Community College Returns for Displaced Workers

The study begins by highlighting the significant and persistent earnings losses experienced by high-tenure displaced workers. This long-term economic consequence has prompted policymakers to substantially increase funding for retraining programs, many of which are delivered through community colleges. The research emphasizes the need to evaluate the effectiveness of these retraining initiatives in mitigating these substantial earnings losses. Existing literature is reviewed, acknowledging studies showing significant long-term earnings declines after job displacement and the limitations of prior research on this specific population. This includes the observation that public sector-sponsored training for disadvantaged individuals may not be directly comparable due to differences in prior work experience and employability. The existing literature on returns to schooling is also acknowledged as potentially inapplicable because displaced workers have unique characteristics, such as being older and having extensive work experience outside of formal education. The study aims to directly address the lack of robust evidence on the impact of community college retraining for displaced workers, providing a more relevant analysis tailored to the specific circumstances of these individuals. This will better inform policy decisions related to resource allocation for retraining programs and their efficacy in offsetting job displacement's financial consequences.

To assess the effectiveness of community college retraining, the researchers linked administrative earnings records with community college transcript data for workers displaced in Washington State between 1990 and 1994. The sample encompassed approximately 97,000 displaced workers, with about 16,000 having earned community college credits by 1996. Low-tenure displaced workers and those with low labor market attachment were excluded to ensure the focus remained on the long-term impact of job loss on experienced workers. Data sources included quarterly earnings and hours worked (1987-1995), demographic information, and details about their displacement (e.g., job tenure, education level, industry). Community college records provided information on courses, credits earned, and grades. This robust dataset allowed for a comprehensive analysis of the relationship between community college credits, course type, and subsequent earnings. The study population comprised mainly prime-aged adults, many of whom had prior college experience and represented a wide range of industries, offering a more complete and realistic understanding of the typical individual undergoing retraining after job loss.

The analysis employed sophisticated regression models to quantify the impact of community college credits on earnings. Several key statistical considerations were addressed: the post-schooling transition period where earnings might be temporarily depressed, potential deviations from strict proportionality between credits and earnings gains, and worker-specific unobserved heterogeneity. The researchers used fixed effects and time trends to manage the unobserved heterogeneity effectively. Their preferred specification revealed a significant positive impact of community college schooling on long-term earnings. For men, an academic year's worth of credits increased earnings by approximately 9 percent, while for women, the increase was around 13 percent. The study acknowledges that this average effect masks considerable variation depending on the type of course taken. These results demonstrate the significant potential of community college retraining in improving the long-term economic prospects of displaced workers, particularly when tailored to specific skills needs as identified by the differences in returns between course types. It highlights the complexity of evaluating training programs and the importance of using appropriate econometric techniques to accurately assess their impact.

A crucial finding is the substantial difference in the effectiveness of different course types. Courses categorized as Group 1 (more technically oriented vocational courses, math, and science) led to significantly larger earnings gains (14% for men, 29% for women) compared to Group 2 courses (less technically oriented vocational, humanities, and social science courses). The Group 1 earnings increase was approximately one-third attributable to wage rate increases, with the rest stemming from increased hours worked. Conversely, the impact of Group 2 courses was much smaller, possibly even zero. This critical finding underscores the importance of course selection and its potential relevance for policy. The high returns of technically oriented training highlight the need to direct resources towards programs that develop skills in high-demand sectors. The findings regarding Group 2 courses highlight the critical need for careful curriculum design in community college retraining programs to align better with market needs and ensure cost-effectiveness.

The study acknowledges several critical statistical challenges in accurately assessing the impact of community college retraining on displaced workers' earnings. First, the researchers address the need to account for a potential post-schooling transition period. During this time, earnings may be temporarily depressed before rising to a new, higher level. Failing to account for this transition period could lead to underestimation of the long-term benefits of retraining. Second, the study challenges the assumption of strict proportionality between credits earned and earnings gains. Previous research often assumes a linear relationship. This study relaxes this assumption, finding that the earnings increase from the first credit is disproportionately higher than subsequent credits, indicating a non-linear relationship that requires careful modeling. The initial credit's impact may be driven by factors such as improved job search networks, rather than solely the knowledge gained from coursework. Thus, the focus shifts to the per-credit earnings increases associated with completing additional credits beyond the initial one. Finally, the researchers emphasize the importance of modeling worker-specific unobserved heterogeneity. This is addressed by incorporating worker-specific fixed effects and, in some specifications, worker-specific time trends into the regression models, to effectively control for individual differences that might confound the relationship between retraining and earnings gains.

The core of the methodological approach involves the construction and application of regression models to analyze the impact of community college credits on earnings. The study carefully considers several model specifications. To address the post-schooling transition effects, interaction terms between a post-training indicator variable and time since the completion of the training were incorporated into the models. This allows for capturing both the short-run and long-run effects, accounting for temporary earnings dips immediately after leaving school. To capture the non-proportional relationship between credits and earnings, the models were specified to include both a binary indicator for having enrolled in community college and a term proportional to the number of credits earned. The choice to focus on this 'hybrid' model follows from findings that the relationship between observable variables and the decision to enroll in training is stronger than the relationship between those variables and the number of credits earned, suggesting that the hybrid model offers more reliable estimates of the effects. Furthermore, the models incorporate time period fixed effects to control for macroeconomic factors influencing all workers and worker-specific fixed effects to control for time-invariant individual characteristics. Worker-specific time trends are also included in some models to account for individual-specific heterogeneity that changes at a constant rate over time. These methods ensure a more robust and accurate estimation of the causal impact of community college training on earnings.

The researchers acknowledge potential biases and limitations in their models. The initial estimates indicating significant earnings increases even with minimal credits might be attributed to self-selection bias. This means more motivated and able workers may have been more likely to enroll in community college in the first place. The inclusion of worker-specific time trends helps alleviate the downward bias stemming from the fact that displaced workers who obtain community college credits, particularly Group 2 credits, tend to have slower earnings growth even before beginning training. The inclusion of worker-specific time trends is crucial to control for unobserved heterogeneity which, if ignored, would lead to biased estimates of the effect of training. The study also examines non-linear effects by investigating the possibility of non-monotonic or non-linear relationships between credits earned and subsequent earnings. Through careful attention to these methodological details, the study attempts to mitigate potential sources of bias and provide more accurate estimates of the true impact of community college retraining on the earnings of displaced workers, improving the reliability and validity of the overall findings.

The study's key finding centers on the significant variation in earnings outcomes associated with different types of community college courses. The analysis distinguishes between two groups of courses: Group 1, consisting of more technically oriented vocational courses, math, and science; and Group 2, encompassing less technically oriented vocational courses, humanities, and social sciences. The results reveal a substantial difference in the impact of these course types on displaced workers' earnings. An academic year of Group 1 courses yielded significantly higher earnings gains, averaging a 14 percent increase for men and a remarkable 29 percent increase for women. This substantial difference in returns highlights the importance of aligning retraining programs with the demands of the labor market. The study suggests that focusing resources on technically oriented training may be a more effective strategy for increasing earnings among displaced workers. In contrast, Group 2 courses showed much smaller and potentially zero returns to training, indicating that not all retraining efforts yield equivalent economic benefits. This underscores the importance of targeted retraining programs aligned with market demands for skills to maximize their impact on post-displacement earnings.

The research goes beyond simply reporting overall earnings gains and decomposes the impact of Group 1 courses on earnings into its constituent parts: wage rates and hours worked. The findings indicate that approximately one-third of the earnings increase associated with Group 1 courses can be attributed to higher hourly wage rates, while the remaining two-thirds resulted from an increase in hours worked. This decomposition provides valuable insights into the mechanisms through which retraining leads to improved economic outcomes. The increase in wage rates suggests that Group 1 courses indeed enhance workers' productivity and make them more valuable to employers. The concurrent increase in hours worked might reflect either increased employment probability or workers' willingness to work more hours at the higher wages afforded by the acquired skills. The difference in the effect on hours versus wages between men and women is also worth noting. For men, there's an effect on both, suggesting the retraining leads both to better-paying jobs and the incentive to work more. For women, it's primarily a wage increase effect, suggesting the retraining opens up opportunities for higher-paying jobs while the number of hours worked remains stable. This detailed breakdown contributes to a more nuanced understanding of how community college retraining affects labor market outcomes.

In contrast to the substantial positive effects of Group 1 courses, the analysis reveals that Group 2 courses have a significantly smaller, and possibly negligible, impact on displaced workers' earnings. The point estimates in models with worker-specific time trends suggest a modest 4 percent increase in earnings for an academic year of Group 2 credits. However, this increase is not statistically significant, implying that these courses may not contribute meaningfully to earnings growth. This finding may reflect several factors. The skills taught in Group 2 courses may not be as in-demand by employers or may not effectively enhance workers' productivity. It may also indicate a mismatch between the skills taught and the skill requirements of available jobs in the market. Alternatively, workers may have already possessed the knowledge and skills covered in the Group 2 courses, diminishing the marginal benefit of the retraining. The lack of significant earnings gains from Group 2 courses underscores the importance of carefully tailoring retraining programs to the specific skills needed in the labor market. Investing in training programs that do not enhance marketable skills is unproductive and represents a misallocation of resources.

The study undertakes a cost-benefit analysis to evaluate the economic efficiency of community college retraining for displaced workers. The analysis focuses on the social benefit-cost ratios and internal rates of return (IRRs) for the different types of community college courses. The computation of social benefits considers only earnings gains, analyzing the expected benefits for a representative participant working for 25 years and completing the average number of Group 1 and Group 2 credits. Future earnings gains are discounted at a rate of 4 percent. The cost of a year of community college schooling is estimated at $8,000 per full-time equivalent student, reflecting tuition and public sector expenditures. The study further accounts for the welfare cost or deadweight loss associated with taxes used to subsidize community college education, estimated at approximately $3,000. This comprehensive approach considers both direct costs and indirect welfare costs associated with the public funding of community college retraining programs, thus providing a complete picture of their economic impact.

The cost-benefit analysis yields significantly different results for Group 1 and Group 2 courses. For Group 1 courses (technically oriented vocational, math, and science), the estimated internal rate of return is approximately 10 percent for men and 17 percent for women. This positive return indicates that these types of training programs are a productive social investment, generating benefits that exceed their costs. This positive return is influenced by the large earnings gains associated with Group 1 courses, which are reflected in the calculation, together with the direct and indirect costs. In contrast, the internal rates of return for Group 2 courses (less technically oriented) are estimated to be negative for both men and women. This suggests that these courses, on average, do not generate sufficient earnings gains to offset their costs, indicating a misallocation of resources for this type of training for the sampled displaced workers. The substantial difference in returns between Group 1 and Group 2 courses emphasizes the importance of directing resources towards training programs that align with labor market demands and develop high-demand skills. The findings highlight the need for careful evaluation and targeting of retraining initiatives to maximize their cost-effectiveness.

The study's sample is drawn from a comprehensive database of individuals who lost their jobs in Washington State between 1990 and 1994, filed unemployment insurance claims, and had at least six quarters of job tenure with their previous employer. The data comprises approximately 97,000 workers. The researchers utilize two types of administrative records: quarterly earnings and unemployment insurance benefits data from 1987 to 1995, and machine-readable transcripts from Washington State's 25 community colleges for the period from Fall 1989 to the end of 1995. The dataset includes rich demographic information and details concerning displacement, such as job tenure, education level, industry, and geographic location. Workers with low job tenure and low labor market attachment were excluded. The approximately 16,000 workers who earned community college credits after job displacement formed a key subset of the sample. This rigorous data collection strategy ensures a robust and reliable dataset for analyzing the effectiveness of community college retraining among displaced workers. The administrative data provided a comprehensive overview of the economic outcomes of the individuals studied. This combination of unemployment insurance and community college data provided a rich data set for analyzing the effectiveness of retraining programs for displaced workers.

The study sample reveals characteristics of the displaced workers who enrolled in community college retraining programs. These workers were typically prime-aged adults, with many having prior college experience. The average age was approximately 36 for men and 37 for women. About half of both male and female participants had some prior college education. A noteworthy observation was the significant representation of women from the non-goods producing sector of the economy, with approximately 70 percent having lost jobs in this sector. The economic context of the study period is important. Washington State, like much of the nation, experienced economic downturn in 1990, with sectors like aerospace and wood products particularly hard hit. While the state legislature increased community college funding for displaced workers in 1993, federal subsidies were not yet substantial, meaning most participants attended college at their own expense. This suggests the studied individuals were motivated to pursue retraining independently, irrespective of significant external financial assistance. The combination of demographic features and economic context provides valuable context for interpreting the study's findings.

The analysis explores earnings trends both before and after job displacement. Figure 1 illustrates the real quarterly earnings (1995 dollars) of displaced workers who earned community college credits and a comparison group who did not. The data reveals a decline in real quarterly earnings in the two years prior to displacement, followed by a sharp drop after job loss. Even after four years, earnings remained approximately $2,000 below pre-displacement peak levels. The analysis distinguishes between workers who earned Group 1 and Group 2 credits. While Group 1 participants’ earnings were similar to the comparison group, Group 2 participants’ earnings were consistently lower both before and after job loss. This difference highlights the importance of carefully considering the type of training undertaken. It also illustrates the significance of not making simple comparisons between those who received training versus those who didn't, since there could be selection bias at play. However, simple comparisons do not account for other factors that might influence earnings trajectories. This emphasizes the necessity of using econometric models that control for other factors such as gender, race, age, prior education, and industry of origin, as used in this study.