VI. Assessing the Efficacy of “Union Organizing” Strategies in Union Revitalization Projects in Australia, Britain, Canada, and the United States

Union Organizing and Union Revitalization in the United States

Jack Fiorito
Florida State University and University of Hertfordshire
Paul Jarley
University of Nevada Las Vegas

Abstract

This paper examines evidence on union organizing activity for 1990–95 and 1999–2004. Despite increased organizing rhetoric and prominence during this time, the data show little change in aggregate organizing activity levels. Disaggregation to the national union level, however, reveals considerable variation. Many national unions increased both organizing activity and organizing success. A concluding section calls for more attention to organizational and environmental differences among national unions, and renewed efforts to model union organizing efforts.

Union organizing via National Labor Relations Board (NLRB) certification elections dropped sharply in 2006 to 1,648 elections, down from 2,142 in 2005 (Bureau of National Affairs 2007). This is one of the steepest drops in a decline dating to the early 1980s, when unions participated in over 6,000 elections annually (Chaison and Dhavale 1990). When Congress created the NLRB in 1935, elections were intended to be the primary means by which private-sector workers could decide on union representation. Explanations for the decline in elections are especially needed in that union leaders, notably AFL-CIO President John Sweeney, have publicly championed intentions to boost organizing for more than ten years. Yet neither organizing nor membership trends clearly show a turnaround (also see U.S. Department of Labor 2007).

Beyond a bleak aggregate picture, however, it is hard to know the reasons behind what has happened. Unions’ membership reports are often suspiciously inconsistent across sources and time. Data on union organizing activity are scattered across at least three federal and numerous state agencies. These sources record votes but not membership. It is unclear how much organizing takes place outside “official channels” (for example, the NLRB). Self-reports from the AFL-CIO’s now-defunct Work in Progress suggest, however, that organizing via elections still represents a large share of organizing activity, roughly 85 percent.

Fuller understanding requires attention to the role of national unions, their decision processes, structures, and environments. Some assessments have recognized this (see, for example, Chaison 2007; Hurd 2007), but most analyses still rely heavily on impressionistic, indirect, and anecdotal evidence. More systematic analyses of national union organizing records may illuminate the nature and causes of continuing union decline. This paper systematically examines national union data from the late “Kirkland era” (1990–1995) and the first years of the “Sweeney era” (1999–2004). After establishing “the facts” as best we can, our attention turns to inferences on union organizing and revitalization.

Table 1 shows indicators of membership, membership growth, NLRB organizing win rates, NLRB elections and relative organizing effort, and leader views on organizing effectiveness, commitment, and expenditures during 1990–2004 for national unions with 50,000 or more members in 2004. Other than NLRB-based data, these data are from self-reports. Data of this kind may be self-serving, as suggested by an average organizing effectiveness score of 3.6 on a 5-point scale for 1990 and 2.5 on a 4-point scale for 1997 (figures differ slightly for Table 1’s large unions). Yet, self-reports may have some advantages over objective data in netting out environmental influences (for example, employer resistance, employment changes).

Membership data reveal diverse union experience and reflect many influences, including mergers, absorptions, “automatic” growth and decline through union security clauses and bargaining unit changes, as well as effects of organizing and decertification. Membership growth can reflect organizing, but it is by no means a clear indicator. In any case, on average unions increased their membership by 3.1 percent over the 1990–2004 period; however, membership declined during the Kirkland era by 5.3 percent and increased during the Sweeney era by 2.7 percent. Consolidation, rather than organizing, however, is the principal explanation for this “growth.” Overall membership declined from 16.7 million in 1990 to 15.5 million in 2004 (Hirsch and Macpherson 2007), but average national union size increased from 142.6 thousand to 172.9 thousand.

NLRB win rates are more precisely about organizing success but are limited to elections actually held. They do not reflect non-NLRB jurisdiction organizing, card checks, or differing propensities to pursue campaigns to elections (for example, based on authorization card signings). Win rates may connote strategies as well as effectiveness. Caveats noted, there are clearly NLRB win rate differences across unions and time. The SEIU nears the top of the list in both eras and improved its win rate noticeably, from 64.6 to 72.4 percent. The IBT shows a low win rate in both eras and improved its rate less than most unions, from 43.3 to 45.2 percent. The average win rate for the Kirkland era was 51 percent, versus 61 percent for the Sweeney era. Trade-offs between NLRB organizing activity levels and win rates are apparent in the SEIU vs. IBT comparison, and for unions generally in the correlation between activity level indicators and win rates for the Kirkland era (r = –0.43, p<0.01) but not for the Sweeney era (r = 0.06, NS).

Correlations among alternate organizing success indicators are generally modest or weak, for example, r = 0.08, NS, for Kirkland era win rate and membership growth, and r = 0.24, p<0.10 for the same indicators in the Sweeney era. This is disappointing but underscores differing information content and perhaps data limitations.

Although the NLRB win rate and membership growth rate improvements between the Kirkland and Sweeney eras suggest success in refocusing unions on organizing, the organizing self-rating on a same-unions basis dropped significantly. Also, organizing activity is arguably more relevant, and the evidence on it is mixed. The number of NLRB certification elections rose during Sweeney’s early years then seemingly returned to a long-term downtrend. The average number of elections per union dropped from about 260 in the Kirkland era to about 139 in the Sweeney era. These figures overstate the decline due to sample composition changes, but even a same-unions comparison shows that the average union participated in about 59 fewer elections in the latter period.

Comparing numbers of elections across unions is of limited value. Organizing requires resources, and unions vary greatly in size and resources. Consequently, emphasis here is on relative organizing activity, that is, numbers of workers “tried” via NLRB elections per thousand current members. Relative organizing activity rose from an average of 62.4 eligible voters tried per thousand members to 73.4, but it is virtually unchanged on a same-unions basis. For several seemingly “organizing-active” unions such as the IBT, UFCW, USWA, and LIUNA, relative organizing activity fell. On average, although win rates rose by five to six percentage points on a same-unions basis, relative organizing activity did not change significantly on that basis. Organizing budgets averaged only about 14.5 percent of total union spending (for a relatively small number of reporting unions, N = 44), in contrast to Sweeney’s 30 percent goal, and showed a mild upward trend over three years, 1996–98 (Fiorito, Jarley, and Delaney 2007). Here, too, there is considerable variation across unions, consistent with distinctions drawn between “organizing,” “organizing-driven,” or “organizing-active” unions (or similar terms) and other unions (see, for example, Bronfenbrenner and Hickey 2004). Relative variability in organizing activity is far greater than that in win rates. The coefficient of variation (CV) for NLRB elections is in the 600–900 percent vicinity, and even after adjusting for union size and taking into account size of units targeted for organizing, relative organizing activity’s CV ranges from 85 percent (Kirkland era) to 194 percent (Sweeney era), but for win rates the CV range is only from 24 percent (Kirkland era) to 28 percent.

Is there intra-union consistency over time, that is, evidence of stable types? Although there is little consistency over time in membership growth (r = 0.12, NS), this indicator is contaminated by mergers, absorptions, and such. Consistency over time for win rates (r = 0.58, p<0.01 ), relative organizing effort (r = 0.53, p<0.01), and self-rated organizing effectiveness (r = 0.41) do suggest this possibility. There has been some shuffling of the ranks. Some unions’ win rates changed by 10 percentage points, and some unions’ growth rates changed dramatically. Relative organizing effort rose or fell 30 percent or more for some nationals.

Obviously, there are many indicators one might consider, and unfortunately they sometimes diverge and some may involve trade-offs that complicate interpretation. NLRB win rates and relative organizing effort measures represent conceptually and practically important indicators in a potential trade-off relation. A union might be able to organize more extensively, but with lower success. Conversely, it may scale back its efforts but improve its success rate, that is, focus its efforts more intensively. If a union increased both its organizing effort and its win rate, that would seem unambiguously positive, and, in contrast, if its effort and win rate both fell, that would seem decidedly negative. Figure 1 plots changes in relative organizing effort and win rates for the 11 largest U.S. unions, those with 500,000 or more members in 2004. Most of these “giants” improved both their relative organizing effort and their win rates between the Kirkland and Sweeney eras. Only one had decreases in both, three improved win rates while diminishing organizing effort, and none expanded efforts with reduced success. The trade-off referenced earlier is not apparent in Figure 1; rather, win rates and effort improved together on average. This positive relation does not hold in a similar plot for “all” unions (all those for which data are available), but even among all unions, a plurality of unions fall in the first quadrant, coupling increased efforts and improved success.

These plots put a rather different light on changes in organizing efforts. Although the aggregate figures show improved win rates, they show diminished or unchanged efforts. In contrast, Figure 1’s data emphasize diversity among unions and suggest that many unions, particularly larger unions, have coupled increased efforts with greater success. How does one reconcile these seemingly inconsistent perspectives? One answer is “Blame the Teamsters.” As Table 1 shows, the Teamsters “drove” about 1,600 fewer elections in the Sweeney era than in the Kirkland era. Some other large unions also showed large decreases in organizing effort, and collectively these cases offset increases by many national unions.

The organizing record underscores that there are important differences among unions. We do not have a clear understanding of the observed differences or reliable models to aid understanding. There have been some notable attempts to describe, analyze, and interpret aggregate union organizing activity (for example, Farber and Western 2002), or to explain U.S.-Canadian differences (for example, Rose and Chaison 2001). These have highlighted the importance of the supply side for union growth and noted the comparatively low level, stagnation, and decline of union organizing efforts. Others have focused on why unions are not organizing more and whether and how organizational change within unions can boost organizing (for example, Hurd 2007). Key issues are why some unions seem far more committed to organizing than others, reasons for differences in organizing strategies, and how high commitment to organizing can be diffused. Much of this work is descriptive and impressionistic, but it offers insights toward a more systematic understanding.

There have been previous efforts to model differential union organizing efforts. Both Block (1980) and Voos (1987) were able to explain only small portions of inter-union variance in organizing activity. It may be that organizing efforts reflect many random or idiosyncratic influences such as the ideologies or strategies of particular union leaders. It is hard to interpret these modeling efforts as highly successful. Indeed, they offer “thin” conceptual models, low “explained” variance, and anomalous results and are based on data from the 1960s and 1970s.

More recent and mainly qualitative assessments have been offered by several scholars. Bronfenbrenner (2001) stressed a harsh environment, political distractions, union education cuts, polling excesses, and failure to undertake cultural change. Farber and Western (2002) noted that “[E]xplanation is difficult to come by” (398) but suggested that key factors might include a harsh political or economic environment and differential union- and non-union-sector growth. Hurd (2004) suggested that contributing influences include excess emphasis on the “organizing model,” dysfunctional and jealously guarded union autonomy, excess focus on numeric goals, and too little focus on “spirit and purpose.”

One can quibble about the adequacy of NLRB data and other indicators of union organizing activity, growth, and vitality. Despite data limitations, however, it is hard to escape the view that aggregate U.S. union organizing activity and union vitality continue long-term downtrends. There are, however, encouraging exceptions to this gloomy generalization. Many national unions have increased organizing efforts and success, no small accomplishment in the political economy of the early twenty-first century.

Despite growing recognition of and rhetoric about the importance of organizing, modeling efforts designed to establish fuller understanding seem to have been largely abandoned. This results in a “doubly grim vineyard” in addressing a somewhat necrotic topic with limited explanatory success. Better data would help. Trite as this may sound, it is important. Consider that when Sweeney called for unions to boost organizing budgets, no one could offer a solid estimate of a baseline of organizing expenditures. It is hard to assess progress when we do not know the starting point. Broader conceptualizations are also needed. Amidst the focus on quantitative organizing goals and the confusion that has often accompanied “organizing model” rhetoric, means and ends have often been confounded. Organizing is a means for improving worker well-being, not an end in itself. Union density may be the single best indicator of union capacity to improve worker well-being, but it has limitations (Sullivan 2007) and is not synonymous with vitality. Broader conceptualizations are needed as well in terms of union environments. It is said that a national union leader vowed to fire any organizer who organized a workplace in the union’s (declining) traditional jurisdiction. Is this refocusing gone berserk or a rational response to the futility of organizing workplaces almost certain to disappear in the face of daunting global competition? Similarly, union decisions to organize or not have to be considered within specific worker and employer attitudinal climates. We cannot gain a solid understanding when we simply compare Union A and Union B, or even Federation A and Federation C, without considering the differing environments they face.

Finally, as Hurd (2007) suggests, unions and those who wish to understand them must give more attention to “spirit and purpose.” This alludes to ongoing debates about social movement unionism and business unionism, or, more broadly, what type of unionism, and how does it resonate with worker representation desires? A “Great Myth” of organizing in the United Kingdom and United States is that millions are ready to join unions and just need to be asked. Polls do indeed show that millions say they would vote for a union in a hypothetical election, or possibly even join one—but again, this is in the abstract. Even though unions enjoy their support to a degree, for many workers, unions are still a long way from providing something that they want to be part of badly enough to commit their votes in real campaigns, much less their dues and their continuing activism. National and local unions are experimenting, as they should. Both success and failure can be instructive. In the long term, it will be unions that create an internal climate that fosters innovation and accepts occasional failures that are likely to find the revitalization formula that works for their circumstances. Scholars would do well to study these experiments, not only in careful case studies and impressionist assessments but also in broader attempts to model union processes and outcomes.

Court Gifford of the Bureau of National Affairs and various staff members at the U.S. Department of Labor and the National Labor Relations Board provided assistance in obtaining data. Vickie Coleman Gallagher and M. Todd Royle provided research assistance.