[Thesis]. Manchester, UK: The University of Manchester; 2019.
In this thesis, I propose an alternative simple and accurate measure of market integration
during financial crises based on Pukthuanthong and Roll (2009). I examine how financial
crises affect market integration at the aggregate level in equity markets and in bond
markets, and at the country and industry level in industry portfolios.
The first essay investigates the determinants of explanatory power in a multi-factor
model during global crises. We find that explanatory power is determined by three
elements: factor heteroscedasticity, changes in factor loadings and residual heteroscedasticity.
Using a counterfactual analysis, we establish the effects of each element on integration
for 53 financial markets during six recent crisis periods: the 1987 US crisis, the
1994-1995 Mexican crisis, the 1997 Asian crisis, the 1998 Russian/LTCM crisis, the
2007-2009 Global Financial crisis (GFC) and the 2009-2014 European Sovereign Debt
crisis (ESDC). We find that the unconditional market integration is much lower for
most markets during crisis periods than implied. Moreover, high factor volatility
and changes in factor loadings during most crises typically causes upward changes
in the percentage of one marketĂ˘s return explained by global risk factors. The influence
of residual heteroscedasticity is negative on explanatory power and after adjusting
for the bias caused by this factor, explanatory power becomes larger. We also find
contagion exists worldwide in most crises except for the 1994-1995 Mexican crisis
and 2009-2014 ESDC. Besides, there is strong evidence of increasing market integration
in most markets,
The second essay decomposes market integration and investigates the degree and dynamics
of industry-level and country-level market integration in 640 industry portfolios.
The market integration is estimated by the explanatory power of global country or
industry risk factors on industry portfoliosĂ˘ returns during stable periods. Explanatory
power is adjusted by the bias caused by factor and residual heteroscedasticity and
changes in factor loadings during financial crises. Country-level market integration
is much higher than industry-level market integration over time during stable periods,
but the differences of the two types of market integration become small during financial
crises in most cases. Besides, country-level market integration has an increasing
trend over time but industry-level market integration illustrates different trend
across different industries. Finally, the country effects dominant industry effects
during normal periods but during crises, the industry effects become strong and play
an indispensable role in many industries, including Consumer Goods, Financials, Industrials
and Oil & Gas.
The third essay studies the dynamics of market integration in government bond markets.
It investigates how the method proposed by Pukthuanthong and Roll (2009) suffers from
bias in measuring market integration during financial crises, the differences in market
integration across markets, whether markets become more integrated over time and the
effects of maturities on market integration. We argue that market integration can
be measured by the explanatory power of global risk factors on bond index returns
during stable periods but this measure must be corrected for bias during the periods
of financial crisis. The results indicate that the method of Pukthuanthong and Roll
(2009) underestimates market integration during crises and the bias-adjusted method
provides a more accurate measure of market integration. Developed markets experience
increasing market integration over time, more than emerging markets. Most of the emerging
markets provide no evidence of greater market integration. The EMU markets become
almost fully integrated after the introduction of the Euro. Market integration also
increases with maturity.