Unbundling Macroeconomics via Heterogeneous Agents and Input-Output Networks -- by David Rezza Baqaee, Emmanuel Farhi
- 25 июня 2018, 15:54
- NBER (National Bureau of Economic Research)
The goal of this paper is to simultaneously unbundle two interacting reduced-form building blocks of traditional macroeconomic models: the representative agent and the aggregate production function. We introduce a broad class of disaggregated general equilibrium models with Heterogeneous Agents and Input-Output networks (HA-IO). We elucidate their properties through two sets of results describing the propagation and aggregation of shocks. First, we characterize how shocks affect prices and quantities of goods and factors. Even with purely microeconomic shocks, the mapping from structural primitives to observed effects is complicated by "local" general equilibrium forces. Our framework shows how to account for these forces, and helps interpret IV-based cross-sectional regression results. We also uncover a surprising property of a large class of efficient representative agent models: they feature symmetric propagation in that a shock to producer i affects the sales of producer j in exactly the same way that a shock to j affects the sales of i. This improbable symmetry breaks in the presence of heterogeneous agents or distortions. Second, we provide aggregation results characterizing the responses of industry-level variables such as markups and productivity. The behavior of these aggregates is particularly delicate in inefficient economies: they respond to microeconomic shocks outside of the industry; and they can give rise to fallacies of composition whereby aggregates move in the opposite direction of their microeconomic counterparts. Our results shed light on many seemingly disparate applied questions, such as: sectoral co-movement in business cycles; factor-biased technical change in task-based models; structural transformation; the effects of corporate taxation; and the dependence of fiscal multipliers on the composition of government spending.
НОВОСТИ ПО ТЕМЕ
15 апреля, 23:16
14 апреля, 14:40
12 апреля, 20:22