Contagion Detection with Switching Regime Models: a Short and Long Run Analysis

The paper evaluates the potentialities of Markov switching models (MS) in contagion analysis. We intend contagion as a break that produces non-linearities in the linkages among financial markets. The MS approach allows the detection of contagion in a more general framework since, differently from the previous literature, (i) the crisis period are endogenously defined by the MS model rather than arbitrarily and are specific for each country, (ii) we investigate the flight to quality effect, i.e. when the non-linear relationship among markets who implies a significant reduction of the link among markets during a crisis period, and (iii) we distinguish between short and long run breaks using Markov switching ECM models. We analyse the period of the Hong Kong stock market crash in 1997. The results show that (i) the relationship between developed markets strengthens, as that between the Hong Kong market and the US and European markets (i.e. contagion) and (ii) the factor loading of the error correction term shows a flight to quality effect suggesting that investors during crisis potentially ignore economic fundamentals.