Ridderprint

Chapter 6 134 organizations may want to explore more inclusive approaches to talent management (e.g., Meyers, De Boeck, & Dries, 2017). Third, organizations could use STIA to prepare young talents for future global leadership, as long as close attention is paid to the repatriation process. Otherwise, STIA may merely delay or even boost talent loss. Rigorous workforce, career, and succession planning should help to identify suitable positions for repatriates. Better expectation management could prevent false expectations regarding the benefits of STIA for career advancement. Here, employees could take on a proactive role themselves in revealing their personal expectations. On top of this, organizations could experiment with shorter assignments instead of sending the same employees on long or consecutive STIA as our results and others (e.g., Kraimer et al., 2009) hint this may improve talent retention. Finally, organizations may seek to replicate the current analysis within their own population to examine how, when, andwhere they can improve the sustainability of talent pipelines in their own context. The capability to pinpoint when and why turnover risk is high may permit timely intervention and more sustainable investment in young talent. Here, extensions of the current model with internal data – for instance, on selection tests, compensation, development, and organizational changes – and external information – such as (labor) market characteristics – could prove valuable. 6.6.6 Future Research and Limitations This study is subject to several limitations. First, despite our longitudinal data, establishing causality remains difficult. For instance, instead of low performance causing voluntary turnover, both may be the result of trainees’ early withdrawal behaviors from the job and organization. Additionally, the strong and complex relationships among employees’ performance, potential, international experiences, and other HRM variables (e.g., salary, promotions) warrant more focused longitudinal research to establish their precise effects on turnover. Second, archival data have their limitations. Inherently, employees do not have performance and potential ratings until they undergo the respective evaluation or assessment. This implied that the HRIS lacked certain data for the early period of all trainees’ careers. We opted to replace these early missing data points with the modal rating in each organization; an alternative was to disregard all observations during this early period. While the results of our hypothesis test remained the same, either approach introduced its own potential biases. A replication of our study using other data sources may be valuable. Third, our results demonstrated that turnover is affected by contextual factors and a more detailed exploration is warranted here. While we controlled for direct country- and organizational-level effects on turnover, we did not model country-level moderations related to labor market characteristics or cultural preferences for and attitudes to talent strategies (Farndale, Scullion, & Sparrow, 2010). Similarly, STIA may have different implications because of home and host country locations or the differences between them. Future research should consider more closely how turnover and its predictors differ across contexts.