Modelling and testing interconnections in dynamic multivariate systems
(2022)
Files
FINAL_THESIS2.pdf
Restricted Access - Adobe PDF
- 4.96 MB
Details
- Authors
- Supervisors
- Candelon , Bertrand
- Abstract
- In a globalized world, modelling interconnections among countries/markets is fundamental for correct inferences. The literature in dynamic multivariate modelling is often divided into data- and assumption-driven models. While the former ones fit rather well the data, the latter ones feature more intuitive economic interpretation. The purpose of the thesis is to (i) build a bridge between those two strands of the literature and (ii) offer new perspectives regarding the degrees of integration/fragmentation of a system. Starting from modelling the term structure of interest rate in the euro area, we show that a model featuring both common and idiosyncratic risk factors captures most of the observed variation in realized excess returns. Moreover, it allows to assess fragmentation risk (i.e., the risk of divergent paths of local sovereign bond yields) not only as independence (i.e., predominance of local risk factors) but also as heterogeneous exposure to common shocks. Secondly, we propose a validity test for ex ante proposed channels of interconnection, showing that not only the tightness of the interconnection matters, but also its sign. Focusing on the long-term sovereign bond market, we show that sign-adjusted models document fragmentation risk in the euro area well before the outburst of the sovereign debt crisis in late 2009, and indicate a rise in fragmentation risk during the COVID-pandemic period. Lastly, to analyse the co-movements in global house prices, we propose and implement a new method for identifying the common components among structural shocks. We show that expectation and mortgage rate shocks account for most of the observed co-movement in global house prices. Moreover, while supply and demand shocks tend to account for the observed trends, mortgage and expectation shocks account most of the boom-bust cycles observed in house prices.
- Affiliations
Citations
Luisi, A. (2022). Modelling and testing interconnections in dynamic multivariate systems. https://hdl.handle.net/2078.5/102748