Cardiff University | Prifysgol Caerdydd ORCA
Online Research @ Cardiff 
WelshClear Cookie - decide language by browser settings

Applications of state dependent models in finance

Guan, Bo ORCID: 2020. Applications of state dependent models in finance. PhD Thesis, Cardiff University.
Item availability restricted.

[thumbnail of 2021GuanBphd.pdf]
PDF - Accepted Post-Print Version
Download (7MB) | Preview
[thumbnail of Cardiff University Electronic Publication Form] PDF (Cardiff University Electronic Publication Form) - Supplemental Material
Restricted to Repository staff only

Download (645kB)


The State-Dependent Model (SDM) prescribes specific types of nonlinearity with linearity as special cases and can identify structural breaks within a time series. In a simulation study of various time series models, the SDM technique was able to capture the true type of linearity/non-linearity in the data. This thesis makes among the first attempts to apply the SDM to business, economics, and financial data. Its application to business cycle indicators suggests the presence of significant nonlinearity in most industrial production sectors, but the results are inconclusive in terms of symmetric or asymmetric nonlinearity. The SDM was also used to test Purchasing Power Parity. The study found that the real exchange rates (against the US dollar) for the Pound, Euro, Yen, are globally mean reverting with ESTAR characteristics, Brazilian Real as random walk, and that the PPP holds consistently for GBP/USD and JPY/USD. Additional analysis indicated that the higher the uncertainty level, the higher the degree of mean-reverting these real exchange rates have, and uncertainty events result in instantaneous shocks in real exchange rates before mean reversion took place. Finally, the forecasting performance of the SDM models was investigated and compared with the linear ARIMA, ETS, and Neural Network Autoregressive models. Employing two sets of real data, the study found that the SDM models possess superior forecasting ability in long-term forecasts for industrial production and Japanese tourism data.

Item Type: Thesis (PhD)
Date Type: Acceptance
Status: Unpublished
Schools: Business (Including Economics)
Uncontrolled Keywords: general state-dependent models, nonlinear models, time series
Date of First Compliant Deposit: 26 April 2021
Date of Acceptance: September 2020
Last Modified: 09 Nov 2022 10:49

Actions (repository staff only)

Edit Item Edit Item


Downloads per month over past year

View more statistics