Discussion of Johnston \Real and Nominal Frictions within the Firm: how Lumpy Investment Matters for Price Adjustment"

Real world observation: The cost of adjusting prices and capital is non-convex in nature: at the rm level, this makes changes in prices rare (and large) and investment lumpy (and large). Conventional ways of modelling costly price and capital adjustment ignore these non-convexities (mostly because of technical diculties), and are somewhat inconsistent with micro/real world evidence. 2 of 21 Practical men are usually slaves of some trick: { For prices, they assume that rms can randomly change their prices only in some period which does not depend on how long ago prices were changed (this yields similar implications to those from a model with convex adjustment costs: Calvo-lumpiness=Rotemberg-convex) { For capital, they resort to models with convex adjustment costs (in traditional NK models, this yields similar implications to those from a model with Calvo-style-lumpy capital adjustment Abel-Hayashi=Sveen-Weinke) 3 of 21 This paper: takes the microeconomic appeal of lumpiness very seri-ously and introduces double lumpiness both in prices and in capital at the level of the rm. On top of this, it studies the general equilibrium implications of these features by considering aggregate responses of real and nominal vari-ables to various shocks. THE MAIN QUESTION: How does the interaction between nominal and real lumpiness aect the sensitivity of real variables to real (tech-nology) and nominal (money growth) disturbances? 4 of 21 A LIST OF FINDINGS (as I see them): The double-lumpiness model has several intriguing properties: 1. Increases the sensitivity of in ation to all shocks and reduces the real eects of money relative to a baseline model without these features 2. Matches the dynamic cross-correlation between output and in a-tion (corr yt; t+k> 0) better than model without these features 3. Shifts the action from consumption to investment in response to technology shocks, from investment to consumption in response to monetary shocks. 4. Delivers realistic hazard functions at rm-level (adjustment hazards for P and K rise with time since last adjustment