Apprenticeships have regained prominence among employment policymakers and scholars during the recent recession for promising a path to labor market success for graduates who do not aim to attain a higher degree. In this voluntary and cooperative arrangement, employers play a major role in operating and funding the training, whereas unions secure the human resource pool and government monitors the training outcomes. As a employer-sponsored program, this characteristic makes the traditional difference-in-difference evaluation for trainee wage not fully applicable in evaluating apprenticeship performance, because from an employers perspective, factors like apprentice retention (Glover & Bilginsoy, 2005; Steward & O’Brien-Turco, 2010) and appropriation (from competitor employers) (Lerman, Eyster, & Chambers, 2009) may outweigh the apprentice's human capital accumulation in determining the success of the program. These two factors significantly influence the willingness to further invest in apprentices for employers, which is the key to the sustainability of apprenticeship arrangement. Evaluation scholars also noticed this nuance when comparing different types of training programs (Heinrich & Holzer, 2011; Van Horn, King, & Smith, 2011), however, we require different analytic angles to identify effective strategy.

To cope with apprentice attrition and appropriation problems, employers have developed cooperative relationships or work with trade unions when initiating apprenticeship programs. Past research (Bilginsoy, 2007) has identified open-shop multiple-employers strategy to be successful in raising the completion rate of apprenticeship program in the construction industry at the State of Washington. This research aims at further applying theory from common-pool resource management (Ostrom et al., 1994)  to examine the effectiveness of cooperative strategy in managing the apprentices resource. Four types of strategies will be tested for their difference in apprentices' survival (completion) rate. Multiple data sets will be used in this paper. The core data include the entire '00-'01 entering cohort for the registered apprentices in Ohio. Unemployment Insurance data from Ohio Department of Job and Family will be used to link with this group of apprentices to track their employment situation over a decade. The strengths of applying these longitudinal data sets are twofold: it can allow researchers to examine the impacts that multiple exposures in labor market have on individual behavior, while insulate the impact of the business cycle has on this behavior (Muehlemann, Wolter, & Wüest, 2009). Event history analysis technique can be a good fit in modeling the change of various intermediate and time-varying factors presented in these data sets. It can deal with the changing effects of these factors have on the final outcome, which is hard to measure and a great challenge to researchers. In sum, this research aims to examine the effectiveness of different strategies for the apprenticeship completion. By analyzing the empirical data, we can not only lay out an effective strategy for apprenticeship management, but also shed light on managing other public-private partnership program.