Three Essays on Financial Markets

Abstract

This thesis comprises three distinct but interconnected studies in the field of financial markets, each exploring different facets of financial markets: asset pricing, sustainable investment and behavioral finance. The first paper derives stock returns for firms producing non-renewable commodities employing the investment-based asset pricing approach. By identifying the appropriate time-varying discount rate the investment-based approach allows an alternative test of the Hotelling Valuation Principle. The empirical results support the principle and enable predicting returns from sorting firms into quintiles by expected return, producing a 16-20 percent realized difference between top and bottom quintile. The return differences cannot be explained by standard risk factors or a commodity-specific factor, suggesting that an important risk factor is still missing from standard models. The approach permits cost-of-capital estimation that circumvents identifying systematic risk factors. The second paper examines whether the carbon pricing risk factor is priced in the cross-section of commodity futures. Analyzing unexpected pricing shocks in carbon emission allowances, it is shown that carbon pricing risk carries a significant positive risk premium in commodity markets. The study reveals that commodity sensitivities to carbon pricing risk vary, influenced by commodity-specific characteristics such as basis and hedging pressure. Additionally, a portfolio of commodity futures constructed based on carbon pricing beta offers superior out-of-sample hedging performance for climate change risk compared to hedge portfolios constructed from equities or ETFs. The third paper investigates the accuracy of target price forecasts made by sell-side analysts, employing machine learning approaches to predict the forecasts’ accuracy. Using a dataset of target price forecasts for U.S. listed companies from 1999 to 2021, ensemble methods incorporating market-level, firm-level, and analyst-level information are used to predict target price accuracy in terms of errors and achievement. A long-short portfolio constructed based on these predictions significantly outperforms the benchmark in terms of cumulative return and Sharpe ratio.