Classification of Monetary Policy Decisions through Text Mining Techniques

Abstract: Ellingsen, Söderström and Masseng (2003) describe an empirical strategy to test their model about how central bank interventions affect asset prices. They manually analyse a Wall Street Journal column in order to determine whether bond traders interpreted a target rate change as reaction to new information about the economy or change of central bank preferences. Instead this thesis tries to replicate their results during another time period by applying Machine Learning and Natural Language Processing techniques. Even though deterministic as well as trained algorithms are applied in order to achieve a consistent classification, not all of their hypotheses can be verified. This can likely be traced back to the extraordinary time period underlying the empirical test in this sample.