Dual peer effects and cross-stock predictability

Abstract

This paper introduces a Peer Index (PI) constructed from economically motivated peer networks that summarizes (i) the strength of a firm’s peers and (ii) the firm’s position within its peer group. PI predicts stock returns and earnings surprises over short and long horizons. Machine-learning models based solely on firm-level characteristics do not subsume PI’s predictive power, supporting the interpretation that it captures genuine cross-stock information. Lag-augmented local projections show that positive PI innovations are followed by higher next-month returns that gradually decay without reversal, consistent with slow diffusion of peer information into prices.