Canadian Worker Earnings Losses After Job Displacement

The study addresses the lack of Canadian data on long-term earnings losses following worker displacement, a topic extensively studied in the US but with limitations. Previous US studies, often using regional data or smaller samples like the Panel Study of Income Dynamics, show substantial and persistent earnings losses, but the evidence is not recent and lacks a nationally representative sample. Canadian studies have only looked at short-term wage changes immediately after displacement, failing to quantify long-term impacts. The research aims to fill this gap using a large, nationally representative sample of Canadian workers to assess the magnitude and duration of earnings losses. This is crucial because Canada's Employment Insurance program provides only one year of coverage. The long-term consequences of displacement bear disproportionately on workers, influencing not only individual earnings but also family stability, and retirement income. Policymakers need this information to develop effective compensation policies or programs to mitigate the impact of resource reallocation in the Canadian economy due to technological changes, international trade, and consumer preferences.

The core data source is Statistics Canada's Longitudinal Worker File (LWF), a 10% random sample of Canadian workers tracked from 1983 to 2002. Linked to the Longitudinal Employment Analysis Program (LEAP), the LWF offers a unique longitudinal perspective on firm births and deaths, enabling precise identification of workers displaced by firm closures. The study uses two definitions of displacement. The first rigorously defines displacement as permanent layoffs in year 't' from firms that closed in that year. The second broader definition includes workers experiencing mass layoffs (at least 50 employees in a firm with a 30% employment drop between year 't-4' and 't+1'). The data also incorporates Records of Employment (ROEs) for separation reasons, T1 and T4 tax files. This allows for a comparison group of workers who never experienced displacement during the observation period. While the LWF lacks details like education levels, its size and national representativeness offer a robust dataset for analyzing both short-term and long-term earnings losses. Econometric models, including fixed-effects models and those with worker-specific time trends, allow for flexible specification of age-earnings profiles, accounting for potential heterogeneity.

The study's principal finding is that long-term earnings losses after displacement are substantial, but significantly more so for high-seniority workers (those with five or more years of tenure). Consistent with US findings (Jacobson, Lalonde, and Sullivan, 1993), high-seniority men experience losses of 18-35% of pre-displacement earnings, while women face 24-35% losses. These losses persist for at least five years after displacement. Analyzing the two definitions of displacement – firm closures and the broader inclusion of mass layoffs – provides a comprehensive view. Results indicate that while average earnings losses from firm closures or mass layoffs are important, the losses for high-seniority workers are considerably greater. The impact of economic conditions is shown by the longer recovery time for workers displaced prior to the 1990-1992 recession. For all workers displaced through firm closures, the analysis shows significant long-term losses based on fixed-effect models. Models with worker-specific trends show that losses become insignificant after three years. The study also uses data to estimate the number of workers displaced, approximately 55,000 annually by firm closures and roughly double that when considering both firm closures and mass layoffs.

The study concludes that significant long-term earnings losses follow worker displacement, especially for high-seniority employees. These losses extend beyond Employment Insurance coverage, suggesting a need for policy interventions to mitigate the economic burden on affected workers. The magnitude of these losses is robust across various model specifications and definitions of displacement. The research emphasizes the substantial disparity in earnings recovery based on the economic climate at the time of displacement, with slower recovery during recessionary periods. The analysis highlights the importance of considering both firm closures and mass layoffs when assessing the scope of worker displacement and its impact. A key limitation is the lack of data on self-employment income, which might partially offset income losses from paid employment. Therefore the study primarily focuses on losses from paid employment.

The study's primary data source is Statistics Canada's Longitudinal Worker File (LWF), a 10% random sample of all Canadian workers covering the period from 1983 to 2002. The LWF's strength lies in its integration of data from four key sources: Records of Employment (ROE) files from Human Resources and Social Development Canada, which provide crucial information on worker separations, including the reason for separation (layoffs, quits, etc.); T1 and T4 files from the Canada Customs and Revenue Agency, offering comprehensive earnings data; and the Longitudinal Employment Analysis Program (LEAP) from Statistics Canada's Business and Labour Market Analysis Division, which provides longitudinal company-level data, including information on firm births and deaths. This multi-source integration is critical for accurately identifying and analyzing worker displacement. LEAP's tracking of firms allows for the precise identification of layoffs resulting from firm closures. This linkage between LWF and LEAP is a key feature, enabling rigorous measurement of displacement as a permanent layoff in year 't' from a firm that ceased operation in the same year. The 20-year longitudinal window of the LWF allows for a comprehensive examination of long-term earnings impacts. Although the LWF lacks information on education, occupation, or immigration status, its size and national representativeness offer significant advantages.

The study employs two distinct definitions of worker displacement to capture the nuances of job loss. The first definition focuses specifically on firm closures, defining displacement as a permanent layoff in year 't' from a firm that ceased operating during that year. This approach ensures rigorous identification of displacement directly linked to company closure. However, it omits job losses from mass layoffs in multi-establishment firms that don't result in complete firm closure. To address this limitation, a second, broader definition is used, encompassing both firm closures and mass layoffs. Mass layoffs are defined as permanent layoffs in year 't' in firms with at least 50 employees in year 't-4' and that experienced a 30% or greater employment drop between year 't-4' and year 't+1'. This broader definition acknowledges that significant job losses can occur even without the complete closure of a company, thereby providing a more complete picture of worker displacement. This dual approach allows for a comparison of results based on different definitions of displacement and for a more robust understanding of the underlying phenomenon. This methodological decision helps to mitigate sample selection bias, a known issue when focusing solely on individually based layoffs.

The analysis utilizes econometric models to quantify earnings losses, taking into account the complexities of age-earnings profiles and potential individual variations. The models allow for flexibility in specifying these profiles, recognizing that different workers may have varying intercepts (starting earnings) and slopes (earnings growth rates) over their careers. The study employs fixed-effects models and models incorporating worker-specific time trends. These approaches enable the researchers to control for unobserved individual heterogeneity that might influence earnings, providing a more accurate measure of the impact of displacement. The study uses two samples of displaced workers: a narrow sample of high-seniority workers (those with five or more years of tenure at the firm before displacement) and a broader sample including all workers displaced through firm closures or mass layoffs. The selection of high-seniority workers is justified by the concern that job loss in this group might involve significant loss of firm-specific skills and pose greater challenges in re-employment. Age restrictions (25-49 at the time of displacement) are imposed on the high-seniority samples to avoid confounding factors from early retirement. The study explicitly chooses not to control for subsequent displacements, focusing instead on the long-term impact of the initial displacement experienced between 1988 and 1997.

The analysis focusing on high-seniority workers (those with five or more years of tenure) reveals substantial and persistent earnings losses following displacement due to firm closures. Using both fixed-effects models and models that incorporate person-specific trends in age-earnings profiles, the study finds that high-seniority men experience long-term (five-year) earnings losses representing 19% to 28% of their pre-displacement earnings in a narrower sample and 25% to 34% in a broader sample. Similarly, high-seniority women experience losses ranging from 24% to 35% of pre-displacement earnings, depending on the sample and model used. These findings align with previous research by Jacobson, Lalonde, and Sullivan (1993) and highlight the disproportionate impact of firm closures on experienced workers. The substantial and enduring nature of these losses underscores the significant long-term economic consequences for this group. The fact that pre-displacement earnings for high-seniority males averaged $34,487 (narrow sample) and $34,715 (broader sample) provides context for the magnitude of these losses in dollar terms. Estimates of 5-year earnings losses range from $6,600 to $11,600 depending on the model and sample used.

Expanding the analysis beyond high-seniority workers to encompass all workers displaced due to firm closures reveals more nuanced results that vary based on the econometric model employed. Fixed-effects models suggest significant five-year earnings losses for both men (16% to 22% of pre-displacement earnings) and women (25% to 34%). However, models that allow for worker-specific time trends in age-earnings profiles show that earnings losses become statistically insignificant after three years. This discrepancy highlights the importance of considering the potential for varying age-earnings profiles among displaced and non-displaced workers. The study's findings underscore the need to account for worker heterogeneity and choose appropriate econometric models when interpreting the duration and magnitude of earnings losses. The substantial difference in findings between fixed-effects models and models with worker-specific trends emphasizes the importance of methodological choices in analyzing this type of data. The average 5-year earnings losses for women, as a percentage of pre-displacement earnings, are similar (24-26% in the narrow sample, and around 35% in the broader sample) regardless of the chosen model.

The analysis demonstrates that the speed of earnings recovery after displacement is significantly influenced by prevailing labor market conditions. A comparison of earnings trajectories for cohorts displaced before and during periods of economic expansion (e.g., comparing the 1989 cohort, displaced just before the 1990-1992 recession, with the 1997 cohort) reveals substantial differences. The 1989 cohort exhibits markedly slower earnings recovery, with earnings five years after displacement not exceeding pre-displacement levels. Conversely, the 1997 cohort shows a faster recovery. This highlights the importance of accounting for macroeconomic factors when assessing the long-term impacts of job displacement and indicates that results are sensitive to the period studied. The findings also reveal that even five years after displacement, the earnings losses for those displaced during the 1987-1992 period remain considerably larger than those displaced during 1993-1997. The difference is estimated to be between $5,300 and $5,800 for high-seniority workers, underscoring the influence of the business cycle on earnings recovery. The study incorporates province-year interactions in its estimation procedure to account for variations in labor market conditions across different regions and over time.

This section extends the analysis beyond firm closures to encompass a broader definition of displacement, incorporating both firm closures and mass layoffs. This is crucial because mass layoffs can cause substantial job losses without necessarily leading to the complete closure of a firm. The study's findings using this broader definition show that while average long-term earnings losses are considerable, they are still less severe than those experienced by high-seniority workers. Including mass layoffs significantly increases the estimated number of displaced workers. The analysis reveals that the average number of workers displaced annually (ages 25-64) nearly doubles when mass layoffs are included, rising from approximately 55,000 (firm closures only) to around 110,000 (firm closures and mass layoffs). This highlights the importance of considering mass layoffs when assessing the overall economic impact of job displacement. The findings suggest that while both firm closures and mass layoffs contribute to substantial earnings losses, the magnitude of these losses remains disproportionately high for high-seniority workers.

A key aspect of this analysis is the assessment of the robustness of the findings across different model specifications and definitions of displacement. The study compares results from fixed-effects models and models with worker-specific time trends. It finds that for high-seniority workers, the estimates of long-term earnings losses are relatively consistent across both models and definitions of displacement. This indicates a high degree of robustness. However, for all displaced workers (not just high-seniority), the choice of model significantly influences the estimated magnitude of long-term earnings losses. Specifically, using models with worker-specific trends reduces the magnitude of estimated long-term losses, especially when using the narrower definition of displacement (firm closures only). This underscores the sensitivity of results to model choices when analyzing all displaced workers. In contrast, the results concerning high-seniority workers are consistently robust regardless of the econometric model chosen. The broader definition of displacement, incorporating both firm closures and mass layoffs, proves more robust to variations in model specification, particularly when analyzing all displaced workers, suggesting that this definition may be conceptually superior for a comprehensive assessment of earnings losses from job displacement.

The analysis also reveals that the magnitude of earnings losses is significantly influenced by prevailing economic conditions at the time of displacement. Workers displaced during the 1987-1992 period (including a recession) experienced considerably larger earnings losses than those displaced during the 1993-1997 period. This difference is consistent across various samples and model specifications, highlighting the importance of business cycle considerations. High-seniority male and female workers displaced during the earlier period (1987-1992) experienced earnings losses five years post-displacement that were $5,300 to $5,800 higher than those displaced during the later period. For example, in the narrow sample, high-seniority males displaced between 1987 and 1992 experienced average long-term losses of approximately $11,300, compared to about $6,100 for those displaced between 1993 and 1997. The findings clearly emphasize the significant impact of macroeconomic factors on the long-term economic consequences of job displacement. The substantial difference in earnings losses underlines the crucial role of economic conditions in determining the severity and duration of the earnings impact from job displacement.

The study's core finding is the significant and persistent long-term earnings losses experienced by Canadian workers following displacement due to firm closures or mass layoffs. The magnitude of these losses is notably higher for high-seniority workers, with those having five or more years of tenure at their previous firm facing significantly greater impacts. The research shows that these losses are not merely short-term effects of unemployment; they persist for several years after the displacement event. The study highlights the substantial economic consequences for both men and women, particularly high-seniority individuals. For high-seniority workers, losses ranged from 18% to 35% of pre-displacement earnings for men and 24% to 35% for women. This underscores the need for policies that address these long-term financial impacts, going beyond the one-year coverage offered by Canada's Employment Insurance program. The study contributes valuable data on this critical area, which was previously under-researched in the Canadian context.

Despite the study's significant contributions, several limitations are acknowledged. Primarily, the absence of data on self-employment income in the Longitudinal Worker File (LWF) prevents a complete assessment of how displaced workers mitigate income losses. While the study sheds light on the recovery of income from paid employment, the potential for self-employment to offset earnings losses remains unexplored. This represents a significant area for future research. Furthermore, while the study establishes the existence of substantial earnings losses and their persistence, it does not delve into the mechanisms underlying the slow earnings recovery. Further investigation into factors like multiple job losses (as suggested by Stevens, 1997), industry mobility, firm size changes, and the effectiveness of various job search methods is warranted to gain a more complete understanding of the post-displacement experience. These limitations do not diminish the value of the study's findings, but they suggest directions for future research to further refine our understanding of long-term economic consequences of job displacement.