The researchers developed a new model called multivariate stochastic volatility with cross leverage, which extends a previous model to multiple variables. They used a method called Bayesian MCMC to estimate the model. The researchers sampled state variables using the Metropolis-Hastings algorithm and computed conditional modes of volatility variables using a scoring method. They also used an auxiliary particle filter to calculate the likelihood function. The model and techniques were tested using daily stock returns data from the Tokyo Stock Exchange.