Econometric modelling in a changing, globalised world

This thesis takes the literature on multi-country Bayesian Panel Vector Autoregressions as its starting point. In three self-contained but related essays, we refine and apply the econometric methods and modelling assumptions necessary to objectively consider different aspects of globalisation. The first essay analyses the Asia Pacific's possible decoupling from the US. We use Bayesian variable selection methods to model empirically relevant interdependencies between the US and eleven Asia Pacific countries. This allows us to capture regional interdependencies and bidirectional spillovers between the US and Asia Pacific. We show that shocks to US economic conditions, financial markets and uncertainty are important but regional shocks play a larger role in a typical Asia Pacific country. We also detect substantive spillovers from the Asia Pacific to US financial markets.;The second essay devises a novel econometric strategy to distinguish between: interdependence, contagion through interdependence and abrupt contagion. Appealing to multiple definitions of contagion, we allow the nature and magnitude of interdependencies and transmission channels selected for inclusion to changeover time. Using our approach, we analyse crisis episodes in Latin America. We only detect abrupt contagion during the global financial crisis from the US to Argentina and Brazil. During crises, results also show that macroeconomic and uncertainty channels play a role not just financial channels. The third essay uses professional forecast data to analyse spillovers in five components of uncertainty across advanced and emerging economies. Uncertainty surrounding output growth, inflation, the interest rate, exchange rate and current account is considered. While the US affects other economies through interest rate, exchange rate and current account uncertainty, interest rate and inflation uncertainty spillovers are also seen from the Eurozone and UK. Uncertainty spillovers are more frequently observed, but smaller when forecaster disagreement rather than the variance of mean forecast errors is used to proxy uncertainty.

Name dc:type.qualificationname

phd

Level dc:type.qualificationlevel

doctoral-pg

Grantor dc:publisher.institution

University of Strathclyde

Year dc:date.issued

2020

Author dc:creator

• Davidson, Sharada Nia

Advisors dc:contributor.advisor

• Darby, Julia
• Koop, Gary

DOI

10.48730/1wyx-hg80

Identifier

T15738

Author Identifier

201666627

OAI identifier oai:identifier

oai:strathclyde:sx61dm28w