This paper develops and tests a unique model of asymmetric employer learning. The previous literature on asymmetric learning assumes that a worker's employer is perfectly informed while outside firms possess only public information. This paper relaxes that assumption, allowing firms to profitably bid for employed workers under conditions that were not profitable in previous models. The model in this paper is the first in the literature to predict either wage growth without promotions or mobility between firms without firm- or match-specific productivity. The bidding through which firms compete for a worker produces a sequence of wages that converges to the current employer's conditional expectation of the worker's productivity. This convergence of wages allows the model to be tested using an extension of existing work on employer learning. Wage regressions estimated on a sample of men from the NLSY produce strong evidence of asymmetric learning.