In-house production versus specific forms of supplier governance: testing predictions of the global value chains model