Job Market Paper
We argue that productive firms share rents with workers only in occupations where workers have individual hold-up power. We present a model of wage determination where firms produce using a novel generalization of Kremer (1993)'s O-ring production function. Workers have individual hold-up power if (i) labor is organized into distinct, differentiated positions (ii) the output of positions is individually complementary or "critical" in the production process, and (iii) skills are position-specific, i.e., skills are acquired on the job and are not transferable across positions or firms. If output losses from an unfilled position are larger at productive firms, incomplete contracts and on-the-job search incentivize productive firms to pay differentially higher wages. We estimate individual worker hold-up power by occupation using the effect of worker deaths on firm profits in Danish administrative data and using a measure of within-firm, across-position task differentiation from US job posting data. High "hold-up" occupations exhibit higher long-run passthrough of permanent firm productivity innovations to wages, supporting the main model prediction. Accounting for heterogeneity in hold-up power across occupations has numerous implications for wage inequality: (1) greater employment of men in high hold-up occupations can account for one fifth of the Danish gender wage gap; (2) rising "superstar firms" increases wage inequality; (3) hold-up power decreases the responsiveness of wages to labor market slack.
We argue that secular change in both the production and composition of investment goods has weakened private investment's role in the transmission of monetary policy to labor earnings and consumption. We show analytically that fluctuations in the production of investment goods amplify the response of consumption to monetary policy shocks by varying labor income for hand-to-mouth agents. We document three secular changes that weaken this channel: (i) labor's share of value added in investment goods production has declined, (ii) the import share of investment goods has risen, and (iii) the composition of investment has shifted towards components that are less responsive to monetary policy. A small open economy, two agent New Keynesian model calibrated to match these facts implies a 38% and 26% weaker response of labor income and aggregate consumption, respectively, to real interest rate shocks in a 2010's economy relative to a 1960's economy.