Feb 26, 2019

CEF Learning Events Boost Confidence

We are happy to share reflections by Miha Trošt from the Economic Policy Analysis Unit at the Slovenian Ministry of Finance about his participation at the learning event where he was hoping to deepen his understanding of time series modelling and forecasting (especially VARs and BVARs).

By attending the CEF blended learning event on time series forecasting and data quality in October-November 2018, I refreshed my existing knowledge of the topic, filled some gaps in my understanding and, most importantly, strengthened confidence in applying time series methods at work.

The blended learning event consisted of two pre-workshop online webinars, including an extensive assignment in statistical package EViews, and a workshop in Ljubljana, with some hands-on exercises using the same package. The workshop also covered the theoretical foundations of macroeconomic forecasting and taught participants estimation and use of univariate and multivariate macroeconomic forecasting models. Univariate models covered Auto Regressive Integrated Moving Average and multivariate consisted of Auto Regressive Distributed Lag and Vector Auto Regressive.

Pre-workshop webinars were highly beneficial for me. Until then, I had been scatteredly applying time series methods at work and my theoretical understanding was a bit rusty, since my primary work task is to fully automate data related tasks. So, I could refresh and deepen my theoretical knowledge of univariate time series forecasting.

However, the main benefits for me came from attending a three day in-person workshop, where I could ask as many questions as I could think of. I took advantage of that, participated actively and tried to fill gaps in my understanding. I had read about it in books before, had some experience from current and previous jobs, but it was during the workshop that I finally understood what it meant that there was no best statistical model and that a forecasting model was only a convenient, scalable tool to summarize data at hand. Moreover, I can automate easily, and it can be used with many limitations and careful interpretation for production of future values.

I gained the much-needed confidence to scale time series modelling at work and to add statistical forecasting methods to my frequently used data analysis toolbox. Furthermore, being very active in the workshop also brought me a nice surprise: at the end of the workshop I was granted “The most active participant” award.