How does the correlation of international markets evolve during crisis?
2020, the year of pandemic and health disasters, but also of high volatility and crash of the financial markets. Comparable to 2018, between February 19 and March 9, 2020, the CAC lost nearly 23% of its value. On March 23, the S&P 500 lost 2.9% of its value and the VIX is at 62 points. On March 17, 2020, short selling is banned in France, and on April 20, 2020, oil prices became negative. The equity markets touched the ground while the public debt scraped the sky.
French version :
FINETUDES.CORRELATION.FRANCAIS
.pdf
Download PDF • 1000KB
English version :
FINETUDES.CORRELATION
.ENGLISH
Download ENGLISH • 910KB
It is for such moments that the theory of portfolio diversification has been adopted, correlations between markets are taken into account in asset allocations alongside returns and volatilities, but many works have proved that correlations change according to the regime of the markets: the correlation between assets in bull markets differs from the correlation in bear markets.
The assertions of Andrew Ang and Geert Bekaert in 1999 that 'if you can't rely on diversification in times of crisis, you may have to change the rules of the game' shed light on a major point in diversification theory which implies that correlations are stable regardless of the market regime.
Arriving in 2001, Chesnay and Jondeau study international correlation in several markets such as the S&P, DAX and FTSE between 1988 and 1999 and conclude that correlations increase in periods of turbulence.
This paper will draw on Chesnay and Jondeau's studies and project their results over the last ten years in order to understand, using the Markov Switching Model, the behavior of international market correlation during periods of turbulence.
Beginning with a review of the literature, we will move on to the Markov Switching Model theory to conduct a practical case that will be used to answer a specific question of : In view of the turbulence of the last ten years, could we reject the hypothesis of constancy of correlations and affirm their dependence on volatility regimes?
