How are wages and employment of manufacturing workers a¤ected by globalization and, in particular, by the sourcing strategies of their employing rms ? What is the role of unions in this process ? I attempt to answer these questions using unique French data that matches rm-level information on imports, value-added, employment,..., rm-level data on unions and negotiations, with individual worker data; at a unique moment the rst years of implementation of the Single Market Program (SMP) within the European Community (1986-1992) when French manufacturing formerly protected and high-rents sectors, faced increased competition and outsourcing options. A simple bargaining model, particularly well-suited to the French institutional setup, allows me to capture the mechanisms by which outsourcing of nal goods (o¤shoring) can directly affect wages: changes in the rms ability to pay (as measured by the quasi-rent, the total pie shared between the workers and the rm), mostly by altering the rmsthreat points. The model shows that, when there is imperfect competition on the product market, rms that face strong unions should outsource more of their production of nal goods because such o¤shoring causes a decrease of their quasi-rent and disciplines workers. However, o¤shoring does not necessarily decrease wages. Then, estimation results show that, indeed, o¤shoring decreases the size of quasi-rents. A regression discontinuity approach shows that bargaining institutions a¤ect outsourcing. Employment is also shown to decrease when o¤shoring increases. IV estimates of the structural wage function derived from the model show that rms that bargained on employment faced strong unions (the bargaining power being equal to 0.5, workers employed at these rms received in average 50% of the quasirent). All other rms faced weaker unions with virtually no bargaining power. Then, I show that the presence of such strong unions led these rms to increase o¤shoring with an associated decline in employment and rents, as predicted by the model. As a result, in those rms where rents were initially high, unionsstrength appears to have back red. JEL codes: F3, F4, J30
Stored only when both papers are in the corpus — 26 citations exist upstream.