A number of papers have documented the finding that foreign firms pay higher wages and, to the best of our knowledge, no study has found the opposite. There are (at least) two approaches to explaining these finding that we will compare here. One we will call the worker learning model, and the other we will refer to as the heterogeneous-worker model. We think it is fair to say that empirical evidence on the former is substantial but that the model’s theoretical foundations are under developed. There also exists evidence supporting the heterogeneous-worker model, with good theory behind it.
The two approaches are by no means mutually exclusive or competing hypotheses. The first purpose of this paper is to develop a tractable theory model that allows for each approach to emerge as a special case. The second purpose is to conduct an empirical investigation that assess the relative contribution of each approach to explaining the foreign firm wage premium.
A good statement of the heterogeneous-worker model is in Yeaple (2005), with some similar components in Ekholm and Midelfart (2001) and Manasse and Turrini (2001). Yeaple assumes that an economy consists of a continuum of workers with different skill levels and ex ante identical firms that can choose one of two technologies, differing in their scale intensity. More skilled workers have an absolute advantage in both technologies, but a comparative advantage in the more scale-intensive technology. In equilibrium, the firms that choose the more scale-intensive technology are larger and employ more-skilled workers and hence pay a higher average wage. Firms using the scale intensive technology are also the exporting firms; this could easily be extended to the establishment of foreign affiliates, with the implication that, within a given country, affiliates of foreign multinationals pay higher wages than the average wage across all domestic firms (which include both scale-intensive firms that produce abroad and strictly local firms that have lower average skill levels and pay lower average wages).
Our model builds on Melitz’s (2003) (also Helpman et al., 2004) model of industry structure with heterogeneous firms, and blends this with the learning-on-the-job models of Ethier and Markusen (1996), Markusen (2001), Fosfuri et. al. (2001), and Glass and Saggi (2002). In our model, foreign firms on average are larger and have higher productivity than domestic firms. But it also provides the hypothesis that, corrected for firm size, foreign firms are not more productive than domestic firms.
Our model also captures the essence of Yeaple’s approach, allowing firms free entry into two alternative technologies, along with two types of workers differing in skills. One group of scarce highly-skilled workers is entirely employed in the scale-intensive, high-productivity technology. Another group is employed in both technologies. The latter workers are ex ante identical and earn the same present value of income over a two-period working lifetime. Skills learned in the first period when employed by a high-productivity firm are transferable to other high-productivity firms.
