The Firm Boundaries of Capital-Skill Complementarity - Measuring Welfare by Matching Households Across Time
We argue production networks are critical to capture the skill bias of new machinery. Linking firm data on machine investments, buyer-supplier transactions, and worker skills in Belgium, we show adoption of new machines by manufacturing firms strongly increases the skills demanded from external suppliers. By contrast, the machines do not alter the skill mix of in-house employment. The skill bias of suppliers reflects machine integration and skilled inputs complementarity to automated production. We develop theory and provide supporting evidence that machine integration features scale economies, driving the outsourcing of these skilled services. Rising machine-labor ratios in manufacturing have increased the aggregate demand for skills by 10 percentage points, entirely driven by external suppliers. Our analysis reconciles recent firm-level studies focusing on in-house employment (Aghion et al. (2022), Curtis et al. (2022), Hirvonen et al. (2022)) with long-standing theories of capital-skill complementarities (Griliches (1969), Goldin and Katz (1998)).