Skip to content

The Determinants of U.S. Labor Disputes

Literature › Works

Peter C. Cramton; Joseph S. Tracy (1994). The Determinants of U.S. Labor Disputes. Journal of Labor Economics 12(2). University of Chicago Press (Society of Labor Economists / NORC).
Type journal article
Scope Comparative and methods
Method econometric
Period studied 1970–1989
Regions United States
Themes unions strikes, comparative theory, methods data sources, labour markets wages
Citation key cramton1994

Argument

The paper presents a strategic wage-bargaining model with private information in which a union facing an expired contract chooses between two threats, a strike or a holdout (working under the expired contract). Strike incidence is driven by shifts in dispute composition between holdout and strike, which respond to the real wage under the prior contract and to outside employment options. Overall dispute incidence is driven mainly by uncertainty over the firm's willingness to pay. The authors calibrate a benchmark model to US contract data and then estimate logit models of dispute incidence and composition for 1970-89. The model's key predictions are supported, but the associations weaken after 1981, which the authors interpret as the possible effect of the threat of permanent replacement workers.

Evidence and method

BLS data on 5,002 major US contract negotiations (bargaining units over 1,000 workers) from 1970 to 1989, with a logit sample of about 1,100-1,200 negotiations. Wage profiles are built from BLS Current Wage Developments and BNA base wages. Other inputs: stock-return uncertainty measures, BLS state and industry employment series, import penetration data from Abowd and Lemieux, union density, and Compustat capital intensity.

Findings

  • Most US contract disputes are holdouts rather than strikes, so strike data alone cover only a small share of dispute activity. (p. 11)

    Holdout 47 32 Strike 10 27 Dispute 57 37

  • The model implies that composition shifts between strike and holdout are highly sensitive to the relative attractiveness of the threats, while dispute incidence is nearly inelastic. (p. 12)

    The most prominent feature of the table is how sensitive the composition of disputes is to changes in the parameters.

  • The model's predicted opposite movements in strike and holdout incidence appear in the time series. (p. 12)

    In 65% of the years, an above (below) average holdout incidence was coupled with a below (above) average strike incidence.

  • The model predicts that the union strikes only when the current wage is below a threshold, and holds out otherwise. (p. 9)

    If w0 < zw, the union strikes; if w0 2 iv-, the union chooses to hold out

  • Higher real wage growth under the prior contract shifts the composition of disputes away from strikes, as predicted. (p. 25)

    A 1 standard deviation increase in the real wage measure is associated with nearly a 3 percentage point shift in the composition of disputes away from strikes.

  • Disputes are more likely in tight local labour markets and after little real wage erosion. (p. 25)

    We find that disputes are more likely to occur in tight local labor markets.

  • The link between firm-specific uncertainty and disputes holds before 1982 but disappears afterwards. (p. 24)

    post-1981 there is no evidence of a positive connection between variability in excess returns and disputes.

  • There is no evidence that inflation uncertainty raises labour disputes, contrary to earlier work. (p. 24)

    we find no evidence of a positive association between inflation uncertainty and labor disputes for either the 1970s or the 1980s.

  • The responsiveness of threat choice to real wage changes shifted after 1981, consistent with a replacement-worker effect, though the overall composition did not shift clearly away from strikes. (p. 28)

    there is strong evidence that the responsiveness of the threat choice to the underlying economic environment changed after 1981

  • Bargaining unit size is the largest single correlate of dispute incidence. (p. 24)

    A 1 standard deviation increase in the size of the bargaining unit is associated with about a 30 percentage point increase in the likelihood of a dispute.