Finsif tukee vuosittain vastuullisen sijoittamisen tutkimusta stipendein
Finsif jakaa stipendejä vastuullisesta sijoittamisesta kiinnostuneille opiskelijoille ja tutkijoille. Stipendien jaon tarkoituksena on edistää vastuullisen sijoittamisen tutkimustyötä. Stipendejä jaetaan vuosittain enintään 8000 euron summalla. Vuoden 2024 stipendihaku alkaa keväällä. Lisätietoa
Studying the possible relationships between different ESG metrics and stock returns by utilizing machine learning methods
Socially responsible investing has become increasingly popular during recent decades, and even more businesses and individuals are incorporating different socially responsible investment strategies into their business practices and investment decisions.
In my master’s thesis Utilizing machine learning to examine the profitability of SRI strategies: a cross-sectional study, I analyze the possible relationships between companies’ ESG metrics* and stock returns by considering monthly returns of 2177 North American companies between December 2016 and December 2022.
Through my thesis, I bring two new and significant perspectives to the discussion on the topic. First, I analyze a comprehensive set of 35 separate ESG metrics instead of, for example, the total ESG scores or even the separate E, S, and G scores. The methodology of my study is also very progressive, as I utilize machine learning methods in addition to the Fama-Macbeth two-pass regression method which is more commonly used in traditional asset pricing literature. I also further assess these results with sorted portfolios based on the significant ESG characteristics.
*In this blog article, I discuss ESG metrics interchangeably with e.g. SRI/CSR/other equivalent responsibility metrics.
Can we identify ESG metrics that influence stock returns?
Based on my initial analysis with the Fama-Macbeth method, there exists evidence of certain ESG metrics influencing stock returns positively: the ESG reporting scope, sustainability compensation incentives score, and product responsibility score. These results also suggest that the ESG metric social pillar score influences stock returns negatively.
When further assessing the significance of these initial results with five-factor regressions on the value-weighted and equally weighted sorted portfolios, my overall results suggest that the only ESG metric influencing stock returns significantly is the product responsibility score, although the sign of this relationship remains unclear due to mixed results.
Consequently, the simple answer to the question “can we identify ESG metrics that influence stock returns?” seems to be yes, although further research on the topic would be needed to obtain more generalizable and robust results. When also considering the two previously mentioned new perspectives on the topic that my study provides, I believe it is an important addition to the literature on socially responsible investing in both Finland and abroad.
As the availability of ESG data continues to grow, I anticipate the possibility of conducting a corresponding study with Finnish stock market data. I also hope my study paves the way for future researchers by demonstrating that different machine learning methods can be utilized to conduct increasingly extensive and complex studies on ESG-related topics as well.
Suvi Hautsalo
M.Sc (Economics and Business Administration), Finance, Hanken, 2023
The study can be found here.
