In the new CAMP working paper 10/2018, Canova and Matthes describe how to use the composite likelihood to ameliorate estimation, computational, and inferential problems in dynamic stochastic general equilibrium models. They propose a method that helps to solve these difficulties and provides parameter estimates and policy analyses that formally combine the information present in different models using a shrinkage-type procedure.
Canova and Matthes present examples indicating that a composite likelihood constructed using the information present in distinct models helps
- To ameliorate population and sample identification problems,
- To solve singularity problems,
- To produce more stable estimates of the parameters of large scale structural models
- To robustly the estimation of parameters appearing in multiple models and to rank models with different observables, and
- To combine information coming from different sources, frequencies, and levels of aggregation.