The Granular Drag on Growth

This paper uncovers a novel mechanism through which market structure shapes future productivity growth. Using a micro-founded exogenous growth model in which granular firms experience random productivity shocks, I characterize sectoral and economy-wide productivity growth conditional on the current market structure. I test the model's predictions using firm-level data from Sweden, complemented by industry data from the United States and other European economies. In efficient industries, the model predicts that higher sales concentration lowers expected productivity growth due to limited reallocation. In the data, a 10-percentage-point increase in the Herfindahl index of sales concentration is associated with a 3-percentage-point lower productivity growth rate over a five-year period. Moreover, in line with the model's predictions for distorted economies, a similar increase in the gap between the Herfindahl indices of sales and cost shares is linked to a stronger decline of about 13 percentage points. The model generates persistent cross-sectional growth heterogeneity consistent with the empirical evidence, even though all firms follow identically distributed productivity processes. I conclude that the interaction between micro-reallocation and market concentration, a mechanism I term the granular drag, is important for understanding productivity growth across industries and possibly entire economies.

Towards Granular Krusell-Smith Equilibria: An Endogenous Growth Application

I study an endogenous growth model with a finite but potentially large number of granular firms that choose innovation effort and face idiosyncratic productivity shocks. A Markov perfect equilibrium suffers from the curse of dimensionality because firms must track the full distribution of rival productivities. I overcome this problem by using deep learning to compute a Krusell-Smith-style approximate equilibrium in which firms use as state variables their own sales share together with a small set of concentration moments whose evolution they can approximately forecast. Despite the model's simplicity, it generates substantial growth heterogeneity across sectors with different levels of concentration. More broadly, the approach developed here can be used to study forward-looking firm behavior in environments populated by granular firms with empirically realistic market share distributions.

Inflation Persistence and a New Phillips Curve

Inflation exhibits substantial persistence in the data, yet the standard New Keynesian Phillips Curve (NKPC) fails to generate this persistence without resorting to ad hoc assumptions like inflation indexation. This paper demonstrates that menu-cost models with state-dependent pricing naturally produce inflation persistence consistent with empirical evidence. The key insight is that menu-cost models feature both intensive and extensive margins of price adjustment. In response to shocks to the growth rate of nominal demand, the intensive margin generates the standard marginal cost channel as in the NKPC, whereas the extensive margin generates history dependence that is captured by the lagged inflation rate. Using a calibrated menu-cost model with idiosyncratic productivity and stochastic adjustment costs, we show that when nominal demand growth is autocorrelated (as in the data), firms optimally delay price adjustments, generating history-dependent inflation dynamics. In Phillips Curve regressions, lagged inflation exhibits a coefficient of 0.50 when controlling for expected marginal costs alone—consistent with empirical estimates. However, this coefficient drops to 0.05 when we include lagged nominal demand growth, revealing that the persistence primarily stems from the extensive margin channel. Our findings suggest that inflation persistence emerges endogenously from firms' optimal price-setting behavior under menu costs, without invoking the Lucas critique concerns associated with mechanical indexation assumptions.