On the performances of Dynamic Conditional Correlation models in the Sovereign CDS market and the corresponding bond market

Saker Sabkha (1) , Christian de Peretti (1)

1 SAF - Laboratoire de Sciences Actuarielle et Financière

The study of an efficient financial assets' modeling method is still an open hot issue especially during recent crises. Using credit risk data from 33 worldwide countries, this paper investigates the performance of 9 Dynamic Conditional Correlation models taking into account different properties of financial markets (long memory behavior, asymmetry and/or leverage effects...). This comparative study is based on the results of several multivariate diagnostic tests. Findings show that no model outperforms the others in all situations, though, the straightforward DCC-GARCH model seems to provide the most relevant estimator parameters. Yet, the innovations distributions assumption significantly impacts the statistical fit of the model. Our work is useful for financial markets' participants so as to making decision in terms of arbitrage, hedging or speculation. JEL Classification G11, G12, F02, C58

Mots clés: DCC-class models; Multivariate diagnostic tests; Time-varying correlation; Sovereign credit market

Domaines: Finance [q-fin.GN]

Saker Sabkha, Christian de Peretti. On the performances of Dynamic Conditional Correlation models in the Sovereign CDS market and the corresponding bond market. Financial and Economic Systems, WORLD SCIENTIFIC (EUROPE), pp.187-212, 2022, ⟨10.1142/9781786349507_0008⟩. ⟨hal-01710398⟩