Makroekonominių rodiklių netikėtumų poveikio akcijų grąžai modeliavimas

Abstract

Literature analysis suggests that there is a lack of research regarding modelling the surprise effect of macroeconomic indicators on stock returns in the short run. To fill this gap in the literature, the major purpose of this dissertation is to research the phenomenon representing the surprise effect of macroeconomic indicators and its relationship with stock returns, to develop and empirically test the models for forecasting the surprise effect of macroeconomic indicators on stock returns by employing traditional and machine learning methods. The scientific problem of the research: what is the essence of the surprise effect of macroeconomic indicators and how to evaluate and model the surprise effect of macroeconomic indicators on stock returns by employing traditional and machine learning methods? The empirical research revealed that most surprises of macroeconomic indicators do not have any statistically significant effect on the return on the EURO STOXX 50 index under different scenarios in the short run. The study with statistically significant models and their data sets representing the potential surprise effect of macroeconomic indicators on the return of the EURO STOXX 50 index disclosed that machine learning methods can provide several times more accurate models for forecasting the return of the EURO STOXX index than traditional methods.