Aristotle University Of Thessaloniki (AUTH)
Τροποποιήσεις χρηματοοικονομικών αριθμοδεικτών ως προς την ενίσχυση της προβλεψιμότητας των χρηματιστηριακών αποδόσεων
Abstract
dc:descriptionReturn predictability has attracted huge debates among academia and practitioners and remains till today one of the key research areas in the finance field. The notion that the stock market followed a random walk and thus, evidence of predictability (if any) were essentially interpreted as market inefficiencies, was succeeded by the intense research efforts to enhance the accuracy of forecasting stock market returns through a thorough examination of specific financial ratios, economic indicators and other behavioural factors with predictive capacity. In this thesis, the primary focus lies on the classical financial ratios dividend-price and earnings-price and based on long-run equilibrium relationships between log dividends, earnings and prices, their modified counterparts are econometrically produced which are essentially stationary versions of the original non-stationary variables. It is shown that in all forecasting horizons our modified variables are more capable to explain variability in time-varying expected returns. The explanatory capacity of a new predictor is similarly explored based on the cointegration relationship between the log 12-month summed-up dividends (d), log prices (p) and log 12-month summed-up earnings (e). Through dpe, the ability to explain stock return variability is increased both in and out-of-sample. Our first published article titled “Modified Ratios and the Cyclically-Adjusted Price-earnings Ratio” focuses on the modifications of the simple dp and ep and compares outcomes in and out-of-sample against Campbell and Shiller’s (2001) well-known CAPE. Our second paper titled “Interpreting Return Variability via the Dividend-Price-Earnings Ratio” stresses on the introduction of a new predictor of returns based on the long-run equilibrium relationship between dividends, prices and earnings (dpe for short). Results are compared to the classical dividend-price and its modified version (mdp based on the cointegration relationship between dividends and prices). Overall, the key findings of the thesis include the following. First, the modified dp seems a more capable predictor of returns regardless the forecasting horizon examined. Second, the modified ep is also able to explain variations in the US returns. Third, both mdp and mep produce more enhanced results both in and out-of-sample against the investment community’s favorite CAPE. Fourth, our suggested dpe and mdp surpass the performance of the well-known dp, by simultaneously improving the quality of forecasting in nominal, excess and real returns. The implications of our study are the following. First, the proposal of the use of new predictors whose construction does not require the knowledge of any complex econometric techniques. Second, an investor who employs our suggested dpe, mdp and mep can increase his forecasting gains both in and out-of-sample, and therefore, boost his ability to manage portfolio risks and address market-timing and asset allocation strategic issues. Our work is strongly related to the daily challenges faced by financial analysts, portfolio and risk managers and investors. Our main aim is to shed more light to the predictive components of time-varying expected returns and underline the increased explanatory potentials of identified variables under certain econometric modifications.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Georgiou, Aikaterini
- Γεωργίου, Αικατερίνη
Subjects
dc:subject × 14- Χρηματοοικονομικοί αριθμοδείκτες
- Τροποποιημένοι αριθμοδείκτες
- Προβλεψιμότητα εντός δείγματος
- Προβλεψιμότητα εκτός δείγματος
- Financial ratios
- Modified ratios
- In-sample predictability
- Out-of-sample predictability
- Οικονομικά και Επιχειρήσεις
- Κοινωνικές Επιστήμες
- Χρηματοοικονομικά
- Economics and Business
- Social Sciences
- Finance
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/54325