Overview of VAR objects
VAR objects can be constructed as plain VARs or simple panel VARs (with
fixed effect), and estimated without or with prior dummy observations
(quasi-bayesian VARs). VAR objects are reduced-form models but they are also
the point of departure for identifying structural VARs
VAR models in IRIS have the following form:
\[
y_t = \sum_{k=1}^{p} A_k\, y_{t-k} + K + J g_t + \epsilon_t
\]
where
- \(y_t\) is an \(n\)-by-1 vector of endogenous variables;
- \(A_k\) are transition matrices at lags 1, ..., k;
- \(K\) is a vector of intercepts;
- \(g_t\) is a vector of exogenous variables;
- \(J\) is the impact matrix of exogenous variables;
- \(\epsilon_t\) is a vector of forecast (reduced-form) errors, with \(\Omega=\mathrm{E}[\epsilon_t \epsilon_t']\).
Categorical list
Constructor
Function |
Description |
VAR |
Create new empty reduced-form VAR object |
Properties Directly Accessible
Function |
Description |
addToDatabank |
Add VAR parameters to databank or create new databank |
comment |
Get or set user comments in an IRIS object |
companion |
Matrices of first-order companion VAR |
eig |
Eigenvalues of a VAR process |
fprintf |
Write VAR model as formatted model code to text file |
get |
Query VAR object properties |
testCompatible |
True if two VAR objects can occur together on the LHS and RHS in an assignment |
isexplosive |
True if any eigenvalue is outside unit circle |
isstationary |
True if all eigenvalues are within unit circle |
length |
Number of parameter variants in VAR object |
mean |
Asymptotic mean of VAR process |
nfitted |
Number of data points fitted in VAR estimation |
rngcmp |
True if two VAR objects have been estimated using the same dates |
sprintf |
Print VAR model as formatted model code |
sspace |
Quasi-triangular state-space representation of VAR |
userdata |
Get or set user data in an IRIS object |
Referencing VAR Objects
Function |
Description |
group |
Retrieve VAR object from panel VAR for specified group of data |
subsasgn |
Subscripted assignment for VAR objects |
subsref |
Subscripted reference for VAR objects |
Simulation, Forecasting and Filtering
Function |
Description |
ferf |
Forecast error response function |
filter |
Filter data using a VAR model |
forecast |
Unconditional or conditional VAR forecasts |
instrument |
Define forecast conditioning instruments in VAR models |
resample |
Resample from a VAR object |
simulate |
Simulate VAR model |
Manipulating VARs
Function |
Description |
assign |
Manually assign system matrices to VAR object |
alter |
Expand or reduce the number of alternative parameterisations within a VAR object |
backward |
Backward VAR process |
demean |
Remove constant and the effect of exogenous inputs from VAR object |
horzcat |
Combine two compatible VAR objects in one object with multiple parameterisations integrate - Integrate VAR process and data associated with it |
xasymptote |
Set or get asymptotic assumptions for exogenous inputs |
Stochastic Properties
Function |
Description |
acf |
Autocovariance and autocorrelation functions for VAR variables |
fmse |
Forecast mean square error matrices |
vma |
Matrices describing the VMA representation of a VAR process |
xsf |
Power spectrum and spectral density functions for VAR variables |
Estimation, Identification, and Statistical Tests
Function |
Description |
estimate |
Estimate a reduced-form VAR or BVAR |
infocrit |
Populate information criteria for a parameterised VAR |
lrtest |
Likelihood ratio test for VAR models |
portest |
Portmanteau test for autocorrelation in VAR residuals |
schur |
Compute and store triangular representation of VAR |