The peer effects in asset price models: evidences from emerging and developed countries

• Autor(a) principal: Selan, Beatriz
• Data de Publicação: 2019
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: http://www.teses.usp.br/teses/disponiveis/18/18157/tde-13062019-093914/
• topic: Developed economies; Economias desenvolvidas; Efeitos pares; Emerging markets; Mercados emergentes; Peer effects; Retorno de ações; Stock return

This study investigates the peer effect in the asset pricing models in the international stock market. The peer effect theory proposes a dependence between individual decisions due to interactions that create a social network structure. The idea is that we need to understand the correlation between outcomes of individuals that interact in an environment and which could lead to a homogenous pattern of movement especially on asset pricing models. We use a sample of almost 7,000 companies listed on fourteen countries from 2006 to 2016 and arrange them in four peer groups. Since the peer effect has a reflection problem, we divide our empirical models in two aspects. First, we analyze the relationship between stock return from the firm, its financial aspects and the financial aspects for the peer group using a fixed effect regressor. Then, we try to understand the relationship between stock return from a firm, the stock return from the peer firms, the financial aspects from the firm and the financial aspects for the peer group by estimating a 2SLS model with an instrumental variable. Our findings show the existence of peer effects on stock return for all the peer groups. Also, the effects are always positive regardless if we select emerging or developed markets. Moreover, there is exogenous peer effect from the characteristics of the peer firms in the stock return that depends on the indicator and the peer group. Market-to-book ratio of the peers presents a positive relationship with the stock return. As a robustness test, we re-estimate the models for two subsamples and find that the results are consistent to the previous ones.