Back to search

Πανεπιστήμιο Μακεδονίας

Μέθοδοι μη γραμμικής συνολοκλήρωσης και αιτιότητας σε μικροοικονομικά και μακροοικονομικά μεγέθη της ελληνικής οικονομίας

Abstract

dc:description

This thesis proposes and presents extensively two methods of testing for stable non-linear short and long-term relationships between Economic variables of interest, also known as Cointegration. As investigated in detail within this framework, there are a number of causes as described in the pre-existing theoretical and empirical literature that deem necessary to transition the notion of Cointegration to the non-linear framework in such a way that uninterpreted by linear methods behaviors are included in the modelling process in order to improve the efficiency of the utilized models. The thought process for the first of the proposed methods is based on the hypothesis that the basic characteristics of homogeneous or asymptotically homogeneous relationships that may link the equilibrium relationships of the variables considered are maintained . Since these non-linear relations are linearized without losing any information on the formation of long-term relations, then we can apply the already existing methodology of testing for linear Cointegration. It's also important to perform a review of the stationary tests using Bootstrapping techniques to calculate and use appropriate critical values. This method builds on and expands on the same methodological approach of Engle and Granger (1987) following exactly the two stages of implementation of this method. The second method is essentially an extension of the previous one allowing Threshold adjustment to the correction term of the error correction model due to the use of the linearized long-term relationship, allowing for the inclusion of non-linearities that remained uninterpreted after the non-linear transformation of the original data in the attempt to linearize the equilibrium vector. The proposed method is essentially an extension of the first to Threshold forms of integration by incorporating the methodological approach of Balke and Fomby (1997) by following similar steps in the Threshold non-linearity check and its application to the data. The linearization of nonlinear relations between the economic variables is considered in a series of possible transformations (including the Box-Cox transformation). The thesis beyond the analysis of existing literature in non-linear Cointegration and the examination of specific methods in a number of alternative variables of interest within the literature proceeds to the examination of the relevant literature in a number of data sets (Macroeconomics and Finance) where it applies a series of existing Linear and Non-linear methods compared to the proposed methodology to examine its effectiveness. From the empirical analysis performed, using several known software packages, non-linear long-term behaviors as well as corresponding correction mechanisms were derived. The possibility of using this technique with a multitude of analogous commercial econometric computing environments is implicit. Also in the context of this thesis, specific sizes of interest (indicatively Defense Spending, Spread, unemployment) are considered to examine and find the peculiarities that can be attributed and captured with Non-linear tools and techniques. Finally, the possible extensions are illustrated by using other possible transformations that can be exploited accordingly.

Degree

thesis:*
Grantor dc:publisher
Πανεπιστήμιο Μακεδονίας
Year dc:date
2023

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Tserkezos, Eustratios
  • Τσερκέζος, Ευστράτιος

Subjects

dc:subject × 16

Rights

Language dc:language
gre

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/53773

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Tserkezos, Eustratios; Τσερκέζος, Ευστράτιος. Μέθοδοι μη γραμμικής συνολοκλήρωσης και αιτιότητας σε μικροοικονομικά και μακροοικονομικά μεγέθη της ελληνικής οικονομίας. Πανεπιστήμιο Μακεδονίας, 2023. http://hdl.handle.net/10442/hedi/53773