There has been renewed interest in revitalizing manufacturing, yet policy often confronts a circular challenge: firms hesitate to expand because they cannot reliably find suitably skilled workers (e.g., STEM-trained), while workers are reluctant to acquire those skills when jobs remain limited. This raises a policy question: intervene at the firm margin or the worker margin, or both? We study this question by extending Acemoglu and Shimer (1999) to a two-sector open-economy. The key friction is capital holdup: firms invest upfront to create jobs, but sunk investment weakens their wage bargaining positions, discouraging investment ex-ante. Because manufacturing is more capital intensive, holdup is more severe, leaving manufacturing employment inefficiently low. In the calibrated model, this inefficiency-induced industrial employment shortfall is about 1 percent of total employment – roughly one-fifth of LAC-East Asia gap. When Hosios condition holds, the optimal policy can be solely on the firm side: an investment subsidy financed by an employment tax on firms. When Hosios condition fails, an additional wedge distorting workers’ sectoral choices emerges, and targeted training subsidies become welfare-improving.