Skip to main content

Forecasting Economic Time Series in Changing Environments

Final Report Summary - BREAKMETRICS (Forecasting Economic Time Series in Changing Environments)

What is management, whether it be a company, a not-for-profit association, or a municipality? It consists of anticipating events in the first place and taking appropriate action. This research project aims to innovate at the highest scientific level the way we think about anticipating events, or, so-called "forecasting".

Currently, forecasters (individuals, companies, governments,...) often disregard gradual or abrupt changes in society. They often rely on their intuition, which tends to be anchored on the past (and is often influenced by others' viewpoints). Of course, this may cause severe forecast errors that may entail disruptions such as an excess workload for employees, bankruptcy, unavailable payment systems, higher unemployment, excessive health expenditures etc. We develop tools for systematic analysis of future events. Using data sets, we put forward forecasting techniques that explicitly incorporate changes in our environment.

The general objective of this proposal is to improve economic forecasts with the help of evolutionary econometrics models, i.e. models that adapt to abrupt (or structural) changes in the economic environment, also called structural break or change-point models.

The need for adaptive modelling is obvious in the light of the current economic conditions. Economic forecasting is essential for decision making with respect to fiscal and monetary policy, public spending, and investment. For example, the monetary transmission mechanism has long and uncertain lags. Therefore, monetary policy should be forward-looking. To effectively ensure price-stability, the central bank for example needs to make forecasts about the evolution of prices and output among others. When a structural break happens, conventional