Calculate a decomposition of the observed variables into the influence of the initial state, the mean (or steady-state), the structural shocks, and the measurement errors. You can choose between bar chart decompositions, scatter-plot decompositions (includes also the state variable), and paths. It is also possible to compute the decompositions using the prior and the posterior draws. The mean and the median of the decompositions are computed for these two sets of parameter draws.
Decompositions
The decompositions are carried out at initial or posterior mode parameter values using smooth estimation of state variables, structural shocks, and measurement errors. The observed variables can either be decomposed in their original form or using the annualization data in the data construction file.
Shock groups can optionally used in the decomposition. A shock group is simply a collection of structural shocks, where each structural shock belongs to exactly one group. The shock groups are set on Actions menu. The contributions of the individual shocks to each shock group can also be plotted in separate graphs.
Scatter-plot decompostions involve two variables, where one or both can be a state variable, and rely on the same data as the observed variable and state variable decompositions. Each point in the plots are given by either the pair of observed data or smooth estimate for the two variables of interest, along with pairs of shares in the decompositions of the two variables due to the individual shocks or shock groups.
Provided that shock groups are not plotted but individual shocks, it is possible to limit the selected number of structural shocks to at least one. Measurement errors and the initial state are excluded from the set of scatter-plots. In the graph below, the user has selected not to include the fiscal (government spending) shocks, but to limit attention to the monetary policy and technology shocks.
Paths
The path decompositions can be performed at initial or posterior mode parameter values as well as using draws from the prior or the posterior distribution. Smooth estimates of states, shocks and measurement errors underlie the computations. The path decompositions are available for either individual shocks or shock groups in original data or in annualized data. In contrast with the bar charts, the decomposed term for the paths is now given by the sum of the initial state term, the mean (or steady-state) term, the measurement error term and the individual shock or shock group term.
Additional Information
• | A more detailed description about observed variable decompositions can also be found in Section 11.8 of the YADA Manual. |
Page url: http://www.texlips.net/yada/help/index.html?observed_variable_decompositio.htm