Research on employer learning has provided important insights into the dynamic process that determines individual wages, especially during the early part of a worker's career. However, the recent evidence on the absence of employer learning for college graduates by Arcidiacono et al. (2008) and results that economic conditions at labor market entry have persistent effects on wages (for example Oreopoulos et al. (2008)) cast doubt on the model’s validity. This paper extends the employer learning model with the theory of rational inattention introduced by Sims (2006). In the model firms optimally allocate attention (=information processing capacity) to learning about the productivity of different worker groups. I find that firms allocate more attention to learning about the productivities of workers who have a higher impact on profits. Furthermore, firms learn about workers’ productivities as quickly as possible. Taken together these results resolve the discrepancy between the data and the employer learning model.