stepan
/
dynare
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #1424 from JohannesPfeifer/manual 

Manual changes related shock_decomposition
time-shift
Houtan Bastani 2017-03-28 11:52:51 +02:00 committed by GitHub
parent 37dd5fd077 b5f989ccb1
commit ad2e1ffc8d
1 changed files with 148 additions and 98 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

246
doc/dynare.texi
View File

@ -2,6 +2,8 @@
@c %**start of header
@setfilename dynare.info
@documentencoding UTF-8
@codequoteundirected on
@settitle Dynare Reference Manual
@afourwide
@dircategory Math
@ -206,6 +208,9 @@ The Model file
* Deterministic simulation::
* Stochastic solution and simulation::
* Estimation::
* Model Comparison::
* Shock Decomposition::
* Calibrated Smoother::
* Forecasting::
* Optimal policy::
* Sensitivity and identification analysis::
@ -1065,6 +1070,9 @@ end of line one and the parser would continue processing.
* Deterministic simulation::
* Stochastic solution and simulation::
* Estimation::
* Model Comparison::
* Shock Decomposition::
* Calibrated Smoother::
* Forecasting::
* Optimal policy::
* Sensitivity and identification analysis::
@ -1104,6 +1112,9 @@ mutually exclusive arguments are separated by vertical bars: @samp{|};
@item
@var{INTEGER} indicates an integer number;
@item
@var{INTEGER_VECTOR} indicates a vector of integer numbers [@var{INTEGER_1} ... @var{INTEGER_N}]
@item
@var{DOUBLE} indicates a double precision number. The following syntaxes
are valid: @code{1.1e3}, @code{1.1E3}, @code{1.1d3}, @code{1.1D3}. In
@ -6970,6 +6981,24 @@ estimates using a higher tapering are usually more reliable.
@end table
@end defvr
@deffn Command unit_root_vars @var{VARIABLE_NAME}@dots{};
This command is deprecated. Use @code{estimation} option @code{diffuse_filter} instead for estimating a model with non-stationary observed variables or @code{steady} option @code{nocheck} to prevent @code{steady} to check the steady state returned by your steady state file.
@end deffn
Dynare also has the ability to estimate Bayesian VARs:
@deffn Command bvar_density ;
Computes the marginal density of an estimated BVAR model, using
Minnesota priors.
See @file{bvar-a-la-sims.pdf}, which comes with Dynare distribution,
for more information on this command.
@end deffn
@node Model Comparison
@section Model Comparison
@deffn Command model_comparison @var{FILENAME}[(@var{DOUBLE})]@dots{};
@deffnx Command model_comparison (marginal_density = laplace | modifiedharmonicmean) @var{FILENAME}[(@var{DOUBLE})]@dots{};
@anchor{model_comparison}
@ -7045,6 +7074,8 @@ Posterior probability of the respective model
@end defvr
@node Shock Decomposition
@section Shock Decomposition
@deffn Command shock_decomposition [@var{VARIABLE_NAME}]@dots{};
@deffnx Command shock_decomposition (@var{OPTIONS}@dots{}) [@var{VARIABLE_NAME}]@dots{};
@ -7083,8 +7114,8 @@ calibrated model.
@xref{nobs}.
@item use_shock_groups [= @var{STRING}]
@anchor{use_shock_groups} Uses groups of shocks instead of individual shocks in
the decomposition. Groups of shocks are defined in the @ref{shock_groups} block.
@anchor{use_shock_groups} Uses shock grouping defined by the string instead of individual shocks in
the decomposition. The groups of shocks are defined in the @ref{shock_groups} block.
@item colormap = @var{STRING}
@anchor{colormap} Controls the colormap used for the shocks decomposition
@ -7092,12 +7123,11 @@ graphs. See @code{colormap} in Matlab/Octave manual for valid arguments.
@item nograph
@xref{nograph}. Suppresses the display and creation only within the
@code{shock_decomposition}-command but does not affect other commands.
@code{shock_decomposition}-command, but does not affect other commands.
@xref{plot_shock_decomposition} for plotting graphs.
@item init_state = @var{INTEGER}
@anchor{init_state} It can take values of @math{0} or @math{1}. If equal to
@math{0}, the shock decomposition is computed conditional on the smoothed state
@item init_state = @var{BOOLEAN}
@anchor{init_state} If equal to @math{0}, the shock decomposition is computed conditional on the smoothed state
variables in period @math{0}, @i{i.e.} the smoothed shocks starting in period
@math{1} are used. If equal to @math{1}, the shock decomposition is computed
conditional on the smoothed state variables in period @math{1}. Default:
@ -7106,10 +7136,12 @@ conditional on the smoothed state variables in period @math{1}. Default:
@outputhead
@defvr {MATLAB/Octave variable} oo_.shock_decomposition
@vindex oo_.shock_decomposition
@anchor{oo_.shock_decomposition}
The results are stored in the field @code{oo_.shock_decomposition}, which is a three
dimensional array. The first dimension contains the @code{M_.endo_nbr} endogenous variables.
The second dimension stores
The second dimension stores
in the first @code{M_.exo_nbr} columns the contribution of the respective shocks.
Column @code{M_.exo_nbr+1} stores the contribution of the initial conditions,
while column @code{M_.exo_nbr+2} stores the smoothed value of the respective
@ -7117,6 +7149,7 @@ endogenous variable in deviations from their steady state, @i{i.e.} the mean and
subtracted. The third dimension stores the time periods. Both the variables
and shocks are stored in the order of declaration, @i{i.e.} @code{M_.endo_names} and
@code{M_.exo_names}, respectively.
@end defvr
@end deffn
@ -7127,11 +7160,11 @@ and shocks are stored in the order of declaration, @i{i.e.} @code{M_.endo_names}
of the shock groups is written in a block delimited by @code{shock_groups} and
@code{end}.
Each line defines a group of shock as a list of exogenous variables:
Each line defines a group of shocks as a list of exogenous variables:
@example
SHOCK_GROUP_NAME = VARIABLE_1 [[,] VARIABLE_2 [,]@dots{}];
`SHOCK GROUP NAME' = VARIABLE_1 [[,] VARIABLE_2 [,]@dots{}];
'SHOCK GROUP NAME' = VARIABLE_1 [[,] VARIABLE_2 [,]@dots{}];
@end example
@optionshead
@ -7155,12 +7188,13 @@ varexo e_a, e_b, e_c, e_d;
shock_groups(name=group1);
supply = e_a, e_b;
`aggregate demand' = e_c, e_d;
'aggregate demand' = e_c, e_d;
end;
shocks_decomposition(use_shock_groups=group1);
@end example
This example defines a shock grouping with the name @code{group1}, containing a set of supply and demand shocks
and conducts the shock decomposition for these two groups.
@end deffn
@deffn Command realtime_shock_decomposition [@var{VARIABLE_NAME}]@dots{};
@ -7171,16 +7205,22 @@ shocks_decomposition(use_shock_groups=group1);
This command computes the realtime historical shock decomposition for a given
sample based on the Kalman smoother. For each period
@math{T=[@code{presample}@dots{}@code{nobs}]}, it computes the:
@math{T=[@code{presample},@dots{},@code{nobs}]}, it recursively computes three objects:
@itemize @bullet
@item
realtime historical shock decomposition @math{Y(t|T)} for @math{t=[1@dots{}T]},
@i{i.e.} without observing data in @math{[T+1@dots{}@code{nobs}]};
realtime historical shock decomposition @math{Y(t|T)} for @math{t=[1,@dots{},T]},
@i{i.e.} without observing data in @math{[T+1,@dots{},@code{nobs}]}. This results in a standard
shock decomposition being computed for each additional datapoint becoming available after @code{presample}.
@item
conditional shock decomposition @math{Y(T|T)} conditional on @math{Y(T|T-1)},
@i{i.e.} @math{Y(t|T)} for @math{t=[T-1@dots{}T]};
forecast shock decomposition @math{Y(T+k|T)} for @math{k=[1,@dots{},forecast]}, @i{i.e.} the @math{k}-step
ahead forecast made for every @math{T} is decomposed in its shock contributions.
@item
forecast shock decomposition @math{Y(T|T-1)}.
realtime conditional shock decomposition of the difference between the realtime historical shock decomposition and the
forecast shock decomposition. If @ref{vintage} is equal to @math{0}, it computes the effect of shocks realizing in period
@math{T}, @i{i.e.} decomposes @math{Y(T|T)-Y(T|T-1)}. Put differently it conducts a @math{1}-period ahead shock decomposition from
@math{T-1} to @math{T}, by decomposing the update step of the Kalman filter. If @code{vintage>0} and smaller than @code{nobs},
the decomposition is conducted of the forecast revision @math{Y(T+k|T+k)-Y(T+k|T)}.
@end itemize
Like @ref{shock_decomposition} it decomposes the historical deviations of the endogenous
@ -7228,40 +7268,79 @@ realtime shock decompositions are computed, @i{i.e.} for
@anchor{forecast_shock_decomposition} Compute shock decompositions up to
@math{T+k} periods, @i{i.e.} get shock contributions to k-step ahead forecasts.
@item save_realtime = [@var{integer1} ... @var{integern}]
@item save_realtime = [@var{INTEGER_VECTOR}]
@anchor{save_realtime} Choose for which vintages to save the full realtime
shock decomposition. Default: @math{0}.
@end table
@outputhead
@defvr {MATLAB/Octave variable} oo_.realtime_shock_decomposition
@vindex oo_.realtime_shock_decomposition
The results of realtime historical decompositions are stored in the field
@code{oo_.realtime_shock_decomposition}, which is a structure. Field
@code{pool} stores the pooled decomposition (@xref{plot_shock_decomposition}).
Fields @code{time_*} store the vintages of realtime historical shock
decompositions.
Structure storing the results of realtime historical decompositions. Fields are three-dimensional arrays with
the first two dimension equal to the ones of @ref{oo_.shock_decomposition}. The third dimension stores the time
periods and is therefore of size @code{T+forecast}. Fields are of the form:
@example
@code{oo_.realtime_shock_decomposition.@var{OBJECT}}
@end example
where @var{OBJECT} is one of the following:
@vindex oo_.conditional_shock_decomposition
The results of realtime conditional decompositions are stored in the field
@code{oo_.conditional_shock_decomposition}, which is a structure. Field
@code{pool} stores the pooled decomposition @math{Y(t|T)} for
@math{t=T-1@dots{}T} @xref{plot_shock_decomposition}. Conditional shock
decomposition sets the initial condition in @math{T-1}, so only computes the
effect of shocks in period @math{T}, @i{i.e.} it is just a @math{1}-period
shock decomposition from @math{T-1} to @math{T}. In practice it decomposes the
update step of the Kalman filter. Fields @code{time_*} store the vintages of
@math{k}-step conditional forecast shock decompositions @math{Y(t|T+k)}, for
@math{t=[T@dots{}T+k}. @xref{vintage}.
@table @code
@item pool
Stores the pooled decomposition, @i{i.e.} for every realtime shock decomposition terminal period
@math{T=[@code{presample},@dots{},@code{nobs}]} it collects the last period's decomposition @math{Y(T|T)}
(see also @ref{plot_shock_decomposition}). The third dimension of the array will have size
@code{nobs+forecast}.
@item time_*
Stores the vintages of realtime historical shock decompositions if @code{save_realtime} is used. For example, if
@code{save_realtime=[5]} and @code{forecast=8}, the third dimension will be of size 13.
@end table
@end defvr
@defvr {MATLAB/Octave variable} oo_.realtime_conditional_shock_decomposition
@vindex oo_.realtime_conditional_shock_decomposition
Structure storing the results of realtime conditional decompositions. Fields are of the form:
@example
@code{oo_.realtime_conditional_shock_decomposition.@var{OBJECT}}
@end example
where @var{OBJECT} is one of the following:
@table @code
@item pool
Stores the pooled realtime conditional shock decomposition, @i{i.e.} collects the decompositions of
@math{Y(T|T)-Y(T|T-1)} for the terminal periods @math{T=[@code{presample},@dots{},@code{nobs}]}.
The third dimension is of size @code{nobs}.
@item time_*
Store the vintages of @math{k}-step conditional forecast shock decompositions @math{Y(t|T+k)}, for
@math{t=[T@dots{}T+k]}. @xref{vintage}. The third dimension is of size @code{1+forecast}.
@end table
@end defvr
@defvr {MATLAB/Octave variable} oo_.realtime_forecast_shock_decomposition
@vindex oo_.realtime_forecast_shock_decomposition
The results of realtime forecast decompositions are stored in the field
@code{oo_.realtime_forecast_shock_decomposition}, which is a structure. Field
@code{pool} stores the pooled decomposition @xref{plot_shock_decomposition}.
Forecast shock decomposition computes the @math{1}-step ahead effect of shocks
on the @math{1}-step ahead prediction, @i{i.e.} @math{Y(T|T-1)}. Fields
@code{time_*} store the vintages of @math{k}-step out-of-sample forecast shock
Structure storing the results of realtime forecast decompositions. Fields are of the form:
@example
@code{oo_.realtime_forecast_shock_decomposition.@var{OBJECT}}
@end example
where @var{OBJECT} is one of the following:
@table @code
@item pool
Stores the pooled realtime forecast decomposition of the @math{1}-step ahead effect of shocks
on the @math{1}-step ahead prediction, @i{i.e.} @math{Y(T|T-1)}.
@item time_*
Stores the vintages of @math{k}-step out-of-sample forecast shock
decompositions, @i{i.e.} @math{Y(t|T)}, for @math{t=[T@dots{}T+k]}. @xref{vintage}.
@end table
@end defvr
@end deffn
@ -7272,12 +7351,11 @@ decompositions, @i{i.e.} @math{Y(t|T)}, for @math{t=[T@dots{}T+k]}. @xref{vintag
@descriptionhead
This command plots the historical shock decomposition already computed by
@code{shock_decomposition}. The @code{variable_names} provided govern which
@code{shock_decomposition} or @code{realtime_shock_decomposition}. For that reason,
it must come after one of these commands. The @code{variable_names} provided govern which
variables the decomposition is plotted for.
Note that this command must come after @code{shock_decomposition} or @code{realtime_shock_decomposition}.
Further note that, unlike the majority of dynare commands, the options
Further note that, unlike the majority of Dynare commands, the options
specified below are overwritten with their defaults before every call to
@code{plot_shock_decomposition}. Hence, if you want to reuse an option in a
subsequent call to @code{plot_shock_decomposition}, you must pass it to the
@ -7300,14 +7378,14 @@ command again.
@itemx graph_format = ( @var{FORMAT}, @var{FORMAT}@dots{} )
@xref{graph_format}.
@item detail_plot = @var{INT_NUMBER}
@item detail_plot
Plots shock contributions using subplots, one per shock (or group of
shocks). Pass @math{1} to turn it on and @math{0} to turn it off. Default:
@math{0}
not activated
@item interactive = @var{INT_NUMBER}
Under MATLAB, add uimenu's for detailed group plots. Pass @math{1} to turn it
on and @math{0} to turn it off. Default: @math{0}
@item interactive
Under MATLAB, add uimenus for detailed group plots. Pass @math{1} to turn it
on and @math{0} to turn it off. Default: not activated
@item screen_shocks
@anchor{screen_shcoks} For large models (@i{i.e.} for models with more than @math{16}
@ -7315,10 +7393,10 @@ shocks), plots only the shocks that have the largest historical contribution
for chosen selected @code{variable_names}. Historical contribution is ranked
by the mean absolute value of all historical contributions.
@item steadystate = @var{INTEGER}
@item steadystate
@anchor{steadystate} If equal to @math{1}, the the @math{y}-axis value of the
zero line in the shock decomposition plot is translated to the steady state
level. Default: @math{0}
level. Default: not activated
@item type = @code{qoq} | @code{yoy} | @code{aoa}
@anchor{type} For quarterly data, valid arguments are: @code{qoq} for
@ -7333,82 +7411,54 @@ default figure name set by @code{plot_shock_decomposition}. This can avoid to
overwrite plots in case of sequential calls to @code{plot_shock_decomposition}.
@item write_xls
@anchor{write_xls} Saves shock decompositions to excel.
@anchor{write_xls} Saves shock decompositions to Excel-file in the main directory, named
@code{FILENAME_shock_decomposition_TYPE_FIG_NAME.xls}. This option requires your system to be
configured to be able to write Excel files.@footnote{In case of Excel not being installed,
@url{https://mathworks.com/matlabcentral/fileexchange/38591-xlwrite--generate-xls-x--files-without-excel-on-mac-linux-win} may be helpful.}
@item realtime = @var{INTEGER}
@anchor{realtime} Which kind of shock decomposition to plot. @var{INTEGER} can take following values:
@itemize @bullet
@item
@code{0}: historical shock decomposition: @math{Y(t|T)} for
@math{t=[1@dots{}T]}, @math{T=} @code{nobs} full sample
@code{0}: standard historical shock decomposition. @xref{shock_decomposition}.
@item
@code{1}: realtime historical shock decomposition: for
@math{T=[1@dots{}@code{nobs}]}, realtime shock decomposition @math{Y(t|T)} for
@math{t=[1@dots{}T]}
@code{1}: realtime historical shock decomposition. @xref{realtime_shock_decomposition}.
@item
@code{2}: conditional shock decomposition: for @code{T=1:nobs}, realtime shock
decomposition of @math{Y(T|T)} conditional on @math{Y(T|T-1)}, @i{i.e.}
@math{Y(t|T)} for @math{t=[T-1@dots{}T]}
@code{2}: conditional realtime shock decomposition. @xref{realtime_shock_decomposition}.
@item
@code{3}: forecast shock decomposition: for @math{T=[1@dots{}@code{nobs}]},
realtime shock decomposition of @math{Y(T|T-1)}
@code{3}: realtime forecast shock decomposition. @xref{realtime_shock_decomposition}.
@end itemize
If no @ref{vintage} is requested, @i{i.e.} @code{vintage=0} then the pooled objects from @ref{realtime_shock_decomposition}
will be plotted and the respective vintage otherwise.
Default: @math{0}
@item vintage = @var{INTEGER}
@anchor{vintage} Applies if @code{realtime}@math{>0}. Can take following values:
@itemize @bullet
@item
@code{0}: plots @math{1}-step pooled shock decompositions
@item
@code{1}: pooled realtime shock decomposition. For @math{T=[1@dots{}@code{nobs}]}, plots last
time point @math{Y(T|T)} of each vintage shock decomposition @math{Y(1:T|T)}
@item
@code{2}: pooled conditional shock decomposition. For
@math{T=[1@dots{}@code{nobs}]}, realtime @math{1}-step shock decomposition of
@math{Y(T|T)} conditional on @math{Y(T|T-1)} (@i{i.e.} decomposition of
@math{1}-step filter updates of each vintage @math{T})
@item
@code{3}: pooled forecast shock decomposition. For
@math{T=[1@dots{}@code{nobs}]}, realtime @math{1}-step ahead shock
decomposition of @math{Y(T|T-1)} (@i{i.e.} decomposition of shock
contributions to @math{1}-step ahead forecasts of each vintage @math{T})
@end itemize
When the value passed is greater than @math{0}, it plots the shock
@anchor{vintage} Selects a particular data vintage in @math{[presample,@dots{},nobs]} for which to plot the results from
@ref{realtime_shock_decomposition} selected via the @ref{realtime} option. If the standard
historical shock decomposition is selected (@code{realtime=0}), @code{vintage} will have no effect. If @code{vintage=0}
the pooled objects from @ref{realtime_shock_decomposition} will be plotted. If @code{vintage>0}, it plots the shock
decompositions for vintage @math{T=@code{vintage}} under the following scenarios:
@itemize @bullet
@item
@code{realtime=1}: the full vintage shock decomposition @math{Y(t|T)} for
@math{t=[1@dots{}T]}
@math{t=[1,@dots{},T]}
@item
@code{realtime=2}: the conditional forecast shock decomposition from @math{T},
@i{i.e.} plots @math{Y(T+j|T+j)} and the shock contributions needed to get to
the data @math{Y(T+j)} conditional on @math{T=}@code{vintage}, with
@math{j=[0@dots{}@code{forecast}]}.
@math{j=[0,@dots{},@code{forecast}]}.
@item
@code{realtime=3}: plots unconditional forecast shock decomposition from
@math{T}, @i{i.e.} @math{Y(T+j|T)}, where @math{T=@code{vintage}} and
@math{j=[0@dots{}@code{forecast}]}.
@math{j=[0,@dots{},@code{forecast}]}.
@end itemize
Default: @math{0}
@end table
@end deffn
@deffn Command unit_root_vars @var{VARIABLE_NAME}@dots{};
This command is deprecated. Use @code{estimation} option @code{diffuse_filter} instead for estimating a model with non-stationary observed variables or @code{steady} option @code{nocheck} to prevent @code{steady} to check the steady state returned by your steady state file.
@end deffn
Dynare also has the ability to estimate Bayesian VARs:
@deffn Command bvar_density ;
Computes the marginal density of an estimated BVAR model, using
Minnesota priors.
See @file{bvar-a-la-sims.pdf}, which comes with Dynare distribution,
for more information on this command.
@end deffn
@node Calibrated Smoother
@section Calibrated Smoother
Dynare can also run the smoother on a calibrated model: