Project

sentometrics; regime-switching; factor models; textual analysis; sentiment; forecasting; counting processes

Lead
Une multitude d’articles traitant de sujets économiques et financiers sont aujourd’hui disponibles à grande échelle et à moindre coût. Ces textes expriment un ton positif, négatif, ou neutre, appelé "sentiment textuel", qui fournit des informations complémentaires aux données quantitatives traditionnelles. Notre projet a pour objectif de déterminer si une compréhension plus fine du sentiment textuel augmente sa valeur informationnelle dans le contexte économique et financier. Nos analyses permettront potentiellement d’améliorer la protection des risques de réputation des entreprises, augmentant ainsi la stabilité des marchés financiers et de l'économie.
Lay summary
Nous mettons en avant trois questions de recherche qui ciblent chacune une caractéristique distincte du sentiment textuel qui n’a pas été étudiée jusqu’à présent. Premièrement, nous cherchons à comprendre si la dimensionnalité du sentiment et son évolution fournissent des informations supplémentaires par rapport à une estimation de sentiment unique. Deuxièmement, nous testons si une mesure plus précise du sentiment améliore son caractère informatif. Troisièmement, nous étudions le lien entre la cyclicité du sentiment et les variables économiques. Notre hypothèse générale est que la prise en compte de la dimensionnalité, de la précision et de la cyclicité du sentiment textuel améliorent sa valeur informationnelle, et permettent donc une meilleure décision pour les acteurs économiques et financiers. Afin de répondre à ces questions, nous développons un cadre méthodologique appelé "Sentometrics". A mi-chemin entre l’intelligence artificielle et l’économétrie, "Sentometrics" propose une collection d’outils pour extraire et exploiter les informations sur la dimensionnalité, la dynamique, et l'incertitude du sentiment extrait de textes.

Direct link to Lay Summary

Last update: 21.04.2018

Natural persons

This research project contributes to the analysis of sentiment in texts by asking whether “a deeper understanding of textual sentiment maximizes its information value”. A wealth of various forms of texts written on connected economic and financial topics have become cheaply available at large scale, presenting an increasingly important driver of the economy. Texts express either a positive, a negative or a neutral tone, textual sentiment, providing incremental information to quantitative data. Natural language processing and machine learning techniques have made significant progress in associating sentiment to texts. What lacks, however, is a decomposition of sentiment, knowledge about the underlying multivariate process of how sentiment is generated, and a framework for accurate statistical analysis. Understanding sentiment at a deeper level is key to grasping the full information potential of the big data sets texts are, with a focus on news articles and corporate disclosures. We put forward three research questions, that each target a distinct feature of textual sentiment. First, we ask if the dimensionality of sentiment and its evolution provides supplementary information as opposed to single sentiment estimates. Second, we question if a more precise estimation of textual sentiment improves its informativeness. Third, we research the added value of linking the cyclicality of textual sentiment to other cyclical variables. Our overarching hypothesis states that accounting for the dimensionality, precision, and cyclicality of textual sentiment enhances its information value.Our methodology is to develop a three-layered framework called “Sentometrics” that addresses the research questions, intertwined with a desire to obtain a deeper understanding of textual sentiment. The first layer consists of a dynamic factor model that reveals the common components of textual sentiment. The second layer is a time-varying stochastic model underlying the generation of positive and negative words in texts. Such a parametric approach is entirely new to the textual sentiment literature. The third layer integrates textual sentiment into a regime-switching model. To validate our main hypothesis, we will deploy this enhanced textual sentiment modeling in three specific economic and financial applications, focusing on the supplementary information value that is obtained. The validation exercises will cover respectively the prediction of macroeconomic indicators, trading performance, and covariance matrices. We analyze extensive firm-specific textual data, that is, media coverage, annual reports, and quarterly earnings press releases, across the largest European countries. A successful development of the “Sentometrics” toolbox fills in the gap of missing econometric tools to decipher textual sentiment, making it possible to extract and exploit more detailed information on the sources, dynamics, and uncertainty of sentiment in texts. Our applications may improve risk protection of companies, banks, and financial institutions, potentially increasing the stability of the financial markets and the economy. As the modelling is set up application-free, it can be deployed in multiple subsequent research settings, including marketing, politics, and web intelligence, effectively providing a means for many fields to use textual sentiment as an optimally informative variable.