Multi-Asset Factor Investing Strategies and Controversy Screening using Natural Language Processing

Ranganathan, Ananthalakshmi and Nolte, Sandra and Lohre, Harald (2023) Multi-Asset Factor Investing Strategies and Controversy Screening using Natural Language Processing. PhD thesis, Lancaster University.

Text (2023RanganathanPhD)
Phd_thesis_revisions_Dec23.pdf - Published Version
Download (2MB)

Abstract

Factor investing strategies have revolutionized the landscape of equity investing, and continues to be heavily researched by academics and practitioners, leading to the documentation of more than 450 factors. However, from a practical investment perspective, much of the factor evidence documented by academics may be more apparent than real. The performance of many factors has found to be dependent on the inclusion of small- and micro-cap stocks in academic studies, although such stocks would likely be excluded from the real investment universe due to illiquidity and transaction costs. We take the perspective of an institutional investor and navigate this zoo of factors by focusing on the evidence relevant to the practicalities of factor-based investment strategies. Establishing a sound theoretical rationale is key to identifying “true” factors, and we emphasize the need to recognize data-mining concerns that may cast doubt on the relevance of many factors. Nevertheless, a parsimonious set of factors emerges in equities and other asset classes, including currencies, fixed income and commodities. Since these factors can serve as meaningful ingredients to factor-based portfolio construction, we build currency factor strategies using the G10 currencies. We show that parametric portfolio policies can help guide an optimal currency strategy when tilting towards cross-sectional factor characteristics. While currency carry serves as the main return generator in this tilting strategy, momentum and value are implicit diversifiers to potentially balance the downside of carry investing in flight-to-quality shifts of foreign exchange investors. Drawing insights from a currency timing strategy, according to time series predictors, we further examine the parametric portfolio policy’s ability to mitigate the downside of the carry trade by incorporating an explicit currency factor timing element. This integrated approach to currency factor investing outperforms a naive equally weighted benchmark as well as univariate and multivariate parametric portfolio policies. Whilst factor investing continues to grow in popularity, investors have expressed interest in aligning their investments with social values in order to maximize positive social impact. Hence, for any company, involvement in socially unethical practices not only leads to reputational damage but also financial consequences, anecdotally. To quantify the consequence of such controversial behaviour, we investigate the price impact of involvement in social controversies and find that the returns drop, on an average, by over 200 basis in the days around the outbreak of news on social violations. We identify companies following socially unethical practices from news headlines with the help of state-of-the-art language modelling approaches. Using a large sample of 1 million news headlines, we further train and fine-tune a DistilRoBERTa model to identify reports of controversial incidents in daily news feed. We map the price reaction of such controversial events using an event study approach and document negative price impact for companies with poor social practices measured via increased controversial behaviour, largely driven by small to medium market capitalization companies. Amongst the eight different social dimensions we examine, controversies surrounding violations of product safety standards, online scams and data privacy breaches significantly impact firm returns. Dissecting this result by geographies, the U.S, Australia, Europe and Emerging Market react very negatively to social controversies.