Unfettered: Fishback 25 Years Later

A quarter century ago, economist Price Fishback published “Operations of ‘Unfettered’ Labor Markets: Exit and Voice in American Labor Markets at the Turn of the Century” in the prestigious Journal of Economic Literature. Fishback’s article is packed with insight… and understatement.

But let’s back up. Virtually every standard history textbook describes U.S. labor markets before the 1920s as a Dickensian hell. Ray Billington, one of my assigned AP U.S. History textbook authors back in 1987-8, unironically opens with the heading “The Plight of the Workers,” followed by lurid details. Why then were immigrants pouring into the U.S. throughout the open borders era? Blank-out. Political slant only marginally matters on this issue; as Klein and Stern show (pp.279-81), historians who self-identify as “conservative” heavily support labor market regulation, too.

Fishback’s JEL piece amusingly ignores this historical orthodoxy, and instead speaks as if there’s an ongoing lively debate that he is ready to adjudicate. Three revealing sentences:

Yet this was no golden age in which markets worked perfectly.

and

Economic historians over the past two decades have studied many of these issues, and have found that the labor markets circa 1900 functioned better than progressive era reformers described but were by no means functioning perfectly smoothly.

and

Removal of workplace regulation would be neither the panacea expected by conservatives nor the disaster predicted by liberals.

Price Fishback

When the normal view is that markets worked horribly, denying that they worked “perfectly” is funny. When the normal view of both liberals and conservatives is that removal of workplace regulation would be a disaster, talking as if the “panacea” position is common is hilarious. Especially given Fishback’s own heretical verdict:

If we were to place the labor markets of the early 1900s on a continuum between complete market failure and smooth market operation, these markets would probably be located roughly three-fourths of the way toward the smooth operation end.

Despite his (willfully?) bizarre framing, Fishback’s article is packed with insight. Highlights:

1. U.S. wages during the late 19th- and early 20th-century U.S. labor markets varied only modestly by region, and workers migrated towards higher-wage areas, two strong signs of national competition rather than local monopsony. Even the South was, in technical terms, “well-integrated” with the national economy:

   The wage gaps between the North and the South were substantially smaller than those between the North and Europe. Further, information in Rosenbloom (forthcoming) suggests that the gap between northern and southern wages was much smaller at higher skill levels than at lower skill levels. Finally, the rise in the North-South wage gap during the 1910s contributed to the sharp rise in northbound migration during that decade.

2. Workers during this era often voluntarily quit their jobs to pursue better options. Even in seemingly isolated areas:

   Turnover rates in coal mining were similar to those found in manufacturing despite the fact that many miners were in more isolated areas where there were likely to be far fewer employers within a one-mile radius. Evidence from the U.S. Coal Commission in 1921 showed that separation rates-the number of separations from the payroll lasting longer than a month divided by the average number of workers on the pay- roll-averaged around 115 percent in coal mining, a level almost identical to the average for manufacturing firms with fewer than 1000 workers in 1913-14.

3. Real wages rose rapidly and workweeks fell rapidly during this period.

   Workers shared in the gains from economic growth during the period, as real earnings trended upward. Albert Rees's (1961, pp. 3-5) estimates of average hourly earnings in manufacturing rose an average of 1.48 percent per year from 1900 to 1914, with the average growth rate rising to 2.3 percent per year from 1914 to 1930 (Rees 1960, p. 3; U.S. Bureau of Census 1975, p. 211). Average annual earnings for all manufacturing workers rose at a slower pace of 1.43 percent annually from 1900 to 1929. The growth in annual earnings was slower than the 2.35 percent per year growth between 1948 and 1973, but substantially more rapid than the 0.46 percent annual growth experienced from 1973 into the early 1990s (Goldin forthcoming, p. 11). One reason that annual earnings grew more slowly than hourly earnings, was that workers succeeded in obtaining a shorter workweek. Average hours worked per week fell from around 59 in 1890 to 48 in 1929 (Robert Whaples 1990b, pp. 33_35).

4. Most though not all studies detect compensating differentials for living conditions and worker safety. All else equal, worse jobs paid more:

   Most of the empirical tests for labor markets circa 1900 suggest that workers received at least partial compensation for negative features of employment. Table 4 summarizes estimates of compensating differentials for accident risk, unemployment risk, risk of illness, cost-of-living differences, and postaccident benefits.

   For example:

   Rosenbloom (1996) finds that manufacturing wages were higher in cities with higher cost of living, higher housing costs, and possibly overcrowding, but he finds no evidence of compensating differentials for differences in infant mortality rates or weather.

5. Regulation can’t explain much of this progress, because so little regulation actually existed throughout this era!

   Reformers at the turn of the century proposed much of the workplace legislation we have today. They succeeded in obtaining limits on child labor and on working time for women, minimum wages for women, and safety legislation and workers' compensation laws. However, minimum wage laws for men, unemployment insurance laws, mandatory health insurance, old-age pensions, and state insurance for workers' compensation either failed to pass or were adopted in only a small number of states.

   

   The Pullman Company

6. The regulations that did pass were often pushed by employers who were already voluntarily following the proposed regulations! Why? To hobble the competition, naturally.

   Other laws, like child labor laws and safety legislation, codified the practices of a significant sub-group of employers. These employers may then have supported the legislation to ensure that the remaining employers followed suit, which reduced any competitive advantages gained by less progressive employers.

7. So-called “company towns” were fine by the standards of their time. Yes, the workers were isolated once they arrived, but employers had to offer a competitive package to attract workers in the first place. And as usual in business, reputation mattered. You won’t recruit many new workers if your existing workers write home to warn, “Life is terrible here.”

   The mobility of the miners also readjusts our picture of why company stores and mines may have developed. Were these designed to limit mobility and prevent unionization, or did they have other purposes? Fishback (1992, 1992b) reexamined the issue and suggests that the company town was a solution to the problem of isolated mines from which both workers and employers gained. Companies generally did not charge monopolistic rents. Rents on housing were similar to or lower than those in many cities, while the internal rates of return on housing investments were similar to those on alternative investments. By owning housing, coal employers could eliminate the transaction costs of contracting with independent housing agents and deny them the quasi-rents from charging high housing prices in an isolated area. Employers owned housing in part to limit the workers' ability to block new hiring by staying in the house, particularly during strikes. Yet workers also had incentives not to own homes near an isolated mine. Renting allowed the worker to avoid giving the employer short-run monopsony power over his labor and to escape capital losses in a risky industry. The timing and location of coal company towns is consistent with this analysis. Because collective action and strikes were ubiquitous in union and nonunion areas, we might expect that all mines would have sought to establish company housing if the primary goal was to limit unionization. Yet there were a number of coal areas where company housing was not dominant. Company towns were found in areas where there was less agricultural activity, smaller communities, and more rapid growth in coal mining employment.

8. Fishback goes out of his way to acknowledge when labor market performance was anything less than stellar. In his view, market forces largely eliminated discrimination within jobs, but not between jobs:

   Competition among employers eroded the most obvious form of discrimination, unequal payment in unskilled and entry-level jobs. However, the competition was not powerful enough to eliminate the discrimination that contributed to the low proportions of Black workers in more-skilled and high-wage jobs.

   Even so, Fishback carefully notes that standard regressions fail to distinguish market from government discrimination, and that government discrimination was a major force:

   The impact of most forms of governmental discrimination cannot generally be measured in straightforward labor market studies. In many cases the impact shows up in the residuals assigned to racial discrimination, as well as the coefficients on some human capital variables… Competition was not as effective at eliminating discrimination that prevented Blacks from obtaining the skills that would allow their advancement, in part because there were information costs about worker productivity that were not easily overcome and in part because the competitive forces were not effective at eliminating governmental discrimination.

If you read his entire article, Fishback’s tone is Panglossian rather than staunchly pro-market. While labor markets worked fairly well in the past with far less regulation than we have today…

speculation about the elimination of labor regulation is probably moot. The political economy section of this essay shows that many of the progressive-era labor regulations were beneficial both to workers and a significant subset of employers. Given that many employers as well as workers have a stake in the set of regulations that have been put in place, it is unlikely that we will see drastic regulatory changes.

While I’m a fan of this article — and Fishback’s work generally — I would have ended on a radically different note. When the normal historical view is grossly in error, I highlight the normal view, then clearly state that it is grossly in error. Something along the lines of:

The performance of U.S. labor markets prior to the Progressive Era and the New Deal was not merely excellent; it was practically unprecedented. Markets with token regulation combined textbook static efficiency with rapid improvements in productivity, wages, and working conditions. The dystopian perspective promoted by the vast majority of labor historians reflects not the facts, but dogmatic anti-market ideology. Labor market regulation is, of course, extremely popular, and that is unlikely to change anytime soon. But the reason labor market regulation is popular is that most people, left and right, casually accept Dickensian and Marxist misconceptions about what laissez-faire means for workers. If economic historians care about our own value-added, we should make discrediting these misconceptions a top priority. Laissez-faire for labor worked well in its heyday, and would work even better today.