Seminar: "Testing for cointegration using ARDL models in R"

AUEB STATISTICS SEMINAR SERIES 2024-2025

Presenter: Kleanthis Natsiopoulos, Lecturer, Accounting and Finance, School of Business, University of Dundee, UK

Testing for cointegration using ARDL models in R

ROOM Τ102

ABSTRACT

Cointegration is an important concept in time series analysis, especially when working with economic or financial data. While tests like Johansen's are commonly used, ARDL (Autoregressive Distributed Lag) models provide a flexible and practical alternative, particularly when dealing with variables of mixed integration order. In this seminar, I will discuss cointegration testing using ARDL models, focusing on how this approach differs from more traditional methods in both theory and practice. I’ll also go over some of the key assumptions behind the test that are often overlooked in applied work and demonstrate how to implement them in R using the ARDL package, which I developed. I’ll cover how it compares to other R packages as well as proprietary software such as Stata and EViews and will discuss future directions for the package.