Yesterday, the National Bank hosted the 32nd Research Club session. Vice Governor Ana Mitreska delivered the keynote address that was followed by presentation of research papers: "Analysis of the features of export-based companies in the Macedonian economy using microdata" by Biljana Jovanovic and "Current GDP estimation for the Macedonian economy through empirical methods using various time-frequency data" by Gani Ramadani and Magdalena Petrovska.
In the first paper, Jovanovic premised its paper on the theory that export companies have better operating results, compared with companies that primarily produce and operate on the domestic market. The purpose of the research is to identify reasons behind the differences among companies, taking into account their export status. Analyzing microdata on a sample of 1.900 companies for the 2013-2017 period, the paper tests two hypotheses - self-selection hypothesis and learning by doing hypothesis. According to the first hypothesis, companies that have better performance can sell on international markets, that is, their superiority is an ex-ante feature. According to the second hypothesis, exporters are more successful as a result of the learning while conducting export activities. Jovanovic also emphasized that the results obtained, just like most of the previous research on this topic, confirm the self-selection hypothesis. Also, the analysis confirms to a certain extent the learning by doing hypothesis, that is, the exports itself explain part of the successful results of the exporting companies.
The second paper, presented by Magdalena Petrovska on behalf of the two authors, focuses on models for forecasting economic activity in the short term, which is of particular importance for the forecast and decision-making processes in the central banks. The paper gives an overview of two relatively new empirical methods for direct modeling of various time-frequency data and unbalanced datasets. Namely, this is about the method known as MIDAS (mixed-data sampling), as well as the newer extension in the literature in this area, known as Bayesian MF-VAR (Bayesian mixed-frequency vector autoregression). As emphasized in the paper, the results indicate that the two models show similar forecasting performance in terms of the dynamics of the Macedonian GDP on a one-quarter forecast horizon. During the presentation, it was emphasized that both approaches are rather complementary than exclusive, since their combined forecast is better in terms of anticipating the milestones of the business cycle.
At the end of the session, encouraged by the presented views and conclusions, the participants discussed the issues and aspects of the two papers.
The next Research Club session will be held in June this year.
All rights reserved © 2017
National Bank of the Republic of North Macedonia