This paper considers a class of C-convolution-based Markov models in which we assume that the error term is dependent on the first lagged state variable and the dependence structure is modeled by a copula function. Such models appear suitable for studying nonlinearity in time series. We show...