Investor Sentiment and European Stock Returns, Extending the Fama-French Five-Factor Model
Author
Kriegbaum, Sebastian Frigast
Term
4. semester
Publication year
2024
Pages
55
Abstract
Dette speciale undersøger, om et mål for investorsentiment kan forbedre forklaringen af europæiske aktieafkast i en udvidelse af Fama–Frenchs femfaktormodel. Med udgangspunkt i Baker og Wurglers metode, tilpasset til europæiske forhold og informeret af europæisk litteratur, konstrueres en sentimentsfaktor, der indgår i en revideret model. Modellen testes med OLS-tidsserieregressioner på både porteføljer og enkeltaktier og evalueres op mod den oprindelige femfaktormodel ved hjælp af R2, justeret R2, intercepts (alfa), GRS-statistik samt t- og p-værdier. Analysen omfatter også tidsvarierende, dynamiske betas. Porteføljetestene giver svage resultater og forfølges ikke yderligere. For enkeltaktier ændrer introduktionen af dynamiske, betingede betas modellen markant og peger på, at sentimentsfaktoren kan øge forklaringskraften i forhold til femfaktormodellen, målt ved R2 og justeret R2. Samtidig indikerer den gennemsnitlige absolutte alfa, at modellen med sentimentsfaktoren indfanger en mindre del af overafkastene. Samlet set peger resultaterne på potentiale for sentimentsfaktoren ved anvendelse af tidsvarierende betas, men også på behov for yderligere forskning.
This thesis examines whether an investor sentiment measure can improve the explanation of European stock returns by extending the Fama–French five-factor model. Following the Baker and Wurgler approach, adapted to European markets and informed by European evidence, a sentiment factor is constructed and added to a revised model. The model is tested with OLS time-series regressions on portfolios and individual stocks and benchmarked against the original five-factor model using R2, adjusted R2, intercepts (alphas), the GRS statistic, and t- and p-values. The analysis also incorporates time-varying, dynamic betas. Portfolio tests are weak and not pursued further. For individual stocks, introducing dynamic conditional betas materially affects fit and suggests that the sentiment factor can enhance explanatory power relative to the five-factor benchmark, as indicated by R2 and adjusted R2. At the same time, the average absolute alpha indicates that the model with the sentiment factor captures less of the excess returns. Overall, the findings point to potential benefits of the sentiment factor when time-varying betas are used, while motivating further research.
[This summary has been generated with the help of AI directly from the project (PDF)]
Documents