WHAT COULD EARLY UNIONS DO FOR WORKERS? THE CASE OF ILLINOIS COAL MINING IN THE 1880S William M. Boala Drake University May 2016 ABSTRACT: This study asks what unions did for Illinois coal miners in the 1880s. It measures the outcomes provided by a traditional union and the Knights of Labor, primarily via differencein-differences applied to a panel of coal mining counties in Illinois. Neither the traditional union nor the Knights of Labor was able to raise wages, provide benefits, or reduce hours per workday, but the two unions may have been able to produce other outcomes valued by workers. Circumstantial evidence suggests the unions were able to ensure prompt payment of wages due. JEL CODES: J51, L71, N31, N51. KEYWORDS: coal industry; early unions; Knights of Labor. ACKNOWLEDGEMENTS: This draft has benefited from comments by Peter Orazem, Jeff Owen, and John Pencavel, and by participants in seminars at Drake University and Iowa State University. a Address: College of Business and Public Administration, Drake University, 2507 University Avenue, Des Moines, IA 50311. Email: [email protected] . 1 WHAT COULD EARLY UNIONS DO FOR WORKERS? THE CASE OF ILLINOIS COAL MINING IN THE 1880S 1. Introduction What outcomes can unions deliver for workers? The most frequent answer given by economists is higher wages and other compensation.1 The importance of wages to economists is demonstrated in both empirical and theoretical studies of trade unionism. A large2 purely empirical literature attempts to measure the effect of unions on wages, and recent estimates usually show an average wage gap of 10 to 20 percent, though with considerable variation across industries and occupations.3 A smaller literature demonstrates that union members receive more non-wage compensation as well.4 A theoretical literature modeling trade union behavior always includes wages as an argument of the union’s objective function.5 What other answers are given? A second possible answer is shorter hours and fewer days of work. Unions often negotiate for reductions in hours or days of work per year.6 A third possible answer is better working conditions—for example, more agreeable and safer surroundings. A fourth possible answer is better treatment by employers and protection from arbitrary actions by supervisors.7 To what extent were unions in the early labor movement able to deliver these outcomes? Historians of the labor movement have emphasized the struggle by unions both for immediate material gain—“less work and more pay”8—and for long-term political goals. Some have acknowledged that individual workers were sometimes just as interested in “friendly society” benefits provided by unions like sickness insurance and funeral funds.9 Yet little quantitative evidence has emerged on what outcomes these early unions were actually able to provide. 1 “The historical record seems to suggest that American workers form and join unions largely to win higher wages and shorter hours,” according to Rees (1977, p. 25). “The standard view of trade unions is that they are organizations whose purpose is to improve the material welfare of members, principally by raising wages above the competitive wage level,” according to Booth (1995, p. 7). 2 Other commonly applied adjectives include “numerous” (Hirsch and Addison 1986, p. 116), “large” (Lewis 1986, p. 1), “voluminous” (Pencavel 1991, p. 11), and “enormous” (Booth 1995, p. 7). In any event, “everyone ‘knows’ that unions raise wages” (Freeman and Medoff 1984, p. 43). 3 See Blanchflower and Bryson (2007), Lewis (1986), and Pencavel (1991). 4 See Budd 2007 for a survey. 5 Booth 1995, pp. 87-94. Hirsch and Addison 1986, pp. 10-18. Pencavel 1991, pp. 54-94. 6 Earle and Pencavel (1990), comparing union and nonunion sectors of the U.S. economy, found that over time, unionism tended to reduce full-time hours of work, but with considerable variation across occupations and industries. 7 Slichter (1941, p. 2) emphasized “the desire of the workers for protection against the arbitrary and uncontrolled discretion of management..” An often-cited study by Seidman, London, and Karsh (1951) found that this reason for joining the union was cited by 39 percent of local union leaders, 21 percent of active members, and 15 percent of rank-and-file inactive members. 8 Slogan of 25,000 marchers at parade organized by the International Labor Union, May 8, 1878 during the textile workers’ strike at Fall River (Foner 1962, vol. 1, p. 504). 9 Webb and Webb (1920, p. 158), Booth (1995, p. 71), Boyer (1988), Hatton, Boyer and Bailey (1994, pp. 442-444). 2 This study investigates what early unions could deliver for coal miners in Illinois in the 1880s. It takes advantage of surprisingly detailed data available for that industry, state, and time period, collected mostly by the then-new Illinois Bureau of Labor Statistics (hereafter, IBLS). An interesting feature of these data is that unionism spread rapidly but unevenly during this period, which facilitates plausibly causal estimates via difference-in-differences. Another interesting feature is that two unions coexisted in the same industry—a traditional coal miners’ union and the Knights of Labor—and their effects on coal mining can often be measured separately. I find little evidence that either union was able to raise wages, but some evidence that the unions were able to deliver other outcomes that would appeal to workers. In particular, I find circumstantial evidence that unions were able to assure that coal miners were paid the wages they were promised—an important benefit of unions not reported elsewhere, to my knowledge. 2. Historical setting Growth of Illinois coal mining. Coal was mined in Illinois as early as 1833. Output grew to just over 6 million tons in 1880, eight percent of the U.S. total (U.S. Geological Survey 1902, pp. 308-313). Output continued to grow rapidly in the early 1880s, then slumped in the middle of the decade before resuming growth toward the end of the decade (see figure 1). The slump in output was undoubtedly related to the long national recession from March 1882 to May 188510 and produced a large drop in coal prices and a smaller drop in wages (see figure 2). Most Illinois coal was mined underground in shaft mines. Workers were lowered down a shaft to reach the coal seam, often hundreds of feet below the surface, where they moved horizontally to work the coal seam. Most mines extracted coal using the room-and-pillar method but some used the longwall method. Workers at the coal face (“miners” in the narrow sense) were paid on piece, per ton of coal loaded, because of the impossibility of close supervision. Other employees laid track for coal cars, attended the hoisting machinery, and did maintenance work. These workers, a minority of the mine work force, were paid a daily wage. Machines for cutting coal were introduced at a few mines in this period with mixed success. By 1888, 29 mines in Illinois were using machines, employing about 3000 machine workers who were mostly paid a daily wage (IBLS 1888, pp. 337-342). Coal mining accidents were frequent by modern standards, killing about one worker in 500 or 600 in Illinois in most years. About 60 percent of fatalities were caused by falling coal or rock, which killed one or two workers at a time (IBLS 1894, pp. LI-LIII). Occasional disasters killed many workers simultaneously. Illinois suffered two disasters during this period, both in 1883: 10 NBER Business Cycle Dating Committee, http://nber.org/cycles.html , accessed August 2012. 3 ten miners were killed in an explosion at Coulterville in Randolph County, and 69 workers were killed in a sudden flood at Braidwood in Will County.11 A law providing for mine safety inspections was passed by the Illinois legislature in 1872 and amended in 1877. It required each county board to appoint and pay a mine inspector. However, inspection was lax, according to the IBLS, because county inspectors were underpaid, usually worked only part-time or part-year, and were easily influenced by coal operators (IBLS 1881, pp. 221-222). Smaller counties sometimes left the position vacant (IBLS 1882, pp. 107-111). To improve the effectiveness of mine inspections, the legislature created an Office of the State Inspector of Mines in June 1883. The state was divided into five districts, each with a full-time mine inspector employed by the state (IBLS 1884, pp. 418-21). In addition to checking mines for safety violations, the inspectors collected a great deal of data, some of which will be used in the analysis below. Early unionism. Strike activity in Illinois coal mining was reported as early as 1861. In that year, a series of strikes near the city of LaSalle, in northern Illinois, prompted the state legislature to pass the so-called “LaSalle Black Laws,” rendering it a crime to influence employees through threat or intimidation to cease work. According to the IBLS, these laws made it “virtually a crime to belong to a trades union,” but were rarely enforced after 1865 (IBLS 1881, pp. 8, 230; David 1936, pp. 43-44). The earliest record of formal unionism in Illinois coal is also from 1861, when miners from Illinois and Missouri formed the American Miners’ Association (McBride 1887, pp. 245-246; Roy 1907, pp. 60-67). This organization lasted until 1868. Later, in 1871, a convention at Bloomington organized the Illinois Miners’ Benevolent and Protective Association to provide benefits to disabled miners, but apparently it did not last long (McBride 1887, p. 248). The Miners’ National Association, organized at Youngstown, Ohio in 1873, attracted some members from Illinois and over 20,000 members nationwide, but after costly strikes and legal battles defending its leaders from prosecution, it collapsed in debt in 1876.12 Union activity in Illinois coal mining apparently remained very weak at least until the early 1880s. Some local coal unions were organized in Illinois in the spring of 1877, but did not survive long (Evans 1920, vol. 1 p. 85). In 1881, the IBLS observed that, for the Illinois union movement as a whole, “there seems to have been little progress made in the past five years” (1881, p. 229). Occasional local bargaining by informal organizations of miners was reported in 1881 and 1882 (IBLS 1882, pp. 79-81; IBLS 1883, p. 76) as well as frequent strikes (IBLS 1883, p. 79; IBLS 1888, pp. 246-313). Nevertheless, as late as 1884, the IBLS noted that formal unionism had since 1877 been “practically inoperative” in Belleville, a major coal mining center 11 The Coulterville disaster is described briefly in IBLS 1883, p. 89. The Braidwood disaster is described in detail in IBLS 1883, pp. 97-109, and Roy 1907 pp. 190-194. 12 On the Miners’ National Association, see Evans 1920, vol. 1, pp. 46-85; McBride 1886, pp. 249-251; Roy 1907, pp. 169-176; Suffern 1926, pp. 25-31. 4 near St. Louis, and that several unions had been started “with only partial success” in Streator, a major center in northern Illinois (1884, pp. 433-436). Meanwhile, a different type of labor organization was achieving some success. The Knights of Labor, founded by Philadelphia garment workers in 1869 (Wright et al. 1887, p. 398; Perlman 1918, p. 197), differed from traditional trade unions in several respects. First, the Knights operated as a secret organization, to shield members from employer retaliation. Second, the Knights admitted members from all occupations. At the local level, the Knights were organized into both “Mixed Local Assemblies” of multiple occupations, and (beginning in 1882) “Trades Assemblies” of particular occupations.13 Third, the Knights officially favored arbitration over strikes. Fourth, the Knights sought legislation and government regulation to improve wages and working conditions for all workers, not just their own members. In practice, the differences were less substantial. Like traditional trade unions, the Knights were in fact primarily composed of skilled workers and did occasionally use strikes as a weapon.14 Like the Knights, traditional trade unions often tried to settle disputes without strikes and also sought protective labor legislation.15 As the Knights were beginning to grow, a new, more effective traditional trade union appeared in Illinois coal mining. The Miners and Mine Laborers Benevolent Protective Association (hereafter, Miners Association) was organized in the late 1870s (IBLS 1886, p. 167). Its representatives participated in an interstate convention at Pittsburgh in May 1883 and in a second convention at Indianapolis in September 1885. The latter convention organized the National Federation of Miners and Mine Laborers, a loose confederation of state miners’ unions. Yet the Miners Association was apparently still weak then. Its president, Daniel McLaughlin, told the Indianapolis convention that he had only a small number of members, confined to northern Illinois.16 Rapid growth of unionism in 1885-1886. The Miners Association began an “energetic” organizing drive in 1885 which resulted in remarkable growth (Evans 1920, vol. 1, p. 145; IBLS 1886, p. 167). A survey by the IBLS in 1886 discovered that most of the Miners Association’s lodges, or locals, had been founded in late 1885 or early 1886 (see figure 3). From early 1885 to 13 Trades assemblies were encouraged by the Knights General Assembly of 1882 (Foner, 1962, vol. 1, p. 512). An analysis of occupations of Knights by the IBLS classified 70 percent as skilled, 26 percent as unskilled, and 4 percent as “non-producers.” (1886, p. 169). The same IBLS report listed 15 coal strikes led by traditional trade unions from 1883 to 1886 (1886, table XXXIX, p. 405) and 33 coal strikes led by the Knights of Labor during the same period (IBLS 1886, table XL, p. 409). With respect to strikes and other concerns, “local and district assemblies [of the Knights] were virtually autonomous and did as they pleased” according to Foner (1962, vol. 1, p. 508). 15 See Evans (1920, vol. 1, pp. 87-88), IBLS (1886, pp. 159-163), and Perlman (1918, pp. 335-348). 16 Evans 1920, vol. 1, pp. 108, 137-138; McBride 1887, pp. 253-255; Perlman 1918, p. 425; Roy 1907, pp. 227-228. In view of the McLaughlin’s statement, it is curious that the IBLS was able to find quite a few unionized coal miners. In a survey of consumer expenditures of working families conducted in early 1884, the IBLS found that of 232 coal miners in the sample, 73 (or 31 percent) paid dues to unspecified labor organizations (IBLS 1884, table IX, pp. 328-329). However, it seems likely the sample was not random. 14 5 July 1886, membership rose from a few hundred to over 7000 (see figure 4). A triumphant state convention in February 1886 approved the incorporation of the Miners Association under state law and created district organizations within the state (Evans 1920, vol. 1, pp. 198-199). Meanwhile, the Knights of Labor as a whole were also growing rapidly. Overall national membership grew with “unprecedented rapidity” in 1885-1886 (Wright et al. 1887, p. 423), gaining 600,000 new members according to Foner (1962, vol. 1, p. 509). Figure 5 shows a huge surge in national organizing: at the end of 1887, about two thirds of the Knights’ Local Assemblies had been organized in just the previous three years. Similarly in Illinois, the IBLS reckoned that at the close of 1886, “the present number [of Local Assemblies] is two or three times as large” as in July 1884.17 Membership appears to have grown at a similar rate (see figure 6). However, these numbers include all trades, not just coal miners. The timing of coal miners’ entry into the Knights of Labor is somewhat less clear. Garlock (1982) counted a final total of 126 Local Assemblies in Illinois that were dominated by coal miners (61 Trades Assemblies, 20 Mixed Assemblies, and 45 Assemblies of unknown status).18 Garlock’s data, displayed in figure 7, reveal rapid organization of Illinois coal miners in 1886, but not the huge surge seen in all trades in figures 5 and 6. Nevertheless, in May 1886 the Knights created a special sub-organization devoted to coal miners, National Trade Assembly No. 135, no doubt partly in response to increasing coal miner membership.19 So coal miners’ membership in the Knights of Labor probably grew roughly simultaneously with membership in the Miners Association, but perhaps not as abruptly. The exact timing is difficult to pin down from available sources. Later developments in unionism. The Miners Association and sometimes the Knights participated in joint wage negotiations with fellow unionists and coal operators from Pennsylvania, Ohio, and Indiana, which were initially successful. The Miners Association sent representatives to joint wage negotiations at Chicago in October 1885, at Pittsburgh in December 1885, at Columbus in February 1886, and again at Columbus in February 1887. At first, coal operators from northern Illinois participated actively in these negotiations. Later, they complained that fellow operators in southern Illinois, where the Miners Association was weaker and the Knights were stronger, were not complying with the negotiated wage scale. Finally, in 17 IBLS 1886, p. 192. According to an IBLS survey, of 204 local assemblies of the Knights of Labor in Illinois in late 1886, 144 local assemblies (or 71 percent) had formed in 1885 or 1886. Of 29,034 employed members of the Knights of Labor in all occupations in late 1886, 19,371 (or 67 percent) were members of Local Assemblies that had formed in 1885 or 1886 (IBLS 1886, table V “Recapitulation,” p. 210). 18 By contrast, the IBLS counted only seven Knights trades assemblies composed strictly of coal miners: the first was organized in 1883, the second in 1884, three more in 1885, and two more in early 1886 (IBLS 1886, table V, pp. 203-210). The disparity between Garlock’s count and the IBLS count might be partly explained by the latter’s exclusion of mixed-occupation local assemblies dominated by miners. 19 Evans 1920, vol. 1, p. 205; Perlman 1918, p. 425; Roy 1907, pp. 243-244. According to Perlman, National Trade Assembly No. 135 was created partly in response to the organization of the rival National Federation of Miners and Mine Laborers the previous year. 6 early 1888, operators from northern Illinois withdrew from joint wage negotiations altogether, claiming that they could not compete against southern Illinois operators paying a lower wage.20 A year later, in spring 1889, coal operators in northern Illinois unilaterally reduced wages 10 cents per ton below the union scale, prompting the Miners Association to strike beginning May 1. The strike continued into autumn, when the miners and operators finally compromised on a reduction of 7 ½ cents per ton.21 It seems likely that the union lost membership and influence during the long strike (Evans 1920, vol. 1, pp. 481-482; Roy 1907, p. 252) and the Knights of Labor may have lost membership as well (Evans 1920, vol. 1, p. 428). Yet these losses were probably not catastrophic. The wage reduction was reversed when coal demand increased at the end of 1889 (IBLS 1890, p. XX). Any membership slide was probably temporary: the number of Illinois miners later covered by the union’s so-called “Columbus Wage Scale” of 1894 was nearly the same as the total number of Association members reported in 1886 (IBLS 1894, appendix pp. 7, 20, 22, 24).22 But it is difficult to say how strong the two unions were after the long strike of 1889. At the end of the 1880s, mergers transformed coal unionism. In December 1888, the Miners Association and coal unions in other states abandoned their loose confederation and merged to form the National Progressive Union (Evans 1920, vol. 1, pp. 389-407; Roy 1907, p. 250). Efforts to merge with the Knights, after repeated failures, finally succeeded in January 1890, when the National Progressive Union merged with the Knights National Trade Assembly No. 135 to become the United Mine Workers of America.23 The analysis below will focus on the period ending June 1889, at the start of the long strike and six months before the merger of the Miners Association and the Knights. The rapid growth of both unions in 1885-1886 will be exploited to measure each union’s effects on Illinois coal mining. 3. County-level panel on Illinois coal mining, 1879-1889 What could these two unions—the Miners Association and the Knights of Labor—do for Illinois coal miners? The remainder of this study will seek to answer this question using quantitative evidence. One might not expect much quantitative evidence to be available from this period. 20 See Evans (1920, vol. 1, pp. 145-365), George (1898, p. 198), IBLS (1885, pp. XXIX-XXXII), IBLS (1889, pp. 114-117), McBride (1887, p. 256), Perlman (1918, p. 426), and Roy (1907,pp. 231-241). Interstate bargaining was not resumed until 1898 (Evans 1920, vol. 2, pp. 634-637; IBLS 1898, pp. 2-4). 21 See Evans (1920, pp. 464-482), IBLS (1889, pp. XVII-XXXI), IBLS (1890, pp. XIX-XX), Lloyd (1890; Suffern 1926, pp. 47-50). 22 Even the miners of Spring Valley Coal Company, in Bureau County, who endured a devastating six-month lockout in 1889 (Lloyd 1890; IBLS 1889, pp. XXVIII-XXX, 20-21, 109; U.S. Commissioner of Labor 1894, vol. 1, pp. 1282-1285), were able to mount a strike two years later (IBLS 1891, pp. 74-75, 319). 23 See Evans (1920, vol. 2, pp. 13-17), Perlman (1918, p. 487), and Roy (1907, pp. 249-259). 7 However, remarkably detailed data on Illinois’s coal mines were collected from operators by coal mine inspectors and the staff of the Illinois Bureau of Labor Statistics (IBLS) in the 1880s. Beginning in 1882, the IBLS published data annually on production, employment, wages, prices, and other coal data for up to 50 counties. I combined these data with less complete data for 1879 from county mine inspectors and for 1880 from the U.S. Census (1886). Descriptive statistics of the resulting county panel data set are shown in table 1, panel a, and details of its construction are given in appendix A. Union membership data were collected and reported by the IBLS only in 1886, but union membership in other years can be extrapolated with some confidence in view of historical events described in the previous section, as I now explain. Membership in Miners Association. In 1886, the IBLS conducted a survey of labor unions in all occupations in Illinois. A special section of the Fourth Biennial Report, entitled “Labor Organizations,” reported membership in each local union of the Miners Association (IBLS 1886, table I, pp. 172-178). Total membership was over 7000. For this study, employed union local membership was summed to the county level, where Miners Association density was computed as “members employed” divided by total coal mining employment in 1886. The results are summarized in table 1, panel b and displayed in figure 8. As the figure shows, the Miners Association was strongest in the northeast part of the state. Of 50 locals listed as affiliated with the Miners Association, all but two were organized in 1885 or early 1886. Those two older locals had only 107 charter members (IBLS 1886, table III, pp. 196-198). So while the Miners Association existed in Illinois as early as 1876, it seems fair to assume that the organization had little influence until its rapid growth in late 1885 and early 1886. For this study, therefore, union density for the Miners Association before July 1885 will be assumed to equal zero. Data for the year ending July 1886, the period of rapid union growth, will be excluded from the analysis below. Available historical information described in the previous section suggests that the Miners Association remained strong in the following years, until at least the summer of 1889. Miners Association density from July 1886 through July 1889 will therefore be assumed constant at its July 1886 level. Extrapolation beyond 1889 seems hazardous, however, first because the long strike beginning May 1889 may have reduced union membership and secondly because the merger of the Miners Association with the Knights in January 1890 may have attenuated any differences between the two unions.24 Membership in Knights of Labor. The same section of the Fourth Biennial Report also reported current membership in the Knights of Labor by county and by occupation (IBLS (1886), table 24 How much the merger changed the behavior of union locals is an open question. While the merger brought all locals under the same district and national organization, locals were allowed to continue some of their former practices. For example, locals were allowed to retain their old organizational structure, and to continue as secret or open organizations (Roy 1907, pp. 253-254). 8 XVIII, pp. 273-294). For this study, Knights members whose occupations were "coal miners," "mine laborers," or "miners" were summed to the county level. At the county level, Knights density was computed as coal-miner membership divided by total coal mining employment in 1886. The results are summarized in table 1, panel b and displayed in figure 9. As the figure shows, the Knights were scattered throughout the state but were strongest in the southern counties. They totaled over 3500, about half as numerous as Miners Association members. The dates when coal miners joined the Knights were not given by the IBLS or by Garlock (1982), but, as noted in the previous section, available information suggests that the Knights grew roughly simultaneously with the Miners Association. In the absence of more precise information on timing, Knights density in coal-mining will be assumed to equal zero before July 1885, while Knights density from July 1886 through July 1889 will therefore be assumed constant at its July 1886 level. Data for the year ending July 1886, a year of rapid growth for the Knights nationwide, will be excluded from the analysis. In short, the same timing assumptions will be made for the Knights as for the Miners Association, though with admittedly less confidence. The Fourth Biennial Report also reports the number of coal miners who were members of both unions. Subtracting this number from the total of Association and Knights membership, I computed overall union density in either union. The results are summarized in the bottom row of table 1, panel b. The county panel. The data set to be analyzed is thus an annual panel of counties. The panel includes six years of no unionism (1879, 1880, 1882, 1883, 1884, and 1885), and three years of substantial unionism (1887, 1888, and 1889) with Association and Knights density assumed to be constant at their July 1886 levels. The panel is not quite balanced because some counties are not represented every year—for example, only 16 counties are available for 1879. Also, some data are not available for every year—for example, wage data begin in 1883. Details are given in appendix A. The following sections estimate the effect of unionism on various outcomes by regressing those outcomes on county effects, time trends or year effects, and union density. The effects of the Miners Association and the Knights can in principle be separately identified because the two unions enjoyed different degrees of organizing success in different counties. Moreover, 17 counties (mostly small) never had miners in either union. So the research design is essentially “difference-in-differences,” with the qualification that union density varied across counties and never reached 100 per cent. Exogeneity of unionism. Econometric identification assumes that the surge in unionism in Illinois coal mining in 1885-1886 was not driven by the outcomes to be explained. For example, it assumes unionism was not caused by a downward (or upward) trend in wages at particular counties. Is this assumption reasonable? Certainly the timing of this surge was related to a 9 nationwide increase in labor agitation outside of coal mining. The year 1886 saw a sharp increase in strike activity throughout the United States and Illinois (figure 10) and a nationwide campaign for an eight-hour workday (IBLS 1886, pp. 473-474; Perlman 1918, pp. 375-386; Kemmerer and Wickersham 1950). Unionists like George McNeill (1887, pp. 170-171) were convinced that “the year 1886 [would] be known as the year of the great uprising of labor.” In Illinois, trade unions in many industries experienced rapid growth simultaneously with the coal miners. Of 229 Illinois trade unions counted by the IBLS in July 1886 (excluding coal miners) 96 had been organized in the preceding 18 months (IBLS 1886, table III, pp. 195-199). So the timing of the surge in unionism in Illinois coal was apparently driven largely by trends outside coal mining and even outside Illinois. The geographic distribution of the Miners Association surge may have been driven in part by proximity to Chicago. That city was a major center of union organizing: of 72,945 employed members of labor organizations in the state of Illinois in July 1886, two-thirds (49,052) resided in Cook County (IBLS 1886, table VII, p. 218). Strikes and demonstrations in Chicago in favor of the eight-hour day attracted hundreds of thousands of participants in spring 1886, despite the tragic events at Haymarket Square in May (IBLS 1886, pp. 479-492; Perlman 1921, pp. 384386). As shown in figure 8, Association density was especially high in coal-mining counties close to Chicago: Bureau (55 percent), Grundy (38 percent), LaSalle (42 percent), Vermilion (34 percent) and Will (56 percent).25 In fact, the correlation of 1886 Miners Association density with distance from Cook County is -0.3.26 This pattern again suggests that the Miners Association surge was driven largely by events outside coal mining. Figure 11 offers some visual evidence on the exogeneity of unionism by comparing wage trends in three groups of counties: one group that later enjoyed substantial Miners Association membership, a second group that enjoyed substantial Knights membership, and third group that experienced no unionism. Before unionization in 1986, all three groups appear to follow parallel trends: a slight increase in wages from 1883 to 1884, followed by decreases in 1885 and 1886. For wages, at least, there is no evidence that unionism was driven by the outcome to be explained. 4. Wages and benefits Coal miners would surely have valued increased wages and benefits. Did the Miners Association and the Knights provide them? Wages. Early writers including Lewis (1963, p. 76), Douglas (1930, pp. 142, 562), Ulman (1955, pp. 28-32), and Suffern (1926, pp. 43-52), noting low union density and weak interstate 25 For the state as a whole, Miners Association density was 19.4 percent. Distance between counties taken from NBER County Distance Database, created by Jean Roth from Census data, posted at http://www.nber.org/data/county-distance-database.html, and accessed March 30, 2016. 26 10 bargaining in coal mining in the late nineteenth century, conjectured that there was little union wage premium. Later writers provided empirical estimates confirming these conjectures using cross-sectional data on workers. 27 Dillon and Gang (1987, p. 523), using a national survey of workers from 1889-1890, found a slightly negative union relative wage effect for coal miners. Eichengreen (1987, p. 512), using a survey of Iowa workers 1894, found a substantially negative union effect on wages of “miners and quarrymen”.28 (Both studies found positive union wage effects for most other occupations.) In this section, I attempt to confirm or reject these findings. Table 2 shows ordinary least squares (OLS) regressions of the log average peak-season (winter) wage rate per ton of coal on union density, using the county panel. The piece rate varied with geological conditions, but county fixed effects should largely control for such conditions. The estimated coefficients of union density predict the percent increase in wages if, extrapolating hypothetically, union density in a county were to reach one hundred percent. At actually observed levels of union density, the coefficients measure the average “wage gain” to all workers from unionism, including any threat effects or spillover effects on nonunion workers in the same county. Column (i) shows estimates of the effect of the Miners Association and the Knights, using a time trend and its square to control for unobserved statewide time-varying effects such as supply and demand conditions. The estimated effects of both the Miners Association and the Knights are very small—about 2 percent and zero percent, respectively. However, the cluster-robust standard errors are large, especially for the Knights, so caution is advised. Remaining columns report alternative specifications for the wage equation. Column (ii) reports estimates that additionally control for the output price of coal (which should have a positive effect on labor demand) and the average size of coal mines in the county (which measures any employer size-wage effect). The coefficients of the controls have the expected signs, but the estimated effects of the Miners Association and the Knights remain essentially zero. Columns (iii) and (iv) replace the time trends with year dummies, and the estimated effects of the Miners 27 By contrast, Hatton, Boyer, and Bailey (1994) estimated a union relative wage effect of 12.7% for British coal miners in 1889-90 (p. 449), where union density was about 57 percent (p. 440). They speculate that newlyorganized unions may have had especially large effect on wages due to an “impact effect” (p. 451). For U.S. coal mining, studies have estimated positive union wage effects in the early twentieth century. For 1912-1923, Fishback (1992, pp. 207-209) estimated a union relative wage effect of about 8 percent, though his standard errors were more than half as large as the point estimates. For West Virginia coal mining, Boal and Pencavel (1994, p. 283) estimated union relative wage effects from 5 to 12 percent in 1897-1912, and from negative 1 percent to positive 8 percent in 1913-1920, depending on specification, though again standard errors were more than half the point estimates. 28 Eichengreen estimates separate regressions for union and nonunion workers, but union differentials for any occupation can easily be computed at the sample means. Inserting sample means for age, marital status, foreign birth, and skill learned abroad (from table 1, p. 505) into Eichengreen’s OLS estimates for these two groups of workers (table 5, p. 512, columns (ii) and (iii)) yields wages of unionized “miners and quarrymen” that were 0.27 log points below that of their nonunion counterparts. Inserting the same sample means into Eichengreen’s selectivity-corrected estimates (table 5, p. 512, columns (v) and (vi)) yields wages of unionized “miners and quarrymen” that were 0.09 log points below that of their nonunion counterparts. Standard errors cannot be computed from Eichengreen’s tables. 11 Association and the Knights are again very small. Columns (v) and (vi) measure the effects of overall union density. Again, the estimated effects of unionism are very small. Weighted least squares estimates (not shown) produce roughly similar results except that the point estimate of the Knights’ effect rises to about 0.10 and the standard errors of all the other coefficients increase. In short, the point estimates from the county panel data suggest no wage gain from unionism, though the standard errors are large, especially for the Knights. The estimated wage gain from unionism could be zero because spillover effects from unionism depressed nonunion wages while raising union wages. Or the wage gain could be zero because operators (the source of the panel data) for some reason did not acknowledge paying union workers more. To check these possibilities, I turned to an alternative source of wage data reported by unionists. As part of the 1886 survey of union locals described above, the IBLS asked what wage was received by members (IBLS 1886, table XVII, pp. 272-273). By comparing union–reported wages to the average wage reported by operators in the same county, I can measure the “wage gap” between union workers and the average wage in the same county. Figure 12 plots wages reported by 49 Miners Association locals against average wages reported by all operators in the same county in the same year (1886). Piece rate wages varied across and within counties for many reasons, including differences in difficulty of working coal seams, so it is not surprising that the figure shows considerable scattering around the 45-degree line. Nevertheless, the figure shows no tendency for union wages to be greater than county averages. Indeed, the mean ratio of union wage to county average is 0.995 (with standard error 0.023) and the median ratio is exactly 1. The estimated wage gap, like the wage gain, is essentially zero. Again, the evidence indicates the Miners Association had little or no effect on wages. 29 Benefits. Few coal operators provided benefits in this period,30 but some unions maintained their own accident, sickness, or funeral funds (Saposs 1918 , p. 124-125). In Britain, for example, 77 percent of union members were eligible for funeral benefits and 48 percent for sickness benefits (Boyer 1988, p. 320). The Webbs (1920, pp. 157-158) noted that such funds could attract new members who might not see other advantages of unionism. As part of its 1886 survey of union locals and Knights assemblies, the IBLS asked whether any such funds were maintained. Of 45 Association locals responding, only 8 (that is, 18 percent) had such funds. Of 182 Knights 29 Another piece of evidence on the union relative wage effect is available but probably less useful. The IBLS surveyed workers in all trades in 1884, before the statewide surge in unionism. That survey included 232 coal miners (IBLS 1884, table IX, p. 328). The 73 coal miners who were "connected with trades unions" enjoyed average annual earnings of $400 while the 159 coal miners "not connected with trades unions" enjoyed average annual earnings of $379. This amounts to a 5.5% earnings gap—greater than zero but less than estimates for other occupations by Eichengreen (1987), Dillon and Gang (1987), and Mullin (1998). The usefulness of the IBLS estimate is limited because there were no controls for geographic location or days of work per year. Not even the identity of the union or unions was reported. 30 Later, in the early 1890s, a few coal companies arranged mandatory accident and life insurance policies for their employees (IBLS 1893, pp. 111-114). 12 assemblies responding, only 65 (that is, 36 percent) had such funds.31 So most members of the Miners Association and the Knights did not enjoy the benefits of such funds. In summary, the available evidence indicates that the Miners Association could not offer Illinois miners increased wages or union-provided benefits. The evidence for the Knights is less complete but not much more favorable. I turn now to other workplace characteristics that might have been improved by these unions. 5. Working time Coal miners would likely have valued shorter daily hours of work. Certainly, workers in many other industries were campaigning for shorter hours at this time.32 By contrast, however, they would likely have valued more workdays per year. Coal demand was seasonal, and varied inversely with temperature, because coal was used for heating but was difficult to store (IBLS 1887, pp. 10-11). Work was scarce and wages were often lower in the summer off-peak season. Because of the drop in demand during the summer, miners worked fewer weeks per year than most other workers. An 1884 IBLS survey of 2129 households found that workers in all occupations worked an average of 43.5 weeks per year, but coal miners in the same survey worked an average of only 36 weeks per year, a difference of nearly 20 percent. The typical miner, according to the IBLS, “is usually eager to work, and … it is the irregularity of his employment quite as much as the smallness of his pay which keeps him poor.”33 Did the Miners Association and the Knights provide shorter hours and more regular employment? Hours of work. The IBLS did not report average hours of work annually at the county level. However, the same 1884 IBLS survey mentioned above found that all coal miners worked an average of 9.25 hours per day before the surge in unionism. Two years later, a survey of Miners Association locals found that hours of work varied from 8 to 12, with the majority working 10 hours per day. The average (weighted by workers) was 9.8 hours per day. A survey of Knights Assemblies found that hours of work for coal miners varied from 7 to 14, with a majority also working 10 hours per day. The Knights’ average (again weighted by workers) was 9.6 hours per 31 Miners Association locals: IBLS (1886, table XLV, pp. 436-437). Knights assemblies: IBLS (1886, table XLVI, pp. 438-443). The Knights data include all trades, not just coal mining. Miners Association dues were correspondingly modest. For the 45 locals responding, average annual dues were $2.17 (median dues were even lower at $1.50) (IBLS 1886, table XLV, pp. 436-437.) These dues are well below the “3% or 4% of annual wages” cited by Mullin (1998) for unions in other occupations. 32 See Cahill (1932, pp. 137-163), IBLS (1886, pp. 298-306, 466-472), and Perlman (1918, pp. 376-380). 33 IBLS 1884, pp. 352-355. Opportunities for work continued to be low for coal miners in the early twentieth century, especially in unionized fields (Archbald 1922, pp. 53-68; Drury 1925, p. 187; Fishback 1992, pp. 83-86; Fisher and Bezanson 1932, pp. 124-153; U.S. Bituminous Coal Commission 1920, p. 25). 13 day. Apparently, neither the Miners Association nor the Knights were able to reduce hours of work during this period.34 Days of operation. Table 3 shows OLS regressions of log average days of operation on union density, using the county panel. Column (i) reports estimated effects of the Miners Association and the Knights with no covariates other than a time trend and its square. Both coefficients are small, but their standard errors are large. Column (ii) reports estimates that additionally control for the output price of coal and the average size of coal mines in the county. Small mines tended to operate only in the peak season, so size is expected to have a positive effect on days of operation. The Miners Association is now estimated to have reduced days of operation by about 19 percent, though the standard error remains fairly large. The estimated effect of the Knights is positive, but similar in magnitude to its standard error. Estimates using year effects instead of time trends (not shown) are quite similar. What explains the apparent negative effect of the Miners Association on days of work? In other industries or time periods, one might blame high union wages acting on a fixed labor-demand curve (Lewis 1964)—but the Miners Association had little effect on wages, as noted in the previous section. Alternatively, one might blame union strike activity. 35 The IBLS does not report strikes by county, but the importance of strikes can be checked by excluding 1889, the year of greatest union strike activity in this decade. Column (iii) reports OLS estimates excluding 1889. They are nearly identical to those in the previous column. So the negative effect of the Miners Association on days of work is difficult to explain but seems robust. 36 Thus, the available evidence suggests that the Miners Association reduced and the Knights increased days of work per year, but the estimates are quite imprecise. Neither the Miners Association nor the Knights reduced daily hours of work. 6. Working conditions Though wages and working time were probably coal miners’ chief concerns, they likely cared about other conditions in coal mines. According to Freeman and Medoff (1979, 1984) unions 34 Nonunion hours: IBLS 1884, table I, p. 352 (the average for workers in all 163 occupations was 9.9 hours). Miners Association hours: IBLS 1886, table XIX, p. 299. Knights hours: IBLS 1886, table XX, p. 301. An eighthour day was later won by the United Mine Workers in 1898. 35 Coal operators charged decades later that unionism reduced days of operation through strikes and unauthorized holidays (Bituminous Operators Special Committee 1923, pp. 166-174). 36 A negative effect of unionism on days of operation has also been found in prior research on early twentiethcentury coal mining. Fishback (1992, pp. 207-209) estimated that unionism reduced days of operation from 1912 to 1923 by one or two days per year out of an average of 206 days, though the standard error was more than half as large as the point estimate. Boal and Pencavel (1994, pp. 291-293) estimated that unionism reduced days of operation in West Virginia in 1897-1938 by about 14 to 20 percent, depending on specification, and the estimates were significantly different from zero in most specifications. Most of this reduction was due to a shift in the labor demand curve rather than a movement along the demand curve caused by an increase in wages. 14 exercise “collective voice,” which can be more effective than individual voice for improving workplace conditions. Union voice can yield higher productivity and perhaps other outcomes valued by individual workers. Productivity. Most workers were paid on piece and set their own pace of work, but were constrained by the speed of other mine operations, especially the speed with which management supplied the miners with empty cars and retrieved their loaded cars.37 Holding piece-rate wages constant, any increase in productivity, whatever the cause, would have increased miners’ earnings. So, all else equal, miners would probably have welcomed increased productivity. Did the Miners Association and the Knights increase productivity? Table 4 shows OLS regressions of coal output per worker per day on union density, using the county panel. Column (i) reports estimated effects of the Miners Association and the Knights with no covariates other than a time trend and its square. Column (ii) reports estimates that additionally control for the fraction of coal mined by machine38 and the average size of coal mines in the county. Columns (iii) and (iv) replace the time trends with year dummies. The estimated effects of unionism are roughly similar across specifications. The effect of the Miners Association is positive and large (between 22 percent and 34 percent) but not statistically significant at conventional levels. The effect of the Knights of Labor is negative and even larger (between 46 percent and 54 percent) with borderline statistical significance. So the Miners Association may have increased productivity, but the Knights surely did not.39 Boys in the mines. Many mines employed boys underground, usually assisting their fathers at the coal face or operating ventilation doors. Adult coal miners likely had mixed attitudes toward boys underground. On the one hand, boys increased the total coal-mining labor supply and therefore might have depressed wages. On the other hand, many boys in coal mines worked with their fathers, helping to augment family income (IBLS 1890, p. XXII). 37 IBLS (1890, p. xx); Archbald (1922, pp. 61-63); Goodrich (1925, pp. 32-35). Slichter (1941, p. 316), referring to a later era when the union was more powerful, noted “the coal miners’ union is constantly exerting pressure to have the managements provide more cars and better car service.” 38 It might be argued that machine mining also depended on unionism. Decades later, coal operators claimed that the union delayed the adoption of machine mining (Bituminous Operators Special Committee 1923, pp. 182-185), a claim repeated by Emmet (1924, pp. 31-33) and Slichter (1941, p. 266). Boal (1994) found econometric evidence of a negative effect of unionism on machine mining in West Virginia in the early twentieth century. However, there is no evidence of such a phenomenon in this Illinois panel. In a regression of fraction of coal mined by machine on the unionism variables with county and time controls (not shown here) the estimated effects of the Association, the Knights, and all unionism were positive but less than their standard errors. 39 There is a large prior literature on unionism and productivity, reviewed by by Doucouliagos and Laroche (2003), and Hirsch (2007). The findings of this literature are mixed, but studies of early coal mining generally find negative effects. Pencavel (1977) found a negative effect in British coal mining from 1900 to 1913. Boal (forthcoming) found a negative effect in West Virginia coal mining from 1914 to 1928. However, Pencavel’s and Boal’s estimates are considerably smaller than the estimated effect of the Knights shown here. 15 For the year ending July 1, 1883, a total of 1256 boys under 16 years were reported to be working underground in Illinois coal mines, 5.2 percent of the total coal labor force. In June of that year, mining laws were amended to prohibit boys under 14 from working underground. By 1888, the number of boys under 16 had dropped to 868 (3.0 percent) and by 1889, the number had dropped slightly further to 859 (2.9 percent).40 The 1883 legislation undoubtedly reduced the number of boys underground, but did unionism also play a role? Table 5 shows OLS regressions of the fraction of workers under 16 on union density, using the county panel. The sample size shrinks because the dependent variable is not available from 1884 to 1887. The estimated coefficients are about -0.02 to -0.03 for each union in every specification. The estimated union effects are substantial compared to the mean fraction under 16 (0.033). However, the standard errors are greater than the point estimates, so the results are at best suggestive of a negative union effect on boys in the coal mines. Safety. Miners surely valued safety. Accidents took the lives of 56 Illinois coal miners per year on average from 1882 through 1889. To be sure, unionists lobbied for safety legislation at the state level (Gottlieb 1978) but could individual coal miners enjoy a safer workplace by joining the union or seeking employment at a unionized mine? Table 6 shows Poisson regression estimates of the effect of unionism on fatalities in coal mining, using the county panel. The sample size shrinks because fatalities were not reported in 1880 and because counties that never experienced a fatality in any year drop out of the conditional likelihood function (Cameron and Trivedi 2013, p. 352). All specifications control for the number of workers and the average days of operation. As usual for Poisson regression, the expected number of fatalities is modeled as the exponential of a linear function of regressors, so the coefficients of the union densities should be interpreted as percent changes in expected fatalities. Estimated effects of both the Miners Association and the Knights are negative for all specifications. Many coefficients are large in magnitude, even larger than other estimates for coal mining in the literature. 41 However, the cluster-robust standard errors are even larger for the most part, so the estimates are not significantly different from zero. Many alternate specifications were estimated but are merely summarized here to save space. Observations for 1883, the year of the two mining disasters, was excluded. The coefficient of the log of workers was constrained to equal one. The log of average days of operation was excluded or its coefficient constrained to equal one. In every specification, the estimated coefficient of the Miners Association was negative. The estimated coefficient of the Knights 40 IBLS 1883, pp. 12, 109, 135; 1884, p. 418; 1888, p. 326; 1889, p. II. Boal (2009), for example, found that unionism reduced fatalities in early twentieth-century coal mining by 30 to 40 percent in the early twentieth century. Morantz (2013) found that unionism reduced tramatic injuries in coal mining from 1993 to 2010 by 14 to 32 percent. Boal’s and Morantz’s estimates are significantly different from zero at conventional levels. Fishback (1986) in one study found little effect of unionism on fatalities in early twentieth century coal mining, but in a later study (1987) estimated that unionism reduced fatalities by 46 percent. However, both of Fishback’s studies had very large standard errors. 41 16 was almost always negative but smaller than the coefficient of the Miners Association. However, standard errors were always greater than the point estimates. It might be conjectured that unionism was driven by accidents. That is, an unusually high number of accidents in a particular county during the mid-1880s might have encouraged subsequent union organization. To check this possibility, observations for 1885 were excluded, but the estimated coefficients for unionism became more negative (and implausibly large). Observations for both 1885 and 1884 were excluded with similar results. So it does not appear that reverse causality can explain the negative estimated effects of unionism on accidents. The county panel data provide some—but hardly overwhelming—evidence that the two unions improved miners’ working conditions. The Miners Association may have increased productivity, though the Knights certainly did not. Both unions may have discouraged the employment of boys underground. Both unions may have reduced accidents. Large standard errors discourage firm conclusions. 7. Treatment by employers Yet another concern of coal miners might have been treatment by employers. Faith and Reid (1986, p. 41) emphasize workers’ need to control “opportunistic behavior by employers.” This need must have been strong in the 1880s, when labor legislation was almost nonexistent. In recent periods, records of grievance procedures might demonstrate how a union defended workers from opportunistic employers, but grievance procedures were not implemented in Illinois coal until 1898 (Bloch 1931; Justi 1902). Circumstantial evidence is available, however, in records of strikes. Assurance of payment. Almost all coal miners during this period were paid once a month.42 Prompt payment would have been important to these workers, most of whom apparently had little or no savings (IBLS 1884, pp. 394-402). If an employer delayed or reneged on payment, the courts would have offered little help in enforcing payment of wages—a mechanic’s lien law for coal miners was not passed in Illinois until 1895 (IBLS 1899, pp. 237-238, U.S. Commissioner of Labor 1896, p. 283). Would a union help enforce prompt payment? Circumstantial evidence on this question is provided by strike data collected by the U.S. Commissioner of Labor.43 Table 8 shows causes of 178 coal strikes reported in Illinois for the period 1881-1889. Strikes were frequent, but most were small, involving a single coal mine. Note that 79 strikes, 44 percent of the total, were not ordered by a labor organization. For the 42 IBLS 1886, table XXVIII, pp. 323-324; and table XXIX, p. 325. A law requiring weekly payment was passed in 1891 (IBLS 1991, p. 46) but compliance lagged (IBLS 1992, p. 549). 43 Strike data for 1881-1886 collected by the Commissioner were published by IBLS (1888, pp. 246-313) in a table entitled “Strikes by Years and Industries.” Data for 1887-1889 were published by U.S. Commissioner of Labor (1894, pp. 162-258, table I). Bailey (1991) suggests that the Commissioner’s reports undercount actual strikes. 17 most part the causes of union and nonunion strikes were similar.44 However, there is one interesting exception. Thirty-four percent of the nonunion strikes were for “payment of wages overdue,” while none of the union strikes were for this cause. What could explain this difference, which is statistically highly significant? One possible explanation is selection, that unions tended to organize less opportunistic coal operators, but it is hard to believe that unions would be more successful in organizing more contented workers. Another possible explanation is that unions somehow suppressed miners’ desire for prompt payment, but this is equally hard to believe. The most plausible explanation is that unions were able to enforce prompt payment of wages without a strike. One naturally wonders whether the Miners Association and the Knights might have helped secure better treatment in other aspects of the employment relationship, such as frequency of payment, terms at company stores, and protection from favoritism in workplace assignment. However, quantitative evidence—even circumstantial evidence—on these points is lacking. 8. Conclusion What could unions do for Illinois coal miners in the 1880s? It appears that neither the Miners Association nor the Knights of Labor raised wages, and only a small minority of union locals provided benefits. Neither union was able to enforce a shorter workday. The Miners Association apparently reduced days of work per year, but it seems unlikely coal miners would have valued this result. Thus the usual benefits of union membership were absent in this period. Nevertheless, the data offer hints of other less salient benefits of union membership. The Miners Association may have increased productivity, which would have raised earnings for piece-rate workers, though the Knights surely did not. Point estimates suggest that both unions may have discouraged child labor in the mines and increased safety, but the estimates are imprecise. Finally, a comparison of causes of union and nonunion strikes suggests that unions were effective at ensuring prompt payment of wages due. In short, coal miners’ unions in Illinois were not able to enforce the usual demand for “less work and more pay,” but they may have improved other workplace conditions and they probably enforced more reliable pay. 44 The sources do not give the name of the “labor organization,” so union strikes could have been led by either the Miners Association or the Knights. 18 References Archbald, H. (1922). The four hour day in coal. New York: H.W. Wilson Company. Bailey, G. L. (1991). The Commissioner of Labor's Strikes and Lockouts: a cautionary note. Labor History, 32(3), 432-440. Birdsall, W. C. (1953). The problem of structure in the Knights of Labor. Industrial and Labor Relations Review, 6(4), 532-546. Bituminous Operators' Special Committee (1923). Comparative efficiency of labor in the bituminous coal industry under union and nonunion operation, a brief submitted to the United States Coal Commission. Blanchflower, D. G., & Bryson, A. (2007). What effect do unions have on wages now and would Freeman and Medoff be surprised? In B. E. Kaufman, & J. T. Bennett (Eds.), What do unions do? A Twenty-Year Perspective (pp. 79-113). New Brunswick, N.J.: Transaction Publishers. Boal, W. M. (1994). Union response to mechanization in U.S. bituminous coal. Labour Economics, 1(3/4), 243-267. Boal, W. M. (2009). The effect of unionism on accidents in U.S. coal mining, 1897-1929. Industrial Relations, 48(1), 97-120. Boal, W. M. (2014). Unionism and productivity in West Virginia coal mining: a longer view. Des Moines: Drake University. Boal, W. M., & Pencavel, J. H. (1994). The effects of labor unions on employment, wages, and days of operation: coal mining in West Virginia. Quarterly Journal of Economics, 109(1), 267-298. Booth, A. L. (1995). The economics of the trade union. Cambridge: Cambridge University Press. Boyer, G. R. (1988). What did unions do in nineteenth-century Britain? Journal of Economic History, 1988(2), 319-332. Budd, J. W. (2007). The effect of unions on employee benefits and non-wage compensation: monopoly power, collective voice, and facilitation. In B. E. Kaufman, & J. T. Bennett (Eds.), What do unions do? A Twenty-Year Perspective (pp. 160-192). New Brunswick, N.J.: Transaction Publishers. Cahill, M. C. (1932). Shorter hours: a study of the movement since the Civil War (Columbia University Studies in the Social Sciences). New York, N.Y.: Columbia University Press. Cameron, A. C., & Trivedi, P. K. (2013). Regression analysis of count data (2nd ed., Econometric Society Monograph). Cambridge: Cambridge University Press. David, H. (1936). The history of the Haymarket Affair: a study in the American socialrevolutionary and labor movements. New York: Russell and Russell. Dillon, P., & Gang, I. (1987). Earnings effects of labor organizations in 1890. Industrial and Labor Relations Review, 40(4), 516-527. Doucouliagos, C., & Laroche, P. (2003). What do unions do to productivity? A meta-analysis. [Review]. Industrial Relations, 42(4), 650-691. Douglas, P. H. (1930). Real wages in the United States 1890-1926 (Publications for the Pollak Foundation for Economic Research, Vol. 9). Boston: Houghton Mifflin. Drury, H. B. (1925). Volume of employment and collateral employment. In E. E. Hunt, F. G. Tryon, & J. H. Willits (Eds.), What the Coal Commission found: an authoritative summary by the Staff (pp. 185-194). Baltimore: Williams and Wilkins Company. Earle, J. S., & Pencavel, J. H. (1990). Hours of work and trade unionism. Journal of Labor Economics, 8(1, part 2), S150-S174. 19 Eichengreen, B. (1987). The impact of late nineteenth-century unions on labor earnings and hours: Iowa in 1894. Industrial and Labor Relations Review, 40(4), 501-515. Emmet, B. (1924). Labor relations in the Fairmont, West Virginia bituminous coal field (United States Bureau of Labor Statistics Bulletin, Vol. No. 361). Washington: Government Printing Office. Evans, C. (1920). History of the United Mine Workers of America from the year 1860 to 1890 (Vols. 1 & 2). Indianapolis. Faith, R. L., & Reid, J. D., Jr. (1987). An agency theory of unionism. Journal of Economic Behavior and Organization, 8, 39-60. Fishback, P. V. (1986). Workplace safety during the progressive era: fatal accidents in bituminous coal mining, 1912-1923. [July 1986]. Explorations in Economic History, 23(3), 269-298. Fishback, P. V. (1987). Liability rules and accident prevention in the workplace: empirical evidence from the early twentieth century. Journal of Legal Studies, 16(2), 305-328. Fishback, P. V. (1992). Soft coal, hard choices: the economic welfare of bituminous coal miners, 1890-1930. Oxford: Oxford University Press. Fisher, W. E., & Bezanson, A. (1932). Wage rates and working time in the bituminous coal industry, 1912-1922. Philadelphia: University of Pennsylvania Press. Foner, P. S. (1962). History of the labor movement in the United States (Vol. 1). New York: International Publishers. Freeman, R. B., & Medoff, J. L. (1979). The two faces of unionism. Public Interest(57), 69-93. Freeman, R. B., & Medoff, J. L. (1984). What do unions do? New York: Basic Books. Garlock, J. (1982). Guide to the local assemblies of the Knights of Labor. Westport, CT: Greenwood Press. George, J. E. (1898). The coal miners' strike of 1897. Quarterly Journal of Economics, 12(2), 186-208. Goodrich, C. (1925). The miner's freedom: a study of the working life in a changing industry. Boston: Marshall Jones Company. Gottlieb, A. Z. (1978). The influence of British trade unionists on the regulation of the mining industry in Illinois, 1872. Labor History, 19(3), 397-415. Hatton, T. J., Boyer, G. R., & Bailey, R. E. (1994). The union wage effect in late nineteenth century Britain. Economica, 61(244), 435-456. Hirsch, B. T. (2007). What do unions do for economic performance? In B. E. Kaufman, & J. T. Bennett (Eds.), What do unions do? A Twenty-Year Perspective (pp. 193-237). New Brunswick, N.J.: Transaction Publishers. Hirsch, B. T., & Addison, J. T. (1986). The economic analysis of unions: new approaches and evidence. Boston, MA: Allen and Unwin. Illinois Bureau of Labor Statistics (1881). First biennial report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1883). Statistics of coal production in Illinois: a supplemental report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1884). Third biennial report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1885). Statistics of coal in Illinois: a supplemental report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1886). Fourth biennial report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1887). Statistics of coal in Illinois: a supplemental report. Springfield, Illinois. 20 Illinois Bureau of Labor Statistics (1888). Fifth biennial report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1889). Statistics of coal in Illinois: a supplemental report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1890). Sixth biennial report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1891). Statistics of coal in Illinois: a supplemental report. Springfield, Illinois. Illinois Bureau of Labor Statistics (1894). Eighth biennial report. Springfield, Illinois. Justi, H. (1902). Conciliation and arbitration in the coal mining industry. Publications of the American Economic Association, 3rd Series, 3(1), 279-304. Kemmerer, D., & Wickersham, E. D. (1950). Reasons for growth of the Knights of Labor in 1885-1886. Industrial and Labor Relations Review, 3(2), 213-220. Lewis, H. G. (1963). Unionism and relative wages in the United States. Chicago: University of Chicago Press. Lewis, H. G. (1964). Relative employment effects of unionism. In G. G. Somers (Ed.), Proceedings of the Sixteenth Annual Meeting, Boston, December 27 and 28, 1963 (pp. 104-115, Publications of the Industrial Relations Research Association, Vol. 32). Madison, Wisconsin: Industrial Relations Research Association. Lewis, H. G. (1986). Union relative wage effects: a survey. Chicago: University of Chicago Press. Lloyd, H. D. (1890). A Strike of Millionaires Against Miners or the Story of Spring Valley. Chicago: Belford-Clarke. McBride, J. (1887). Coal miners. In G. E. McNeill (Ed.), The labor movement: the problem of today (pp. 241-267). Boston: A.M. Bridgman and Company. McNeill, G. E. (1887). Progress of the movement from 1861 to 1886. In G. E. McNeill (Ed.), The labor movement: the problem of to-day (pp. 124-171). Boston: A. M. Bridgman and Company. Morantz, A. D. (2013). Coal mine safety: do unions make a difference? Industrial and Labor Relations Review, 66(1), 88-116. Mullin, D. (1998). A new look at the union wage premium during the early years of the AFL. Industrial and Labor Relations Review, 51(2), 253-268. Pencavel, J. H. (1977). The distribution and efficiency effects of trade unions in Britain. British Journal of Industrial Relations, 15(2), 137-156. Pencavel, J. H. (1991). Labor markets under trade unionism: employment, wages, and hours. Cambridge: Basil Blackwell. Perlman, S. (1918). Part VI: Upheaval and reorganization (since 1876). In J. R. Commons (Ed.), History of labour in the United States (Vol. 2, pp. 195-540). New York: Macmillan. Rees, A. (1977). The economics of trade unions (2nd ed.). Chicago: University of Chicago Press. Roy, A. (1907). A history of the coal miners of the United States from the development of the mines to the close of the anthracite strike of 1902, including a brief sketch of early British miners. Columbus, Ohio: J. L. Trauger. Saposs, D. J. (1918). Part I: Colonial and federal beginnings (to 1827). In J. R. Commons (Ed.), History of labour in the United States (Vol. 1, pp. 25-168). New York: Macmillan. Seidman, J., London, J., & Karsh, B. (1951). Why workers join unions. American Academy of Political and Social Science, 274, 75-84. Slichter, S. H. (1941). Union policies and industrial management. Washington, D.C.: Brookings Institution. 21 Suffern, A. E. (1926). The coal miners' struggle for industrial status: a study of the evolution of organized relations and industrial principles in the coal industry. New York: Macmillan. U.S. Bituminous Coal Commission (1920). Majority and minority reports. Washington, D.C.: Government Printing Office. U.S. Census Office, D. o. t. I. (1886). Report on Mining Industries in the United States (exclusive of precious metals), with special investigation into the iron resources of the Republic and into cretaceous coals of the Northwest. Tenth Census (1880) (Vol. 15). Washington D.C. U.S. Commissioner of Labor (1894). Tenth Annual Report: Strikes and Lockouts. Washington, DC: GPO. U.S. Commissioner of Labor (1896). Second Special Report: Labor laws of the United States (Second ed.). Washington, DC: GPO. U.S. Commissioner of Labor (1901). Sixteenth Annual Report, 1901: Strikes and Lockouts. Washington, DC: GPO. U.S. Geological Survey (1902). Mineral resources of the United States. Washington: Government Printing Office. Ulman, L. (1955). The rise of the national trade union: development and significance of its structure, governing institutions, and economic policies (Wertheim Publications in Industrial Relations). Cambridge, Massachusetts: Harvard University Press. Webb, S., & Webb, B. (1920). Industrial democracy. London: Longmans, Green and Co. Wright, J. L., Macauley, R. C., Hilsee, J. M., Cook, W., Keen, R. W., & Kennedy, J. S. (1887). History of the Knights of Labor. In G. E. McNeill (Ed.), The labor movement: the problem of today (pp. 397-428). Boston: A.M. Bridgman and Company. 22 Appendix A: County panel data set This appendix describes construction of the county-level panel data set. The Illinois Bureau of Labor Statistics (hereafter, IBLS) collected and reported data on production, employment, etc. beginning in 1879. Data on union membership were collected only once in 1886. Miners Association membership. Detailed data on union membership were published in the IBLS Fourth Biennial Report (1886). In that year, the Bureau conducted a survey of labor unions in all occupations in Illinois. The data obtained from the Miners and Mine Laborers Benevolent Protective Association (hereafter, the Miners Association) were characterized by the IBLS as especially complete, thanks to the cooperation of the union leadership, but nevertheless, 6 locals and 804 members are missing for unexplained reasons (IBLS 1886, p. 181). A special section of the Report, entitled “Labor Organizations,” reported the survey results in tabular form. For each lodge or local union the following information was provided: month and year organized, charter members, total present members, and present members currently employed (IBLS 1886, table I, pp. 172-178). “Present” meant as of July 1886 (see IBLS 1886, p. 221, first paragraph). For this study, local employed membership data were summed to the county level. The IBLS itself reported membership at the county level (IBLS 1886, tables VII and VIII, pp. 214-219), but the IBLS county totals seem to contain errors. For example, the IBLS reported 552 members in Adams County, which contained no coal mines. Also, the IBLS apparently misallocated the membership of the local at Illiana, in Vermilion County, to Edgar County, perhaps because the latter contains Vermilion City. So instead of using the county totals reported by the IBLS, local employed membership data were summed to the county level by hand. Locals were assigned to counties using each local’s city address and Bing Maps. In three counties (Marshall, St. Clair, and Saline) employed membership was not reported by the IBLS. Now, excluding these three counties, the ratio of employed members to total members for the entire state was 0.795. So for these counties employed membership was imputed as total members times 0.795. Knights membership. In the same volume, the IBLS (1886, table XVIII, pp. 273-294) also reported current membership in the Knights of Labor by county and occupation as of mid-1886. For this study, only Knights members whose occupations were "coal miners," "mine laborers," or "miners" were included. They totaled over 3500, about half as numerous as Association members. For the Knights, only total membership, not employed membership, is available. Other data. Beginning in 1883, the IBLS published annual data on production, employment, wages, prices, and other coal data from coal operators. Wages and employment sometimes varied seasonally, in which case the winter (peak season) values were used. All data were reported by county for up to 50 counties. In even years, these data were included in the IBLS Biennial Report. In odd years, a separate report usually entitled Statistics of Coal in Illinois was 23 published. Both sources reported nearly the same data for the same time periods (years ending July 1, IBLS 1883, p. 5) and are easily combined into a county panel. However, the panel is not quite balanced because some counties with very small production do not appear every year. Moreover, some items were not reported every year (see table A1). The IBLS also published some data for earlier years, collected by county mine inspectors, apparently for calendar years (IBLS 1882, p. 42). These include reasonably complete data for 1882, just sixteen counties for 1879, but nothing for 1880 or 1881. Months (rather than days) of operation were reported in these years (and for 1883) so I imputed days of operation assuming 20 days of operation per month. This seemed satisfactory because average imputed days were then roughly similar to average days reported for later years. Note that the choice of assumed days per month (20) does not affect the OLS estimates if days are entered in logarithmic form and fixed effects for years are included, but any imputation does of course create measurement error. For 1880 only, limited data were published by the U.S. Census (1886), including coal output, employment, the price of output, and boys in mines, apparently for the year ending June 1, 1880 (p. xiii). Also, figures for “production capacity” were published, and I used them to impute days of operation. For some small counties, production capacity seems to have been calculated by the Census as actual production divided by (unpublished) actual months of operation, times twelve. For these counties, I backed out days of operation using the Census’s apparent assumption of “twelve months of twenty-five working days each” (U.S. Census 1886, p. 684): 𝑖𝑚𝑝𝑢𝑡𝑒𝑑 𝑑𝑎𝑦𝑠 = 𝑎𝑐𝑡𝑢𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 × 300 . 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 I am reasonably confident in this procedure because the result was always a multiple of 25. For other counties, production capacity seems to have been estimated by operators, but the Census characterized the operators’ capacity estimates as excessively “sanguine” in view of (unpublished) data on days of operation (p. 682). For Illinois as a whole, the Census reported that the ratio of actual production to production capacity was only 44.10 per cent while the days of operation were 78.15 percent of the 300-day maximum (table 35, p. 683, columns 10 and 15). So for these counties, I applied a correction factor: 𝑖𝑚𝑝𝑢𝑡𝑒𝑑 𝑑𝑎𝑦𝑠 = 𝑎𝑐𝑡𝑢𝑎𝑙 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 78.15 × 300 × . 𝑝𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦 44.10 Data availability for all years in the sample (with imputations) is summarized in table A1 below. Calculation of union density. Union density at the county level was computed as follows. For 1886, density for the Miners Association was computed as employed membership divided by total employment in coal mining. Similarly, union density for the Knights was computed as total membership in coal-related occupations divided by total employment in coal mining. The historical record suggests that both unions were weak or nonexistent in the early 1880s, grew 24 rapidly in 1885 and 1886, and remained fairly strong until at least the summer of 1889. Union density was therefore assumed to equal zero for all counties before 1886. Union density was assumed constant and equal to its 1886 value after 1886. Observations for the year ending July 1886, the year of rapid union growth, were dropped from the analysis. Mine-level data. I originally intended to construct and analyze a panel at the level of the individual coal mine, but this proved to be impossible. The IBLS did report a wealth of minelevel data on production, employment, etc. Unfortunately, these data cannot be matched to union membership because the IBLS reported only the postal addresses of local unions and not union members’ places of employment. Even without union data, construction of a panel of mines would have been difficult because mine names changed frequently during this period, most likely due to changes in ownership. Table A1: Data availability--Illinois coal mining counties, 1879-1889 Source of data Maximum number of observations Winter (peak season) wage for hand-mining (cents per ton) Average days of operation a Coal output (tons) per worker per day Boys under 16 as fraction of mine workers Fatalities Price of coal at the mine (cents per ton) Fraction of coal mined by machine b Number of workers in and around mines Number of coal mines Average size of coal mine (workers per mine) 1879 1880 1882 IBLS Census IBLS 16 45 43 X X X X X X X X X X X X X X X X X X X X X X X X 1883 IBLS 46 X 1884 IBLS 49 X 1885 IBLS 50 X 1886 IBLS 50 X 1887 IBLS 50 X 1888 IBLS 50 X 1889 IBLS 49 X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X NOTES: Year 1886 excluded from analysis because of rapidly changing levels of unionism in late 1885 and early 1886. a Average days of operation estimated for 1879, 1882, and 1883 as months of operation times 20. Average days of operation for 1880 imputed using actual production and production capacity (see data appendix). b Fraction of coal mined by machine available in 1882 for only 3 counties (all zero). 25 Figure 1 Growth of Illinois coal mining, 1880-1890 30,000 30 Coal output 0 0 1890 5 1889 5,000 1888 10 1887 10,000 1886 15 1885 15,000 1884 20 1883 20,000 1882 25 1881 25,000 Output: millions of tons 35 Employment 1880 Employment: workers 35,000 SOURCE: IBLS 1890, pp. 285, 288. Figure 2 Coal prices and wages in Illinois, 1882-1890 $1.60 $1.40 $1.20 $1.00 $0.80 $0.60 $0.40 Price of coal (per ton) at mines $0.20 SOURCE: IBLS 1890, pp. 291, 292. 1890 1889 1888 1887 1886 1885 1883 1882 1884 Wage paid (per ton) to hand miners $0.00 26 Figure 3 ? 1876 ? 1885 Jan. 1885 Feb. 1885 Mar. 1885 Apr. 1885 May 1885 June 1885 July 1885 Aug. 1885 Sept. 1885 Oct. 1885 Nov. 1885 Dec. 1885 Jan. 1886 Feb. 1886 Mar. 1886 Apr. 1886 May 1886 June 1886 July 1886 Aug. 1886 Sept. 1886 Oct. 1886 Nov. 1886 Dec. 1886 500 400 300 200 100 0 Charter members of Miners and Mine Laborers Benevolent Protective Association, by date local was organized SOURCE: IBLS 1886, pp. 173-174. Month of organization not reported for one local founded in 1876 at LaSalle and one local founded in 1885 at Springfield. Figure 4 Total members as of July 1886 Employed members as of July 1886 Jan. 1885 Feb. 1885 Mar. 1885 Apr. 1885 May 1885 June 1885 July 1885 Aug. 1885 Sept. 1885 Oct. 1885 Nov. 1885 Dec. 1885 Jan. 1886 Feb. 1886 Mar. 1886 Apr. 1886 May 1886 June 1886 July 1886 Aug. 1886 Sept. 1886 Oct. 1886 Nov. 1886 Dec. 1886 8000 7000 6000 5000 4000 3000 2000 1000 0 Cumulative membership in Miners and Mine Laborers Benevolent Protective Association, by date local was Charter members organized SOURCE: IBLS 1886, pp. 173-174. 27 Figure 5 5000 100 4000 80 3000 60 2000 40 1000 20 All U.S. 1887 1886 1885 1884 1883 1882 1881 1880 1879 1878 1877 1876 1875 1873 1872 0 1869 0 Illinois only All U..S. Local Assemblies of Knights of Labor, by date of organization Illinois SOURCE: All U.S.: Garlock 1982, p. xix. Illinois only (1877-1886): IBLS 1886, p. 210. Figure 6 Cumulative membership in Knights of Labor (all trades, Illinois only), by date Assembly was organized 40000 Charter members 35000 Total members as of 1886 30000 Employed members as of 1886 25000 20000 15000 10000 5000 0 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 SOURCE: IBLS 1886, p. 210. 28 Figure 7 Local Assemblies dominated by coal miners, by date of organization, Illinois only 20 15 10 5 0 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 SOURCE: Garlock 1982, pp. 63-92. 29 Figure 8 SOURCE: Author’s calculations from IBLS 1886. Omitted counties contained no coal mines. 30 Figure 9 SOURCE: Author’s calculations from IBLS 1886. Omitted counties contained no coal mines. 31 Figure 10 Number of workers striking, all industries, by year 600,000 500,000 400,000 300,000 Illinois 200,000 United States 100,000 1890 1889 1888 1887 1886 1885 1884 1883 1882 1881 1880 0 SOURCE: United States Commissioner of Labor 1901, pp. 63, 343. Figure 11 Trends in peak-season wage for three groups of counties 120 Cents per ton 100 80 Miners Association density > 0.3 in 1986 60 Knights of Labor density > 0.3 in 1986 40 No unionism in 1986 20 0 1883 1884 1885 1886 1887 1888 1889 SOURCE: IBLS (various issues) and author’s calculations. Average wages are weighted by employment. 32 Figure 12 Wages for hand mining (per ton), 1886 $1.00 Union wage, as reported by Miners Association locals $0.90 $0.80 $0.70 $0.60 $0.50 $0.40 $0.30 $1.00 $0.90 $0.80 $0.70 $0.60 $0.50 $0.40 $0.30 County average wage, as reported by operators SOURCE: Union wage: IBLS 1886, table XVII, pp. 272-273. County average wage: IBLS 1886, pp. 567-622. Solid line is 45-degree line. 33 Table 1: Descriptive statistics of sample a. Panel of Illinois coal-mining counties, 1879-1889 Winter (peak season) wage for hand-mining (cents per ton) Average days of operationa Coal output (tons) per worker per day Boys under 16 as fraction of mine workers Fatalities Price of coal at the mine (cents per ton) Fraction of coal mined by machine Number of workers in and around mines Number of coal mines Average size of coal mine (workers per mine) Obs. 287 377 376 247 352 392 255 395 397 395 Mean Std. Dev. 83.7 23.9 205.5 1.856 0.033 1.142 139.9 0.063 518.2 15.7 53.3 52.2 1.199 0.038 4.164 36.2 0.184 662 19 63 Min. 47.0 Max. 200.0 21.3 0.081 0.000 0.000 68.9 0.000 1 1.000 0.500 420.3 12.113 0.200 70 254 1.000 3491 87 361 SOURCE: Data for 1879, 1882, 1883, 1884, 1885, 1887, 1888, and 1889 from Illinois Bureau of Labor Statistics (various issues). Data for 1880 from U.S. Census (1886). Unit of observation is county. Year 1886 excluded because of rapidly changing levels of unionism in late 1885 and early 1886. a Average days of operation estimated for 1879, 1882, and 1883 as months of operation times 20. Average days of operation for 1880 imputed using actual production and production capacity (see data appendix). b. Union density in Illinois coal-mining counties, 1886 Fraction of workers who are employed Obs. 49 Mean Std. Dev. 0.114 0.189 Min. 0 Max. 0.791 a members of Miners Association Fraction of workers who are members of Knights of Labor Fraction of workers who are members of either union SOURCE: Illinois Bureau of Labor Statistics (1886). a 49 0.107 0.151 0 0.642 49 0.164 0.182 0 0.656 Miners and Mine Laborers Benevolent Protective Association. 34 Table 2: Estimated effect of unionism on wages (i) (ii) (iii) (iv) Fraction of workers who are employed members of Miners Associationa 0.0215 (0.0655) -0.0029 (0.0713) 0.0431 (0.0663) 0.0194 (0.0700) Fraction of workers who are members of Knights of Labor -0.0011 (0.1240) -0.0173 (0.1177) 0.0246 (0.1347) 0.0034 (0.1293) Fraction of workers who are members of either union (v) (vi) 0.0357 (0.0958) 0.0079 (0.0896) Log price of coal at the mine 0.2911 *** (0.0988) 0.2729 *** (0.0993) 0.2746 *** (0.0977) Log average size of coal mine 0.0189 (0.0186) 0.0162 (0.0181) 0.0166 (0.0173) Annual time trend -0.0978 *** (0.0229) -0.0577 ** (0.0274) Time trend squared 0.0059 *** (0.0018) 0.0036 * (0.0020) Fixed effects for counties Fixed effects for years Adjusted R-square Number of observations Yes No 0.8992 279 Yes No 0.9093 277 Yes Yes 0.9082 279 Yes Yes 0.9163 277 NOTES: OLS estimates. Dependent variable is log of winter (peak season) wage for hand-mining. Standard errors (in parentheses) are cluster-robust. a Miners and Mine Laborers Benevolent Protective Association. * Significantly different from zero at 10% in two-tailed test. ** Significantly different from zero at 5% in two-tailed test. *** Significantly different from zero at 1% in two-tailed test. Yes Yes 0.9084 279 Yes Yes 0.9167 277 35 Table 3: Estimated effect of unionism on days of operation (i) (ii) (iii) Fraction of workers who are employed members of Miners Associationa -0.0443 (0.2232) -0.1929 (0.1537) -0.1909 (0.1882) Fraction of workers who are members of Knights of Labor 0.0749 (0.1432) 0.1294 (0.1205) 0.1781 (0.1323) Log price of coal at the mine -0.0506 (0.1564) -0.0673 (0.1732) Log average size of coal mine 0.1951 ** (0.0758) 0.2065 ** (0.0855) Annual time trend -0.0085 (0.0272) -0.0252 (0.0229) -0.0280 (0.0249) Time trend squared -0.0005 (0.0027) 0.0007 (0.0021) 0.0009 (0.0026) Fixed effects for counties Fixed effects for years Adjusted R-square Yes No 0.237 Yes No 0.3381 Yes No 0.3129 Number of observations 369 367 321 NOTES: OLS estimates. Dependent variable is log of average days of operation. Standard errors (in parentheses) are cluster-robust. Column (iii) omits observations for 1889. a Miners and Mine Laborers Benevolent Protective Association. * Significantly different from zero at 10% in two-tailed test. ** Significantly different from zero at 5% in two-tailed test. *** Significantly different from zero at 1% in two-tailed test. 36 Table 4: Estimated effect of unionism on productivity (i) (ii) (iii) (iv) Fraction of workers who are employed members of Miners Associationa 0.2901 (0.2327) 0.3394 (0.3012) 0.2314 (0.2493) 0.2192 (0.2919) Fraction of workers who are members of Knights of Labor -0.4590 * (0.2613) -0.3723 * (0.2230) -0.5429 * (0.3010) -0.4705 ** (0.2223) Fraction of workers who are members of either union Fraction of coal mined by machine 0.4221 *** (0.1523) 0.4182 *** (0.1442) Log average size of coal mine -0.2067 *** (0.0680) -0.2070 *** (0.0675) Annual time trend 0.0078 (0.0263) -0.1942 (0.1308) Time trend squared 0.0005 (0.0028) 0.0155 (0.0097) Fixed effects for counties Fixed effects for years Adjusted R-square Number of observations Yes No 0.4284 368 Yes No 0.6734 244 Yes Yes 0.4298 368 NOTES: OLS estimates. Dependent variable is log of coal output per worker per day. Standard errors (in parentheses) are cluster-robust. a Miners and Mine Laborers Benevolent Protective Association. * Significantly different from zero at 10% in two-tailed test. ** Significantly different from zero at 5% in two-tailed test. *** Significantly different from zero at 1% in two-tailed test. Yes Yes 0.6908 244 37 Table 5: Estimated effect of unionism on boys in coal mines (i) (ii) Fraction of workers who are employed a members of Miners Association -0.0224 (0.0248) -0.0199 (0.0253) Fraction of workers who are members of Knights of Labor -0.0323 (0.0344) -0.0268 (0.0356) Fraction of workers who are members of either union (iii) (iv) -0.0261 (0.0298) -0.0206 (0.0307) Annual time trend -0.0012 (0.0031) -0.0012 (0.0031) Time trend squared 0.0000 (0.0003) -0.0001 (0.0003) Fixed effects for counties Fixed effects for years Adjusted R-square Number of observations Yes No 0.2994 241 Yes Yes 0.3148 241 Yes No 0.2995 241 Yes Yes 0.3155 241 NOTES: OLS estimates. Dependent variable is fraction of mine workers who are under 16 years. Standard errors (in parentheses) are cluster-robust. a Miners and Mine Laborers Benevolent Protective Association. * Significantly different from zero at 10% in two-tailed test. ** Significantly different from zero at 5% in two-tailed test. *** Significantly different from zero at 1% in two-tailed test. 38 Table 6: Estimated effect of unionism on fatalities (i) (ii) (iii) Fraction of workers who are employed a members of Miners Association -0.0589 (0.8009) -0.6502 (0.8764) Fraction of workers who are members of Knights of Labor -0.2706 (0.8760) -0.5590 (1.0829) Fraction of workers who are members of either union -0.3539 (0.6539) (iv) -1.1577 (0.9843) Log average days of operation 0.9245 (1.5069) 0.9187 (1.5242) 0.0643 (0.8889) 0.0723 (0.8958) Log number of workers 1.7283 *** (0.4100) 1.7076 *** (0.3885) 1.6260 (0.3735) 1.6175 *** (0.3637) Annual time trend 0.1611 (0.1479) 0.1586 (0.1463) Time trend squared -0.0246 (0.0161) -0.0233 (0.0156) Fixed effects for counties Fixed effects for years Log likelihood Number of observations Number of countiesb Yes No -327.076 246 35 Yes No -326.978 246 35 Yes Yes -288.987 246 35 Yes Yes -288.374 246 35 NOTES: Panel poisson regression estimates. Dependent variable is log of coal output per worker per day. Standard errors (in parentheses) are cluster-robust. a Miners and Mine Laborers Benevolent Protective Association. b Includes only counties with at least one fatality in at least one year. Counties with no fatalities in any year drop out of the conditional likelihood function (Cameron and Trivedi 2013, p. ???). * Significantly different from zero at 10% in two-tailed test. ** Significantly different from zero at 5% in two-tailed test. *** Significantly different from zero at 1% in two-tailed test. 39 Table 7: Causes of strikes in Illinois coal mining, 1881-1889 Cause For increase of wages Against reduction of wages For payment of wages overdue For reinstatement of employees For change of screena In sympathy with strike elsewhere Against work rules For employment of checkweighman For equal wages, winter and summer Other TOTAL All strikes Number Percent 52 35 27 15 6 4 3 2 1 33 178 29% 20% 15% 8% 3% 2% 2% 1% 1% 19% 100% Strikes ordered by labor organization Number Percent 35 25 0 9 5 2 2 1 0 20 99 35% 25% 0% 9% 5% 2% 2% 1% 0% 20% 100% Strikes not ordered by labor organization Number Percent 17 10 27 5 1 2 1 1 1 14 79 22% 13% 34% 6% 1% 3% 1% 1% 1% 18% 100% Test of differenceb Chi-sq p-value 2.88 3.54 33.84 0.46 1.87 0.05 0.15 0.03 1.25 0.18 0.090 0.060 0.000 0.496 0.172 0.821 0.700 0.873 0.263 0.675 SOURCES: 1881-1886: IBLS (1888, pp. 246-313, table "Strikes by Years and Industries"). 1887-1889: U.S. Commissioner of Labor (1894), pp. 162-258, table I. a Screens were used to sort coal into lumps, for which the miner was paid, and smaller pieces of lower market value, for which the miner was not paid. Thus the larger the holes in the screen, the fewer lumps retained, and the less the miner was paid. A law requiring payment before screening went into effect in 1891 (IBLS 1891, p. 46). b Chi-square test of homogeneity (1 degree of freedom). [end of manuscript]
© Copyright 2026 Paperzz