Abstract: The goal of this paper is to quantify the importance of imperfect competition in the U.S. labor market by estimating the size of rents earned by American employers and workers from ongoing employment relationships. To this end, we construct matched employer-employee data by combining the universe of U.S. business and worker tax records for the period 2001-2015. Using this panel data, we provide two empirical findings on the role that firms play in the wage determination in the U.S. The first finding is that idiosyncratic productivity shocks to a firm transmit significantly to the earnings of its workers. The second finding is that little of the variation in earnings is due to workers being employed in different firms. Estimating a two-way (worker and firm) fixed effect model, we find that firm effects explain no more than 4 percent of the variation in workers' earnings. To interpret these two findings, we develop a model of the labor market where multiple employers compete with one another for workers who have heterogeneous preferences over non-wage job characteristics or amenities. These heterogeneous preferences give rise to imperfect competition and rents. The model suggests a significant amount of rents and imperfect competition in the U.S. labor market. Comparing these worker rents to those earned by employers suggests that total rents are divided relatively equally between firms and workers. The model also reveals that the finding of small firm effects do not imply that labor markets are competitive or that rents are negligible. Instead, firm effects are small because productive firms tend to have good amenities, which pushes down the wages that these firms have to pay.