Abstract This paper utilizes a new panel data set of workers employed by the Canadian Pacific Railway between 1904 and 1929 to analyse the incidence of long-term employment and reasons for changes in job durations after World War I. The hazard function estimates indicate that individuals have only a small probability of staying for more than ten years at the firm, although at any point in time long-term workers constitute a sizable fraction of employees. Increasing post-war spell lengths reflect the changing composition of the workforce rather than differences in macro-economic conditions or changes in employment relationships. Specific human capital and job matching were also important: Long-term employees were shielded from both firm-specific and economy-wide downturns.
B. Hamilton, M. Mackinnon