Modelling country default risk as a latent variable: a multiple indicators multiple causes approach

We study the determinants of country default risk by applying a Multiple Indicators Multiple Causes (MIMIC) model. This accounts for the fact that country default risk is an unobservable variable. Whereas existing (regression-based) approaches typically use only one of several possible country default risk indicators as the dependent variable, the MIMIC model enables us to consider several indicators at once. The simultaneous consideration of sovereign yield spreads and Standard and Poor (S&P) ratings may help to improve the identification of the latent country default risk. Our results confirm most of the literature's main findings regarding important determinants of country default risk, refute others and provide new evidence to controversial questions.