Are equity returns still predictable? The case of industry portfolios and ARIMA models
Author
Hagelskjær, Hans Lauge
Term
4. semester
Publication year
2023
Submitted on
2023-05-31
Pages
93
Abstract
This thesis examines whether returns on Swedish industry portfolios remain predictable and what this implies for weak-form market efficiency. Using the Box–Jenkins framework, ARIMA models are estimated and evaluated on log returns for self-constructed Swedish industry portfolios over 2010–2022. The approach includes tests for stationarity and randomness (including Augmented Dickey–Fuller and Ljung–Box), ACF/PACF analysis, model selection via AIC, and forecast and residual diagnostics. The results indicate a rejection of weak-form efficiency for these industry portfolios, suggesting that past prices and returns are not fully incorporated into current prices. At the same time, the economic value added by ARIMA-based forecasts is inconsistent across industries, implying that predictability varies by sector. The thesis contributes to the debate on market efficiency and return predictability and offers practical insights for researchers and practitioners considering active strategies based on historical information.
Denne afhandling undersøger, om afkast på svenske industrisektorsporteføljer fortsat er forudsigelige, og hvad det betyder for den svage form for markeds-efficiens. Med udgangspunkt i Box–Jenkins-metoden estimeres og evalueres ARIMA-modeller på logafkast for selvkonstruerede svenske industrisektorsporteføljer i perioden 2010–2022. Tilgangen omfatter tests for stationaritet og tilfældighed (bl.a. Augmented Dickey–Fuller- og Ljung–Box-tests), analyse af ACF/PACF, modelvalg via AIC samt prognose- og residualdiagnostik. Resultaterne indikerer et afslag af den svage form for efficiens for disse industrisektorsporteføljer, hvilket antyder, at historiske priser og afkast ikke er fuldt indarbejdet i aktuelle priser. Samtidig er den økonomiske merværdi ved at bruge ARIMA-baserede prognoser inkonsistent på tværs af industrier, hvilket peger på, at forudsigeligheden varierer mellem sektorer. Afhandlingen bidrager dermed til debatten om markeds-efficiens og returnforudsigelighed og giver praktiske implikationer for forskere og praktikere, der overvejer aktive strategier baseret på historiske informationer.
[This apstract has been generated with the help of AI directly from the project full text]