University of North Texas
Implied Cost of Capital as an Improved Measure of Equity Premium: A Deep Learning Approach
Abstract
dc:descriptionCorrectly estimating expected equity returns is critical for resolving the equity premium puzzle and understanding the true cost of equity. Traditional methods that rely on historical market returns over extended periods can produce misleading estimates. Although Fama and French’s (2002) dividend and earnings growth model mitigates reliance on past returns, it introduces its own sources of bias. Our out-of-sample evaluation for 2001 to 2024 demonstrates that this fundamental approach yields higher standard errors and lower precision compared to realized returns. The implied cost of capital (ICC), theoretically a more accurate proxy for investors’ expected returns, is often inflated due to upward-biased analyst earnings forecasts. As a result, prior literature has tended to overstate both the ICC and the equity premium. In this paper, I propose a novel machine learning framework to forecast future earnings, which I then use to calculate the implied cost of equity. The proposed method reduces the forecasting biases inherent in traditional approaches and provides a more precise, consistent, and robust estimate of the equity premium. This better measure of equity premium is lower than previous researchers have concluded and helps partially solve the equity premium puzzle.
Degree
thesis:*- Grantor dc:publisher
- University of North Texas
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Jaiswal, Umesh Kumar
- Contributors dc:contributor
-
- Liu, Ian
- Ay, Lezgin
- Owen, Stephen R.
- Zhou, Yang
Subjects
dc:subject × 5Rights
dc:rights- Statement dc:rights
-
- Public
- Jaiswal, Umesh Kumar
- Copyright
- Copyright is held by the author, unless otherwise noted. All rights Reserved.
- Language dc:language
- English
Identifiers
dc:identifier.*- Identifier
-
local-cont-no: submission_4521
https://digital.library.unt.edu/ark:/67531/metadc2481297/
ark: ark:/67531/metadc2481297