Marginal likelihood

Hi all,

I am trying to estimate a small open economy model using Bayesian method. I obtain some reasonable estimates of parameters but the marginal likelihood obtained (using Laplace approximation or harmonic mean estimator) is positive rather than negative. Does this imply something is wrong with the model or estimation?

Thank you so much!


Hi Musmic, The answer is no. Dynare reports the logarithm of the marginal density. The (logged) density of the sample, depending on the model and/or the sample, can be positive or negative. The absolute level of this density is not really informative.

Best, St├ęphane.

Hi Stephane,

Many thanks! Your information is very useful.