This paper describes the simulation of a propagation mechanism of counterparty risk through banks, as a "domino effect" via Credit Default Swap (CDS).
For this purpose, a model that uses Markov and Monte Carlo methods is developed. It creates a typical set of borrowing companies and lending banks in order to analyze interactions between these economic agents. In the first part, samples of companies' initial assets sizes are uniformly chosen over a given interval. To estimate probability of default (PD), the structural model of Merton, which uses a Geometric Brownian Motion to simulate the net asset values of the firms through time, is implemented. Depending on these PDs, a "first rank" bank chooses whether it will keep the loan in its portfolio or sell its risk through securitization to another bank. For this second option, a CDS is priced and sold to a "second rank" bank and banks are thus linked with each other. As the assets and the liabilities of the firms are made to evolve, the number of defaults and the amount of loss that propagate to other banks are determined. Default may happen to a bank of first rank, or of second rank, or the ultimate banker, the central national bank. The model developed explores this "domino effect" through several years. Thousands of iterations are run for different scenarios of macroeconomic and microeconomic sets of parameters. We try to illustrate by approximation how banks' portfolios react to interbank CDS. The result sought is: -That a bank should not exceed a certain ratio of CDS of a given risk level in its portfolio in order to ensure strong capacities to meet its financial commitments; -And that this ratio, and speed of propagation are dependent on the concentration ratio of the banking system and initial balance sheet structures.