Gerd Altmann from Pixabay
Prof. Yvonne Hofstetter, CEO and co-Founder of 21strategies, argues for applying scenario techniques of Bayes theory for hedging financial market risks.
How can hedging of currency exposure be optimized? Prof. Hofstetter and her co-authors Christian Held, former Head of Treasury at Bayer AG, and Carsten Jäkel, Head of Global Treasury Services for the DACH region at Ernst & Young, provide a comprehensive insight into the mathematical constructs on which predictions for exchange rate moves are based. They are recognized experts in the fields of artificial intelligence (AI) and the hedging of financial market risks and share their experiences. One question comes to mind: Why do so few treasurers use probabilistic models for rate predictions?
The authors take the scenario technique using Bayesian theory as an example and illustrate its effectiveness in hedging foreign currency exposure. Finally, an answer is found to the question of whether fragmented and subjective assessments in treasury departments can be more effective in applying and predicting changes than proven mathematical models.
When Black Swans and Dragon Kings Appear
Latest with the Corona pandemic, everyone could become aware of how difficult and complex reliable planning and forecasting can be. Plans suddenly became obsolete which suggests that planning is not possible in a complex, hyper-connected world. Not only do individual factors change, but structures can also be turned upside down at short notice. Finance departments are concerned with reliability on a single planning path. Is it not rather the case that scenario techniques can better support treasurers?
Artificial neural networks and scenario techniques – how they work
Artificial intelligence in the strict sense is based on neural networks that bottom-up evaluate collected mass data. In this way, correlations can be determined. However, the authors also point out that spurious effects can occur - correlations which do not hold true in reality. An alternative is to model an environment using Bayes theory, the authors say. Such probabilistic model provides possible statements about future developments. It works top-down, meaning that a model is first modeled with specific knowledge of the environment before it is calibrated with mass data data. By inserting real-time data, hypotheses can first be changed and statements about probabilities of occurrence can be made.
Automated scenarios in Foreign Currency Hedging
People tend to view an increase in the flood of data as an increase in complexity. Interpretations are therefore even more difficult. It is precisely here that the authors see the advantages over the classic AI approach in treasury departments. An automated use of scenario techniques based on Bayes theory is thus able to make reliable statements under uncertainty about future developments. Forecasts about developments of exchange rates and thus hedging recommendations are just waiting to be used in treasury departments. Prof. Yvonne Hofstetter, CEO and co-Founder of 21strategies, says: "Bayes as a systematic scenario technique for treasury is enormously powerful, although complicated. The model development alone takes several years."
The article appeared in April 2021 in REthinking Finance. With a focus on the technological and organizational transformation of finance functions, it discusses digital transformation issues in financial planning.