This thesis examines the integration of environmental, social, and governance (ESG) factors into quantitative portfolio optimisation through a dynamic, rolling-window framework. Building on the classical mean–variance model, it investigates how ESG-based constraints reshape the efficient frontier, alter diversification, and influence the persistence of portfolio efficiency over time. The empirical analysis uses monthly data for Euro Stoxx 600 constituents from 2002 to 2024, with sustainability "universes" defined via a best-in-class (top 30%) and worst-in- class (bottom 30%) screening based on the LSEG ESG Combined Score. Portfolios are optimised separately across unconstrained, ESG-constrained, and long–short configurations, applying identical rebalancing and estimation procedures to ensure comparability. Results reveal that ESG screening consistently raises portfolio sustainability while marginally reducing mean–variance efficiency. The diversification cost, however, is state-dependent and becomes more pronounced during high-volatility or high-correlation regimes. Rolling window regressions between in-sample and out-of-sample Sharpe ratios show a positive but imperfect relationship: portfolios that perform efficiently during estimation tend to preserve their relative efficiency when tested on future data. This pattern suggest that the optimisation process captures stable structural patterns in risk and return rather than fitting random fluctuations. Overall, the analysis supports a constraint-based interpretation of sustainable investing: ESG primarily restricts the feasible investment set rather than creating an autonomous source of excess return. Yet, when implemented within a disciplined quantitative framework, ESG integration proves compatible with portfolio efficiency and can enhance the stability of risk–return profiles over time.
Questa tesi analizza l’integrazione dei fattori ambientali, sociali e di governance (ESG) nell’ambito dell’ottimizzazione quantitativa di portafoglio, utilizzando un approccio dinamico basato su rolling-windows. Partendo dal classico modello mean–variance, si indaga in che modo i vincoli ESG modificano la frontiera efficiente, influenzano la diversificazione e incidono sulla stabilità dell’efficienza dei portafogli nel tempo. L’analisi empirica utilizza dati mensili delle società appartenenti all’indice EURO STOXX 600 nel periodo 2001–2024. Gli "universi" di investimento vengono definiti attraverso due strategie di selezione: una best-in-class (top 30%) e una worst-in-class (bottom 30%), basate sull’ESG Combined Score fornito da LSEG. I portafogli vengono ottimizzati separatamente per tre configurazioni: unconstrained, ESG-constrained e long–short, mantenendo identiche procedure di ribilanciamento e stima, in modo da garantire confronti coerenti. I risultati mostrano che l’applicazione di filtri ESG aumenta in modo sistematico la sostenibilità complessiva dei portafogli, a fronte di una riduzione solo marginale dell’efficienza mean–variance. Tuttavia, il costo di diversificazione associato a tali vincoli dipende dal contesto di mercato e tende ad accentuarsi nei periodi di elevata volatilità o di forte correlazione tra gli asset. Le regressioni rolling-window tra Sharpe ratio in-sample e out-of-sample evidenziano una relazione positiva, seppur non perfetta: i portafogli che mostrano buona efficienza in fase di stima tendono, in media, a mantenere una performance relativamente solida anche nei periodi successivi. Questo suggerisce che il processo di ottimizzazione cattura strutture di rischio–rendimento stabili nel tempo, anziché adattarsi a fluttuazioni casuali. Nel complesso, l’analisi supporta una visione dell’investimento sostenibile basata su vin- coli: l’ESG agisce soprattutto come restrizione del set di investimento, più che come fonte autonoma di rendimento extra. Se applicata in modo rigoroso, l’integrazione ESG risulta comunque compatibile con l’efficienza dei portafogli e può contribuire a stabilizzare il profilo rischio–rendimento nel tempo.
ESG-constrained portfolio optimization: evidence from the Euro Stoxx 600 Index
Riondato, Giovanni Simone
2025/2026
Abstract
This thesis examines the integration of environmental, social, and governance (ESG) factors into quantitative portfolio optimisation through a dynamic, rolling-window framework. Building on the classical mean–variance model, it investigates how ESG-based constraints reshape the efficient frontier, alter diversification, and influence the persistence of portfolio efficiency over time. The empirical analysis uses monthly data for Euro Stoxx 600 constituents from 2002 to 2024, with sustainability "universes" defined via a best-in-class (top 30%) and worst-in- class (bottom 30%) screening based on the LSEG ESG Combined Score. Portfolios are optimised separately across unconstrained, ESG-constrained, and long–short configurations, applying identical rebalancing and estimation procedures to ensure comparability. Results reveal that ESG screening consistently raises portfolio sustainability while marginally reducing mean–variance efficiency. The diversification cost, however, is state-dependent and becomes more pronounced during high-volatility or high-correlation regimes. Rolling window regressions between in-sample and out-of-sample Sharpe ratios show a positive but imperfect relationship: portfolios that perform efficiently during estimation tend to preserve their relative efficiency when tested on future data. This pattern suggest that the optimisation process captures stable structural patterns in risk and return rather than fitting random fluctuations. Overall, the analysis supports a constraint-based interpretation of sustainable investing: ESG primarily restricts the feasible investment set rather than creating an autonomous source of excess return. Yet, when implemented within a disciplined quantitative framework, ESG integration proves compatible with portfolio efficiency and can enhance the stability of risk–return profiles over time.| File | Dimensione | Formato | |
|---|---|---|---|
|
ESG_Constrained_Portfolio_Optimization_Tesi_Riondato.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Testo della tesi
Dimensione
4.14 MB
Formato
Adobe PDF
|
4.14 MB | Adobe PDF | Visualizza/Apri |
|
Executive_Summary_Riondato.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Executive Summary
Dimensione
897.02 kB
Formato
Adobe PDF
|
897.02 kB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/246309