A Bayesian Model Comparison for Trend-Cycle Decompositions of Output

Angelia Grant and Joshua Chan (2017)
Journal of Money, Credit and Banking, 49(2-3): 525-552
[ Journal Version | Working Paper | Code ]

This code estimates the output gaps from six univariate unobserved components models using US GDP data and two bivariate UC models.

The code also computes the marginal likelihood for each model using the improved cross-entropy method developed in Chan and Eisenstat (2015).

The six unobserved components models are:

  1. UC model with orthogonal trend and cycle innovations
  2. UC model with correlated trend and cycle innovations
  3. correlated UC model with one break in trend output growth
  4. correlated UC model with two breaks in trend output growth
  5. UC model with a deterministic trend output
  6. UC model with a deterministic trend output and one break
  7. bivariate UC model of GDP and unemployment with one break in trend output growth
  8. bivariate UC model of GDP and inflation with one break in trend output growth