Explaining the erosion of private-sector unions: How corporate practices and legal changes have undercut the ability of workers to organize and bargain

Download PDF

Press release

Unequal Power

Part of the Unequal Power project, an EPI initiative to reestablish the understanding in law, politics, economics, and philosophy, that equal bargaining power between workers and employers does not exist. Recognizing this inherent workplace inequality will bolster freedom, economic fairness, workplace protections and democracy.

Executive Summary

An increasing volume of research demonstrates that erosion of worker bargaining power and collective bargaining have led to wage suppression and the deterioration of labor’s share of income. At the same time, bold and robust policy proposals to strengthen workers’ bargaining power have risen to a new level of priority for the center-left. President-elect Joe Biden has produced an extensive proposal to strengthen workers’ ability to form unions, and a comprehensive reform of the National Labor Relations Act (NLRA) recently passed the U.S. House of Representatives.

A full appreciation of the need for comprehensive labor law reform requires an understanding of the serious shortcomings in current law and how they have been exploited over the years by employers resisting efforts by their workers to form unions. Structural weaknesses in the law, exacerbated by anti-union amendments to the NLRA in 1947 and aided by a series of rulings by the National Labor Relations Board (NLRB) and courts, have allowed employers to interfere in and defeat efforts by their workers to organize unions and to face no real consequences for doing so. The full effect of these trends can be seen by analyzing how dramatically new unionization fell in the 1970s—a trajectory from which the labor movement has never recovered.

This paper explains what happened to private-sector unionization in the 1970s by examining data on union elections and workers’ ability to achieve an initial collective bargaining agreement. After showing that a dramatically smaller percentage of workers have been successful at forming a union and winning a first contract, the paper examines the changes in employer anti-union behavior that contributed to this result.

Among its key findings, the paper shows that in the 1950s and 1960s more than 1% of those employed participated in an NLRA election each year. In the 1970s that share fell to 0.78% and in the 1980s to 0.29%. In addition, workers began to lose elections at a higher rate in the 1970s in the face of increased employer resistance. In the 1940s, workers won a union in 80 percent of NLRB representation elections, but by 1977 workers were losing more than half of these elections. And, while 86% of workers who chose a union were able to win a first contract in the 1950s, that share declined to less than 70% in the 1970s. By the 1990s, it was down to 56%. Putting these three pieces together—participation in elections, successful elections, and winning a first contract—we show that while 0.46% of the workforce was able to make it across the unionizing finish line in the 1966–1968 period, only 0.17% of the workforce was able to do so by 1978–1980.

In addition, by the 1970s employers were charged with committing significantly more unfair labor practices (ULPs), such as firing union activists during organizing campaigns. ULP charges against employers rose sevenfold between 1950 and 1980. Starting in the 1970s, employers also made greater use of the “free speech” rights included in the Taft-Hartley amendments of 1947, holding mandatory “captive audience” meetings to voice opposition to unions and make thinly veiled threats about what could happen if workers organized. Employers also began far more extensive use of a growing “union avoidance” industry of consultants. While there were just a handful of anti-union consulting firms in the beginning of the 1970s, by decade’s end there were hundreds, and a management consultant told Congress in the late 1970s that his industry had grown tenfold over the preceding decade.

Employers were able to defeat unions so effectively because, over the years, labor law had become heavily tilted against workers and toward employers. Though these employer-friendly laws were on the books in the 1940s, 1950s, and 1960s, it was not until the 1970s that employers began to take full advantage of their power. Several key sources set the stage for this 1970s unraveling of workers’ bargaining power under the law. First, a Republican Congress largely neutered workers’ leverage in passing the 1947 Taft-Hartley Act over President Truman’s veto. Second, Taft-Hartley forced the NLRB to prioritize litigation against unions for engaging in so-called secondary activity over all other cases, including cases involving illegal firings of union supporters. Third, the law’s ineffective remedies became obvious, and the NLRB’s efforts to hold employers accountable for violating the law were stymied in the courts. Fourth, employers increasingly found an ally in the U.S. Supreme Court, which issued a series of decisions restricting workers’ rights and limiting employers’ bargaining obligations. Finally, employers started making greater use of replacement workers during strikes—a trend that grew in the 1970s and 1980s and significantly undermined workers’ right to strike. The cumulative impact of these factors meant that by the 1970s the law did not effectively protect workers’ bargaining power and gave employers a wealth of tools to resist unionization.

Legislative efforts to strengthen the law in the 1960s, 1970s, and 1990s were thwarted by an organized and united business community that stepped up to vigorously oppose and, through a filibuster by a minority of senators, defeat all attempts at legislative reform.

By telling the story through statistics, labor history, and the law about the various factors that resulted in the decline of unionization, this paper provides a more accurate accounting of the decline of unionization than do the frequent assertions of globalization or automation as the driving forces. Our empirical assessment of the role of globalization and automation focuses on the impact of the shrinkage on manufacturing employment. We provide detailed statistical analyses showing that at most one-fifth of the decline is due to manufacturing’s erosion, and provide evidence of the severe declines in union coverage in nonmanufacturing sectors (e.g., utilities, transportation, construction, mining, and communications) and in many nonmanufacturing industries (e.g., grocery stores, bus transportation, newspapers, metal ore mining, and building services). A review of various analyses of wage determination also casts doubt on a dominant role of automation and globalization on private-sector union decline. Last, an examination of international comparisons of union erosion also confirms a minor role for manufacturing decline, finding that the pace, intensity, and timing of union decline does not correspond to manufacturing’s decline.

Survey research confirms that working people want unions: Recent polling shows that nearly half of nonunion workers would vote to have union representation if given an opportunity to do so on their current job.

Labor law has not kept pace with workers’ interests and needs and provides a classic example of “policy drift,” the failure to update the law to reflect changing external circumstances, with the result that and the outcomes of the policy start to shift. Labor law’s support for workers’ ability to pursue union organizing and collective bargaining has declined over many decades, and efforts to remedy this drift have been blocked by a minority of senators despite majority support in both houses of Congress and presidential support for reform.

As policymakers and others concerned about the erosion of workers’ bargaining power and its impacts on today’s workforce debate measures to strengthen the ability of workers to organize, the background information and analysis in this paper should be of assistance in understanding the shortcomings in current law that need to be addressed if workers are to truly have the freedom to form and join unions.


After decades of growing economic inequality, bold and robust policy agendas on labor and union policy have arisen on the center-left of American politics. The candidates in the 2020 Democratic presidential primaries, including the eventual nominee, offered expansive policies (Greenhouse 2019; Biden 2020), and an extensive reform of the National Labor Relations Act (NLRA) recently passed the U.S. House of Representatives (McNicholas 2020). On the research front, there is an increased appreciation of the prominent role that the erosion of union/worker power has played in generating wage suppression and inequality (Stansbury and Summers 2020; Fortin et al. 2019).

This paper provides background for the consideration of changes in labor law by addressing the question: Why have union membership and union coverage (membership plus those covered by collective bargaining, even if not members) declined so precipitously in the private sector? We argue that corporations took advantage of the weak labor law regime in the United States to legally and illegally thwart union organizing and robust bargaining, especially in the 1970s, thus closing off unions’ ability to bring in new members and grow along with the economy.

Also key to unions’ decline were corporate practices and legal changes in the 1970s, 1980s, and 1990s that eroded bargaining power. We use a broad historical context to show how this assault on organizing and bargaining power weaponized weaknesses in the NLRA as interpreted by the National Labor Relations Board (NLRB) and the courts. Examples include the weak penalties established from near the very start for employer violations; a 1938 Supreme Court ruling on striker replacements; and the Taft-Hartley amendments of 1947 recognizing “employer free speech,” allowing right-to-work laws, and banning secondary boycotts.


Unequal Power events • Oct. 7, 2020

Rebuilding Collective Bargaining Back Better

Watch videoDownload slides


As such, the decline of private-sector unionism provides a classic example of what political scientists call “policy drift”:

Drift occurs when a policy or institution is not updated to reflect changing external circumstances, and this lack of updating causes the outcomes of the policy or institution to shift—sometimes dramatically. (Galvin and Hacker 2020)

In the case of labor law, its support for workers’ ability to pursue union organizing and collective bargaining has declined over many decades, and the weaknesses began to be exploited by management extensively in the 1970s.i

Except for an expansion of coverage into health care in the 1970s, all of the legislative changes to the NLRA since its enactment in the mid-1930s have been changes that weakened unions. Efforts during the Great Society period, when Democrats were at their peak congressional power, and under each successive Democratic president—Carter in 1978, Clinton in 1993, and Obama in 2009—to strengthen the NLRA’s protections of workers’ rights to collective bargaining were all defeated, despite majority support in both the House and Senate and by the president. The tool employed in each defeat was the filibuster, spearheaded by a minority of senators representing an even smaller share of the population. The result has been policy drift in labor law, allowing outcomes to shift in favor of corporate employers and their allies.

We find these explanations for union decline more persuasive than the dominant narrative that unions were dinosaurs that did not fit a modern economy characterized by automation and globalization. Indeed, we find only a limited role for automation and globalization on private-sector union decline. We quantify specifically the impact of the decline in manufacturing employment on union membership and coverage and find that less than a fifth of the union decline can be attributed to the loss of manufacturing employment’s importance. The experience of other advanced nations confirms the limited role of manufacturing’s erosion on union decline.

Another common explanation for the decline in unionization is the contention that private-sector unions outlived their roles as workers decided they no longer needed unions, or that unions became complacent and stopped reaching out to organize new workers (Cowie 2010; Moody 1988; Davis 1986; McAlevey 2017). Though we do not substantively engage in this question in this limited forum, we find persuasive scholarship that highlights how union organizing activity and labor activism continued strongly through the key decade of the 1970s, even as success rates dropped. Moreover, the years between 1973 and 1977 represent the peak in absolute numbers of NLRB elections held (see the appendix), and women and people of color were at the forefront of much of this labor activism. Unions began to pull back dramatically from organizing campaigns only in the early 1980s, after facing over a decade of potent employer resistance to organizing (Windham 2017; Brenner, Brenner, and Winslow 2010). In the paper’s final section we review recent studies that reveal that a substantial share of the nonunion workforce desired collective bargaining during the pivotal 1970s, especially Black workers, and that this unmet demand for collective bargaining has escalated in recent years such that nearly half of nonunion workers, including those who are supervisors and ineligible, would vote to have union representation in their current jobs if given an opportunity to do so.

We also are not able to explore fully the argument that unions missed the opportunities opened by the civil and women’s rights movements to diversify their ranks, and that the individual rights framework that undergirded these movements proved stronger than the collective New Deal framework that built unions and labor law (Lichtenstein 2013; Frymer 2008). Labor law effectively excluded many women and people of color because it did not cover the jobs they were most likely to hold, such as in agriculture and domestic work. For decades many unions excluded Black workers from their ranks and were riddled with deep-seated sexism; men and women of color and white women had to use the Equal Employment Opportunity Commission as a tool to force open many unions (Katznelson 2005; Kessler-Harris 2001; Frymer 2008). Yet recent scholarship correctly emphasizes the complexity and diversity of the working-class people who sought to improve their lives through the labor movement. Despite unions’ racism and sexism, women and people of color were the most likely groups to unionize in the decades after the Civil Rights Act’s passage; they led many union organizing drives and inspired others to join them. Though it seemed that the promise of the New Deal and labor law would be open to everyone, when these workers tried to organize they ran into the wall of corporate resistance and weak labor law that we describe in the following pages. The fact that a new wave of women and people of color wanted unions but could not effectively organize in the 1970s is a major and often-overlooked piece of the puzzle of union decline (Windham 2017; MacLean 2006; Jones 2013; Cobble 2004; Deslippe 2000).

This paper also does not account for the increased numbers of workers who may have lost union coverage because their firms moved or closed or else terminated the collective bargaining arrangement for other reasons. The increased erosion in membership among already-represented workers reflects a variety of factors: the regular ebb and flow of facility openings, closings, shrinkages, and expansions; faster-than-average growth in nonunion industries and occupations; and the anti-union animus of firms that increasingly ghettoized their organized operations by closing unionized units and opening nonunion ones.ii Much of the erosion of currently represented union workers reflects the fundamental rules of labor law that establish as a default rule representation at the individual-unit level of enterprises and do nothing to facilitate representation rights at the occupation or industry level, making maintaining union coverage exceedingly difficult, We also do not account for the increase in the numbers of jobs in categories that often fall outside of collective bargaining coverage, such as supervisors and contingent, contractual, or temporary workers.

The paper proceeds through the following points and findings:

  • The big picture of union decline is the dramatic drop in new unionization in both the manufacturing and nonmanufacturing sectors between the late 1960s and early 1980s, based on (1) fewer union elections, (2) a decline in union win rates in the elections that were held, and (3) the inability of the newly organized to obtain a first contract.
  • Employer resistance to union organizing sharply increased over the course of the 1970s, effected through aggressive management opposition, some of it, such as the firing of union activists, illegal; increased use of anti-union consultants; the weaponization of shutdown threats; and delaying tactics.
  • Decisions by courts and the NLRB, many of them emerging after passage of the anti-union Taft-Hartley Act of 1947, weakened bargaining power. These decisions include the prohibition against secondary boycotts; the banning by states of union security agreements; dramatically expanded management rights and the curtailment of the ability of unions to bargain with their employers about contracting-out decisions and plant closings; escalating use of striker replacements, especially after President Reagan normalized this employer behavior in the Professional Air Traffic Controllers Organization (PATCO) strike; the increased use of employer lockouts; and the exploitation of the toothlessness of labor law.
  • Globalization and automation, and the concurrent decline of manufacturing, can explain only a small portion of the decline in unionization. Private-sector unionization eroded rapidly in major sectors and in many detailed industries in nonmanufacturing, frequently to a greater extent than in manufacturing, and detailed statistical analyses show that changing employment patterns across industries can account for less than a fifth of union erosion. Moreover, international comparisons show that manufacturing decline explains very little of cross-country differences in union decline.
  • Polling data show that there has been a large unmet demand for collective bargaining, a finding that belies the argument that union decline has been the result of a lessened interest among workers in seeking collective bargaining.

The big picture of private-sector union decline

Union membership and union coverage (membership plus those covered by collective bargaining, even if not members) in the private sector has declined broadly in the United States. Our focus is on the private sector, since this is where unionization has dropped precipitously.

The long-term erosion of the share of union membership in overall private-sector employment (or “union density”) since 1929 is shown in Figure A. The line is broken at about 1972 because there are two series. The first, for 1929–1972, is based on a series developed from union revenue data by Troy and Sheflin (1985, Appendix A, and published in Hirsch 2008). The second series, based on an analysis of Current Population Survey data by Hirsch and MacPherson (2020), covers the 1973–2019 period.

The data in Figure A are sometimes erroneously interpreted as if there were a steady decline in union membership since the mid-1950s. This is not the case, as the decline greatly accelerated in the 1970s and 1980s.

Figure A

U.S. private-sector union membership rate, 1929–2019

Year Percent union members (1929-1972) Percent union members (1973-2019)
1929 12.4%
1930 13.3%
1931 14.0%
1932 15.2%
1933 15.5%
1934 16.3%
1935 14.2%
1936 15.0%
1937 19.5%
1938 21.9%
1939 22.8%
1940 24.3%
1941 25.9%
1942 28.1%
1943 30.8%
1944 32.4%
1945 33.9%
1946 34.1%
1947 34.9%
1948 34.7%
1949 34.9%
1950 34.6%
1951 34.7%
1952 35.2%
1953 35.7%
1954 35.6%
1955 35.1%
1956 34.7%
1957 34.7%
1958 33.9%
1959 32.3%
1960 31.9%
1961 31.9%
1962 31.6%
1963 31.2%
1964 31.0%
1965 30.8%
1966 30.3%
1967 30.5%
1968 29.9%
1969 29.0%
1970 29.1%
1971 28.2%
1972 27.3%
1973 24.5%
1974 23.6%
1975 21.7%
1976 21.5%
1977 21.7%
1978 20.7%
1979 21.2%
1980 20.1%
1981 18.7%
1982 17.6%
1983 16.5%
1984 15.3%
1985 14.3%
1986 13.8%
1987 13.2%
1988 12.7%
1989 12.3%
1990 11.9%
1991 11.7%
1992 11.4%
1993 11.1%
1994 10.8%
1995 10.3%
1996 10.0%
1997 9.7%
1998 9.5%
1999 9.4%
2000 9.0%
2001 9.0%
2002 8.6%
2003 8.2%
2004 7.9%
2005 7.8%
2006 7.4%
2007 7.5%
2008 7.6%
2009 7.2%
2010 6.9%
2011 6.9%
2012 6.6%
2013 6.7%
2014 6.6%
2015 6.7%
2016 6.4%
2017 6.5%
2018 6.4%
2019 6.2%
ChartData Download data

The data below can be saved or copied directly into Excel.

Notes: There is no definitive, fully time-consistent series on union membership as a share of employment. Figures for 1929–1972 were compiled by Troy and Sheflin (1985, Appendix A) from union financial reports; those for 1973 forward are compiled from Current Population Survey (CPS) household data (Hirsch and Macpherson, 2003, updated at http://www. unionstats.com). 1973–76 CPS figures adjusted to account for association members, who are included both in the Troy-Sheflin series and in the CPS beginning in 1977. 1973 CPS figure adjusted so that 73/77 CPS ratio equals 73/77 ratio in Troy and Sheflin.

Source: Hirsch 2008.

Copy the code below to embed this chart on your website.

Table 1 provides metrics to assess the rate of decline, presented for the decades from 1950 to 2000 and then in the 2000s using the cyclical peak of 2007 as the dividing point; this distinction allows us to separately assess the business cycles of 2000–2007 and 2007–2019. Assessing the trends in the 1970s presents a problem because of the data discontinuity between 1972 and 1973, when there is an implied very large and implausible decline of 2.8 percentage points (if one assumes the series is continuous, which it is not). To deal with this problem, Table 1 presents the periods of 1970–1972 and 1973–1980 separately and constructs an “alternative 1970–1980” metric to reflect the trends in the whole decade using the 1970–1972 and 1973–1980 trends.iii

Table 1

Measures of rate of decline in private-sector union membership, various periods, 1950–2019

Time period Change in union membership rate Annual percentage point change in union membership rate Percent decline in (10 yr) union membership rate
1950–1960 -2.7 -0.27 -7.8%
1960–1970 -2.8 -0.28 -8.8%
1970–1972 -1.8 -0.90 -30.9%
1973–1980 -4.4 -0.62 -25.5%
Alt 1970–1980 -6.8 -0.68 -28.9%
1980–1990 -8.2 -0.82 -40.8%
1990–2000 -2.9 -0.29 -24.4%
2000–2007 -1.5 -0.21 -23.8%
2007–2019 -1.3 -0.11 -14.4%

Source: Hirsch 2008. 

Copy the code below to embed this chart on your website.

For each period, Table 1 shows the absolute change in the union membership rate (column 1) and, to enable better comparisons across time, the annual percentage point change (column 2). These data tell us that the annual decline in private-sector union membership was far faster in the 1970s and 1980s (0.68 and 0.82 percentage points each year) than the relatively slow declines of the 1950s and 1960s (0.27 and 0.28 percentage points each year).

Yet these comparisons alone fail to tell the full story, because the same annual percentage point decline represents a differing scale of erosion when the starting point is, say, 34.6%, as it was in 1950, and when it is 11.9%, as it was in 1990. To account for the differing size of union membership at the start of each period, the metric presented in column 3 is the percent decline in the union membership rate as a 10-year rate of change. This metric reveals that the union membership rate declined by 7.8% and 8.8% in the 1950s and 1960s, respectively, and then the decline more than tripled in the 1970s (to 28.9%) and more than quadrupled in the 1980s (to 40.8%). The rate of decline in union membership slowed to about 25% in the 1990s and early 2000s, a rate still triple the pace of the 1950–1970 period.

What happened in the 1970s and 1980s that powered such a rapid decline in union density? A close look at union elections and first contracts in these years is revealing. In the United States, workers must clear three major hurdles if they want to form a union. First, unless workers can persuade their employers to recognize their union without going through the election process, at least 30% must sign union cards or petitions asking the government to hold a union election. Second, workers must win the government-sponsored election by a majority vote. Only then will the law require their employer to recognize the workers’ union and negotiate a union contract “in good faith.” The third hurdle is getting the employer to sign a first contract. Downward trends at each hurdle in the 1970s and later years relative to the1950–1970 period greatly help explain the dramatic decline of the inflow into unions and the associated erosion of the share of the workforce in private-sector unions.

Building on earlier work (Windham 2017; Goldfield and Bromsen 2013), we present data showing that the combined impact of substantive changes in the 1970s and early 1980s at each stage of this process, i.e., a smaller percentage of the workforce voting in union elections, lower rates of election victories, and reductions in the rate of unions achieving first contracts, greatly reduced the numbers of workers clearing the hurdles.

A smaller percentage of the workforce voting in elections

Figure B presents the trend in worker efforts to clear the first hurdle by participating in an election at their workplace. It shows the share voting overall in private-sector NLRB elections as well as the share voting in elections that were successful, presented as a share of total nonagricultural private-sector wage-and-salary employment (these series and a parallel one that shows trends relative to private-sector production/nonsupervisory employment are presented in the appendix).

Figure B

Workers participating in NLRB elections as a share of nonagricultural employment, 1951–2009

FY Year Workers eligible in all elections Workers eligible in elections won by union
1951 1.59% 1.21%
1952 1.81% 1.37%
1953 1.72% 1.36%
1954 1.21% 0.81%
1955 1.11% 0.81%
1956 1.02% 0.64%
1957 1.00% 0.58%
1958 0.77% 0.43%
1959 0.89% 0.53%
1960 1.02% 0.60%
1961 0.96% 0.49%
1962 1.11% 0.63%
1963 0.89% 0.53%
1964 1.06% 0.57%
1965 1.02% 0.63%
1966 1.06% 0.60%
1967 1.10% 0.63%
1968 0.94% 0.48%
1969 0.98% 0.49%
1970 1.01% 0.53%
1971 0.95% 0.45%
1972 0.94% 0.47%
1973 0.82% 0.35%
1974 0.80% 0.29%
1975 0.87% 0.33%
1976 0.68% 0.24%
1977 0.78% 0.30%
1978 0.60% 0.22%
1979 0.73% 0.26%
1980 0.65% 0.24%
1981 0.54% 0.19%
1982 0.33% 0.11%
1983 0.22% 0.10%
1984 0.26% 0.12%
1985 0.26% 0.10%
1986 0.26% 0.09%
1987 0.24% 0.10%
1988 0.24% 0.10%
1989 0.28% 0.11%
1990 0.26% 0.09%
1991 0.22% 0.08%
1992 0.21% 0.08%
1993 0.23% 0.10%
1994 0.20% 0.08%
1995 0.20% 0.08%
1996 0.20% 0.07%
1997 0.22% 0.09%
1998 0.23% 0.09%
1999 0.22% 0.10%
2000 0.22% 0.10%
2001 0.19% 0.08%
2002 0.16% 0.07%
2003 0.16% 0.07%
2004 0.15% 0.07%
2005 0.14% 0.06%
2006 0.11% 0.06%
2007 0.09% 0.05%
2008 0.10% 0.06%
2009 0.07% 0.05%
ChartData Download data

The data below can be saved or copied directly into Excel.

Source: Authors' analysis of National Labor Relations Board annual reports, Bureau of Labor Statistics.

Copy the code below to embed this chart on your website.

These data were developed from NLRB election data published in NLRB annual reports and employment data from the Bureau of Labor Statistics (see the appendix for details on the development of these NLRB election data and a discussion of measurement issues). By focusing on NLRB elections, these data understate the scale of organizing in the mid-1990s and later years because as many as 40-50% of newly organized workers were organized outside of the NLRB system, i.e., by “card check” or voluntary recognition by employers.

Figure B illustrates that the percentage of the workforce voting in union elections began to lessen between the late 1960s and the late 1970s (as well as in the 50s), plummeted to very low levels by the early 1980s, and remained at very low levels thereafter. In the 1950s and 1960s, more than 1% of those employed (1.2% and 1.0%, respectively) participated in an NLRA election each year. Participation in elections fell to 0.78% of employment in the 1970s and to just 0.29% in the 1980s (Table 2). In the early 2000s just 0.13% of the employed were involved in NLRB elections, 90% below the election rate of the 1950s.

Table 2

Trends in NLRB elections and win rates

Period Total elections Elections won Employees eligible to vote Win rate Share of nonagricultural private wage-and-salary employment
Eligible voters Workers in election victories
Total Elections won Elections Voters
1951–1960 5,155 3,410 517,277 354,649 65.8% 66.7% 1.21% 0.83%
1961–1970 7,071 4,165 523,710 285,857 58.8% 54.5% 1.01% 0.56%
1971–1980 7,712 3,944 502,785 202,082 51.1% 39.9% 0.78% 0.32%
1981–1990 3,782 1,828 231,810 88,075 48.6% 38.4% 0.29% 0.11%
1991–2000 3,002 1,513 204,816 83,753 50.4% 40.7% 0.21% 0.09%
2001⁠–⁠2009 2,044 1,209 140,837 69,867 59.9% 51.9% 0.13% 0.06%


Source: Authors' analysis of National Labor Relations Board (NLRB) data and Bureau of Labor Statistics household employment data. See appendix for details.

Copy the code below to embed this chart on your website.

More workers losing their elections

Over the years fewer workers have been able to successfully clear the second hurdle and win the union election, further dampening the inflow of new union members in the private sector. We can measure union election win rates in two ways, by elections and by voters, and there was a decline in both cases. In the 1940s, workers chose collective bargaining in 80% of the NLRB representation elections (Goldfield and Bromsen 2013). By 1977, however, they were losing more than half of the elections that they themselves had asked the government to hold.

The percentage of eligible voters in NLRB elections who chose collective bargaining also shrank over this time (Table 2). In the 1950s two-thirds of the pool of workers voting in NLRB elections were able to win their union elections, but that rate fell to 55% among involved workers in the 1960s and to 40% in the 1970s, 1980s and 1990s.

Fewer workers ever getting a first contract

After a group of workers have made it over the first two hurdles—triggering an election and then winning that election—the final hurdle is to obtain a first contract with the employer. In the U.S. firm-based labor law system, if the workers are not able to obtain a first contract, then they don’t directly benefit from collective bargaining.

While employers are required by law to bargain in good faith with workers over the contract, under the weak U.S. labor law system many employers drag their feet or only make a show of bargaining, knowing full well they can get away with it without serious penalty. Employers often slow-walk the bargaining process because the lack of a contract within a year can trigger a decertification election, with the workers losing their union representation rights.

The number of workers who were able to obtain a first union contract after going through the NLRB election process has declined over the years. While 86% of workers who chose a union were able to win a first contract in the 1950s, that share declined to less than 70% in the 1970s. By the 1990s, it was down to 56%.iv

Far fewer workers clearing the three hurdles to unionization

By the late 1970s, far fewer workers than before were able to clear all three hurdles and cross the finish line to become union members with a contract. The inflow into unions became a mere trickle. Table 3, which compares the 1966–1968 period to the 1978–1980 period (assessing the years preceding and at the end of the pivotal decade), provides the relevant NLRB data and an analysis showing which factors drove the decline in workers obtaining new union contracts.v

Table 3

Impact of number of elections, win rates, and first contract rates on decline in workers obtaining first union contracts, 1966–1968 versus 1978–1980

Erosion of:
Share of workers in elections Share of workers successful (win rate) First contract rate Workers with new union contract
As % of wage-and-salary employment
1966–68 1.03% 54.91% 81.8% 0.46%
1978–80 0.66% 36.44% 69.9% 0.17%
Percentage-point change 0.37% 18.47% 11.9% 0.30%
Percent contribution 44.2% 40.4% 15.5% 100.0%
As % of production/nonsupervisory employment
1966–68 1.22% 54.91% 81.8% 0.55%
1978–80 0.80% 36.44% 69.9% 0.20%
Change 0.00% 0.00% 0.0% 0.00%
Percentage-point change 0.43% 18.47% 11.9% 0.35%
Percent contribution 43.1% 41.1% 15.8% 100.0%

Source: Authors' analysis of National Labor Relations Board data and Bureau of Labor Statistics household employment data (Table 2). See appendix for details on NLRB data.

Copy the code below to embed this chart on your website.

Table 3 shows that, while 0.46% of the nonagricultural wage-and-salary workforce were able to make it across the unionizing finish line in the 1966-1968 period, only 0.17% were able to do so by 1978–1980. Production/nonsupervisory workers experienced similar results These declines in the percentage of workers voting in elections, win rates, and first contract rates translate to roughly 210,000 fewer workers a year who were able to enter into the collective bargaining relationship by 1978–1980 than we would have expected if rates had held steady at the 1966–1968 level.

Table 3 also presents analyses of the contribution of each hurdle to the deterioration of union membership. Within the nonagricultural wage-and-salary workforce (results are similar for the production/nonsupervisory workforce), 44.2% of the decline in the numbers of newly organized workers was due to a reduced share of workers voting in elections, 40.4% to a lesser win rate, and 15.5% to the decline in the ability to secure a first contract.

The reduced flow of workers into union representation automatically led to a declining union share of employment because new membership is needed to offset the ongoing erosion of established collective bargaining units that occurs even absent anti-union animus (facilities and firms shut down, downsize, relocate, etc.) Narratives that focus on automation and globalization as drivers for union density’s decline ignore the more than 200,000 workers a year who, by the end of the 1970s, were no longer able to enter unions through the unionization process. This process, as we reveal in subsequent pages, became increasingly dysfunctional as employers increased their resistance to workers’ union organizing and first contract efforts in the 1970s.

The fact that workers began to lose their freedom to enter unions had enormous implications for labor’s ability to grow. Figure C portrays the impact of the decline in successful union formation by showing the contribution of elections to overall private-sector union rates (computed at 10-year rate of change) in each period. These data represent workers who survived the first two hurdles—having an election and winning the election—but don’t reflect whether the workers won a first contract. The scale of successful union elections in the 1950s effectively added 8.7 percentage points to the overall share of union members in the private sector, all else equal. The contribution of successful elections to increasing union membership declined rapidly in each ensuing period so that by the 1980s the contribution was only 1.2 percentage points and by 2007–2017 just 0.5 percentage points.vi Even taking into account workers who are newly organized through card check and neutrality agreements does not change the overall picture that entry into unions has been substantially reduced over the last four decades compared to the 1950s and 1960s.

Figure C

Additions to private-sector membership rate from inflow of NLRB elections, 1951–2017

Time period Organizing additions to density, 10–year rate
1951–1960 8.7%
1960–1970 5.5%
1970–1980 3.2%
1980–1990 1.2%
1990–2000 0.9%
2000–2007 0.8%
2007–2017 0.5%
ChartData Download data

The data below can be saved or copied directly into Excel.

Source: National Labor Relations Board data. See appendix. Organizing inflow from number of voters in NLRB units that chose collective bargaining and does not take first contracts into account.

Copy the code below to embed this chart on your website.

These data on union membership and election trends yield two key conclusions. One is that the developments that eroded unions and new unionization did not occur gradually since the mid-1950s; rather, private-sector union erosion greatly intensified in the 1970s and 1980s. Understanding what happened in that particular period is therefore key to explaining private-sector union decline. The second conclusion is that the number of workers who were able to win union elections and get first contracts rapidly declined over the period from the late 1960s and into the early 1980s so that new unionization was not able to contribute much to maintaining the overall rate of union membership in the private sector starting in the 1980s. Understanding the suppression of union organizing is key to understanding union decline. This topic is addressed in the next section, which examines the institutional, political, legal, and management practices and strategies developed in the 1970s and 1980s that can account for the barriers workers faced at each step in the unionizing process.

What happened to union organizing in the U.S. in the 1970s that accelerated unions’ decline?

In the 1970s, U.S. corporations greatly increased their resistance to unions and union organizing, exploiting the weaknesses of U.S. labor laws to effectively squash workers’ right to organize and obtain collective bargaining. Workers in the United States historically have faced more employer opposition and less government support than workers in other nations. Active employer challenges to organizing and strikes, often coupled with government-led strike breaking and injunctions, helped defeat class-based uprisings in the late 19th century. In the early 20th century employers united to break workers’ organizing efforts with an open-shop drive, and they resisted workers’ rights in the making of the New Deal in the 1930s (Fantasia and Voss 2004; Greene 1998; Phillips-Fein 2009). During a window in the mid-20th century, from the 1940s to the 1960s, labor and management in the United States found a kind of uneasy balance that more closely resembled labor relations in European nations. Even then, however, unions were not strong in the South, and millions of women and people of color in domestic work and agriculture were left out because they held jobs not covered by labor law (Lichtenstein 2013).

Starting in the 1970s, employers began to shift this balance once again. They became much more politically active than they had been in the mid-century years, and they began to push to limit government regulation on multiple fronts such as the environment, consumer rights, and labor (Hacker and Pierson 2010). As part of this renewed conservative activism, employers ramped up their resistance to established unions and new union organizing. They did so, in part, because they faced a new economic paradigm created by a variety of emerging trends. First, financialization shifted the locus of economic power from manufacturing to banks and investment firms. Second, U.S. corporations, which were the world’s economic leaders in the years after World War II, faced more global competition as countries like Germany and Japan got back on their feet. Third, the rate of profit for private business fell by 29% between 1965 and 1973, and among manufacturers it fell by more than 40% (Brenner 2006). And finally, U.S. employers were more heavily saddled by social welfare costs, like health care and pensions, than were their global competitors because of the U.S.’s employer-based social welfare system (Hacker 2002). To address social welfare costs, employers began to move to a lower-cost employment model in which they could avoid providing security and social welfare to their employees; these moves included hiring more temporary and part-time workers, shifting to subcontractors, and driving down wage and benefit standards. They also worked to limit new demands from the collective bargaining relationship, including by attacking new organizing efforts. “People began looking for ways to economize and found that…they had given it away in the contract,” remembers Douglas Soutar, co-founder of the Business Roundtable.vii Many corporations sought to limit the number of workers who could access collective bargaining and tried to keep workers from ever forming unions in the first place.

Employers ramped up their resistance to unions as they faced a new wave of union organizing in the private sector, especially among a newly diversified workforce. Worker interest in unions remained high through the 1970s, and working people continued to try to organize in both the private and public sectors. Workers began to try to form unions in traditionally nonunion sectors of the economy, like retail and service, and throughout the South.

Leading these drives were women and people of color who had long been excluded from many of the nation’s best jobs and were outside much of labor law’s purview. The 1964 Civil Rights Act gained them new access. Once they got the coveted jobs, they pushed to unionize (Windham 2017). Three million women joined unions’ ranks between the 1960s and 1970s, and by 1980 28% of union members were women (BLS 1980; Kistler 1984). Black workers were particularly interested in organizing. In 1977, 70% of blue-collar African Americans said that they would vote for a union (Quinn and Staines 1979). In fact, by the 1970s Black and Hispanic workers were the most likely demographic groups to be union members; in 1973, Black and Hispanic men’s unionization rates were 38%, far above the 24% rate of all workers (Table 4).viii Black women were the only group to increase their unionization rate in the 1970s; the peak of 22% in 1979 is double the rate for non-Hispanic white women. Because the question on union membership was changed in the Current Population Survey (the source of data in the table) between 1976 and 1977, the erosion of union membership is likely understated, and it is not possible to assess the understatement for particular race/gender groups (Hirsch, MacPherson, and Vroman 2001).ix

Table 4

Private-sector union membership rates, by race and gender, 1973–1980

Private-sector union membership rate
 All White* Black* Hispanic
All workers
1973 24.2% 23.2% 29.0% 32.6%
1974 23.4 22.7 27.4 28.6
1975 21.5 20.8 26.9 25.1
1976 21.3 20.4 28.7 24.7
1977 21.7 20.8 28.4 25.4
1978 20.7 19.7 27.9 25.0
1979 21.2 20.1 28.9 24.7
1980 20.1 18.9 27.5 25.2
Male workers
1973 31.0% 30.0% 38.1% 37.7%
1974 30.2 29.6 36.2 32.3
1975 27.9 27.1 36.1 28.9
1976 27.7 26.7 37.3 29.4
1977 28.1 27.3 35.4 30.2
1978 26.9 26.0 34.1 31.4
1979 27.4 26.5 34.8 29.3
1980 26.3 25.2 33.5 31.0
Female workers
1973 13.1% 12.0% 16.7% 23.5%
1974 12.4 11.5 15.6 21.9
1975 11.6 10.7 15.3 18.5
1976 11.6 10.7 17.2 16.9
1977 12.2 11.0 19.7 16.8
1978 11.6 10.3 20.3 13.6
1979 12.3 10.8 21.5 16.6
1980 11.4 10.1 20.3 15.8

* Race and ethnicity are mutually exclusive.

Source: May Current Population Survey, 1973–1980.

Copy the code below to embed this chart on your website.

Though people of color and women were particularly interested in unionizing, this diverse group of workers was increasingly unsuccessful in its organizing attempts because employers pushed to new heights their efforts to break and bend labor law. From 1970 to 1980, charges against employers for law breaking more than doubled, as did the number of illegal firings.x Employers also developed and honed a new set of techniques to fight union organizing, promulgated through business schools and the vastly expanded “union-avoidance” industry. This onslaught of resistance to union organizing shifted norms about what was right or fair when it came to workers’ rights. Employer law breaking was remarkably effective. While workers won roughly 80% of their union elections in the 1940s, by the late 1970s they won fewer than half (Goldfield 1987). By the time of the battle around labor law reform legislation during the Carter administration in the late 1970s, employers were using a new level of political activism to defend an emerging status quo that offered companies tremendous latitude to resist workers’ union organizing and establish a union-free environment.

How workers lost the freedom to organize

Over the course of the post-World War II period, U.S. workers’ right to organize in the private sector has evolved to the point where it is now stronger on paper than in practice. Though U.S. labor law guarantees most private-sector workers the freedom to form unions, that right has been effectively curtailed as constant conservative and employer opposition to organized labor swayed the government to gradually cede its role as referee. Legislative changes, court decisions, and judgments by the NLRB have diminished workers’ right to unionize. When employers faced a changed economic paradigm in the 1970s and 1980s, this process accelerated, and employers were able to effectively weaponize this legal regime to block new union organizing and to weaken unions.

Many private-sector workers first gained a right to organize unions with the passage in 1935 of the NLRA, also known as the Wagner Act. The Wagner Act established a national policy of encouraging workers to organize into unions and engage in collective bargaining with their employers.xi Congress mandated that workers had the “full freedom of association” and protected their right to “designation of representatives of their own choosing, for purposes of negotiating the terms and conditions of their employment….” Under the Wagner Act, if the government certified that the workers had a union, then the company was obligated to enter into collective bargaining. The act excluded some categories of private-sector workers, including farmworkers, domestic workers employed by a family in its home, independent contractors, and supervisors.

In its early years, the law’s enforcement agency, the NLRB, required employers to remain neutral on the issue of a union, leaving the choice on collective bargaining solely up to the employees involved. In 1941 the Supreme Court decided that employers could weigh in during elections as long as they were not “coercive,” but the board’s enforcement remained vigorous in this period and workers were still routinely able to form unions.

A Republican Congress, urged on by employers wanting to curtail unions’ power, successfully revised the Wagner Act with the 1947 Taft-Hartley Act, passing it over a presidential veto. The Taft-Hartley Act weakened unions’ power on many fronts, including making it more difficult for workers to form unions and enter into collective bargaining. For example, it added language that became known as the “employer free speech” clause, affirmatively stating that an employer’s expression of views regarding unionization is not prohibited unless the employer’s statement contains an offer of benefit or threat of reprisal if workers choose to unionize..xii Employers relied on this language to engage in anti-union campaigns in ways that were previously found illegal under the Wagner Act, for example through extensive use of forced-attendance company meetings against the union, or “captive audience” meetings, a practice that was professionalized (via a new anti-union consultant industry) and used extensively in the 1970s. Taft-Hartley also added new unfair labor practices against unions and made explicit that workers could refrain from organizing activitiesxiii (Gross 1981; Becker 1991–1993). Taft-Hartley also added provisions allowing employers to file representation petitions to determine whether their employees wanted union representation.xiv Previously, the petition process had only been available to employers faced with organizing drives by competing unions.

Employers tried again to further weaken labor law in the 1960s. A number of the leaders of the nation’s largest corporations, including General Electric, Ford, and US Steel, began in late 1965 an effort to roll back the laws protecting workers’ organizing and bargaining rights through a new alliance, the Labor Law Reform Group (LLRG). The executives circulated a study among members of Congress in 1967 detailing 23 changes in labor law that would benefit employers. The changes professed to strengthen employer “free speech,” insist on “meaningful” bargaining units, and “prevent improper remedies” for employer unfair labor practices during representation campaigns, for example (Gross 1995; Windham 2017). The LLRG aimed to win its changes after the 1968 elections, but the group’s hopes were dashed when Congress remained Democratic. The LLRG soon merged with two other employer groups in 1972 to form the Business Roundtable, which began to successfully win its proposed changes, largely through litigation before the NLRB.xv

The five-member NLRB is appointed by the president and generally reflects the party in power. Under President Nixon, the NLRB quickly began erecting more obstacles in the way of organizing. It allowed employers to tell organizing workers that signing union cards would be “fatal” and cause “turmoil,” that if they chose a union they could lose what they had because bargaining “starts from scratch” and “everything is up for negotiation.” The board decided that employers legally could predict that they would have to close up shop due to financial difficulties if the workers voted yes.xvi The NLRB under the Carter administration did little to reverse the trend of weakening labor law that would persist and deepen in the Reagan years (Gross 1995).

Employers won a major victory in the U.S. Supreme Court in 1974 that allowed them to deny recognition to a union even if a majority of workers signed cards or petitions indicating their support for forming one. The card check method of forming a union, also called majority signup or voluntary recognition, has been a standard feature in U.S. labor relations since before the passage of the Wagner Act, and it is expressly recognized in the act. But the Supreme Court ruled in Linden Lumber (1974) that employers may refuse to recognize unions based on a showing of majority support and insist on an NLRB election. This requirement undermined the ability of workers to form unions because it subjected them to the NLRB election process and the attendant delays and employer anti-union campaigns.

Over the decades, such decisions by the NLRB and the courts steadily increased employers’ power to weigh in on elections and curtailed workers’ right to form unions. Consider the way policy drift in the law created two separate standards for how employers and unions are allowed to communicate with workers. Prior to passage of Taft-Hartley, the NLRB had ruled that employers violated the law and committed an unfair labor practice when they held mandatory meetings of employees to express anti-union views.xvii Within a year of Taft-Hartley’s enactment, the NLRB decided that employers could legally force workers to attend such captive audience meetings. The NLRB announced that it would find the meetings unlawful only if the employers threatened workers or promised some new benefit. The NLRB also reversed its earlier rulings that employers who hold captive audience meetings must allow unions an opportunity to respond. As a result, employers got the green light to hold mandatory anti-union meetings, and unions were denied the legal right to enter the employer’s premises to respond.

The Supreme Court expanded the problem in 1956 when it ruled in NLRB v. Babcock & Wilcox that employers are not required to give union organizers access to parking lots to talk with employees unless the union lacks alternative means of reaching employees.xviii The decision exacerbated an existing imbalance in communications during union organizing campaigns that greatly restricts and limits union organizers’ access to employees and employees’ ability to hear from union organizers at the workplace.xix A series of subsequent decisions gave employers a free hand to ban union supporters from captive audience meetings and even ban employees from speaking during the meetings.

By the 1970s, employers began to put their legal prerogative to hold such meetings to increased use. They routinely cherry-picked the workers who were undecided about the union, forced them to attend coercive meetings against the union, and were never required to allow the union equal say (Becker 1991–1993). The number of employers requiring such meetings increased by a third in the 1970s through the 1990s.xx By the end of the 20th century, 92% of employers held these forced attendance meetings (Bronfenbrenner 2000). The result is that management became able to communicate to employees in one-on-one conversations between direct supervisors and workers and in mandatory group meetings at work sites while union organizers were required to meet employees offsite and after hours, greatly complicating their ability to communicate with employees.

Policy drift in labor law meant that over time employers also gained more freedom to threaten to shut down if the workers voted for a union. At first, the NLRB seemed to support workers on this issue; it softened Taft-Hartley’s impact soon after its General Shoe decision in 1948, requiring that union elections must take place in “laboratory conditions” free from a coercive atmosphere. Nevertheless, in the early 1950s the NLRB decided that an employer was within legal bounds when it predicted it would have to close to meet unions’ wage demands. The board reversed that rule in 1962, deciding that such predictions of company closure were actually threats. But the threats became more potent in 1965 when the U.S. Supreme Court held that a company does not illegally discriminate against union supporters when it shuts down its business entirely in order to avoid unionization.xxi Employers cannot lawfully close one facility in order to chill union organizing at another facility, or transfer work from a union shop to a nonunion shop to avoid the union (the “runaway shop”), but the burden of proving that the decision was motivated by anti-union animus and not business reasons has proven difficult.xxii The court handed management a powerful rhetorical weapon to suggest that if workers successfully unionize then the facility may be closed, diminished, or moved. In the 1970s, the NLRB went even further, making threats of closure legal again as long as the company did not threaten to close solely because of the union.xxiii Employers dramatically expanded their threats of closure in the 1970s, as discussed below. By the 1990s, half of all employers facing worker organizing campaigns threatened to shut down if the workers formed a union (Bronfenbrenner 2000).

Acceleration of employer attacks on union organizing in the 1970s

Even though U.S. labor law left workers vulnerable, most major industrial employers, especially in the more unionized Northeast and Midwest, more or less complied with laws protecting workers’ right to organize from the 1940s through the mid-1960s. Beginning in the 1970s, however, employers tried to limit labor costs, including by restricting workers’ ability to enter into unions and collective bargaining.

When workers tried to exercise their right to vote for a union, employers exploited these previously described weaknesses in the U.S. labor law regime and began to break and circumvent the law at new levels. They learned through experience that labor law violations carried no real penalty and no real public stigma. Employer threats, mandatory anti-union meetings, and illegal firings paid off. By 1977, unionizing workers began to lose more than half of their elections for the first time since the Wagner Act’s inception (Goldfield 1987).

Labor law still prohibits employers from firing or threatening workers for supporting the union, as well as spying on workers, threatening to shut down if the workers vote in favor of a union, or promising workers more money or perks if they reject a union. The NLRB considers such acts “unfair labor practices” (ULPs), and charges against employers for ULPs rose sevenfold between 1950 and 1980xxiv (Figure D). These were not empty charges. Indeed, in 1980 alone the NLRB required employers to pay workers backpay in more than 15,000 cases after illegally firing them or cutting their pay as retribution for union activity, a record level at that point (Goldfield and Bromsen 2013).

Figure D

Number of unfair labor practices (ULP) charges against employers: 1950–1990

1950 4472
1952 4306
1954 4373
1956 3522
1958 6068
1960 7723
1962 9231
1964 10695
1966 10902
1968 11892
1970 13601
1972 17733
1974 17978
1976 23496
1978 27056
1980 31281
1981 31273
1982 27749
1983 28995
1984 24852
1986 24084
1988 22266
1990 24075
ChartData Download data

The data below can be saved or copied directly into Excel.

Source: National Labor Review Board annual reports, 1950–1990 8(a)1 unfair labor practice charges. For years 1954–1990, see Table 2 and Table 4. For 1950, see Table 3A and Table 10. For 1952, see Table 2 and Table 3.

Copy the code below to embed this chart on your website.

Yet the penalties for labor law violations are scant. No fines are levied, no employer goes to jail, and any costs incurred are negligible. Typically, if the NLRB finds that an employer illegally fired a union supporter, for instance, that company simply has to rehire the worker, pay back wages (minus what the worker earned at another job, or could have earned, in the meantime), and post a sign in the breakroom explaining that it broke the law. Workers do not receive monetary damages to compensate them for the economic harms inflicted by their illegal treatment. Unlike other employment laws, workers have no right to bring a lawsuit against the employer for violating their NLRA rights; they are entirely dependent on the agency pursuing their case. In contrast, other employment laws, such as civil rights laws, provide much greater penalties and provide for a private right of action so workers can bring cases on their own and collect attorneys’ fees if they prevail (Weil 2005).

If the employer violates labor law multiple times during a campaign, then the NLRB can order a new election, but even a new election cannot erase the original threats’ effects. Occasionally the NLRB will order a labor law violator to bargain with its workers (a so-called “bargaining order”), but this process usually takes years, and the courts have been resistant to these orders. Bargaining orders issued by the NLRB have dropped from more than a hundred per year to a small fraction of that, and rarely result in a collective bargaining agreement (Brudney 1996, 1581–87).

The fact that labor law is so toothless means that employers have an economic incentive to violate it. The law really protects workers only when employers more or less voluntarily comply with it.

Starting in the 1970s, many more mainstream, large companies became far less willing to act in accordance with the law. Fortune 500 firms with longstanding bargaining relationships ramped up their resistance to union organizing; they skirted the law, delayed at every step, and increasingly spoke out against new union organizing, even when some of their workers were already covered by collective bargaining agreements. “It requires a certain nerve for those companies whose names you see in the batting order of big hitters in the bargaining game to try to keep plants unorganized,” a vice president of BF Goodrich told an industrial relations convention in 1978. “Management is more sophisticated and bolder…and the times ‘they are a-changing'” (Pestillo 1978). Union-busting tactics moved squarely into the industrial sector, the area where unions had traditionally been the strongest and which had long formed the core of the nation’s economy. An analysis of the ratio of the number of ULPs to the number of petitions filed within specific sectors is revealing. Not only did the level of lawbreaking per election shoot up in the 1970s, but industrial-sector workers were more intensely subjected to employer resistance than were workers in the service and retail sectors, which were traditionally less unionized, though resistance greatly increased in all sectors (Figure E).xxv

Figure E

Ratio of ULPs (CA) filed against employers to petitions filed for union certification (RC), by sector: 1950–1980

Year Manufacturing Retail Service
1950 0.5 0.4 0.8
1953 0.5 0.5 0.6
1955 0.6 0.5 0.5
1958 0.8 0.6 0.9
1960 0.8 0.7 1.1
1963 1.0 0.8 1.0
1965 1.0 1.0 1.0
1968 1.0 0.9 0.9
1970 1.4 1.0 1.0
1973 1.5 1.3 1.2
1975 2.2 1.5 1.2
1978 2.8 2.2 2.0
1980 3.7 2.7 2.2
ChartData Download data

The data below can be saved or copied directly into Excel.

Notes: CA cases are charges of unfair labor practices against employers under Section 8(a)1 of the National Labor Relations Act. RC elections are those triggered by workers who are trying to form a union.

Source: National Labor Review Board annual reports, 1950–1980.

Copy the code below to embed this chart on your website.

Normalization of union busting by management consultants and business schools

Managers who had long begrudgingly accommodated unions faced a learning curve when it came to outright avoiding them. In response, a large anti-union consultant industry stepped up to lead the way in the 1970s. These anti-union firms, often hand-in-hand with the nation’s business schools, taught business that “busting” unions was acceptable behavior and that good management meant remaining union-free. Through an avalanche of seminars, trainings, books, and speeches, these new management consultants helped make mainstream a level of anti-unionism that had been extreme in the mid-century labor-management arrangement. “Any management that gets a union deserves it—and they get the kind they deserve,” was the mantra of one sought-after consultant (Hughes 1976). The consultants helped entrench the concept that managers could and should avoid unions in all arenas, and so helped further deactivate labor law’s worker protections.

Anti-union labor consultants weren’t entirely new; the nation’s first anti-union firm, Labor Research Associates (LRA), was formed in 1939 in Chicago. Yet management resistance to unions in the earlier decades was neither as widespread nor as accepted as it would be by the 1970s and 1980s. While there were just a handful of anti-union consulting firms in the beginning of the 1970s, by decade’s end there were hundreds. One management firm founder told a congressional hearing in 1979 that his industry grew tenfold over the preceding decade, and the AFL-CIO estimated that a full 70% of all campaigns involved some sort of management consultant (Logan 2006; U.S. Congress 1980).

Business schools and professors were also key in shifting management’s values on unions. By the 1970s, U.S. business managers were far more likely to have gone to business school than in previous decades, and they were far more likely to do so than managers in other industrialized nations (Locke and Spender 2011). Business schools in the 1970s began to teach students that unions were an unnecessary expense on the cost and balance sheet and tutored them in how to avoid unionization. One frequent contributor to Harvard Business School case studies, for example, instructed that since the NLRB response process to employer unfair labor practices was so lengthy and the penalties “quite mild,” “it is quite possible for management to effectively destroy an organizing effort or, at the very least, signal to employees the relative ineffectiveness of the union in dealing with management” (Fulmer 1982). One California State University business professor asserted that, “In all but the most unusual circumstances it is almost negligent for a company to allow unionization to happen….When one surveys all the things a nonunion employer can do to stay that way…the employer would almost have to try to get itself organized to end up with a union” (Kilgour 1981). Universities themselves began to host the myriad of anti-union seminars made available by union consultants. The University of Delaware, Denver University, the University of San Francisco, the University of Alabama, Clemson, and Wake Forest were among the schools hosting such seminars in the late 1970s. According to the AFL-CIO, one consultant boasted of having taught at 30 universities (McDonald and Wilson 1979).

Management consultants built their businesses by stoking fears based in racism and sexism and by teaching employers how to beat back their diversified workforces’ organizing efforts. One anti-union management consultant warned, “Danger: a union can muster a most potent campaign when it can take advantage of a ‘racial’ or ‘sexist’ theme” (Jackson 1981). Another told a Wake Forest University seminar in 1979 to try to limit the number of African American workers it hired in order to stay union-free. “Blacks tend to be more prone to unionization than whites,” he told the managers in the closed-door session. “If you can keep them at a minimum, you are better off.”xxvi In Confessions of a Union Buster, Marty Levitt spelled out how his firm, Three M (for Modern Management Methods), developed tactics to “awaken within the mostly white supervisor corps a hatred of blacks…contempt for women, mistrust of the poor…” (Levitt and Conrow 1993). Many of the consultants rang alarm bells for the mostly male management class about women’s interest in union organizing. “All indications are that women are now more inclined to vote union than men,” warned one anti-union specialist. “This is entirely consistent with the women’s movement, by whatever name…” (Kilgour 1982).

Anti-union consultants instructed clients in how to avoid unions completely, often by opening nonunion facilities, hiring people who were the least likely to unionize, and by being perfectly clear that the company philosophy was nonunion. The consultant Charles Hughes trained over 27,000 managers and supervisors how to “remain union-free” between 1974 and 1984 (Logan 2006). Stephen Cabot, a Philadelphia lawyer, helped firms decide where to locate in order to remain nonunion, sometimes even identifying specific areas of cities where workers were the least likely to unionize (Wall Street Journal 1979). By 1983, nearly half of firms identified remaining union-free as their most pressing labor relations goal (Freeman and Kleiner 1990).

Much of the anti-union consultants’ work, however, came after workers already showed interest in a union. Once employers realized that their workforces were signing union cards, they often called in consultants to usher them through the union campaigns, step-by-step, in order to defeat the workers’ organizing efforts. Consultants often spent weeks at the worksite, training supervisors and offering advice, though rarely appearing before the workforce.

Consultants made good use of the predictable patterns in an NLRB election process. First, at least 30% of workers had to sign union cards or petitions showing an interest in a union, after which they petitioned the NLRB to hold an election. Consultants advised employers how to discourage card signing. Alfred DeMaria, a high-profile management consultant, advised employers that, “The Board has approved some surprisingly strong employer statements.” As an example, he noted, “One employer was lawful when it told its workers, ‘Don’t sign any cards; they can be fatal to business'” (DeMaria 1982).

The next step in the election process was for the union and company to work out the “bargaining unit,” or the specifics of who could vote. Consultants urged employers to demand a protracted NLRB hearing to determine which workers got to cast ballots. “Always go to hearing…. I have yet to see a situation where time worked against the employers in an election,” urged management consultant Fred Long in an executive meeting captured on tape by a union infiltrator in 1975, a transcript of which surfaced in a 1979 congressional hearing. “Suffice it to say, you have at least 500 issues. So you litigate those issues….You could come up with them for almost a year, as we did in one case” (U.S. Congress 1980, 1:208). Such delays cost organizing workers dearly. One study found that each month of delay between the filing of the petition and the election decreased the workers’ chance of winning the election by 2.5% (Kistler 1984). Consultants also instructed employers how to manipulate the loopholes in the NLRB process in order to seed the voting group with as many no votes as possible. “Hire five of your relatives on a regularly scheduled part-time basis….You have 60 days to hire even a hell of a lot of people if you need to,” continued Long to the gathered executives.

Once the election was finally scheduled, employers launched intense campaigns against the workers’ unionization effort. Some firms developed elaborate systems to track and sway union sentiments among workers. One “highly confidential management document” instructed supervisors at Cannon Mills in 1982 to rate each worker in their department from strongest for the company to the weakest, and to profile employees by race, sex, and age.xxvii

Employers learned how to threaten unionizing workers with loss of benefits and strikes while skirting the legal prohibitions on such threats. Through letters, speeches, and flyers employers made clear to workers that the company would not really have to offer anything new if the workers won the right to collective bargaining. “The Hotel does not have to agree to a single thing the union proposes so long as we bargain in good faith,” asserted the Boardwalk Regency Hotel in Atlantic City.xxviii Warnings about strikes often featured prominently in the anti-union campaigns. “Tell employees that the law permits the hiring of a permanent replacement for anyone who engages in an economic strike,” urged Brandeis University to its supervisors in 1976 when librarians tried to unionize.xxix

Employers learned how to legally threaten workers with plant closure if the union won the election. For example, one General Electric facility in Goldsboro, N.C. faced a union drive in 1978, and issued the following thinly veiled threat: “Cleveland Welds…was represented by the IUE, as were a number of other plants that have closed, including Cleveland Lamp plant, Oakland Lamp plant….Don’t mistake me. I’m not saying we will automatically lose our business if the Union wins the election. But it’s clear that unions…can, and they do, hurt people’s job security.”xxx

The threat of plant closure held enormous sway in the climate of capital flight in the late 1970s and early 1980s. In fact, many U.S. manufacturers used globalization as a weapon against workers’ unionization efforts. For example, the financier David E. Murdock bought Cannon Mills in 1982, and when the workers then tried to form a union he used the threat of global competition in textiles to successfully beat back their unionization attempt: “If I determine that Cannon cannot operate competitively, I can and I will cease to operate Cannon,” he told them in mandatory-attendance meetings. “This is my decision, and mine alone, and no one can stop me—including this Union”xxxi (Windham 2017).

If the workers did manage to win a campaign, employers routinely delayed or avoided actually signing a collective bargaining agreement—the very relationship that the entire election process was designed to facilitate. One AFL-CIO survey found that among workers who won elections, only 63% ever actually got a union contract.xxxii If all else failed, consultants taught employers the ins and outs of decertifying a union already in place. While it is technically illegal for an employer to assist or promote a decertification petition by employees, employers routinely do so (Shawe 1979). The number of decertification elections doubled between 1972 and 1982.xxxiii

The labor movement tried to fight back, and in 1977–1978 it sought to strengthen workers’ rights to form unions and strike through the Labor Law Reform Act. Though the Business Roundtable was at first split on whether to oppose the bill, a broad coalition of American businesses ultimately mounted a massive, coordinated campaign to leave weak labor law unchanged (Stein 2010, 187; Waterhouse 2013). Ironically, some of the same businessmen who had wanted to change labor law as part of the LLRG in the late 1960s later defended the broken status quo because it so aptly suited their purposes. Business had become far more politically active than ever by the late 1970s, and the number of registered lobbyists for business firms increased by fourfold over the decade (Vogel 1989). Business put this new organized power to work against the bill. After a 19-day filibuster and five attempts to get the 60 votes needed for cloture in the Senate, labor and its supporters had to admit defeat. In an “Open Letter to American Business” in the Wall Street Journal, AFL-CIO President George Meany asked business, “Why? What is your motivation?…Where is the moral basis for your attacks? Is not the real intent of this attack the destruction of the uniquely American system of collective bargaining….Do you want to destroy trade unionism?”xxxiv United Auto Workers President Douglas Fraser denounced the “one-sided class war” that broke “and discarded the fragile, unwritten compact” between labor and business.xxxv

The employer campaigns against unionization in the 1970s were remarkably potent. On paper U.S. workers still had the right to organize, but by the end of the decade they were losing it in practice as they faced defeat in more than half of the elections that they themselves had triggered. The AFL-CIO’s assistant organizing director told Congress in 1984, “I’ve been involved in organizing off and on…since 1967 and can assert categorically that the state of the art in employer resistance to employees’ organizing efforts has achieved a level of sophistication and effectiveness far exceeding that of the late ’60s and early ’70s.”xxxvi Doreen Lavasseur, a union organizer who helped university and clerical workers organize throughout the decade, remembers the ground-level impact of the employer campaign on workers: “I would just watch these people go from feeling strong and like we need to do something to feeling like totally terrified to do anything, and paralyzed.”xxxvii The rise in employer law breaking, the spread of employer anti-union campaigns deep into the nation’s core industries, and the tutorials of union consultants coalesced to undermine workers’ freedom to form unions by the end of the 1970s, and it has never recovered.

The unraveling of workers’ bargaining power under the law

By the end of the 1970s, employers had fully exposed the shortcomings and weaknesses of the NLRA and learned how to exploit the union-restrictive provisions of the 1947 Taft-Hartley Act to undermine and defeat union organizing. Employers knew they could vigorously campaign against unions and even break the law by firing union activists without facing any real financial penalties or consequences. Yet employers did not stop there. They also attacked existing unions and curtailed the bargaining power of unionized workers. They began to replace strikers far more frequently, limited what workers could bargain about, began to lock workers out in disputes, and even began bargaining to impasse in order to force strikes. By the early 1980s, they began to demand across-the-board concessions in many contract negotiations. In addition, the shift of power from manufacturing to finance meant that the banks’ and shareholders’ needs often took precedence over those of workers. Though workers were going to the bargaining table with factory owners, the entity with the real power was often a financier on Wall Street (Davis 2009; Stein 2010; Applebaum and Batt 2014).

Employers were able to squeeze unions so effectively because, over the years, labor law had become heavily tilted against workers and toward employers. Though these employer-friendly laws were on the books in the 1940s, 1950s, and 1960s, it was not until the 1970s that employers began to take full advantage of their power. Several key developments set the stage for this 1970s unraveling of workers’ bargaining power under the law. First, a Republican Congress largely neutered workers’ leverage in passing the 1947 Taft-Hartley Act over President Truman’s veto. Second, Taft-Hartley forced the NLRB to prioritize, over all other cases, including cases involving illegal firings of union supporters, litigation against unions for engaging in so-called secondary activity. Third, the law’s ineffective remedies became obvious, and the NLRB’s efforts to hold employers accountable for violating the law were stymied in the courts. Fourth, employers increasingly found an ally in the U.S. Supreme Court, which issued a series of decisions restricting workers’ rights, expanding employer power, and limiting employers’ bargaining obligations. Finally, employers started making greater use of replacement workers during strikes—a trend that grew in the 1970s and 1980s and significantly undermined workers’ right to strike. The cumulative impact of these factors meant that by the 1970s the law did not effectively protect workers’ bargaining power and gave employers a wealth of tools to resist unionization.

The impact of Taft-Hartley

The 1947 Taft-Hartley Act dramatically weakened workers’ bargaining power in several ways. As noted in the previous section, it strengthened employers’ influence on the organizing process by giving employers more leeway on speaking out against the union and allowing employers themselves to file representation petitions. Taft-Hartley also authorized states to ban “union security” agreements, under which employers and unions agree that all represented employees should share in the cost of union representation through either union dues or fair share fees. This change allowed states to pass laws (so-called “right-to-work” laws) allowing workers to obtain the benefits of union representation without contributing toward the costs, creating a free-rider problem designed to undermine unions.xxxviii Recent research has shown that right-to-work laws have had substantial direct and indirect impacts on wages and wage inequality (VanHeuvelen 2020).

Taft-Hartley also imposed new restrictions on “secondary boycotts,” the picketing of other employers to put pressure on the workers’ own employer, discussed next.

NLRB decisively ends union secondary boycotts

As soon as Taft-Hartley became law, employers saw the power and benefit of its new ban on employee secondary activity. Unlike other violations of the NLRA, under Taft Hartley violations of the prohibition on secondary activity against so-called “neutral” employers are subject to civil lawsuits and money damages by employers against unions. The NLRB is statutorily mandated to seek federal court injunctions against unions engaged in secondary boycott activity and to give these cases priority over all other cases, including those alleging illegal conduct by employers against workers forming unions.xxxix

The resulting enforcement disparity was stark and immediate. The ratio of unfair labor practice charges against unions compared to charges against employers grew from one in four in 1948 to half in 1956 (Figure F). Injunctions against unions for alleged secondary activity grew from 17 in 1948 (the first year such injunctions were authorized) to 127 10 years later. Injunctions against unions grew further to 219 by 1960, an astonishing 1,188% increase from 1948. During this same period, the NLRB pursued almost no injunctions against employers for unfair labor practices.xl Taft-Hartley quickly succeeded in shutting down one of workers’ most powerful economic weapons.

Figure F

Growth of charges for secondary boycott violations: 10(l) charges, 1948–1960

1948 17
1949 32
1950 22
1951 32
1952 35
1953 44
1954 66
1955 59
1956 78
1957 98
1958 127
1959 129
1960 219
ChartData Download data

The data below can be saved or copied directly into Excel.

Source: National Labor Review Board annual reports, 1948–1960.

Copy the code below to embed this chart on your website.

In stark contrast to the mandatory federal court injunction procedure for violations by unions of the secondary boycott restrictions, no such mandatory injunction proceedings are available for retaliation or discrimination against workers for supporting a union. And while employers have the ability to sue unions in court and win money damages for violations of the secondary boycott provisions, workers have no similar ability to sue their employers for money damages for violating their NLRA rights (Human Rights Watch 2000).

Erosion of collective bargaining rights by the Supreme Court

The erosion of workers’ bargaining power was further exacerbated by a number of significant decisions by the U.S. Supreme Court. These rulings limited access to the workplace by union organizers; undermined the remedies available to the NLRB for violations of the law; greatly constrained the right of workers and their unions to bargain with employers over contracting, plant closing, and other decisions impacting the bargaining unit; and expanded employers’ economic leverage during labor disputes by allowing them to proactively lock out employees. Each of these decisions significantly weakened workers’ bargaining power. Taken together, they undermined an already weak law and tilted it away from workers and in employers’ favor.xli

Narrowing of the mandatory scope of bargaining

One line of Supreme Court decisions dramatically expanded management rights and curtailed the ability of workers and unions to bargain with their employers about contracting-out decisions, plant closings, and other issues affecting the bargaining unit. At the urging of employers wanting to narrow the scope of topics about which they were required to bargain with their workers’ unions, the Supreme Court deemed these topics “managerial” and beyond the scope of mandatory bargaining.

This series of decisions started with the Supreme Court’s 1964 ruling in Fibreboard Paper Products Corp. v. NLRB, which involved an employer’s decision to contract out the work performed by bargaining unit employees. The Eisenhower NLRB, which was considered employer-friendly, initially ruled that the employer was not legally obligated to bargain over what the board deemed a “basic management” decision. The Kennedy NLRB reversed that decision, and in the ensuing uproar the employer community went to Congress and the Supreme Court for relief (Gross 1995, 172–74). The Supreme Court affirmed the Kennedy NLRB’s ruling that the employer was legally required to bargain in this instance, which the court majority described narrowly as “contracting out of plant maintenance work previously performed by employees in the bargaining unit, which the employees were capable of continuing to perform.”xlii The court noted that requiring bargaining over the decision furthered the policies and purposes of the NLRA:

[A]lthough it is not possible to say whether a satisfactory solution could be reached, national labor policy is founded upon the congressional determination that the chances are good enough to warrant subjecting such issues to the process of collective negotiation.

Still, the Fibreboard majority limited the reach of its decision to the facts presented, explicitly noting that “[o]ur decision need not and does not encompass other forms of ‘contracting out’ or ‘subcontracting’ which arise daily in our complex economy.”

A concurring opinion in Fibreboard by Justice Stewart contained a statement that would come to frame the law governing managerial decisions that have an impact on employees’ jobs. He opined:

Nothing the Court holds today should be understood as imposing a duty to bargain collectively regarding such managerial decisions, which lie at the core of entrepreneurial control. Decisions concerning the commitment of investment capital and the basic scope of the enterprise are not, in themselves, primarily about conditions of employment, though the effect of the decision may be necessarily to terminate employment. If, as I think clear, the purpose of § 8(d) is to describe a limited area subject to the duty of collective bargaining, those management decisions which are fundamental to the basic direction of a corporate enterprise or which impinge only indirectly upon employment security should be excluded from that area.

Seventeen years later, in 1981, Justice Stewart’s views formally became the majority view in First National Maintenance Corp. v. NLRB.xliii There, the Supreme Court ruled that employers have no duty to bargain over a decision to terminate a contract for business—even when that decision results in the layoff of bargaining unit employees. The company involved in First National Maintenance provided cleaning and housekeeping services, and it terminated a contract to provide services to a nursing home without first bargaining with the union representing its housekeeping employees, who then lost their jobs. The Supreme Court characterized the decision made by the employer as “involving a change in the scope and direction of the enterprise,” which the court said “is akin to the decision whether to be in business at all.” The court concluded:

…the harm likely to be done to an employer’s need to operate freely in deciding whether to shut down part of its business purely for economic reasons outweighs the incremental benefit that might be gained through the union’s participation in making the decision, and we hold that the decision itself is not part of 8(d)’s “terms and conditions” over which Congress has mandated bargaining.

Under First National Maintenance, therefore, employers are required to bargain with their employees’ union over the effects of decisions to cancel contracts, restructure, or cease some or all operations, but not over the decisions themselves, which the courts view as “managerial” decisions. These decisions on the scope of bargaining obligations significantly undermine workers’ and unions’ bargaining power and deprive them of the ability to participate in and shape decisions and actions by their employers that affect the employer’s ongoing operations and employment.

Giving employers economic leverage in labor disputes

Allowing striker replacements. Under U.S. labor law, it is illegal for an employer to fire or retaliate against a worker for engaging in “protected, concerted activity” such as a strike, but in a twisted anomaly, employers are legally permitted to hire “permanent replacements” for strikers’ jobs. In an early decision from the first days of the Wagner Act, the Supreme Court indicated in dicta (language not part of the legal holding in the case) that an employer whose employees were engaged in an economic strike, as contrasted with a strike over unfair labor practices, could permanently replace striking employees without violating the NLRA.xliv This rule significantly undermines workers’ legal right to strike, because workers faced with deciding whether to strike over economic issues know that they can be permanently replaced by other workers and lose their jobs.

Until the 1970s, few employers used the practice of replacing strikers because it was considered so confrontational. But the practice of permanently replacing strikers “sharply increased” in 1975 and became a much more prominent practice in the 1980s (LeRoy 1995; Stelzner 2017; Logan 2008). This practice escalated after the very public example of President Reagan replacing striking air traffic controllers and breaking the PATCO strike. Although the air traffic controllers’ strike was illegal, Reagan nevertheless established a new norm for employer behavior (McCartin 2006). Major employers, including Greyhound, Phelps Dodge, Massey, Caterpillar, Colt Industries, Bridgestone/Firestone, and the Chicago Tribune Co., were emboldened to hire, or threaten to hire, permanent replacements during legal strikes.xlv One of the most prominent incidents of permanent replacements took place at International Paper in Jay, Maine and two neighboring plants. More than 2,300 workers participated in a 16-month strike that ended unsuccessfully after the company hired permanent replacements.xlvi The dispute divided a small town and created lingering bitterness between strikers, the company, and the community.xlvii

The General Accounting Office (now the General Accountability Office) reported that employers announced that they would hire permanent replacements in about one-third of the strikes in 1985 and 1989, and actually hired them in approximately 17% of all strikes (GAO 1991).

McCartin (2006) described the evolution of the tactic using LeRoy’s enumeration of casesxlviii:

LeRoy found forty-four cases involving permanent replacements decided under the National Labor Relations Act or the Railway Labor Act in the 1950s. This amounted to only one documented use of permanent replacements per 80 major work stoppages during that decade. In the 1960s, the rate was one per 83 major work stoppages. In the 1970s, a slight increase in employers’ tendency to use permanent replacements was detectable, as the rate rose to one per 66 major work stoppages. LeRoy argues that this shift began around 1975. But it was in the aftermath of the PATCO strike that employers aggressively seized upon the striker replacement tactic. In the first ten years after 1981, employers used permanent replacements in roughly one out of seven major work stoppages. A sea change had clearly occurred in employers’ willingness to replace strikers.

The regular threat and use of permanent replacements dramatically undermined workers’ legal right to strike and gave employers a powerful economic weapon to undermine and defeat unions. As Logan (2008) explained:

[The use of striker replacements] has allowed hostile firms to defeat numerous strikes, undermined unions during contract negotiations, provided powerful antiunion propaganda during organizing campaigns, and enabled firms to instigate strikes, then recruit permanent replacements as a means of unloading unwanted unions.

Legislation to curtail the practice of permanent replacements enjoyed the support of a majority of the House and the Senate, but failed because of numerous Republican filibusters, encouraged by the business community, in 1992 and 1994 (Logan 2008). Efforts to restrict the practice of permanent replacements through executive action in 1995 were struck down by the courts.xlix

Allowing employer lockouts. At the same time Congress and the courts were weakening workers’ leverage, the NLRB and the courts gave employers additional leverage by allowing them to proactively lock out (layoff) their employees, the equivalent of the employer going on strike. Prior to 1964, employers were permitted to proactively lock out employees (called an offensive lockout) in two narrow circumstances: (1) where the employer was part of multiemployer bargaining and the union was striking one employer in an effort to force other employers into agreeing with the union’s demands, a practice known as whipsawing, and (2) in situations where the employer reasonably believed a strike was imminent. But in a 1965 decision in American Ship Building Co. v. NLRB,l the U.S. Supreme Court ruled that employers could proactively lock out their employees once an impasse had been reached in bargaining “for the sole purpose of applying economic pressure in support of [the employer’s] legitimate bargaining position.” In other words, the employer did not need to show that it was at risk of being whipsawed in a multiemployer arrangement, or that a strike was imminent—an employer could proactively lock out its employees simply to create leverage in support of its bargaining demands. American Ship dramatically shifted bargaining power to employers and, not surprisingly, afterwards employers increasingly engaged in proactive lockouts to achieve their bargaining goals.

Lockouts became more prominent as strikes and union membership diminished. Though there are no data on lockouts for the period before 1990, two independent analyses show their increased importance. Marvit (2016), who employed various sources to track lockouts and strikes between 1990 and 2015, found that, though lockouts declined over the period, from 32 in 1990 to 13 in 2015, the decline in strikes was greater. As a result: “In 1990, lockouts represented less than 4% of total work stoppages, whereas in 2015 lockouts represented over 10% of total work stoppages.”

The Bureau of National Affairs has also documented lockouts and strikes since 1990.li In 2012 it found: “The huge plunge in union membership over the past two decades has meant a huge plunge in union-initiated strikes. Yet it hasn’t meant a huge plunge in employer-initiated lockouts” (Coombs 2012a). It calculated that there had been 221 lockouts per 5,431 stoppages in 1990–1999 (4.07%), 164 lockouts per 2,995 stoppages in 2000–2009 (5.48%), and 31 lockouts per 312 stoppages in 2010–2011 (9.64%).

It should be noted that the threat of a lockout, as with a threat of a strike, can have a substantial impact on bargaining outcomes.

The rise in the proportion of lockouts to strikes changed in recent years due to a wave of strikes in 2018 and 2019 (a trend that is likely to hold in 2020 as well). There were more than 150 strikes in each of those two years, while lockouts remained at similar levels as in the 2012–2014 period, around 10 each year.lii

Diminishing the duty to bargain in good faith. We’ve seen that by the 1970s there was growing awareness among employers that the NLRA lacked teeth, and a realization that, even if employers were found to have committed illegal, unfair labor practices, the remedies were weak and ineffective. Employers made use of these weaknesses to strip workers of bargaining power. Consider employers’ failure to bargain in good faith. In 1970, the Supreme Court ruled in H.K. Porter v. NLRB that the NLRB could not require any particular terms in a collective bargaining agreement and could only require that parties go back to the bargaining table—even though the parties in this case had been bargaining for eight years. The decision stripped the NLRB of a potent tool for getting employers and unions to bargain and reach agreement.liii Moreover, in earlier decades, the NLRB had ordered employers who had made frivolous objections to union certification and had failed to bargain in good faith to compensate workers for the wages and benefits they lost because of the bad-faith bargaining. But in 1970, the NLRB changed course and decided it lacked authority to order these remedies, on grounds that such remedies constituted “punishment” that is not authorized by the NLRA.liv With the absence of a meaningful remedy, the legal duty of employers to bargain in good faith with their workers’ union was severely undermined. Employers use this freedom to great advantage when dealing with newly formed unions. As noted above, workers at approximately half of all newly organized shops fail to reach an initial collective bargaining agreement with their employer (Fisk and Pulver 2009). Employers can string out the bargaining process and avoid reaching an agreement, creating a feeling of futility among workers who have recently chosen to organize.

Allowing employers to use the bankruptcy process to gut collective bargaining agreements. Another trend that emerged in the 1970s and undermined unions and bargaining was the practice of employers using the bankruptcy system to shed their wage-and-benefit obligations under collective bargaining agreements. Corporations seized on court decisions finding collective bargaining agreements to be “executory” contracts, meaning contracts that had not yet been fully executed, and they sought and received permission from bankruptcy courts to shed these executory obligations. The U.S. Supreme Court exacerbated the problem in its 1984 decision in NLRB v. Bildisco & Bildisco,lv in which the court affirmed the ability of employers to shed their contractual obligations under collective bargaining agreements. Bildisco & Bildisco had reneged on wage increases and health and welfare contributions that were due to workers under the terms of the collective bargaining agreement that the company had negotiated with the Teamsters. According to news reports, at the time the decision was pending 22 corporations, including major corporations like Continental Airlines and Wilson Foods, had attempted to avoid their collective bargaining agreements through bankruptcy, and 19 had succeeded (Townsend 1984).

Bildisco prompted a huge outcry, and Congress quickly responded by passing legislation to restrict the ability of employers to shed their collective bargaining responsibilities in bankruptcy, but the bankruptcy code still contains provisions allowing employers to void collective bargaining agreements where they make a sufficient showing of need (Sousa 2003).

Ceccotti (2007) notes that in the early 2000s there was a “wave of bankruptcy cases targeting significant reductions in labor costs, pension funding, and retiree health obligations that has surged through the airline industry, the steel industry, auto supply, and other heavily unionized industries in recent years….[D]ebtors have been able to extract substantial labor and benefit costs cuts, either through, or under the threat of, court-ordered relief under sections 1113 and 1114. Many have involved the termination of defined benefit pension plans as well.” Ceccotti notes that in these “transforming business restructurings,” “bankruptcy has once again become a deliberate strategy used to broadly target costs associated with collective bargaining agreements and collectively-bargained pension and retiree health obligations.”

One union president testified in 2013 that bankruptcy proceedings meant a “debtor (company) essentially had a gun to labor’s head—it was a take it or leave it proposition, not a negotiation.”lvi

Automation and globalization are a minor part of the overall picture

The contribution of automation and globalization to the decline of unions is prominent in most discussions, and the view draws on some commonly known facts. First, the rapid decline in union membership in manufacturing occurred at the same time as imports escalated and attention to competitive pressures from foreign producers, increasingly those from low-wage nations, became a prominent concern. Second, because automation has been faster in manufacturing than in other sectors of the economy, the manufacturing share of total employment has been declining. Given that manufacturing was a highly unionized sector, this development mechanically led to an eroded union share. That union erosion has also happened in other advanced nations supports an intuition that trends happening across nations, such as automation and globalization, are the true underlying factors.

In this light, any impact that eroded unionization is having on wage inequality or labor’s share of income, the argument goes, is just a reflection of globalization and automation. In economics discussions this means that union decline is a symptom, not a cause, of wage inequality, with globalization and automation the real culprits behind both wage inequality and deunionization.

This section examines the role of globalization and automation primarily by examining the role of manufacturing decline on unionization trends. The analysis does not examine the impacts of globalization on unions in sectors other than manufacturing because these impacts, like the outsourcing of call centers and white-collar work, primarily developed in the 1990s and later years and not in the turning-point decades for private-sector unionization of the 1970s and 1980s.

Trends in manufacturing versus nonmanufacturing sectors, 1977–2019

Figure G shows union coverage in the manufacturing and nonmanufacturing segments of the nonagricultural private sector in the United States (all references to the private sector hereafter are to the nonagricultural private sector) from 1977 to 2019.lvii “Union coverage” reflects those who are members of the union as well as those who aren’t members but are covered by a collective bargaining agreement. Note that although union coverage in manufacturing started at a higher level and fell more sharply than in the rest of the private sector, the decline was far from unique. Union coverage in nonmanufacturing fell from 17.6% in 1977 to just 6.8% in 2019, a 60% contraction. The manufacturing union coverage contraction was 74%.

Figure G

Trends in private-sector union coverage rates in manufacturing and (nonagricultural) nonmanufacturing, 1977–2019

Manufacturing Nonagriculture, nonmanufacturing
1977 37.6% 17.6%
1978 36.7% 16.9%
1979 38.2% 17.2%
1980 34.8% 16.7%
1981 33.8% 15.4%
1982 32.2% 14.9%
1983 30.5% 14.5%
1984 28.4% 13.2%
1985 26.9% 12.3%
1986 25.8% 11.8%
1987 24.7% 11.3%
1988 23.8% 11.0%
1989 23.1% 10.6%
1990 22.2% 10.6%
1991 21.8% 10.4%
1992 21.0% 10.2%
1993 20.3% 10.0%
1994 19.6% 9.9%
1995 18.7% 9.4%
1996 18.3% 9.2%
1997 17.2% 9.1%
1998 16.8% 8.8%
1999 16.6% 8.8%
2000 15.6% 8.6%
2001 15.5% 8.6%
2002 15.1% 8.3%
2003 14.3% 8.1%
2004 13.9% 7.8%
2005 13.7% 7.6%
2006 12.5% 7.4%
2007 12.0% 7.6%
2008 12.3% 7.8%
2009 11.9% 7.4%
2010 11.6% 7.1%
2011 11.2% 7.1%
2012 10.5% 6.9%
2013 11.0% 7.0%
2014 10.5% 7.0%
2015 10.0% 7.1%
2016 9.6% 7.0%
2017 10.0% 6.9%
2018 9.7% 6.8%
2019 9.4% 6.8%
ChartData Download data

The data below can be saved or copied directly into Excel.

Sources: For the years 1973–1981 the May Current Population Survey (CPS); for the years 1983-2019 the CPS Outgoing Rotation Group (ORG) earnings files. Unionstats.com, Hirsh and Macpherson 2020.

Copy the code below to embed this chart on your website.

The data underlying Figure G allow us a first pass at assessing the role of manufacturing’s decline on union coverage. A simple shift-share analysis indicates that the erosion of the manufacturing share of employment from 1979 to 2019, a decline from 30.2% to 12.6%, is responsible for a decline of private-sector union coverage of 3.3 percentage points, or about a fifth of the overall 15.8 percentage point decline in the private nonagricultural sector.lviii This erosion of manufacturing employment presumably captures the impact of greater imports (and trade deficits in “goods”) and provides our first assessment of their impact: not trivial but not dominant. We will explore more rigorous assessments below that all find an even smaller impact of manufacturing’s erosion on union decline.

Union coverage erosion in major sectors and specific industries

We can obtain a clearer picture of globalization’s impact by examining the erosion of union coverage in specific broad sectors and in more detailed industries. Table 5 presents trends in selected major private-sector industries where the data are comparable for 1979, 1983, and 2019 (this methodology leaves out much of private-sector personal and business services).lix Adding the data for 1979 highlights the very sharp drop in union coverage emerging from an assault on unions amidst the deep recession and a sharp rise in imports in the early 1980s.

Table 5

Union coverage in selected sectors: 1979, 1983, and 2019

Union coverage (%) Change (%)
1979* 1983 2019 1979–2019 1983–2019
Mining 37.2 23.1 4.6 -32.6 -18.5
Construction 35 30.4 14.5 -20.5 -15.9
Nondurable goods manufacturing 32.5 28.4 9.7 -22.8 -18.7
Durable good manufacturing 39.8 32.1 9.2 -30.6 -22.9
Utilities 46.7 43.7 25.1 -21.6 -18.6
Wholesale 12.7 10.9 5 -7.7 -5.9
Retail trade 10.9 9.7 4.7 -6.2 -5
Transportation/warehousing 52 53.6 24.1 -27.9 -29.5
Communications 54.6 51.7 13 -41.6 -38.7

* Data for 1979 are pooled May Current Population Survey data for 1978–1980.

Source: Current Population Survey Outgoing Rotation Group data.

Copy the code below to embed this chart on your website.

Several nonmanufacturing sectors that had substantial union coverage rates in 1979, many comparable to or greater than manufacturing, suffered a larger or nearly comparable erosion of union coverage than manufacturing: They include mining, construction, communications, transportation/warehousing, and utilities. Other sectors such as wholesale trade and retail trade lost more than half their union coverage. Overall, union erosion was pervasive in sectors that had substantial union coverage, beyond just manufacturing, and every one of these sectors lost a large share of its union coverage.

Table 6 adds detail to the analysis by examining union coverage decline in 19 specific detailed industries over the 1979–2019 period.lx There were substantial declines in union coverage in a wide array of specific industries beyond manufacturing: mining, both metal and coal; fishing and hunting; transportation such as air, trucking, buses, and rail; hotels; grocery stores; hospitals; radio and television broadcasting; warehousing; newspapers; and electric power.

Table 6

Changes in union coverage, selected detailed industries, 1979–2019

Union coverage Change
Detailed industry 1979* 1983 2019 1979-2019 1983-2019
Fishing, hunting, and trapping n.a. 17.8 8.0 n.a. -9.8%
Metal ore mining 59.6 42.2 3.2 -56.4% -39.0%
Coal mining 67.8 62.8 9.6 -58.2% -53.2%
Nonmetallic mineral and not specified mining, and quarrying 46.6 34.8 14.2 -32.4% -20.6%
Newspaper publishers 24.8 21.4 6.6 -18.2% -14.8%
Air transportation 47.5 45.7 39.4 -8.1% -6.3%
Rail transportation 83.6 85.4 60.9 -22.7% -24.5%
Truck transportation 48.6 39.6 8.8 -39.8% -30.8%
Bus service and urban transit 49.8 52.3 38.7 -11.1% -13.6%
Warehousing and storage 26.2 28.1 5.7 -20.5% -22.4%
Radio and television broadcasting and cable 19.7 17.4 12.1 -7.6% -5.3%
Electric power generation, transmission, and distribution 51.8 47.8 27.2 -24.6% -20.6%
Hardware stores 6.3 4.3 1.3 -5.0% -3.0%
Grocery stores 38.2 33.5 15.1 -23.1% -18.4%
Vending machine operators 16.4 18.6 4.1 -12.3% -14.5%
Hotels (Traveler accommodation) 18.3 15.4 7.4 -10.9% -8.0%
Services to buildings and dwellings 24.6 15.6 5.9 -18.7% -9.7%
Barber shops n.a. 10.7 3.2 n.a. -7.5%
Hospitals 21.0 22.6 15.1 -5.9% -7.5%

* Data for 1979 are pooled May Current Population Survey (CPS) data for 1978-1980.

Source: CPS Outgoing Rotation Group data.

Copy the code below to embed this chart on your website.

Rosenfeld (2020) and Denice and Rosenfeld (2018) make a similar point by examining union decline in major blue-collar occupations. Rosenfeld compared union decline from 1973 to 2016 in two occupations greatly affected by automation and/or globalization—production and mining— and in two occupations that “either cannot be outsourced or haven’t yet borne the brunt of technological changes”—transportation and construction. Rosenfeld notes that union erosion was widespread in these occupations, with the decline in unionization within transportation and construction occupations falling from 50% in 1973 to less than 20% in 2016.

Statistically assessing the impact of the change in the industrial distribution of employment

It is possible to obtain more exacting analyses of the impact of shifting industrial employment patterns on union coverage trends by employing regression analyses.lxi The starting point is regression analysis explaining union coverage (members plus those covered by a union contract as the dependent variable) that considers many factors including and beyond a worker’s industry.lxii The goal is to assess the impact of the change in the industrial distribution of employment (primarily the shift from manufacturing to the service sector) on the 15.8 percentage point fall (from 22.8% to 7.0%) in private-sector union coverage between 1979 and 2019.lxiii We do this in two ways.

The first methodlxiv decomposes, or breaks down, the changes in union coverage into the role of the changes over time in (1) the “composition” of the workforce—its industrial distribution and its demographic and other characteristics—and (2) the separate effect that each of these characteristics of the workforce has on workers’ likelihood of being in a union.lxv If workers in a particular state are less likely to be in a union, for example, how much of the decline in unionization is due to the fact that there are more workers in that state now than was the case in 1979 (a composition effect) and how much is due to a change in the likelihood that workers in that state will be in a union today than in 1979?lxvi This exercise reveals that the changes in industry employment composition explain 2.5 percentage points of the overall 15.8 percentage point decline in union coverage from 1979 to 2019, or about 16% of the total change. The advantage of this computation is that it examines the impact of the changes in the industrial composition of employment while controlling for changes in the distribution of workers across states and occupations and across various demographic characteristics including gender, race/ethnicity, education, and age.

The second method uses the regressions for each year, 1979 and 2019, to simulate the level of union coverage that would have obtained if the industry structure were swapped between years.lxvii We present the results in Table 7. In 1979, 22.8% of private-sector workers were covered by a union. Our simulation finds that if in 1979 the economy had instead the same industrial structure as in 2019, but all other features of the 1979 economy were held constant, then union coverage would have been 20.3%. This exercise suggests that the changed pattern of sectoral employment lowered union coverage by only 2.5 percentage points. We get similar results when we do the simulation the other way around. In 2019 private-sector union coverage was 7.0%, but it would have been 7.4% had the sectoral composition in 2019 been identical to what it had been in 1979, suggesting that sectoral employment patterns only reduced union coverage by about 0.5 percentage points. The usual interpretation of these results is to bracket the actual impact as somewhere between these two estimates. In this case, the data suggest that the shift out of manufacturing employment has had an impact that accounts for less than a fifth of the overall decline of private-sector union coverage between 1979 and 2019.

Table 7

Impact of change in industry composition: Counterfactuals in 1979 and 2019

1979* 2019
Actual With 2019 industry distribution Actual With 1979 industry distribution
Private-sector union coverage 22.8% 20.3% 7.0% 7.4%
Impact of industry composition changes -2.52% 0.46%
Regression decomposition
Impact of industry composition 2.5%

* Data for 1979 are pooled May Current Population Survey data for 1978–1980. Difference between 1979 and 2019 means evaluated at 1979 'betas' for industry.

Source: Current Population Survey Outgoing Rotation Group data.

Copy the code below to embed this chart on your website.

Union decline as an independent, not dependent, factor

The claim that automation and globalization have driven the erosion of unions means that automation and globalization, not unionism itself or collective bargaining, are what is responsible for the wage stagnation or wage inequality that have flowed from union erosion. In statistical terms, these claims are saying that union decline is not the true factor (independent of other factors) but rather is itself determined by other factors. Because we do not have blind trial experiments or even natural experiments focused on union decline, there is no dispositive evidence upon which to draw. The evidence presented here so far has focused on data indicating that union decline goes much beyond what has occurred in the sector most impacted by globalization, manufacturing.

More can be learned from some of the analyses of the union impact on wages. For instance, Denice and Rosenfeld (2018) quantify the impact of union erosion on the wage levels of nonunion workers. They find:

The size of the union effect is substantial, especially for men. We estimate that weekly wages would be approximately $61—or 6 percent—higher for nonunion private sector men if private sector union densities were as high in 2015 as they were in 1977. Over the course of a year, this would result in a wage gain of $3,172. Among women, the counterfactual predictions reveal that a nonunion worker would earn $18 more weekly had unions not declined since 1977, or $936 annually.

It is useful that Denice and Rosenfeld are examining the decline of unions within occupations as opposed to industries and that they directly control for the risks of automation (“the average routine task content of each occupation to control for that occupation’s risk of automation”) and manufacturing decline (the proportion of each occupation’s employment in manufacturing). They report that their results “are robust to the inclusion of controls for the risk of automation, offshoring.” This leads them to claim that their estimates reflect the “independent influence of private-sector union strength on nonunion private-sector pay.” For our purposes it is important to note that including measures of offshoring and automation risk does not diminish the estimated union impact on nonunion wages.

Newly developed union membership data back to the passage of the NLRA in the 1930s allow Farber et al. (2018) to examine whether the patterns of union rise and decline correspond to the expected patterns if changes in union density simply reflected automation. They do so in response to a set of papers (most prominently Acemoglu et al. 2001) that “argue that any empirical relationship between unions and inequality is spurious, driven by variation of an omitted variable [skill-biased-technological change, i.e., automation] that simultaneously reduces union density and increases inequality.”lxviii Automation, according to Acemoglu et al., is expected to erode “union density because as automation increases unskilled unionized workers become relatively more expensive to employ [i.e., priced out of market] and unions have, by construction, no option to organize the more skilled workers.” Farber et al. note that data from a point in time (i.e., a cross-section analysis of one year’s survey) do not allow a test of the claim that automation is the true driver of union decline, but that the new historical time series on union membership, tracking eras of both union expansion and union contraction, is well suited to examining the co-movement of union density, skill composition, and union premiums predicted by those claiming that automation is the true factor underlying union decline. Farber et al. conclude that “our data reject many of these predictions, suggesting that union density is not merely an artifact of skill-biased technological change,” i.e., automation.

Last, Ahlquist and Downey (2019) directly examine the impact of imports on unionization in manufacturing and overall, building on the influential work of Autor et al. (2013) examining the impact of Chinese imports on wages and employment (they also build on Pierce and Schott 2016).lxix

Ahlquist and Downey note that there has been “a substantial academic literature investigating the link between globalization and deunionization, with a particular focus on trade-related ‘deindustrialization’ and its effects on the relatively unionized manufacturing sector.” They conclude that, “This early literature generally found weak or inconsistent effects of trade on union density.”

Based on their own analysis, Ahlquist and Downey conclude:

We study how import competition affects union membership in the United States, adapting identification strategies from recent work on imports from China. Within manufacturing, union workers are slightly more affected than non-union ones, inducing modest declines in unionization. At the same time, total manufacturing declines are greatest among Right-to-Work states. We provide evidence that firms in Right-to-Work states tend to specialize in lower-quality products, making them more susceptible to competition with Chinese goods. However, while reducing unionization within manufacturing, import competition causes a robust increase in unionization outside of manufacturing, more than offsetting within-manufacturing declines. This appears to be driven by family members of would-be manufacturing workers shifting to higher-wage jobs: for less-educated women, the highest paying opportunities are often in healthcare and education, which are disproportionately unionized. Altogether, we calculate that the decline in US union density would have been 36% larger without Chinese imports.

So, at least in the case of the surge in Chinese imports, there has been an adverse impact of imports on union density in manufacturing, but it accounted for only one-sixth of the decline in manufacturing union density from 1990 to 2014.lxx More important, however, is that the decline in manufacturing unionization was offset by increases in family member unionization in nonmanufacturing. Obviously, this analysis addresses the key import trend of the late 1990s and 2000s but does not address the import surge in the early 1980s.

Overall, these studies cast doubt on the claim that trends in unionization are simply artifacts of the underlying factors of globalization and automation.

International comparisons

Collective bargaining has declined in many advanced nations over the last few decades, a trend that has given credibility to the notion that pervasive changes happening across countries, such as automation, the decline in manufacturing, and globalization, have led to the erosion of collective bargaining. What at quick glance seems obvious and intuitive, however, turns out not to be a primary explanation of union decline.

During the 1970s, the pivotal decade when union decline accelerated in the United States, many other nations saw the opposite trend. In fact, unions swelled across much of the globe, even as U.S. workers faced a wall of employer resistance to new organizing. Of 23 developed nations, 18 saw their union movements grow in that decade (Western 1997, 18-24).

The modest impact of manufacturing decline also emerges from a recent comprehensive analysis of union erosion by the OECD (2019). The report notes the declines in both union membership (from 33% in 1975 to 16% in 2018) and union coverage (from 45% in 1985 to 32% in 2017) among OECD countries, but concludes:

The drivers of the decline in union density are numerous and vary between countries and over time. Contrary to a commonly held belief, the combined contributions of demographic changes and structural shifts, such as the shrinking of the manufacturing sector, are small and leave most of this declining trend unexplained.

The OECD’s analysis reveals the complexity of the underlying trends:

This average downward trend, however, masks important cross-country variations in terms of initial unionisation levels, the actual direction of trends, and, in countries where it happened, the pace, intensity and timing of the decline. First, trade union density in the mid-1970s varied from around 75% in Sweden, to around 20% in France and just above 10% in Korea. Second, while union density declined in a majority of countries, it increased in Iceland and Belgium and was relatively stable over the last four decades in Canada, Korea and Norway. Third, decline was much faster and more abrupt in some countries than in others. In the 1990s, Eastern European countries, Israel, and New Zealand experienced a fall of at least 30% of union density (Turkey in the 2000s is another example) over a relatively short time-span. By contrast, decline was much more gradual (and much smaller) in countries like Denmark, Switzerland or Chile – where it was more akin, in fact, to a progressive erosion than to a drop. Finally, the timing of decline also differs: it starts in the 1980s in several countries, but already in the 1960s in the United States, Austria or the Netherlands, and much later—in the 1990s—in several Nordic countries. Changes in union density accelerated at various points in time over the period, with individual countries exhibiting specific spikes.

This leads to the bottom line:

This heterogeneity of the evolution of union density across countries suggests that it may be the result of a combination of country-specific factors rather than global forces—although some drivers might be common across countries or groups of countries.

The OECD study also presents a detailed statistical decomposition of the factors that led to union erosion in each nation, though differences in data availability make each analysis slightly different for each country. For the United States the OECD uses survey data of individual workers (specifically, Current Population Survey data for 1983 and 2018) to predict union membership using various factors such as demographic and job characteristics.lxxi The changing composition of job characteristics (across three occupations, five industries, and public vs. private sector) contributed 1.73 points of the 9.5 percentage point decline in the unionization rate over the 1983–2018 period, explaining 18% of the decline.lxxii Job composition, therefore, had a modest impact on the erosion of unionization, and the impact of automation and globalization—to the extent they manifest in a shrinking of manufacturing employment—would be even smaller.

Schmitt and Mitukiewicz (2012) in an earlier analysis examined changes in union membership and union coverage across 21 advanced nations between 1960 and 2010, and they also highlighted the heterogeneity of outcomes that suggest a limited role for automation and globalization:

Union coverage (the share of workers whose terms of employment are covered by a collective bargaining agreement) changed little and even rose slightly in a substantial number of countries, including the period since 1980. Union membership (the share of workers who are members of a union) fell in most of the rich economies, with the United States experiencing losses (from a low initial level of unionisation) near the middle of the distribution. These differences across countries exposed to broadly similar levels of globalisation and technological change suggest that neither factor mechanically determines national levels of unionisation. These findings are broadly consistent with other research on long-term trends in unionisation in the world’s rich democracies.

This initial review of the data raises serious doubts about the inevitability of union decline. While union membership rates fell in two-thirds of the sample, membership rates were flat or constant in the other third. While union coverage rates fell in just under half of the 21 countries, coverage rates were flat or rising in the other half. At least at face value, globalisation and technological progress do not appear capable of explaining these observed differences, primarily because all of these economies have been subjected to similar sets of forces in both regards.

Schmitt and Mitukiewicz (2012) point to the “[n]ational political traditions established in the period 1946 through 1980” as the best predictor of unionization trends:

Between 1980 and 2007, the social democratic countries saw coverage rates increase, on average, five percentage points. Over the same period, the continental market economies experienced a small decline in coverage that averaged four percentage points. Meanwhile, coverage rates fell in all of the liberal market economies, with an average decline of 26 percentage points (20 percentage points at the median).

Union decline was not the result of lessened interest of workers in seeking collective bargaining

Some commentary attributes the decline in unionization to an abandonment by workers who have lost interest in collective bargaining, as if the need for collective bargaining has diminished or does not apply to new sectors. It is worth a brief examination of the evidence of workers’ interest in collective bargaining.

Contrary to the claim that nonunion workers lost interest in collective bargaining, the available polling evidence suggests that there is a substantial unmet demand for collective bargaining and that many nonunion workers would rather be covered. This unmet demand has reached its highest levels in recent years. The issue then is to assess why so many workers’ desire for collective bargaining has gone unmet rather than assess any trend of workers refraining from collective bargaining.

Kochan, Kimball, Yang, and Kelly (2018) examined the level of interest in joining a union among unorganized workers, comparing 1977, 1995, and 2017, and found the “demand for unions” has risen substantially since the late 1970s. Kochan and Kimball (2019) summarized those results:

…the 1977 and 1995 results were nearly identical: approximately one third of the non-union workforce indicated they would vote to have union representation if given an opportunity to do so on their current job. In 2017 that number increased to 48 percent. This number translates into an under-representation of unions of approximately 58 million workers.

Freeman (2007) analyzed a different set of surveys and also found a large unmet demand for worker voice and, in particular, collective bargaining. Freeman examined the preferences of union and nonunion nonmanagerial workers and found that workers wanted greater say at their workplace as much or more than they did in the 1990s, and that they wanted unions more than ever before. Specifically, he found that the proportion of workers who wanted unions had risen substantially over the last 10 years at the time of the study, and a majority of nonunion workers in 2005 would vote for union representation if they could,lxxiii up from the roughly 30% in the mid-1980s and the 32-39% in the mid-1990s, depending on the survey. Given that nearly all union workers (90%) desire union representation, the mid-1990s analysis suggested that if all the workers who wanted union representation could achieve it, then 44% of the workforce would have union representation. The rise in the desire for union representation since then suggests that the share of the nonunion workforce wanting union representation in 2005 was 53%. These results, in turn, suggest that if workers were provided the union representation they desired in 2005, then the overall unioniza­tion rate would have been about 58%.

Freeman also examined Gallup polling back to 1947 and found that the majority of the public “approved of unions,” far more than those who “disapprove of unions,” over the entire postwar period. In 2005, 58% approved of unions while 33% disapproved (9% had no opinion). Union approval has increased since Freeman’s analysis, with Gallup (Jones 2019) reporting on its survey in 2019:

The current 64% reading is one of the highest union approval ratings Gallup has recorded over the past 50 years, topped only in March 1999 (66%), August 1999 (65%) and August 2003 (65%) surveys.

The Gallup data do show that the approval of unions deteriorated over the 1950s, 1960s, and 1970s, but even in a relative unfavorable time for unions—1979—55% of people approved of unions and 33% disapproved.


Workers have always had a high level of interest in forming unions at their workplaces to win better pay, benefits, and dignity on the job, and their interest now is higher than ever. The sharp decline of union representation and new union members in the 1970s—a decline from which workers and the labor movement have never recovered—was due not to worker disinterest but rather to a combination of employer tactics and weaknesses in the law that undermined worker organizing. Nor can the decline in unionization be explained by globalization and automation: These factors account for only a small percentage of the decline. The dominant explanation of eroded private-sector unionism is a policy failure, leaving workers and unions powerless to combat an increasingly aggressive set of legal and illegal employer practices. Such a policy failure, what the political scientists call policy drift, can be offset by better policy. Policymakers should take account of these lessons and conclusions as they debate measures to strengthen workers’ ability to organize and bargain in the months ahead.


Melat Kassa and Jori Kandra provided assistance with data gathering and analysis and preparation of tables and figures. John-Paul Ferguson provided computations of NLRB election data and advice on data matters and interpretation. We received helpful comments from Craig Becker, Jim Brudney, Alex Hertel-Fernandez, Dan Galvin, Mike Goldfield, Ellen Kurlansky, Celine McNicholas, Jake Rosenfeld, and John Schmitt.

About the authors

Lawrence Mishel is a distinguished fellow and former president of the Economic Policy Institute. He is the co-author of all 12 editions of The State of Working America. His articles have appeared in a variety of academic and nonacademic journals. His areas of research include labor economics, wage and income distribution, industrial relations, productivity growth, and the economics of education. He holds a Ph.D. in economics from the University of Wisconsin at Madison.

Lynn Rhinehart is a Senior Fellow at EPI, where she works on labor and employment policy, with a focus on collective bargaining.  She previously served as General Counsel of the AFL-CIO, a federation of 55 national and international labor organizations. While at the AFL-CIO, she was the executive director of the Lawyers Coordinating Committee (LCC). Rhinehart has authored publications on enforcement of federal workplace safety laws, the contingent workforce, and other topics. She earned a J.D. from Georgetown University Law Center and a B.A. from the University of Michigan.

Lane Windham is associate director of Georgetown University’s Kalmanovitz Initiative for Labor and the Working Poor and co-director of WILL Empower (Women Innovating Labor Leadership). She is author of Knocking on Labor’s Door: Union Organizing in the 1970s and the Roots of a New Economic Divide (UNC Press, 2017), winner of the 2018 David Montgomery Award.  Windham spent nearly 20 years working in the union movement, including as a union organizer. She earned an M.A. and Ph.D. in U.S. history from the University of Maryland and a B.A. from Duke University.

Appendix: Tracking organizing using NLRB election data

NLRB data, 1950–2009

Data on National Labor Relations Board elections are available in annual NLRB reports for each year from 1950 to 2009. The data include number of eligible voters and number of elections, and offer a comparable breakdown for elections only where workers chose collective bargaining. The data developed for this analysis build on the work of Windham (2017) and Goldfield and Bromsen (2013). The data differ in the years before 1964 from the data for the 1964–2009 years. In the latter period the data in NLRB tables directly presents data for “RC” representation cases (see below), which are what is employed in our analysis. The NLRB annual reports provide data for “collective bargaining” cases in the 1950–1963 period. According to the annual reports: “The term ‘collective-bargaining election’ is used to cover representation elections requested by a union or other candidate for employee representative or by the employer” and is the sum of “RC,” “RM,” and “R” cases. The 1952 NLRB annual report (p. 277) defines these different types of cases. RC cases are: “A petition by a labor organization or employees for certification of a representative for purposes of collective bargaining under section 9 (c) (1) (A) (i).” RM cases are: “A petition by employer for certification of a representative for purposes of collective bargaining under section 9 (c) (1) (B).” R cases are: “A petition for certification of a representative for purposes of collective bargaining with an employer under section 9 of the National Labor Relations Act, prior to amendment.” These are carryovers from pre-Taft-Hartley elections. The last R election was processed in 1951.

RM elections are just a small part of total collective bargaining (CB) cases, accounting for 2-3% of elections and eligible voters in most years, though RM cases were huge in 1950, accounting for 6% of elections but 32% of eligible voters (reflecting elections by the International Union of Electrical Workers (IUE) and United Electrical Workers of America (UE) represented bargaining units). So, CB cases are for the most part comparable to RC cases. The election data for the 1950–1963 years are based on the CB cases less the RM cases, which is simply the RC cases plus the R cases. These counts differ from a straight count of RC cases only in the 1950–1951 years. Thus, our data counts for all years except 1950–1951 are the same as those in Windham (2017), who tracks RC cases. The specific tables used for collective bargaining cases are Table 15 (1950), Tables 12A and 12B (1951), Table 10 (1952), and Tables 13 and 13A (1953–1963). The specific tables used for RM cases are Table 13 (1950), Table 10 (1951), Table 9 (1952), and Table 11 (1953–1964).

Voter and election counts in elections where workers chose collective bargaining are not available for RC cases in the 1950–1963 period but are available for all CB cases. Except for 1950 the “win rates” for CB cases are primarily the result of RC cases because the RM cases are a small fraction. So, for the 1950–1963 period we use the win rates (computed as the number of eligible voters and elections where workers chose collective bargaining as a share of all elections and eligible voters) for the CB cases and apply them to our counts of elections and eligible voters to obtain counts of eligible voters and elections in units choosing collective bargaining. That is, our computations assume that the win rate (per election or eligible voter) is the same in RC as in all CB elections. This is possibly problematic only in 1950, where RM cases are significant, including 32% of voters.

NLRB data, 2007–2017

John-Paul Ferguson of McGill University kindly provided tabulations of NLRB election data for the 2007–2017 period. The data are described in Ferguson (2018).

Scaling to employment

Because the scale of the number of NLRB elections and, especially, the number of eligible voters should be viewed relative to the size of the workforce, we scale the data using two measures of employment. The first employment measure (used in Figure B) is from the Current Population Survey household survey, the employment of wage-and-salary workers in the nonagricultural private sector (BLS series LNS12032189Q, tabulated as the average of the four quarters of each year). The second employment measure is from the establishment survey and is the number of private-sector production and nonsupervisory workers (roughly 82% of all workers). This is what Windham (2017) uses to scale the election data. For the years 1950–1963 we use the BLS series EEU00500003 and for 1964 and later years we use BLS series CEU0500000006.

We present the annual data scaled to both series in  Appendix Table A. The average for each decade are presented in Appendix Table B. Our findings are robust to the selection of the particular employment measure to scale eligible voter data.

Appendix Table A

NLRB election results as a share of employment: Production worker and nonagricultural wage-and-salary employment

As share of production worker employment As share of nonagricultural W&S employment
Year Eligible voters Voters in election victories Eligible voters Voters in election victories
1950 1.76% 1.49% 1.53% 1.30%
1951 1.80% 1.36% 1.59% 1.21%
1952 2.04% 1.54% 1.81% 1.37%
1953 1.93% 1.53% 1.72% 1.36%
1954 1.36% 0.91% 1.21% 0.81%
1955 1.26% 0.92% 1.11% 0.81%
1956 1.16% 0.73% 1.02% 0.64%
1957 1.15% 0.66% 1.00% 0.58%
1958 0.91% 0.51% 0.77% 0.43%
1959 1.04% 0.62% 0.89% 0.53%
1960 1.20% 0.71% 1.02% 0.60%
1961 1.15% 0.58% 0.96% 0.49%
1962 1.32% 0.75% 1.11% 0.63%
1963 1.18% 0.64% 0.89% 0.53%
1964 1.28% 0.69% 1.06% 0.57%
1965 1.21% 0.75% 1.02% 0.63%
1966 1.24% 0.71% 1.06% 0.60%
1967 1.31% 0.75% 1.10% 0.63%
1968 1.11% 0.56% 0.94% 0.48%
1969 1.14% 0.58% 0.98% 0.49%
1970 1.19% 0.62% 1.01% 0.53%
1971 1.14% 0.54% 0.95% 0.45%
1972 1.11% 0.56% 0.94% 0.47%
1973 0.97% 0.42% 0.82% 0.35%
1974 0.96% 0.35% 0.80% 0.29%
1975 1.05% 0.40% 0.87% 0.33%
1976 0.82% 0.29% 0.68% 0.24%
1977 0.94% 0.36% 0.78% 0.30%
1978 0.73% 0.27% 0.60% 0.22%
1979 0.88% 0.32% 0.73% 0.26%
1980 0.78% 0.29% 0.65% 0.24%
1981 0.65% 0.24% 0.54% 0.19%
1982 0.41% 0.14% 0.33% 0.11%
1983 0.27% 0.12% 0.22% 0.10%
1984 0.32% 0.14% 0.26% 0.12%
1985 0.32% 0.12% 0.26% 0.10%
1986 0.32% 0.11% 0.26% 0.09%
1987 0.29% 0.12% 0.24% 0.10%
1988 0.29% 0.12% 0.24% 0.10%
1989 0.33% 0.13% 0.28% 0.11%
1990 0.31% 0.11% 0.26% 0.09%
1991 0.26% 0.10% 0.22% 0.08%
1992 0.25% 0.09% 0.21% 0.08%
1993 0.27% 0.11% 0.23% 0.10%
1994 0.24% 0.10% 0.20% 0.08%
1995 0.24% 0.10% 0.20% 0.08%
1996 0.23% 0.09% 0.20% 0.07%
1997 0.26% 0.11% 0.22% 0.09%
1998 0.26% 0.10% 0.23% 0.09%
1999 0.25% 0.12% 0.22% 0.10%
2000 0.26% 0.12% 0.22% 0.10%
2001 0.23% 0.09% 0.19% 0.08%
2002 0.20% 0.09% 0.16% 0.07%
2003 0.19% 0.09% 0.16% 0.07%
2004 0.18% 0.09% 0.15% 0.07%
2005 0.16% 0.08% 0.14% 0.06%
2006 0.13% 0.07% 0.11% 0.06%
2007 0.11% 0.06% 0.09% 0.05%
2008 0.12% 0.07% 0.10% 0.06%
2009 0.09% 0.06% 0.07% 0.05%

Note: Details described in appendix.

Source: Authors' analysis of National Labor Relations Board annual reports, Bureau of Labor Statistics.

Copy the code below to embed this chart on your website.

Appendix Table B

NLRB election results as a share of production worker and nonagricultural wage-and-salary employment, by decade

As share of production worker employment As share of nonagricultural W&S employment
Time period Eligible voters Voters in election victories Eligible voters Voters in election victories
1951–1960 1.39% 0.95% 1.21% 0.83%
1961–1970 1.21% 0.66% 1.01% 0.56%
1971–1980 0.94% 0.38% 0.78% 0.32%
1981–1990 0.35% 0.13% 0.29% 0.11%
1991–2000 0.25% 0.10% 0.21% 0.09%
2001–2009 0.16% 0.08% 0.13% 0.06%

Note: Details described in appendix.

Source: Authors' analysis of National Labor Relations Board annual reports, Bureau of Labor Statistics.

Copy the code below to embed this chart on your website.

Potential bias of NLRB data relative to all union organizing

There are three flows of new union members into private-sector unions that are not captured by NLRB statistics on private-sector union organizing. First, expanding union membership and new contracts in construction do not take place through NLRB elections. This has been the case the entire period under review, since 1951. So, our data do not reflect developments in construction.

Second, there are sectors in the private sector that are covered by the Railway Labor Act and not the NLRA; specifically, unionization in railroads and airlines takes place through elections overseen by the National Mediation Board (NMB). Annual reports of the National Mediation Board for the 1965–1980 period indicate that new unionization resulted in about 1,900 new member in the 1965–1971 period, falling to 1,540 new members each year in the 1972–1980 period, a 19% decline. We were also able to obtain election data from 2004 to 2011 and from 2013. In these later years there were 3,083 new members each year, though the average was particularly inflated by the 17,000 new members in 2013. Excluding 2013 the average number of new members obtained through NMB elections in the 2000–2011 period was 1,877. It does not appear that adding NMB elections to our analysis would materially change our results: the aggregate numbers are small and the NMB election trends also indicate a decline relative to employment.

Third, the most consequential omission from NLRB election data is the new membership obtained in card checks, frequently through voluntary recognitions, methods of organizing selected by unions to avoid NLRB procedures. Efforts to organize outside of the NLRB picked up in the mid-1990s. Therefore, organizing outside of the NLRB procedures has no bearing on our key results of the erosion of unionization that occurred between the late 1960s and late 1970s. How much does it change our assessment of new unionization since 1995? There is not much firm data on the scale of non-NLRB organizing. Schmitt and Zipperer (2009) identified three sources of data for the 1998–2003 period and found:

Since the data appear to be noisy and show no obvious trend, we take the average over all years of both sets of estimates and conclude that NLRB elections covered about 60 percent of potentially eligible private-sector workers over the period, leaving about 40 percent organized outside of the traditional NLRB-election procedures. We assumed that about 10 percent of workers had been organized outside these procedures in earlier years, and ended up with a scaling factor of 30 percent.

Brudney (2005) also provides an estimate of non-NLRB organizing in the 1998–2003 period, drawing on published information and correspondence from various union officials. Brudney (2005, page 11) notes the importance of non-NLRB organizing:

The Service Employees, Needle Trades Workers, Hotel and Restaurant Workers, and Autoworkers report that a plurality or majority of newly organized members have come in through contractual arrangements rather than traditional Labor Board supervised election campaigns. For these and other unions, neutrality plus card check account for more new recruits than NLRB election victories.

Brudney does not provide an estimate of the share of non-NLRB organizing as a share of all private-sector organizing.

The Schmitt and Zipperer calculations imply that actual combined organizing, inside and outside of NLRB process, was 66.7% (the 40% outside of NLRB relative to the 60% using NLRB) higher in the 1998–2003 period and that non-NLRB organizing raised organizing by 11.1% in earlier years. Adjusting our estimates accordingly to account for non-NLRB organizing shows the number of workers in successful union elections falling from 0.62% of employment (using all nonagriculture wage-and-salary employment as in Table 2) in the 1960s to just 0.11% of employment in the 2001–2009 period, just 17% as much. In contrast, our estimates presented in the paper show that new union members as a share of employment fell from 0.56% to 0.06%, a rate 12% as much. Including non-NLRB organizing does not alter the general picture of a drastic decline in new unionization over the last five decades.


i. Estlund (2002) and Brudney (2005) also detail the changes in legal interpretations and management practices that eroded the effectiveness of the NLRA in the absence of reforms to strengthen workers’ use of collective bargaining.

ii. This is detailed in two presentations: “What Is Needed to Achieve Density Goals in the Next Ten Years?” by Lawrence Mishel and John Schmitt at the AFL-CIO Commission on the Future of Work and Unions, December 17, 2018, and “Private Sector Union Density Dynamics, 1951–2017,” by Lawrence Mishel and John Schmitt at Columbia University, September 12, 2019. These analyses decomposed the changes in union coverage into that due to the inflows from organizing and the ongoing erosion of incumbent union representation, both of which accelerated in the 1970s.

iii. This is based on converting the nine years of available data (1970–1972 and 1973–1980) into a 10-year change (multiplying the nine-year change by 10/9). This simply assumes the 1972–1973 change is the average of what occurred in the other years of the 1970s.

iv. The Dunlop Commission Report (1994) reported a first contract rate of only 55.7% of unions between 1986 and 1993; Weiler (1984) shows a decline in first contract success rates from 86% to 63% from 1960 to 1980 (recent data from Ferguson 2008 cited by Liebman 2008).

v. The decomposition is based on the following relationship. The “share of workers with a new contract” equals the product of “share of workers in elections” times “union NLRB election win rates” times “first contract rate.” We simply take the natural log of each component in each time period and then subtract the earlier period (1966–1968) from the latter period (1978–1980) to obtain the changes. The percent contribution to the total change is obtained by dividing the change for that factor divided by the total change (share of workers with a new contract). All of the data for this calculation are provided in the table.

vi. The data for this period are based on computations of NLRB election data provided by Jean-Paul Ferguson of McGill University.

vii. Douglas Soutar, interview with Shelly Coppock, Litchfield, Ariz., November 16, 1990, transcripts #5843 OHT, NLRB Oral History Project II, Kheel Center for Labor Management Documentation & Archives, Catherwood Library, Cornell University, Ithaca, N.Y.

viii. Tabulations of the May surveys of the Current Population Survey from 1973 to 1980 for workers 16 or older and limited to those in the private sector.

ix. Hirsch, MacPherson, and Vroman (2001, p. 51) note: “the union membership question did not include the phrase ‘or employee association similar to a union.’ Absent any adjustment, union membership density in the CPS is measured as increasing from 22.4 percent in 1976 to 24.1 percent in 1977, despite the fact that membership was falling in years before and after 1977. BLS annual figures based on union financial reports, however, show a 0.4 -percentage point decline in union membership density between 1976 and 1977, from 24.5 percent to (coincidentally) the same 24.1 percent found in the CPS. Assuming that a time consistent CPS series would have fallen by 0.4 percentage points, a multiple of 1.094 is required to adjust upward pre-1977 figures to the post-1977 CPS definition including employee association members (that is, 1.094 times 22.4 percent equals 24.5 percent).”

x. NLRB annual reports, 1970-1980, Tables 2 and 4. See also Figure 1.

xi. 29 USC 151, “It is hereby declared to be the policy of the United States to eliminate the causes of certain substantial obstructions to the free flow of commerce and to mitigate and eliminate these obstructions when they have occurred by encouraging the practice and procedure of collective bargaining and by protecting the exercise.…”

xii. 29 USC 158(c).

xiii. 29 USC 158(b), 29 USC 157.

xiv. 29 USC 159(c)(1)(B).

xv. The other groups included a small group of politically active executives known as the March Group—founded by ALCOA and General Electric executives—and another employer organization, the Construction Users’ Anti-Inflation Roundtable (CUAIR).

xvi. Airporter Inn Hotel, 215 NLRB, 824 (1974); Stumpf Motor Co, 208 NLRB 431, 432 (1974); Birdsall Construction Company, 198 NLRB 163,163 (1972).

xvii. Clark Brothers, 70 NLRB 802 (1946). The decision read “[W]e must perform our function of protecting employees against that use of the employer’s economic power which is inherent in his ability to control their actions during working hours.”

xviii. NLRB v. Babcock & Wilcox, 351 U.S. 105 (1956).

xix. Following Babcock & Wilcox, the NLRB tried to develop an approach that recognized the importance of workplace communications to workers’ organizing rights, but the Supreme Court shut the NLRB down in its 1992 decision in Lechmere v. NLRB (Brudney 1996).

xx. The AFL-CIO estimated that 70% of employers held captive audience meetings in 1967; statement of William L. Kircher to the Special Labor Subcommittee to the House Education and Labor Committee on HR 11725, “A Bill to Amend the NLRA to Increase the Effectiveness of Remedies,” August 7, 1967, box 2, AFL-CIO ODR.

xxi. Textile Workers Union v. Darlington Mfg. Co., 380 U.S. 263 (1965).

xxii. Id., see also Estlund (1993) describing the inadequacy of current labor law in evaluating and addressing anti-union animus in investment decisions.

xxiii. Birdsall Construction Co, 198 NLRB 163, 163 (1972); Leggett and Pratt, Inc., 230 NLRB 463 (1977).

xxiv. NLRB annual reports, 1950-1990. For years 1954–1990, see Tables 2 and 4. For 1950, see Table 3A and Table 10. For 1952, see Tables 2 and 3.

xxv. Table 5, NLRB annual reports, 1950–1980. This chart is based on CA cases, charges of unfair labor practices against employers under section 8(a)1 of the NLRA. The union elections referenced in the chart are RC elections, those elections trigged by employees at a workplace who are trying to form a union.

xxvi. “Report on Union Buster (RUB)” sheet, September 1979, 8-9, box 11, Alan Kistler Papers, George Meany Memorial AFL-CIO Archive (GMMA). See also Wall Street Journal (1979).

xxvii. “Instructions for Use of the Block 30” sheet, Cannon Mills Supervisory Training Session, 1982, box 111, Acc 5619-017, Amalgamated Clothing and Textile Workers Union (ACTWU), Kheel Center for Labor-Management Documentation & Archives, Catherwood Library, Cornell University, Ithaca, N.Y.

xxviii. Boardwalk Regency Hotel, “Questions and Answer Fact Sheet,” September 1, 1980, master campaign file, box 6, The Papers of Earle K. Shawe, University of Virginia School of Law Special Collections, Arthur J. Morris Law Library, Charlottesville, Va.

xxix. “Guidelines for Supervisors and Managerial Employees During a Union Organizing Drive,” Brandeis University, January 12, 1976, folder 38, box 7, SEIU District 925 papers, Walter P. Reuther Library, Archives of Labor History and Urban Affairs, Wayne State University, Detroit, Mich. (hereafter WPR).

xxx. Speech by E.D. Smith, GA, Goldsboro, N.C., November 8, 1979, file 19, box 8, Alan Kistler Papers, GMMA.

xxxi. “Cannon Employee Questions Answered by Mr. Murdock,” n.d., 5619-017, box 111, Amalgamated Clothing and Textile Workers Papers, Kheel Center for Labor-Management Documentation and Archives, Catherwood Library, Cornell University, Ithaca, N.Y.

xxxii. Alan Kistler speech to secretary-treasurers meeting, St. Louis, Mo., March 1983, box 2, Alan Kistler Papers, GMMA. The survey was among units with over 100 employees.

xxxiii. “Decertification Activity,” n.d., c 1982, folder 10, box 4, Alan Kistler Papers, GMMA.

xxxiv. AFL-CIO News Release, May 4, 1978, box 85, series 6, Office of the President, George Meany files, GMMA.

xxxv. Douglas A. Fraser to Labor-Management Group Members, July 17, 1978, file: Labor O/A, box 231, Eizenstat papers, Domestic Policy Staff, Jimmy Carter Presidential Library & Museum, Atlanta, Ga.

xxxvi. Statement of Charles McDonald to the Labor and Management Subcommittee of the House Education and Labor Committee oversight hearings on the Landrum-Griffin Act, Labor Management Consultants, February 7, 1984, folder 2, box 2, Alan Kistler Papers, GMMA.

xxxvii. Doreen Levasseur interview by Ann Froines, February 23, 2005, Braintree, Mass., SEIU District 925 Legacy Project, oral history transcripts, WPR.

xxxviii. 29 USC 164(b).

xxxix. 29 USC 160(l).

xl. Authors’ analysis of data from NLRB annual reports, found at www.nlrb.gov.

xli. This shift has been quantified by scholarly research. Between 1940 and1969, the Supreme Court ruled in favor of the unions’ position nearly 80% of the time. After 1970, that percentage has dropped to only 50% (Brudney 1996, 1572-73).

xlii. Fibreboard Paper Products Corp. v. NLRB, 379 U.S. 203 (1964).

xliii. First National Maintenance Corp. v. NLRB, 452 U.S. 666 (1981).

xliv. NLRB v. Mackay Radio & Telegraph Co., 304 U.S. 333 (1938).

xlv. Id. at 176-77.

xlvi. Alan R.Gold, “Maine Paper Mill Workers End Bitter 16-Month Strike,” New York Times, October 11, 1988, https://www.nytimes.com/1988/10/11/us/maine-paper-mill-workers-end-bitter-16-month-strike.html.

xlvii. Id.

xlviii. McCartin’s documentation notes: M. H. LeRoy, “Regulating Employer Use of Permanent Striker Replacements: Empirical Analysis of NLRA and RLA Strikes 1935–1991,” Berkeley Journal of Employment and Labor Law 16 (1995): 169–207. Calculations on the frequency of striker replacement usage are based on the data in LeRoy’s appendix and Bureau of Labor Statistics data on work stoppages.

xlix. President Clinton issued an executive order to prohibit federal contractors from hiring permanent replacements in 1995, and it was challenged by the Chamber of Commerce and struck down by the D.C. Circuit as preempted by the NLRA; Chamber of Commerce v. Reich, 74 F.3d 1322 (D.C. Cir. 1996).

l. 380 U.S. 300 (1965).

li. Bloomberg Law’s Work Stoppages database covers strikes, lockouts, sickouts, and other instances of labor unrest of all sizes. Bloomberg Law researchers glean the information daily from published resources, including news sources, union publications, and government reports.

lii. Data on lockouts and strikes kindly provided by BNA’s Robert Coombs.

liii. H.K. Porter Co. v. NLRB, 397 U.S. 99 (1970).

liv. Ex-Cell-O, 185 NLRB 107 (1970), aff’d in relevant part, 449 F.2d 1058 (D.C. Cir).

lv. 465 U.S. 513 (1984).

lvi. “Oral Testimony of James Campbell Little: ACB Field Hearing Before the ABI Commission to Study the Reform of Chapter 11, at 35, Mar. 14, 2013 (ASM Transcript).

lvii. The analysis starts in 1977 because that is when the CPS data provided by Hirsch and MacPherson at www.unionstats.com begins having data on union coverage and not just membership.

lviii. The data for “1979” is the average of 1978–1980 to account for the volatility of the data due to relying on data from just one month’s CPS, the May CPS in those years. We multiply the change in manufacturing employment share by the average union coverage rate for the 1978–2019 period (18.8%) to obtain the 3.3 percentage point impact.

lix. The 1983 and 2019 data are from Hirsch and MacPherson (2020) tabulations of CPS data. The 1979 data are tabulations of the pooled May CPS surveys for 1978–1980. This allows us to gauge the sharp union coverage declines in 1979–1983. Communications in 2019 from aggregate of: radio and television and cable; internet publishing and broadcasting; wired telecommunications; and other communications. Hotels in 1983 are matched to traveler accommodations in 2019.

lx. These were chosen from among those detailed industries that had the same names in both 1983 and 2019.

lxi. John Schmitt suggested this approach.

lxii. We do so for “1979,” the pooled May CPS surveys of 1978–1980, and for 2019 (CPS-ORG data). The dependent variable is union coverage and the explanatory variables (mostly dummy variables) are gender (and marital status); occupation (22 categories); race/ethnicity; education (five categories); age (six categories); state (50 plus D.C.); and industry (12 categories). The sample is private-sector wage-and-salary workers.

lxiii. Means of the dependent variables, union coverage, in the regressions for 1979 and 2019.

lxiv. This method is a traditional Oaxaca decomposition.

lxv. In other words, changes in the coefficients in the separate regressions for 1979 and 2019.

lxvi. Specifically, the decomposition examines: (1) the changes in the coefficients between the 1979 and 2019 regressions evaluated at the mean of the variables in 2019; and (2) the changes in the mean of the variables between 1979 and 2019 evaluated with the coefficients of each factor in 1979.

lxvii. Specifically, we start with the regression results for each year. The mean of union coverage is the sum of each variable’s coefficient times the mean of that variable. We then substitute the means of the industrial sector dummy variable for the alternative year (i.e., the 2019 means plugged into the 1979 regression, keeping the coefficients for 1979 and the mean of variables for factors other than industry employment at their 1979 means). This gives us our counterfactual estimate for that year.

lxviii. As noted in Schmitt and Mitukiewicz (2012), “Acemoglu et al. (2001), for example, argue that skill-biased technical change (SBTC) is ‘at the root of deunionisation.’ SBTC ‘undermines the coalition between skilled and unskilled workers underlying unions’ by increasing the relative competitive-market pay of skilled workers, which ‘weakens their incentives to join the unionised sector’ (p. 231).”

lxix. The strength of Autor et al. is that in their identification strategy (how to isolate “causality”) they “use industry-level Chinese exports to other OECD countries as an instrument for industry-level exports from China to the U.S.”

lxx. “We estimate that exposure to import competition induced a 2 percentage point decline in union density within manufacturing, roughly one-sixth of the observed decline between 1990 and 2014.”

lxxi. “Multivariate decompositions analysis based on probit regressions including control for sex (female), age groups, education, migrant workers, job tenure, type of contract (part-time), contract duration (temporary jobs), occupation, industry, quintiles of the hourly earnings, sector (public sector) and firm size” (Annex Table 2.C.2.). For the United States there is no control for firm size or contract duration.

lxxii. Data drawn from Annex Table 2.B.3., Annex Table 2.C.2., and Annex Table 2.C.1.

lxxiii. The question asked was, “If an election were held today to decide whether employees like you should be represented by a union, would you vote for the union or against the union?”


Acemoglu, D., P. Aghion, and G.L. Violante. 2001. “Deunionization, Technical Change and Inequality.” In Carnegie-Rochester Conference Series on Public Policy, vol. 55, 229-64. New York: Elsevier.

Ahlquist, John S., and Mitch Downey. 2019. “Import Exposure and Unionization in the United States.”

Applebaum, Eileen, and Rosemary Batt. 2014. Private Equity at Work: When Wall Street Manages Main Street. New York: Russell Sage Foundation.

Autor, D.H., D. Dorn, and G.H. Hanson. 2013. “The China Syndrome: Local Labor Market Effects of Import Competition in the United States.” American Economic Review 103, no. 6: 2121-68.

Becker, Craig. 1991–1993. “Democracy in the Workplace: Union Representation Elections and Federal Labor Law.” Minnesota Law Review 77: 495-603.

Biden for President. 2020. “The Biden Plan for Strengthening Worker Organizing, Collective Bargaining, and Unions.”

Brenner, Aaron, Robert Brenner, and Calvin Winslow, eds. 2010. Rebel Rank and File: Labor Militancy and Revolt from Below in the Long 1970s. London: Verso.

Brenner, Robert. 2006. The Economics of Global Turbulence: The Advanced Capitalist Economies from Long Boom to Long Downturn, 1945-2005. London: Verso.

Bronfenbrenner, Kate. 2000. “Uneasy Terrain: The Impact of Capital Mobility on Workers, Wages, and Union Organizing.” Submitted to the U.S. Trade Deficit Review Commission. <http://digitalcommons.ilr.cornell.edu/reports/3/ >.

Brown, Michael K. 1999. Race, Money, and the American Welfare State. Ithaca, NY: Cornell University Press.

Brudney, James J. 1996. “Reflections on Group Action and the Law of the Workplace,” Texas Law Review 74, no. 1563: 1577-78.

Brudney, James J. 2005. “Neutrality Agreements and Card Check Recognition: Prospects for Changing Paradigms.” Iowa Law Review 90.

Bureau of Labor Statistics (BLS), Department of Labor. 1980. Perspectives on Working Women: A Databook. Bulletin 2080. Washington, D.C.: U.S. Government Printing Office.

Ceccotti, Babette A. 2007. “Lost in Transformation: The Disappearance of Labor Policies in Applying Section 1113 of the Bankruptcy Code.” American Bankruptcy Institute Law Review 15: 415.

Cobble, Dorothy Sue. 2004. The Other Women’s Movement: Workplace Justice and Social Rights in Modern America. Princeton, N.J.: Princeton University Press.

Coombs, Robert. 2012a. Labor Stats and Facts: Why 2011 Was the Biggest Year Yet for Lockouts. Washington, D.C.: Bureau of National Affairs.

Coombs, Robert. 2012b. Labor Stats and Facts: Lockout Rates Continue to Surge. Washington, D.C.: Bureau of National Affairs.

Cowie, Jefferson. 2010. Stayin’ Alive: The 1970s and the Last Days of the Working Class. New York: New Press.

Davis, Gerald F. 2009. Managed by the Markets: How Finance Reshaped America. New York: Oxford University Press.

Davis, Mike. 1986. Prisoners of the American Dream: Politics and Economy in the History of the U.S. Working Class. London: Verso.

DeMaria, Alfred T. 1982. How Management Wins Union Organizing Campaigns. New York: Executive Enterprises Publications Co., Prentice-Hall.

Denice, Patrick, and Jake Rosenfeld. 2018. “Unions and Nonunion Pay in the United States, 1977-2015.” Sociological Science 5: 541-61.

Deslippe, Dennis. 2000. Rights, Not Roses: Unions and the Rise of Working-Class Feminism. Urbana: University of Illinois Press.

Dickens, William T. 1983, “The Effect of Company Campaigns on Certification Elections: Law and Reality Once Again.” Industrial & Labor Relations Review 36, no. 4: 560-75.

Eidelson, Josh. 2012. “Rise of the Lockout: Another Sign of Labor’s Slide.” Salon, September 28.

Estlund, Cynthia. 1993. “Economic Rationality and Union Avoidance: Misunderstanding the National Labor Relations Act.” Texas Law Review 71, no. 921: 937-38.

Estlund, Cynthia. 2002. “The Ossification of American Labor Law.” Columbia Law Review 102: 1527.

Ferguson, John-Paul. 2008. “The Eyes of the Needles: A Sequential Model of Union Organizing Drives, 1999-2004.” Industrial & Labor Relations Review 62, no. 1: 3–21.

Ferguson, John-Paul. 2018. “Organizing Trends in CATS and NextGen NLRB Data.” Technical Report. Washington, D.C.: AFL-CIO.

Fisk, Catherine L., and Adam R. Pulver. 2009. “First Contract Arbitration and the Employee Free Choice Act.” Louisiana Law Review 70, no. 1.

Fantasia, Rick, and Kim Voss. 2004. Hard Work: Remaking the American Labor Movement. Berkeley: University of California Press.

Farber, Henry S., Daniel Herbst, Ilyana Kuziemko, and Suresh Naidu. 2018. “Unions and Inequality over the Twentieth Century: New Evidence from Survey Data.” Working Paper No. 24587. Cambridge, Mass.: National Bureau of Economic Research.

Fortin, Nicole M., Thomas Lemieux, and Neil Lloyd. 2019. “Labor Market Institutions and the Distribution of Wages: The Role of Spillover Effects.” Vancouver, Canada: Vancouver School of Economics, University of British Columbia.

Freeman, Richard. 2007. “Do Workers Still Want Unions? More Than Ever.” Briefing Paper 182. Washington, D.C. Economic Policy Institute.

Freeman, Richard B., and Morris M. Kleiner. 1990. “Employer Behavior in the Face of Union Organizing Drives.” Industrial & Labor Relations Review 43, no. 4: 351-65.

Frymer, Paul. 2008. Black and Blue: African-Americans, the Labor Movement, and the Decline of the Democratic Party. Princeton, N.J.: Princeton University Press.

Fulmer, William E. 1982. Union Organizing: Management and Labor Conflict. New York: Praeger.

Galvin, Daniel J., and Jacob S. Hacker. 2020. “The Political Effects of Policy Drift: Policy Stalemate and American Political Development.” Studies in American Political Development, 1–23.

General Accounting Office (GAO). 1991. “Strikes and Striker Replacements in the 1970s and 1980s.” GAO/HRD-91-2. Washington, D.C.: GAO.

Gregory, David. 1985. “The NLRB and the Politics of Labor Law.” Boston College Law Review 27: 39-40.

Goldfield, Michael. 1987. The Decline of Organized Labor in the United States. Chicago: University of Chicago Press.

Goldfield, Michael, and Amy Bromsen. 2013. “The Changing Landscape of U.S. Unions in Historical and Theoretical Perspective.” Annual Review of Political Science 16: 231-57.

Greene, Julie. 1998. Pure and Simple Politics: The American Federation of Labor and Political Activism, 1881-1917. Cambridge, England: Cambridge University Press.

Greenhouse, Steven. 2012. “More Lockouts as Companies Battle Unions.” New York Times, January 23.

Greenhouse, Steven. 2019. “Worker Rights Are Shaping Up a Key Issue in 2020. Who Has the Best Ideas?: A to F+: Grading the Presidential Candidates on Their Labor Plans.” New York Magazine, September 23.

Groshen, Erica L., and Harry J. Holzer. 2019. “Improving Employment and Earnings in Twenty-First Century Labor Markets.” Russell Sage Foundation Journal of the Social Sciences 5, no. 5.

Gross, James. 1995. Broken Promise: The Subversion of U.S. Labor Relations Policy. Philadelphia, Pa.: Temple University Press.

Gross, James. 1981. The Reshaping of the National Labor Relations Board: National Labor Policy in Transition, 1937-1947. Albany: State University of New York Press.

Hacker, Jacob S. 2002. The Divided Welfare State: The Battle over Public and Private Social Benefits in the United States. New York: Cambridge University Press.

Hacker, Jacob S., and Paul Pierson. 2010. Winner-Take-All Politics: How Washington Made the Rich Richer—and Turned Its Back on the Middle-Class. New York: Simon and Schuster.

Harner, Michelle M. 2014. Final Report of the ABI Commission to Study the Reform of Chapter 11. Alexandria, Va.: American Bankruptcy Institute.

Hirsch, Barry. 2008. “Sluggish Institutions in a Dynamic World: Can Unions and Industrial Competition Coexist?” Journal of Economic Perspectives (Winter).

Hirsch, Barry T., and David A. Macpherson. 2003. “Union Membership and Coverage Database from the Current Population Survey: Note.” Industrial and Labor Relations Review 56, no. 2 (January): 349-54.

Hirsch, Barry T., and David A. Macpherson. 2020. Hirsch and Macpherson (2003) updated at unionstats.com.

Hirsch, Barry T., David A. Macpherson, and Wayne G. Vroman, 2001. “Estimates of Union Density by State.” Monthly Labor Review 124, no. 7.

Hughes, Charles L. 1976. Making Unions Unnecessary. New York: Executive Enterprises Publications.

Human Rights Watch. 2000. Unfair Advantage: Workers’ Freedom of Association in the United States Under International Human Rights Standards. New York: Human Rights Watch.

Jackson, Gordon E. 1981. When Labor Trouble Strikes: An Action Handbook. Englewood Cliffs, N.J.: Prentice-Hall.

Jones, Jeffrey M. 2019. “As Labor Day Turns 125, Union Approval Near 50-Year High.” Gallup, August 28.

Jones, William Powell. 2013. The March on Washington: Jobs, Freedom, and the Forgotten History of Civil Rights. New York: W.W. Norton.

Katznelson, Ira. 2005. When Affirmative Action Was White: An Untold History of Racial Inequality in Twentieth-Century America. New York: W.W. Norton.

Kessler-Harris, Alice. 2001. In Pursuit of Equity: Women, Men, and the Quest for Economic Citizenship in 20th Century America. New York: Oxford University Press.

Kilgour, John G. 1981. Preventative Labor Relations. New York: Amacom.

Kilgour, John G. 1982. “Office Unions: Keeping the Threat Small.” Administrative Management (November).

Kistler, Alan. Papers, RG95-009. College Park, Md.: George Meany Memorial AFL-CIO Archive, University of Maryland.

Kistler, Alan. 1984. “Union Organizing: New Challenges and Prospects.” Annals of the American Academy of Political and Social Science 473 (May): 96-107.

Kochan, Thomas A., and Will Kimball. 2019. “Unions, Worker Voice, and Management Practices: Implications for a High-Productivity, High-Wage Economy.” In Groshen and Holzer 2019.

Kochan, Thomas A., Will Kimball, Duanyi Yang, and Erin L. Kelly. 2018. “Worker Voice in America: Is There a Gap Between What Workers Expect and What They Get?” Working Paper. Cambridge, Mass.: MIT Institute for Work and Employment Research. https://doi.org/10.1177/0019793918806250.

LeRoy, Michael H. 1995. “Regulating Employer Use of Permanent Striker Replacements: Empirical Analysis of NLRA and RLA Strikes 1935–1991.” Berkeley Journal of Employment and Labor Law 16: 169–207.

Levitt, Martin Jay, and Terry Conrow. 1993. Confessions of a Union Buster. New York: Crown Publishers.

Lichtenstein, Nelson. 2013. State of the Union: A Century of American Labor (Revised and Expanded Edition). Princeton, N.J.: Princeton University Press.

Liebman, Wilma. 2008. “National Labor Relations Board Representation Elections and Initial Collective Bargaining Agreements: Safeguarding Workers’ Rights?” Testimony before the U.S Senate Subcommittee on Labor, Health and Human Services, Education, and Related Agencies of the Senate Committee on Appropriations, Washington, D.C., April 2.

Linder, Marc. 1999. Wars of Attrition: Vietnam, the Business Roundtable, and the Decline of Construction Unions. Iowa City, Iowa: Fanpihua Press.

Locke, Robert R., and J.-C. Spender. 2011. Confronting Managerialism: How the Business Elite and Their Schools Threw Our Lives Out of Balance. London: Zed Books.

Logan, John. 2006. “The Union Avoidance Industry in the United States.” BJIR British Journal of Industrial Relations 44, no. 4: 651-75.

Logan, John. 2008. “Permanent Replacements and the End of Labor’s ‘Only True Weapon.’” International Labor and Working-Class History 74, no.1: 171-92.

MacLean, Nancy. 2006. Freedom Is Not Enough: The Opening of the American Workplace. New York: Russell Sage Foundation and Cambridge, Mass.: Harvard University Press.

Marvit, Moshe. 2016. “Is It Time for the Courts to End Labor Lockouts?” The Century Foundation, June 30.

McAlevey, Jane. 2017. No Shortcuts: Organizing for Power in the New Gilded Age. New York: Oxford University Press.

McCartin, Joseph A. 2006. “PATCO, Permanent Replacement, and the Loss of Labor’s Strike Weapon.” Perspectives on Work 10, no. 1: 17-19.

McCartin, Joseph A. 2011. Collision Course: Ronald Reagan, the Air Traffic Controllers, and the Strike that Changed America. New York: Oxford University Press.

McDonald, Charles, and Dick Wilson. 1979. “Peddling the Union-Free Guarantee,” AFL-CIO Federationist (April).

McNicholas, Celine. 2020. “EPI Applauds House Passage of Comprehensive Labor Reform Legislation—The PRO Act.” Economic Policy Institute statement, February.

Moody, Kim. 1988. An Injury to All: The Decline of American Unionism. London: Verso.

National Labor Relations Board. Annual Reports. Washington, D.C.: U.S. Government Printing Office.

Organization for Economic Cooperation and Development (OECD). 2019. Negotiating Our Way Up: Collective Bargaining in a Changing World of Work. Paris: OECD Publishing.

Pestillo, Peter J. 1978. “Learning to Live Without the Union.” Paper presented at the Annual Meeting of the Industrial Relations Research Association, Madison, Wis., August 30.

Phillips-Fein, Kim. 2009. Invisible Hands: The Making of the Conservative Movement from the New Deal to Reagan. New York: W.W. Norton.

Pierce, J.R. and P.K. Schott. 2016. “Trade Liberalization and Mortality: Evidence from U.S. Counties.” Working Paper No. 22849. Cambridge, Mass.: National Bureau of Economic Research. https://doi: 10.3386/w22849.

Prosten, Richard. 1978. “The Longest Season: Union Organizing in the Last Decade, a/k/a How One Team Has to Play with Its Shoelaces Tied Together.” Proceedings of the 31st Annual Meeting of the Industrial Relations Research Association, Madison, Wis., 240-49.

Quinn, Robert P., and Graham L. Staines. 1979. The 1977 Quality of Employment Survey: Descriptive Statistics, with Comparison Data from the 1969-70 and the 1972-73 Surveys. Ann Arbor: Survey Research Center, Institute for Social Research, University of Michigan.

Rosenberg, Rosalind. 1984. “Bankruptcy and the Collective Bargaining Agreement: A Brief Lesson in the Use of the Constitutional System of Checks and Balances.” American Bankruptcy Law Journal 58, no. 316: 293-306.

Rosenfeld, Jake. 2020 (forthcoming). You’re Paid What You’re Worth and Other Myths of the Modern Economy. Cambridge, Mass.: Harvard University Press.

Schmitt, John, and Alexandra Mitukiewicz. 2012. “Politics Matter: Changes in Unionisation Rates in Rich Countries, 1960–2010.” Industrial Relations Journal 43, no. 3: 260–80.

Schmitt, John, and Ben Zipperer. 2009. “Dropping the Ax: Illegal Firings During Union Election Campaigns, 1951-2007.” Washington, D.C.: Center for Economic and Policy Research.

Seeber, Ronald L., and William Cooke. 1983. “The Decline in Union Success in NLRB Representation Elections.” Industrial Relations 22, no. 1.

Shawe, Earle K. 1979. “The Move Towards De-Unionization.” The Answer (Fall).

Sousa, Michael D. 2003. “Reconciling the Otherwise Irreconcilable: The Rejection of Collective Bargaining Agreements Under Section 1113 of the Bankruptcy Code.” Labor Lawyer 18, no. 3: 453-83.

Stansbury, Anna, and Lawrence Summers. 2020. “The Declining Worker Power Hypothesis: An Explanation for the Recent Evolution of the American Economy.” Brookings Papers on Economic Activity Conference, Washington, D.C., 2020.

Stein, Judith. 2010. Pivotal Decade: How the United States Traded Factories for Finance in the Seventies. New Haven, Conn.: Yale University Press.

Stelzner, Mark. 2017. “The New American Way: How Changes in Labour Law Are Increasing Inequality.” Industrial Relations Journal 48, no. 3: 241-42.

Townsend, Ed. 1984. “Bankruptcy Ruling Gives Employer an Edge in Labor Bargaining,” Christian Science Monitor, February 24.

Troy, Leo, and Neil Sheflin. 1985. U.S. Union Sourcebook: Membership, Structure, Finance, Directory. West Orange, N.J.: Industrial Relations Data and Information Services.

U.S. Congress, House of Representatives, Oversight Hearings Before the Subcommittee on Labor-Management Relations of the Committee on Education and Labor, 96th Congress, First and Second Sessions. 1980. Pressure in Today’s Workplace, Volumes I-III. Washington, D.C.: U.S. Government Printing Office.

U.S. Department of Labor. 1994. Commission on the Future of Worker-Management Relations. Washington, D.C.: U.S. Government Printing Office.

VanHeuvelen, Tom. 2020. “The Right to Work, Power Resources, and Economic Inequality.” American Journal of Sociology 125, no. 5.

Vogel, David. 1989. Fluctuating Fortunes: The Political Power of Business in America. New York: Basic Books.

Wall Street Journal. 1979. “Labor Nemesis: When the Boss Calls in This Expert, the Union May Be in Real Trouble,” November 19.

Waterhouse, Benjamin C. 2013. Lobbying America: The Politics of Business from Nixon to NAFTA. Princeton, N.J.: Princeton University Press.

Weil, David. 2005. “Public Enforcement/Private Monitoring: Evaluating a New Approach to Regulating the Minimum Wage.” Industrial & Labor Relations Review 58, no. 2: 238-57.

Weiler, Paul. 1984. “Striking a New Balance: Freedom of Contract and the Prospects for Union Representation.” Harvard Law Review 98: 351, 353.

Western, Bruce. 1997. Between Class and Markets: Postwar Unionization in the Capitalist Economies. Princeton, N.J.: Princeton University Press.

Windham, Lane. 2017. Knocking on Labor’s Door: Union Organizing and the Roots of a New Economic Divide. Chapel Hill: University of North Carolina Press.

See more work by Lawrence Mishel, Lynn Rhinehart, and Lane Windham