Hasty Briefsbeta

The invisible hand screws up your regression

a day ago
https://www.economicforces.xyz/p/the-invisible-hand-screws-up-your
Copy Link
  • #empirical_research
  • #economics
  • #price_theory
  • The 'China Shock' research by Autor, Dorn, and Hanson estimates that Chinese import competition accounted for 21% of U.S. manufacturing job losses from 1990 to 2007, roughly 1.5 million jobs.
  • Price theory complicates empirical economics because prices coordinate behavior across markets, making it hard to isolate causal effects when shocks affect interconnected systems.
  • Robert Minton and Casey Mulligan's 'market interpretation of treatment effects' distinguishes between Difference-in-Differences (DiD), Treatment on the Treated (ToT), and Scale Effect, highlighting how spillovers distort estimates.
  • DiD measures the gap between treated and control groups, ToT includes spillover effects on the treated, and Scale Effect reflects the impact when the entire market is treated—these are not the same.
  • The 'missing intercept' problem arises because cross-sectional regressions capture relative effects (slopes) but miss common shocks absorbed by time fixed effects, leading to incomplete aggregate estimates.
  • In the China Shock case, the naive scale-up assumes no spillovers, but price theory suggests spillovers are pervasive, making aggregate effects hard to infer from regional comparisons.
  • Recent studies (e.g., Adão et al., Lyon and Waugh) show the missing intercept can either amplify or dampen aggregate effects, depending on labor market responses and general equilibrium adjustments.
  • The problem extends beyond trade: airline merger studies face similar issues, as network-wide effects (e.g., fare changes) are invisible to cross-sectional DiD designs.
  • Structural models can help interpret cross-sectional estimates but cannot fully resolve the missing intercept problem without additional economic assumptions.
  • The key takeaway: market interconnectedness means distributional effects (DiD) and aggregate effects (Scale Effect) require separate analysis, with the latter needing economic models to account for spillovers.
AboutLogin