-field/

Freeman, R. B. (2007). America works: Critical thoughts on the exceptional US labor market. Russell Sage Foundation.

Hirsch, B. T., & Macpherson, D. A. (2003). Union membership and coverage database from the Current Population Survey: Note. Industrial and Labor Relations Review, 56(2), 349–354. https://doi.org/10.1177/001979390305600208

OECD. (2023). OECD/AIAS ICTWSS database on institutional characteristics of trade unions, wage setting, state intervention and social pacts. https://www.oecd.org/employment/ictwss-database.htm

OECD. (2024). Income distribution database [Gini coefficients, disposable income]. https://stats.oecd.org/Index.aspx?DataSetCode=IDD

Western, B., & Rosenfeld, J. (2011). Unions, norms, and the rise in U.S. wage inequality. American Sociological Review, 76(4), 513–537. https://doi.org/10.1177/0003122411414817