Esteemed US labour economist Richard Freeman puts his case for open source unionism, and argues for a wider definition of union membership and a key role for the Internet in union renaissance and worker well being.

The Road To Union Renascence And Worker Well-Being
In The US

Conferences on US unionism have turned into wakes, as year after year the numbers on membership and density show a long term downward trajectory in the rate of unionization. Private sector union density has fallen to single digits – comparable to the level at the turn of the 20th century. Density in the public sector is four times the private sector level and holding steady, but the public sector accounts for just about 15% of the US work force. Total density may stabilize at a rate higher than the 10% that Medoff and I projected in What Do Unions Do? (1984, p 241) but it may drop even further. In the state with the lowest rate of public and private unionization, North Carolina, density fell from 5.3% in 1990 to 3.2% in 2002.[1]

  • Early analyses attributed the decline in unionism to shifts in the
    distribution of employment among industries – the decline of traditional
    union sectors such as manufacturing and mining. But density has fallen within
    virtually every private sector industry. Later analyses attributed the trend to
    management opposition operating under the National Labor Relations Act, which
    has institutionalized a process that effectively gives management near veto
    power of whether or not workers become organized. Since unions raise wages and
    benefits, which reduces profits, and challenge management power at workplaces,
    most American managements spend considerable money and time to keep their
    workplace union-free. Looking at the union side of the organizing process,
    union analysts have attributed part of the downward trend to union failure to
    increase organizing resources commensurate with management opposition.
  • In many, but not all, other advanced countries, union membership is also
    falling, and union influence with it. But labor laws throughout much of the
    European Union enshrine collective bargaining as a mode of wage-setting by
    extending contracts from organized to non-organized firms, so that rates of
    collective bargaining coverage remain high. In France, union density is a bare
    6%–considerably below the US density – but collective bargaining
    determines the pay and conditions for some 92% of the work force, By contrast,
    as density falls in the US, so too does collective bargaining coverage, which
    weakens the ability of unions to gain benefits for their members. In no
    advanced democracy do unions face as bleak a future as in the US.
  • Has the time come to give last rites to unions as a significant force in the US? Should labor relations experts turn their subject over to the morticians, archaeologists, and economic historians? Or can labor find a way to reverse the process of de-unionization and become again a major force in the economy?

    The labor movement has sought to reverse the process of de-unionization in two ways. First, some unions have increased the share of their budgets to organizing. Most leaders and activists see organizing workers and gaining collective contracts as the only way to increase union membership. Second, unions have devoted resources in elections, especially at the national level. Unions endorse candidates who support Labor issues; get union members out to vote on election day; and hold officials accountable once they are elected.[2] Many union leaders and activists see a union friendly President and Congress as the only way to create a more favorable legal and administrative environment and to improve the well-being of all workers, non-union as well as union. Neither of these strategies – increased organizing effort or increased national political activity – has yet succeeded in stemming the tide of decline.

    In this essay I argue that the statistics of union organizing and the statistics of union voting make it difficult for unions to recover through either of these conventional channels. Unions need something different – the proverbial out of the box solution – if they are to break the downward trend in union density and influence. I suggest two innovations that could help turn membership numbers around: development of a new looser union form; and devolution of labor laws and union political activity to state and local communities.

    Normal Organizing Cannot Reverse the Decline

    In response to the fall in union density , the “New Voice” leadership team that the AFL-CIO elected to head the organization in 1996 embarked on a campaign to convince unions to increase their organizing resources from modest amounts to 30% or so of dues money. In contrast to unionization in many other countries, unionization in the US requires continual organizing to keep from eroding. The natural ebb and flow of business means that unionized and nonunionized firms or establishments within firms go out of business every year. Since new firms and establishments are born nonunion, maintaining union membership requires unions to organize the same number of workers as they lose through the normal turnover of firms. To maintain union density in a growing work force, unions must also organize the same proportion of the growth of employment that they hold of the existing work force.

    Responding to this reality, some unions have increased their organizing activity considerably. UNITE raised its organizing effort from 20% of its budget to 40% from 1995 to 2002. SEIU has gone on a huge campaign to organize workers in the health care sector. The UAW invested resources in organizing graduate students around the country. Three of the most active unions in organizing chose their organizing directors to lead their union: HERE, UNITE, SEIU. But most unions did not come close to spending the 30% of dues money on organizing that the AFL-CIO wanted them to do, possibly because they did not have good organizing targets. Some unions, like the Teamsters, cut their organizing budgets. Others, like the Steelworkers, had trouble finding campaigns where large expenditures will pay off. Still overall, aggregate union organizing budgets rose in the 1990s, though in the absence of an AFL-CIO census of union organizing activity, it is difficult to know by how much.

    New emphasis on organizing notwithstanding, union density fell in the late 1990s/early 2000s. Arguably the increased organizing effort helped staunch the rate of union decline. From 1996 to 2002, private sector density fell by 1.3 percentage points (– 9 %), compared to 1.6 percentage points (–10%) in the preceding six years. As many factors affect union density, these raw statistics do not isolate the impact of the commitment to organizing. Absent the new organizing effort, union density might have fallen more rapidly than in earlier years. But whatever its impact, the new effort did not produce a rebound in membership.

    The statistics of organizing suggest that almost no conceivable increase in normal organizing activity can reverse the downward trend in unionization. To maintain its 2002 8.9% share of the private workforce, unions must add about 500,000 new members annually. To add a point of density, unions must organize close to 1 million new members. National Labor Relations Board (NLRB) election statistics show that unions win less than 100,000 workers annually, of whom at most two-thirds eventually end up with collective bargaining contracts. The AFL-CIO’s Work in Progress reports higher numbers, but these figures are a mixture of affiliations of previously organized workers into the Federation, new unionization of public sector workers, or of workers in the health sector where the state funds the employers, as well as unionization of private sector workers covered by the NLRB. Of the 230,000 new members reported in 2002 only 67,000 came from private industries exclusive of health care; 84,000 were health care workers working for private groups but under state contracts; and another 18,000 were also in the health sector.[3] The numbers of newly organized workers reported in Work in Progres have fallen far short of the one half million newly organized members needed to stabilize density in all but one year. The report estimated that unions organized 161,000 new members in 2000; 230,000 in 2002, and were on a pace through March 2003 to organize about 200,000 new members in 2003. The exceptional year was 2001, when Work in Progress reported 446,000 new members. But the jump in 2001 was due to affiliation of existing unions – the California State Employees Association (175,000) and the United American Nurses (100,000) to the federation – not to organization of previously non-union workers. .

    To fund an organizing campaign sufficiently massive to reverse the downward trend in density would take huge union resources. Turning Paula Voos’s (1983) estimates of the marginal cost of organizing a new member into 2001 dollars, the cost of organizing a new member would appear to be on the order of $2,000 – though it could be as low as the $1,000 that is the rule of thumb for some unions and as high as $3,000.[4] Adding ½ million new members annually at $2,000 per member would then require spending $1 billion, or about 20 percent of total annual union dues. Adding 1 million members would take about 40 percent of total dues.[5] These figures bracket the AFL-CIO’s goal that unions dedicate 30 percent of revenues to organizing.

    Making the problem more difficult, however, I would expect that increased union organizing would raise the cost of organizing per new member. Assuming that unions target the most winnable campaigns first and then move onto difficult campaigns, the cost curve for organizing is likely to be upward sloping: more money needed per new member as the number organized rises. In addition, if union campaigns were to score some large successes, management might increase its expenditures on anti-union activities, which in turn would require greater union resources to gain new members. The main beneficiary from such an “arms race” would be the union busting firms.

    As union membership falls, moreover, another factor kicks in to affect the cost calculus even if the cost of organizing an additional worker is constant. The expense to existing union members of organizing new workers rises because the cost of each newly organized worker is shared by fewer members, raising the amount that each member. A 10% drop in membership makes it 10% more expensive to remaining members to fund an organizing campaign to organize a new worker. This creates a vicious spiral in the organizing process: fewer members –> less resources available for organizing –> less organizing success –> fewer members, and so on.

    In sum, continued decline in private sector density in the “organize or die” period and the statistics of organizing shows that even huge increases in normal organizing activity have little chance of producing a labor renascence.

    Normal Politics Cannot Reverse the

    Unable to gain additional members by organizing, unions have sought to strengthen their influence on economic outcomes by devoting resources to politics. By getting out the union vote and influencing union voters to support selected candidates, unions play a major role in national politics. In election years, many unions put aside organizing and allot their discretionary resources to politics. Although forty percent of union members typically vote Republican, most union political resources goes to the Democrats, albeit with some notable exceptions, such as in the 2002 NY governor’s race. Union resources are critical to the Democratic Party. Absent union support, the Democrats would have to develop new ways of operating or a policy agenda more favorable to the upper and middle classes who contribute most money to campaigns and make up a disproportionate share of the electorate.

    Statistics from exit polls on the union affiliation of voters suggest that the union political effort has been remarkably successful (Rosenthal, 2003) From 1992 to 2000, Voting News Service (VNS) exit polls show that the share of households with union member rose from 19% to 26% of the electorate. Over this period the VNS data show that the number of voters from union households increased by 4.8 million while the number of voters from non-union households fell by 15.5 million. To increase the union share of voters in the face of the downward trend in the union share of the work force would seem to be nothing short of miraculous. If correct, these statistics suggest a huge payoff to the union effort to increase turnout and its political clout.

    But the exit poll data is not correct. Two other data sets that give information on unionization and voting – the November Current Population Survey and the National Election Study – show no rise in the union share of voters from 1992 to 2000. Instead, both data sets show a drop of 1 percentage point in the proportion of voters from union households over the period. The inconsistency reflects a problem with the VNS data. Between 1992 and 2000, the VNS changed the question regarding union affiliation on the exit poll in ways that produced a misleading picture of union success. The 2000 question asked if the respondent or someone else in the household was in a union and probed whether the individual, someone else, or both were in the union. By probing for who was in the union, the 2000 VNS presumably prompted a higher union response than if it simply asked for a yes/no response. In 1996 the question was also a reasonably good one, “do you or someone in your household belong to a labor union? This was a separate item with places for the respondent to mark yes or no, but no probing for whether the union person was the voter or someone else..

    In 1992, however, the union question was something utterly different. The VNS asked the respondent to check if they were in a union household in the middle of a “Grabbag” of 14 items – whether the person was gay/lesbian, environmentalist, religious right, etc . There was no space to check no, so the data do not differentiate persons who left the space blank from those who answered “no”. The item was on the second sheet of the questionnaire, which some persons did not read at all and did not respond to the grabbag. This design necessarily underestimated the union share of persons who were asked the exit poll in that year. The VNS warns: “Exit poll users are cautioned against comparing estimates from the Grabbag with those from full questions because the Grabbag significantly underestimates the population values”.[7] Indicative of the problem in 1988 when CBS conducted its own exit poll, it asked the question “are you or is anyone in the household a member of a union?” and obtained a 27% positive response – similar to the 2000 VNS response. By contrast, in 1988 ABC listed being a member of a union household on a grabbag set of items under the question “do any of the following apply to you?” and received a 10% positive response. Adding to the unreliability of the VNS for analyzing trends over time, the response rate to the VNS exit poll fell from 60% in 1992 to 50% in 2000, and shows a marked difference in the number voting from administrative voting records.

    That unions have not performed the political miracle indicated by the raw VNS data does not mean that their political activity has failed. To the contrary, unions have succeeded in increasing turnout and inducing members to vote for union-preferred candidates. My analysis of November CPS data from 1988 to 2000 and of National Election Study (NES) data from 1948 to 2000 suggest that unions increased the turnout of members and their families compared to demographically equivalent non-union members (Freeman, 2003). Both data sets suggest that unions increase the likelihood of voting by 4-5 percentage points. In addition, the NES shows that union members have a more liberal orientation and are 9-10 percentage points more likely to support the Democrats (taken as the more pro-union party) than non-union members with the same demographic characteristics and income. The union impact appears to operate through union members having more liberal and pro-Democratic party attitudes than otherwise comparable non-members Since the bulk of union members come from the upper half of the US earnings distribution, they could be expected to favor Republicans more than Democrats. That they do not reflects in part the impact of their unions on their political orientation.

    But union success in raising turnout and influencing members’ voting has not translated into national policies that can reverse the downward spiral in density. The fall in density per year during the Clinton Administration is nearly as great as the fall during the years Republicans controlled the White House. Clinton invited union leaders to the White House but battled against the AFL-CIO to enact the NAFTA legislation that business favored, and made numerous other policy decisions favorable to business groups, particularly those in high tech. The Commission on the Future of Labor/Management Relations was mandated to seek policies or institutions that “enhance work-place productivity” not policies that would improve democracy at workplaces, and organized under the direction of the Secretaries of Labor and Commerce, rather than being set up as a Presidential Commission. Even if the Administration had tried to deliver more for unions, however, conservatives in the House and Senate would have blocked meaningful reforms, as they did in 1978, when unions had much greater voting power. Both business and labor can block almost any labor reform, so that barring some remarkable confluence of interest or deal, the only changes that can be made are “least common denominator” changes.

    Not Dead Yet: Unions Grow In Spurts

    Then Pat MaloneFergot that he was deadHe raised himself an’ shouted from th bed

    (The Irish Wake,

    Density down. Organizing strategy seemingly incapable of turning the tide. Share of voters stable, but likely to fall as density falls. Political strategy seemingly incapable of turning the tide. Consensus projections of further decline. For friends of labor, perhaps it would seem time to start the keening.

    Projections of the death of unions have gone awry in the past, and they could very well prove wrong again. In 1933, right before the Depression era growth of unionism, the President of the American Economic Association, George Barnett, declared that unions had no future, much as many analysts of the US scene do today. In the 1950s, before unionism came to the public sector, the President of the AFL-CIO, George Meany, declared public sector workers non-organizable in the US, despite unionization of the public sector in other countries.

    The reason analysts have failed to forecast union grwoth is that unions in the US and elsewhere rarely grow at modest rates under normal organizing conditions. Rather, they grow in sudden sharp “spurts” in periods of social crisis (Freeman, 1998). Union density grew in nearly all advanced countries during World War I when governments needed labor cooperation to prosecute the war. Density grew in advanced countries in the aftermath of the Great Depression, when workers lost faith in business leadership and the market system. Density grew in World War II in advanced countries that were not conquered by the Nazis. During the 1970s oil crisis inflation, density grew in most EU countries and in the US public sector. Projections and analyses based on extrapolation of the recent past inevitably miss these spurts.

    What can we learn from these spurts about the possible future growth of unions? The configurations of social and economic factors associated with past union spurts does not produce any easy generalization or prediction. Union growth during World Wars I and II occurred during a tight wartime labor market that shifted the balance of power from business to workers and that induced governments to help unions to avoid disruption in war-time production. Union growth during the Depression reflected an entirely different dynamic: loss of faith in the business leadership and economic desperation among workers. Union organizing efforts escalated as workers desperately sought to gain some control over their working lives, while management battled against unions as hard as it ever has. The 1970s oil shock crisis led to rapid growth of unions in many European countries, as workers sought protection against inflation through collective bargaining, but had no impact on unionization in the US. Perhaps US division over the Viet Nam War, which pitted many natural supporters of unionism against the AFL-CIO, diverted activists from organizing in response to the oil price-induced inflation.

    Difficult to predict though union growth spurts are, they have some identifiable characteristics that provide guides to the nature of any future spurt. Spurts in the US have been associated with changes in union form that attract previously non-organizable groups of workers or with institutional or legal changes that greatly weaken employer opposition. In the Depression Era spurt, the new form was the industrial union, which the CIO pushed and which the AFL eventually adopted as well. In the 1880s, the new form was the geographic lodges associated with the Knights of Labor. In the World War I and World War II period, the new institutional form or legal change was the formation of various war labor boards and compulsory arbitration to deal with disputes. The public sector spurt transformed existing associations, most notably the National Education Association, from nonunion groups into collective bargaining organizations. The associated legal change was the enactment of public sector collective bargaining laws that forced public sector employers to bargain collectively, usually pushed by the existing associations..

    What new form of union organization might provide a “spurt” in
    membership in the new information based economy?

    What institutional or legal change might reduce employer opposition and make
    unionization easier for workers who want unions at their workplace?

    Toward A New Union Form: Open Source Unionism

    In a series of articles, Joel Rogers and I have argued for a new form of unionism that provides services and organizes workers who cannot gain majority status at their workplace (Freeman and Rogers, 2002a, 2002b). The argument for non-collective bargaining or minority unionism has two parts. First is the presence of large numbers of workers who desire union representation at workplaces where unions cannot gain the requisite 50+% of votes in NLRB elections or collective bargaining contracts. Second is the development of modern computer technology, particularly the advent of the Internet, as a cost effective means of communicating with and delivering union services to workers outside of collective bargaining. Our analysis suggests that a new union form that combined worker desire for representation and information and communication over the Web in an innovative way could grow rapidly, albeit with little or no dues and a wider definition of membership. Since any union organization needs face human interactions to build the trust and solidarity which lie at the heart of any collective organization, we suggested that the new union form would need a strong local basis – be it through city labor councils or community organizations or some other geographic units that transcend particular workplaces.

    Our proposed new form would expand union membership from workers in a given unit that has a collective contract to include a diverse forms of membership and fees associated with membership. Someone who supports the union by providing it with contact detail such as email or who backs the union in a particular campaign – such as student or community activists – or who who receives information and advice through the Internet at low cost to the union and who pays nominal dues, or an activist who obtains a nonunion job or who votes for a union in a losing organizing drive would be part of an open source union. The worker and the union, not the employer, would decide who is organized and who is not organized.

    Defining union membership outside a majority collective bargaining situation has historical precedence in the US. In the 1880s the Knights of Labor established “mixed lodges” that included workers from many occupations and firms in the same geographic area. This form underlies the1880s spurt in unionization. The first constitution of the AFL, adopted at its founding in 1886, stated that membership was open to any “seven wage workers of good character and favorable to Trade Unions, and not members of any body affiliated with this Federation.” Many such groups received direct affiliation with the national federation, migrating later to one or another international union. Direct affiliation and membership in minority or members only unions was also important in the 1930s, when workers who did not fit well into established forms sought to join unions. In the early organizing of the industrial trades, of the mineworkers and steelworkers, such non-majority unions provided a union presence in the workplace well before an employer recognized a collective-bargaining unit. As noted, such groups were important in the 1960s and 1970s public-sector union spurt, which developed from minority associations that did not bargain collectively, often under meet and confer provisions that did not impose binding duties to bargain on employers. Over one million employees in the federal government have joined unions even though the federal government does not bargain over wages and working conditions. Union density was higher in 2002 among federal employees (32.3%) than among private sector workers (8.5%).

    An open source union would give its members services outside of collective bargaining, including helping them at their workplaces with much the same legal status as a majority status union.[8] The union would use the Internet to provide members with accurate and up-to-date information about economic conditions affecting them. At firm-based locals, the union would gather information from members as well as from the regular business and economic news services about the firm. At locals in a given occupation or industry, the union would gather information about wage and employment developments in the specified job markets. Since the union’s first product would be information, the union web site would have to provide utterly trustworthy information, and develop other features that would make it a valuable site for members. It could also give information about unionism in the world writ large, for instance through access to Going beyond provision of information, the open source union would provide personalized advice to members about economic problems: what workers might do if an employer mistreated them; how to get an employer to offer more options for defined contribution pension moneys; the range of pay in their area or occupation; and so on.

    Since open source union would not have the clout inside the firm of a majority-status collective bargaining union at particular workplaces, it would have to use different tactics to impact obstreperous employers. It could use its reputation for providing accurate information to set the agenda in disputes with management, to shape public discourse, and to marshal resources from outside a unit to support workers in disputes. When young law associates felt that they were getting an unfair deal from major law firms in 1998, they set up a web site that provided information to potential recruits to firms about salaries and working conditions. Fearful that complaints on the site could adversely affect future recruitment, the major law firms quickly improved conditions, raising pay massively in some cities to remedy one of the major problems – differential rates of pay between associates working in Silicon Valley and those working in other locales. In the space of one year, the law associates won larger gains in pay than almost any union has won through collective bargaining (Tara and Gesser, 2002). Since workers seeking employment often use the Internet to find out about a prospective employer, open source unions could provide the venue for current workers to inform potential new employees about working conditions and problems at the workplace. If existing workers were upset about a problem, it could harm recruitment, pressuring firms to resolve the issue. If workers felt management treated them fairly, this would help the firm in its recruitment

    The absence of bargaining strength inside the firm would induce open source union to forge meaningful alliances with community groups outside itself and engage in local political activity, much as public-sector unions did when they were governed by meet-and-confer legislation. Unlike unions based on collective bargaining open source unions would have to look outward in struggles with mangement. Open source unions would gain the political and social influence that comes from playing a broader public role, as a substitute for traditional sources of muscle in disagreements with management.

    Does all this sound “pie in the sky”? If I had offered this vision of a new union form when I was on the Commission on the Future of Labor/Management Relations in the early 1990s, the Chair of the Commission, John Dunlop, would undoubtedly have called for the white coats to cart me away. Labor relations is a highly practical world, which disdains blue sky speculations.

    But the pie is real in many places. Throughout the latter part of the 1990s through the early 2000s, innovative unions in the United States and countries with similar labor relations systems – the United Kingdom, Australia, and Canada — have developed cutting edge practices that exploit modern technology and enlarge the meaning of membership along the lines of open source unions. These include[9]:Alliance@IBM (, a chartered local of the Communication Workers of America (CWA) at the historically non-union (at least in the US) IBM; WashTech (, also associated with the CWA, the Australian IT Workers Alliance (;the National Writers Union (NWU,, which is part of the United Automobile Workers; the CWA ‘s WAGE (Workers at GE,, for workers not covered by collective bargaining at GE; the King County Labor Council in Seattle ( has developed a Web site that increasingly functions as a broader community means of communication around political programs, organizing, and other activities. The United Food and Commercial Workers describes uses its site, as a major organizing. The Hotel and Restaurant Employees Union site provides data on Sodexho-Marriot’s record in food safety and labor relations that will help gain community support in any organizing. Closer yet to the open source union form are the worker centers that have been set up in various cities to help temporary workers, particularly immigrants, with their workplace problems ( The centers provide information and assistance, make use of the Internet to inform workers about issues, but do not themselves negotiate collective contracts

    These examples are the tip of an iceberg of what innovative unions, union activists, and union-associated groups are doing in the 2000s. They show that some unions are moving in the direction of open source unionism, using the Internet to mobilize and organize workers and labor and broader communities and deliver union services that did not exist even a few years ago. While no single union has put all the pieces together to create the full-fledged new form, the picture of that form that I have laid out is grounded in reality.

    Helping Enron’s Non-union Workers

    The Enron scandal in the US in 2002 galvanized the AFL-CIO into a set of actions that fits well the open source union form, save in one critical respect. Top officials of the Federation and constituent union presidents were among the most outspoken supporters of Enron’s laid-off nonunion employees. They spent money and staff resources on behalf of these workers. Supported by the federation, the Harris county (Houston, Texas the headquarters of Enron) central labor council provided direct support to former Enron employees. The AFL-CIO paid for lawyers to fight for a larger severance package for the workers; organized a public campaign to support this action; and pressured other companies to remove from their boards members of the Enron board; and lobbied Congress to change its laws to protect employees with 401k moneys.

    Since neither the AFL-CIO nor any constituent union had a collective bargaining contract for almost all of the laid-off workers, the union campaign was based on providing information, galvanizing community action, and giving legal support. The federation provided a highly informative website on the Enron situation It called on its membership countrywide to support the Enron workers by leafleting creditor banks in cities throughout the country. Using an email database with over half a million persons, it encouraged activists to send some 55,000 faxes and phone calls to creditor committees and Enron executives. Its legal staff submitted petitions to the SEC to tighten its rules and investigate diverse aspects of the Enron disaster. In June 2002, former Enron employees and their court appointed Employee Committee reached an agreement with the firm that increased the amount of severance pay going to the workers by 32 million dollars amount. In addition, the AFL-CIO pressured and shamed other companies where the directors at Enron sat on the board to the extent that these men resigned directorships at other firms and institutions, including Harvard University. All this on behalf of nonunion workers.

    The critical factor separating the AFL-CIO effort from open source unionism is that the federation did not sign up the ex-Enron workers into any sort of non-collective bargaining organization, nor use the Enron rescue to try to organize any of the millions of other workers around the country troubled over the safety of their retirement moneys or jobs due to top management malfeasance or criminal behavior. The Federation could not enroll as union members the workers it supported because there is still no place for such workers in majority status collective bargaining unions! Open source unionism would say to these workers, “we have helped you, we will help you and workers like you in the future. Join our movement. There is a place (at nominal dues) for laid off workers and others without collective contracts in the house of labor”.

    The number of workers who might be attracted to a new form of unionism is uncertain. Surveys of worker opinions suggest the numbers could be large (Freeman and Rodgers, 2002), but survey responses may be inaccurate. Perhaps most workers who say they want some form of union representation or services would reject joining an explicit non-collective bargaining organization. Going beyond surveys, however, are millions of workers with strong pro-union sentiments that have led them to contact unions, to support unions during organizing drives, to vote union in NLRB elections. In addtion, there are millions of workers who had postive union experiences who switch to nonunion jobs as part of normal labor turnover. If only a small proportion of these groups responded to a new union form, the reach of trade unions would expand greatly.

    The 2000 National Election Study asked respondents to give a thermometer score to unions, rating them from 0 (cold and unfavorable) to 100 (hot and faborable). My tabulations of these data show that 80% of voters who rate unions in the 81-100 thermometer range and 73% of those who rate unions at 100 in the thermometer are non-members. The reason that the bulk of persons with pro-union attitudes are non-members is not that non-members have more favorable views of unions. Members have more favorable views than non-members, but there are so few members that the vast majority of persons with high thermometer scores for unions are non-members. Some of these pro-union non-members are in households where another person is a union member but 91% are in households where no one is a union member. These pro-union persons are more likely to be black, aged, less educated, with lower household income than the average non-union person, particularly likely to view themselves as average working class persons.[10] They are less likely to vote than union members, which makes them a potentially good target for get-out-the-vote drives.[11] They would seem to be ideal candidates for open source union activity. Unions should seek sign up these voters and others who organizers would normally turn away because of the low probability of being able to attain a majority status collective bargaining contract. The AFL-CIO or some affiliates ought to experiment with developing some way the ex-Enron employees and their peers throughout the US could attach themselves to the union movement – at minimal cost.

    State Labor Law And Policies

    The federal government regulates private sector labor relations through the NLRA and its successor legislation. State governments regulate state and local government labor relations. Unions and labor lawyers favorable to the purposes of the NLRA have traditionally preferred national labor legislation and NLRB administrative decision-making to state labor legislation and court interpretation of the Act. Given the history of judicial rulings on labor issues in the US, one can understand that proponents of collective bargaining would prefer power to reside with a dedicated national agency than with non-specialized state or federal courts.[12] While states rights and federalism are usually associated with conservative political thinking, the success of unions in the US public sector and of unions in Canada, where provinces have the dominant role in labor law, has convinced many analysts that unions might do better if states had greater leeway to adopt laws of their own choosing. Absent federal preemption states that do not look favorably on unions might pass laws making union organizing harder, but other states would likely provide stronger protection for workers seeking to unionize. The outcome would be wide diversity among states in laws regulating private sector unionization, as exists in the public sector. Since union membership is slight in the states likely to enact legislation unfavorable to unions, unions would have little to lose in those areas, whereas they could gain considerable membership from favorable legislation in “pro-union” states. Some union activists might worry about “low road” competition between areas with more favorable union regulations and those with less favorable regulations, but in an information economy states with better labor practices are likely to win out.

    Accepting this analysis, are there ways to weaken federal preemption of state labor law short of changing national labor legislation?

    Not readily. Since the Supreme Court decision in Garmon (1959), it has generally been held that states cannot regulate activity that the NLRA protects or prohibits. The Warren Supreme Court favored federal preemption over states’ rights to the extent that it ruled that the NLRB had exclusive jurisdiction over conducted regulated by the NLRA. In the early 1980s Wisconsin tested this interpretation of the law. The state directed its Department of Industry, Labor and Human Relations to list firms who violated the NLRA over a five year period and to bar those firms from state procurement. In 1986 the Supreme Court unanimously ruled that by prohibiting state purchases from repeat labor law violators, the state was acting as a regulator rather than as a purchaser and held that the NLRA preempted the Wisconsin statute (

    In 2000 California tried to get around this rule by enacting a law that prohibited employers from using funds received by the state to oppose union organizing (AB 1889). The state act stipulates that every contractor’s request for reimbursement from state funds must include a certification that the contractor is not billing for costs incurred to assist, promote or deter union organizing. Recipients of grants from the state are also prohibited from such use of state funds and employers cannot hold anti-union meetings on state property. The emphasis was on the state’s rights as a purchaser rather than as an entity seeking to enforce the NLRA. In 2002 New York enacted similar legislation, and in 2003 the Hawaii AFL-CIO Federation proposed a similar statute for Hawaii.

    Employers challenged the California law on the notion that it limited the legal rights of employers under the NLRA to speak against unionization and that the legislation by the NLRA. In 2002 the US District Court for California decided in the employers’ favor, interpreting employer free speech to include the right to use state contract moneys to fight unionization, even when state legislatures explicitly passed laws that said they do not want taxpayers moneys spent in that way. To a free market economist, this is Alice in Wonderland thinking, limiting what governments can do with their own moneys. Higher courts may find the reasoning behind the District Court decision less convincing. Much of the impetus for federal preemption came from liberal Supreme Court justices, influenced by pro-labor legal scholars such as Archibald Cox, who argued for federal preemption although there is no explicit Congressional statement in this regard (Gottesman(1990)). In employment law, notably discrimination, states have their own regulations and do much of the enforcement, though the Congress has passed strong anti-discrimination legislation. The 1993 unanimous decision by the Rhenquist Supreme Court that upheld the right of the state government to exclude non-union contractors in the Boston Harbor construction case (Building & Construction Trades Council v. Associated Builders & Contractors of Massachusetts, 507 U.S. 218 (1993)) suggests that the Court might favor more states rights and less federal preemption in the California case as well.

    If higher courts uphold the preemption doctrine, unions and their allies could still use state law to make it more difficult for employers to break labor laws with impunity. Employers who regularly violate NLRA are also likely to violate many other statutes in the labor area, where presumably states could make it difficult for the violators to win state contracts. In many cases of violations of NLRA, the offending firm is also likely to violate some aspect of employment law protecting the individual, where state regulatory bodies operate. Finally, unions and state regulators could improve the social climate for workers to form unions and negotiate with employers in diverse other ways,using state labor codes, along the lines that the California AFL-CIO has been pioneering (Rosenfeld, 2003).

    Conclusion: What Next For Labor Movement?

    Economists and others who believe that the Invisible Hand needs no assistance in creating desirable economic outcomes rejoice in the decline of unions. Absent this market impediment, the economy will perform better in their view. The facts tell us something quite different. Capitalism without unions will produce different results than capitalism with a vibrant union movement, but probably the only certainties are that: economic inequality will be greater, the national savings rate will fall (union pension funds are a major factor in national savings – Freeman, 1985), and the US will be slower in developing policies to help the disadvantaged and poor on the one hand and to protect consumers, workers, and shareholders from business crime and dishonesty on the other hand. The analysis in What Do Unions Do (1983) and succeeding work (Mishel and Voos, 1992) shows that the Invisible Hand needs an organized finger, or two or three. The US has a stake and union leaders and activists have a responsibility to find the road to union renascence.

    The analysis in this essay suggests that the road requires unions to innovate with new organizational forms and to make greater use of state and local laws and policies. But the open source form and the use of state and local laws and policies are unlikely to be sufficient to turn the tide in unionisation. They may be the necessary form for a new spurt in unionism, but there also has to be new substance as well. Unions cannot be reactive to market-based reforms, as they have been in the past two to three decades in the EU was well as in US. They need a realistic assessment of the modern economy, of what workers want in that economy, and a clear set of longterm goals – five or seven or some such number of goals for the future around which nonunion as well as union labor can coalesce. They have to win the intellectual battle for setting the economic agenda. In What Workers Want, Rogers and I tried to determine what workers wanted at their workplace, but we did not examine the broader issue of what kind of market based economy would best suit workers. In the modern economy workers have diverse roles as employees, shareholders, pension fund owners. They work at home using computers and the Internet as well as at workplaces. They compete with other workers around the world. Their needs and those of the economy they work in go beyond what collective contracts and legal regulations usually cover. It is easy for practical folk to deride the “vision thing”, but unions will need new vision as well as new form to make the contribution to capitalist economies that our economies sorely need.


    Diamond, Wayne and Richard B. Freeman “Will Unionism Prosper in Cyber-Space? The Promise of the Internet for Employee Organization,” National Bureau of Economic Research, working paper 8483 (

    Freeman Richard B. (and James Medoff) What Do Unions Do? (Basic Books, NY 1984)

    Freeman, Richard B. “Unions, Pensions, and Union Pension Funds.” Pensions,
    Labor, and Individual Choice, edited by David A. Wise. Chicago: University of
    Chicago Press, (1985), pp. 89-121.)

    Freeman, Richard B. “Spurts in Union Growth: Defining Moments and Social Processes”, Bordo, Goldin, and White (eds) The Defining Moment: the great depression and the American economy in the twentieth century Chicago, University of Chicago Press, 1998

    Freeman, Richard B. and Joel Rogers, What Workers Want (Ithaca:
    Cornell University Press, 1999).

    Freeman, Richard B. and Joel Rogers, A Proposal to American Labor (The
    Nation, June 24, 2002)

    Freeman, Richard B. and Joel Rogers, Open Source Unionism (WorkingUSA,
    spring 2002, )

    Gottesman Michael “Rethinking Labor Law Preemption” Yale Journal on Regulation

    Summer 1990, 7 Yale J on Reg 355

    Guyer, Linda, ““Real World Experiences of Online Organizing,”” Unions and the Internet Conference, London, May 11––12, 2001 (

    Hartford, Bruce , ““Riding the Internet,”” presented at Unions and the Internet Conference, London, May 21––22, 2001 (

    Harvard Trade Union Program, the London School of Economics, and the British Trade Unions Congress, on “Unions and the Internet,” available at

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

    Mishel, Lawrence and Paula Voos Unions and Economic Competitiveness Economic Policy Institute, 1992)

    Morris, Charles Democracy at Work: Restoring the Right of Association in the American Workplace (Ithaca: Cornell University Press, forthcoming).

    Rosenfeld, David Using California’s Labor Laws Offensively: Organizing through enforcement of state employment laws 2003

    Summers, Clyde , “Unions Without Majority——A Black Hole?” Chicago-Kent Law Review 66

    (1990): 531––48.

    Taras, Daphne and A. Gesser How New Lawyers Use E-Voice to Drive Firm Compensation:

    The “Greedy Associates” PhenomenonJournal of Labor Research, 2002

    Trades Union Congress (TUC) s Promoting Trade Unionism Task Group, “Reaching the missing millions,” Trade Union Congress, College Hill Press, August 2001.

    Voos, Paula “Union Organizing: Costs and Benefits“ Industrial and Labor Relations Review vol 36, no 4, July 1983, PP 576-591

    Speech by Richard B. Freeman, Harvard University and NBER, Centre for Economic Performance, LSE

    [1] For 1990 see

    State%20Union%20Membership%20Density%201964-2001.xls. For 2002, see


    [3] We tabulated all of the reported numbers organized for 2001 from

    [4] Voos’ estimates vary depending on her model. Average costs were $729 in 1967 dollars. Her marginal cost figures range from $176 to $579 in 1967 dollars. This translates into a range in 2001 dollars of $933 to $3100. Taking the midpoint and rounding, I use $2000 as my estimate of the marginal cost of gaining a new member with a collective bargaining contract.

    [5] Organizing at this scale would likely change the cost rules, but the direction of that is unclear. On the one hand, economies of scale could reduce per new- member costs . On the other, the likely anti-union management campaign that such an effort would stimulate might easily raise the per capita costs. In any case, the money needed vastly exceeds what is currently being spent or that unions as a whole show any intention of spending.

    [6] Much of this section summarizes Richard Freeman. “What do Unions Do … to Voting?” Paper presented at Russell Sage, March 2003.

    [7] Voter News Service General Election Exit Polls, ICPSR 2780, 1988 VNS Exit Poll Methodology, p 8,

    [8] US labor law protects all workers engaged in concerted action. Labor law scholars are uncertain about the legal requirement for employers to meet with a minority union.

    [9] For detailed discussion see Richard B. Freeman and Wayne J. Diamond, “Will Unionism Prosper in Cyber-Space? The Promise of the Internet for Employee Organization,” National Bureau of Economic Research, working paper 8483 ( See also reports on and proceedings from the May 2001 conference, jointly sponsored by the Harvard Trade Union Program, the London School of Economics, and the British Trade Unions Congress, on ““Unions and the Internet,”” available at

    [10] Persons 60 and over have a modestly higher chance of giving unions a high thermometer score (81-100); blacks are more than twice as likely to give unions a high thermometer score, 58% of those giving unions a thermometer score are high school graduates are less, who make up just 37% of the NES sample. Similarly 49% of those giving unions a high thermometer rate themselves as working class while they make up 35% of the sample. As for nonunion persons in a union household, 9.2% give unions a high thermometer, whereas they make up just 6.2% of the sample.

    [11] Nonunion persons who give unions an 81-100 thermometer score have a voting rate of 69% compared to 72% for all nonunion persons who answered the thermometer question and 84% of union persons who answered the question. These high rates show that upward bias in the NES survey.

    [12] Courts made numerous decisions throughout US history against efforts to protect labor and unions, often declaring laws unconstitutional under the due process clause of the 14th Amendment.