An Application of Autoregressive Distributed Lag-Models for Earnings per Share Forecasting in Poland

This investigation delves into the significance of precise earnings forecasts for publicly traded companies in achieving investment success. It emphasises the importance of this aspect, particularly in markets with limited analyst coverage, such as emerging markets including Poland. The study assesses the accuracy of predictions generated using the Autoregressive Distributed Lag (ARDL) framework with the XGBoost type of modelling and various methods of combining forecasts compared to the seasonal random walk model. Positioned as an intermediate step between time series and multivariate forecasting, these models are applied to earnings per share (EPS) data of companies listed on the Warsaw Stock Exchange from 2009 to 2019, i.e. the last financial crisis and the pandemic shock. The seasonal random walk model attained the lowest error rates based on the Mean Arctangent Absolute Percentage Error (MAAPE) metric, a conclusion substantiated by rigorous statistical tests and robustness checks employing different periods and error metrics. The enhanced performance of the simpler seasonal random walk model may be ascribed to the relatively uncomplicated nature of the Polish stock market.