University of Thessaly (UTH)
Computational statistics and econometrics: economic applications and model development in R- language
Abstract
dc:descriptionThis thesis focuses on the theory, implementation, and empirical application of Autoregressive Distributed Lag (ARDL) and Conditional Error Correction (CECM) models, with a particular emphasis on the bounds testing procedure for cointegration. While ARDL models have become a cornerstone of applied econometric analysis due to their flexibility in handling variables of mixed integration orders, their practical implementation in statistical software has often been inconsistent or incomplete. The first part of the thesis provides a detailed theoretical and methodological overview of ARDL and ECM models, including their underlying assumptions, estimation, inference, and cointegration testing procedures. Building on the foundational work of Pesaran, Shin, and Smith (2001, Journal of Applied Econometrics, 16(3), 289-326), it clarifies the relationships among ARDL, VAR, and VECM frameworks and examines how the bounds testing approach can be understood within a conditional ECM representation. The second part presents the development of the open-source ARDL package for R, designed to offer a comprehensive and transparent implementation of ARDL estimation, automatic model selection, bounds testing, dynamic multipliers, and related tools. The package aims to bridge the gap between theory and practice, providing researchers with a reliable, reproducible, and user-friendly platform for applied time series analysis. Its design, functionality, and computational efficiency have been discussed in Natsiopoulos and Tzeremes (2022, Journal of Open Source Software, 7(79), 3496) and further elaborated in Natsiopoulos and Tzeremes (2024, Computational Economics, 64, 1757-1773), while the package itself has been officially released on CRAN (Natsiopoulos and Tzeremes, 2023). In addition, its design and performance are compared against alternative implementations available in other statistical environments, highlighting both its methodological rigor and practical advantages. These works collectively correspond to the methodological and software contributions presented in this thesis. The final part applies the developed tools to a replication and extension of the UK wage equation originally analyzed in Pesaran, Shin, and Smith (2001). The empirical analysis, published as Natsiopoulos and Tzeremes (2022, Journal of Applied Econometrics, 37(5), 1079-1090), includes a narrow sense replication that reproduces the original findings, validating the package’s accuracy, while a wide sense replication extends the dataset and incorporates updated sources. The analysis reveals that the long-run relationships between wages, productivity, and unemployment remain robust over time, although the influence of tax wedges and union power has weakened in the extended sample. An extended wide sense replication and sensitivity analysis further assess the stability and interpretation of the long-run coefficients across model specifications. The results and software outputs of this dissertation have been disseminated through peer-reviewed publications, international conferences, and invited seminars, including the International Joint Conference CFE-CMStatistics (2024, 2025), and invited talks at the Department of Statistics, Athens University of Economics and Business (2025), the Economics Department, University of Patras (2025), and the Mathematics Division, University of Dundee (2025). Collectively, this body of work unifies theoretical understanding, computational implementation, and empirical practice within a coherent and accessible framework for ARDL modeling, while contributing new tools and empirical insights that have been adopted by academic, governmental, and financial institutions worldwide.
Degree
thesis:*- Grantor dc:publisher
- University of Thessaly (UTH)
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Natsiopoulos, Kleanthis
- Νατσιόπουλος, Κλεάνθης
Subjects
dc:subject × 22- Μοντέλα αυτοπαλίνδρομων χρονικών υστερήσεων (ARDL)
- Μοντέλα διόρθωσης σφαλμάτων (ECM)
- Συνολοκλήρωση
- Έλεγχος bounds
- Ανάπτυξη οικονομετρικού λογισμικού ανοικτού κώδικα
- Οικονομετρία χρονολογικών σειρών
- Προγραμματισμός στη γλώσσα R
- Autoregressive distributed lag (ARDL) models
- Error correction models (ECM)
- Cointegration
- Bounds testing
- Open-source econometric software development
- Time series econometrics
- R programming
- Οικονομικά και Επιχειρήσεις
- Κοινωνικές Επιστήμες
- Οικονομετρία
- Οικονομικά
- Economics and Business
- Social Sciences
- Econometrics
- Ecomonics
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/60591