Udo Hahn

How to Do Economics with Words – The Emerging Impact of Natural Language (Processing) for Economic Theory and Practice


For a long time, the explanation and prediction of economic processes was based on metrical indicators that were intrinsically quantitative (such as GDP, rates of inflation or unemployment). With the advent of behavioural economics, psychological, cognitive, emotional, cultural and social factors complemented classical economic theory with variables that are more qualitative in nature. In my talk, I will focus on a subclass of these behavioural indicators and emphasise the role of natural language in economics and finance by reviewing recent advances in natural language processing (NLP), a field that deals with the computational analysis of natural languages.

Shamelessly borrowing parts of the title of my presentation from John L. Austin’s seminal work “How to Do Things with Words” (1955), the impetus of pragmatics and the behavioural implications of natural language for economics and finance are my major concerns. I will talk about the importance and (often un)conscious effects of linguistic signals exchanged among economic actors (companies, customers, the banking system, public administration and political institutions). A bird’s eye view is provided of fundamental methods used to uncover explicit and implicit messages of verbal signals by NLP systems and some prominent applications are featured to illustrate their functionality, e.g., NLP-based financial performance prediction for companies and effects on the stock market, product reviewing and framing via social media, forecasting of future (trans)national macro-economic development.