Research

I introduce a discrete Taylor rule that mirrors the institutional practice of adjusting policy rates in 25-basis-point steps. Re-estimating several benchmark specifications from the literature shows that this rule reduces residuals by up to one-third in a reduced-form setting. Embedding the rule in a New-Keynesian model calibrated to the U.S. economy, and additionally restricting interest rate adjustments to two predetermined dates per quarter, yields three further insights. First, with the period length calibrated to only 30 minutes, the model directly replicates monetary policy surprises as studied in the high-frequency identification literature. A decomposition shows that such surprises are substantially distorted by discrete rates, with monetary policy shocks explaining only 45% of their variance. Surprises about rates expected at later meetings are less contaminated by demand and markup shocks, which supports identification designs based on longer-horizon futures. Second, impulse responses of all model variables are affected significantly by discrete rates and infrequent rate adjustments, exhibiting pronounced movements after interest rate adjustments and cyclical patterns between meeting dates. Third, in a simulated U.S. business cycle, the interest rate deviates on average by 22 basis points from the linear-rule benchmark. However, effects on output and inflation volatility remain muted, which lends support to currently implemented designs of monetary policy.

We introduce a task-based framework for modeling production in which certain tasks are too complex for many workers to perform. In such an environment, workers’ wages may significantly diverge from their relative productivities: Workers with marginally higher skill levels may obtain a large additional wage premium on top of the skill premium, which we call complexity premium. We apply our framework to explain past employment and wage polarization and estimate model parameters for the U.S. labor market between 2001 and 2019. Beyond a rising skill level and skill premium, we find that the complexity of tasks increases and employees performing more complex tasks earn a significant complexity premium, which accounts for up to 43 percent of their wages. Finally, we explore the effects of artificial intelligence and find it may aggravate wage inequality, with an ambiguous effect on complexity premia.