Ts Dyn

Package ‘tsDyn’
February 1, 2024
Type Package
Title Nonlinear Time Series Models with Regime Switching
Version 11.0.4.1
Date 2023-01-26
Depends R (>= 3.5.0)
Imports mnormt, mgcv, nnet, tseriesChaos, tseries, utils, vars, urca,
forecast, MASS, Matrix, foreach, methods
Suggests sm, scatterplot3d, rgl, tidyverse, rugarch
Maintainer Matthieu Stigler <[email protected]>
Description Implements nonlinear autoregressive (AR) time series models. For univariate se-
ries, a non-parametric approach is available through additive nonlinear AR. Parametric model-
ing and testing for regime switching dynamics is available when the transition is either di-
rect (TAR: threshold AR) or smooth (STAR: smooth transition AR, LSTAR). For multivariate se-
ries, one can estimate a range of TVAR or threshold cointegration TVECM mod-
els with two or three regimes. Tests can be con-
ducted for TVAR as well as for TVECM (Hansen and Seo 2002 and Seo 2006).
License GPL (>= 2)
URL https://github.com/MatthieuStigler/tsDyn/wiki
BugReports https://github.com/MatthieuStigler/tsDyn/issues
RoxygenNote 7.2.1
LazyData true
NeedsCompilation yes
Author Antonio Fabio Di Narzo [aut] (<https://orcid.org/0000-0002-4033-5038>),
Jose Luis Aznarte [ctb] (<https://orcid.org/0000-0002-1636-244X>),
Matthieu Stigler [aut, cre] (<https://orcid.org/0000-0002-6802-4290>)
Repository CRAN
Date/Publication 2024-02-01 09:50:58 UTC
1
2 R topics documented:
R topics documented:
tsDyn-package . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
aar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
accuracy_stat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
addRegime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
ar_mean . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
autopairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
autotriples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
autotriples.rgl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
availableModels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
barry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
BBCTest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
charac_root . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
coefB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
delta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
delta.lin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
fevd.nlVar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
fitted.nlVar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
getTh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
GIRF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
IIPUs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
irf.linear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
isLinear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
KapShinTest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
lags.select . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
LINEAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
lineVar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
llar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
logLik.nlVar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
LSTAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
m.unrate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
MakeThSpec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
MAPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
mse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
nlar-methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
NNET . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
plot methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
plot_ECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
predict.nlar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
predict.TVAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
predict_rolling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
rank.select . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
rank.test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
regime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
resample_vec . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
resVar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
selectHyperParms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
tsDyn-package 3
selectSETAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
SETAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
setar.sim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
setarTest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
setarTest_IIPUs_results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
sigmoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
STAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
toLatex.setar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
TVAR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
TVAR.LRtest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
TVAR.sim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
TVECM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
TVECM.HStest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
TVECM.SeoTest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
TVECM.sim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
UsUnemp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
VAR.sim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
VARrep . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
VECM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
VECM_symbolic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
zeroyld . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Index 104
tsDyn-package Getting started with the tsDyn package
Description
Getting started with the tsDyn package
Details
This package provide some tools inspired by nonlinear dynamics for the analysis-modelling of ob-
served time series.
For loading the package, type:
library(tsDyn)
A good place to start learning the package usage, is the vignette. It contains a more detailed guide
on package contents, and an applied case study. At the R prompt, write:
vignette("tsDyn")
For a full list of functions exported by the package, type:
ls("package:tsDyn")
There is also an experimental GUI for built-in NLAR models. Call it with:
nlarDialog(timeSeries)
where timeSeries is an available time series object.
Each exported function has a corresponding man page (some man pages are in common to more
functions). Display it by typing
help(functionName)
4 aar
Author(s)
Antonio, Fabio Di Narzo
See Also
availableModels for listing all currently available NLAR models
autopairs,autotriples,autotriples.rgl for graphical exploratory functions
llar, delta, delta.lin for nonlinearity checking tools
aar Additive nonlinear autoregressive model
Description
Additive nonlinear autoregressive model.
Usage
aar(x, m, d=1, steps=d, series)
Arguments
x time series
m, d, steps embedding dimension, time delay, forecasting steps
series time series name (optional)
Details
Nonparametric additive autoregressive model of the form:
m
X
xt+s = µ + sj (xt−(j−1)d )
j=1
where sj are nonparametric univariate functions of lagged time series values. They are represented
by cubic regression splines. sj are estimated together with their level of smoothing using routines
in the mgcv package (see references).
Value
An object of class nlar, subclass aar, i.e. a list with mostly internal structures for the fitted gam
object.
Author(s)
accuracy_stat 5
References
Wood, mgcv:GAMs and Generalized Ridge Regression for R. R News 1(2):20-25 (2001)
Wood and Augustin, GAMs with integrated model selection using penalized regression splines and
applications to environmental modelling. Ecological Modelling 157:157-177 (2002)
Examples
#fit an AAR model:
mod <- aar(log(lynx), m=3)
#Summary informations:
summary(mod)
#Diagnostic plots:
plot(mod)
accuracy_stat Forecasting accuracy measures.
Description
Compute forecasting accuracies. This is very similar to the accuracy method form forecast.
Usage
accuracy_stat(object, ...)
## Default S3 method:
accuracy_stat(object, true, ...)
## S3 method for class 'pred_roll'

accuracy_stat(object, w, ...)
Arguments
object A data-frame, matrix, or object of class pred_roll
... Not used currently.
true If object is just a matrix or data-frame, true values to be compared to should be
supplied
w Optional. For objects of class pred_roll containing multiple variables, user
can specify the way to aggregate the specific x-step-ahead into the ‘all’ category
Details
The function works either for a simple data.frame or for objects pred_roll. For simple data.frames,
the argument true, i.e. a data frame containing the true values, has to be provided. For pred_roll
objects, the true values are contained in the object, so no need (nor possibility) to provide the true
values.
6 addRegime
Value
A data-frame containing the forecasting accuracy measures.
Author(s)
Matthieu Stigler
Examples
## univariate:
mod_ar <- linear(lynx[1:100], m=1)
mod_ar_pred <- predict_rolling(mod_ar, newdata=lynx[101:114])
accuracy_stat(object=mod_ar_pred$pred, true=mod_ar_pred$true)
## multivariate
data(barry)
mod_var <- lineVar(barry, lag=1)
mod_var_pred <-predict_rolling(object=mod_var, nroll=10, n.ahead=1:3)

accuracy_stat(object=mod_var_pred)
accuracy_stat(object=mod_var_pred, w=c(0.7, 0.2, 0.1))
addRegime addRegime test
Description
addRegime test
Usage
addRegime(object, ...)
Arguments
object fitted model object with at least 2 regimes
... arguments to and from other methods
Value
A list containing the p-value of the F statistic and a boolean, true if there is some remaining nonlin-
earity and false otherwise.
ar_mean 7
Author(s)
J. L. Aznarte
References
TODO
See Also
star
Examples
##TODO
ar_mean Long-term mean of an AR(p) process
Description
Computes the long term mean of an AR process
Usage
ar_mean(object, ...)
## S3 method for class 'linear'

## S3 method for class 'setar'

## S3 method for class 'lstar'

Arguments
object an object of class linear, setar or lstar

... unused argument
8 autopairs
Details
The function computes the long-term mean of an AR(p) process, or of the corresponding sub-
regimes in SETAR or LSTAR model. There are three possible cases:
No constant nor trend The LT mean is 0

constant The LT mean is given by const/(1-sum(AR coefs))
Trend The LT mean is not defined
Examples
## estimate a (linear) AR, a SETAR and a LSTAR
lin_cst_l1 <- linear(lh, m = 1, include = "const")
set_cst_l1 <- setar(lh, m = 1, include = "const")
lst_cst_l1 <- lstar(lh, m = 1, include = "const", trace = FALSE)
ar_mean(lin_cst_l1)
ar_mean(set_cst_l1)
ar_mean(lst_cst_l1)
autopairs Bivariate time series plots
Description
Bivariate time series plots: scatterplots, directed lines and kernel density estimations using functions
in the sm package.
Usage
autopairs(
x,
lag = 1,
h,
type = c("levels", "persp", "image", "lines", "points", "regression")
)
Arguments
x time series
lag time lag
h kernel window (useful only for kernel estimations)
type type of plot: contour levels, perspective plots, image, directed lines, points or
points with superposed kernel regression
autotriples 9
Details
Bivariate time series plots: scatterplots, directed lines and kernel density and regression func-
tions estimations using functions in the package sm. In particular, for kernel density estimation
sm.density is used, with smoothing parameter h defaulting to hnorm. For kernel regression,
sm.regression is used.
Value
None. Plots are produced on the default graphical device.
Author(s)
Wrappers to sm by Antonio, Fabio Di Narzo
See Also
For finer control on density estimation, consider using directly sm.density and, especially, sm.ts.pdf
from package sm.
Examples
x <- log10(lynx)
autopairs(x, lag=2, type="lines")
autotriples Trivariate time series plots
Description
Trivariate time series plots: kernel autoregression using functions in the sm package
Usage
autotriples(
x,
lags = 1:2,
h,
type = c("levels", "persp", "image", "lines", "points")
)
Arguments
x time series
lags vector of regressors lags
h kernel window
type type of plot: contour levels, perspective plots, image
10 autotriples.rgl
Details
This function displays trivariate time series plots, i.e. kernel regression of x[t − lags[1]], x[t −
lags[2]] against x[t] using functions in the package sm. In particular, sm.regression is used, with
smoothing parameter defaulting to hnorm(x).
Value
None. Plots are produced on the default graphical device.
Author(s)
Wrappers to sm by Antonio, Fabio Di Narzo
See Also
For finer control on kernel regression, consider using directly sm.regression and, especially,
sm.autoregression in package sm.
Examples
autotriples(log(lynx))
autotriples(log(lynx), type="persp")
autotriples(log(lynx), type="image")
autotriples.rgl Interactive trivariate time series plots
Description
Interactive trivariate time series plots
Usage
autotriples.rgl(x, lags = 1:2, type = c("lines", "points"))
Arguments
x time series
lags vector of regressors lags
type type of plot: contour levels, perspective plots, image
Details
This function displays interactive trivariate time series plots x[t-lags[1]], x[t-lags[2]] against
x[t] using the interactive rgl device.
availableModels 11
Value
None. A plot is produced on the current rgl device.
Author(s)
Wrapper to ’sm’ and GUI by Antonio, Fabio Di Narzo
See Also
autotriples for 3d visualization via scatterplot3d package and for kernel post-processing of the
cloud for nonparametric autoregression functions estimates.
Examples
if(interactive())
autotriples.rgl(log(lynx))
availableModels Available models
Description
Available built-in time series models
Usage
availableModels()
Details
Return the list of built-in available ‘nlar’ time series models
Value
A character vector containing built-in time series models. For help on a specific model, type:
help(modelName).
Author(s)
Examples
availableModels()
12 BBCTest
barry Time series of PPI used as example in Bierens and Martins (2010)
Description
This data set contains the series used by Bierens and Martins for testing for PPI between Canada
and US.
Usage
data(barry)
Format
A data frame with 324 monthly observations, ranging from 1973:M1 until 1999:M12.
dolcan Exchange rate US/Can dollar.

cpiUSA US Consumer Price Index.
cpiCAN Canada Consumer Price Index.
Author(s)
Matthieu Stigler
Source
Bierens, H. and Martins, L. (2010), Time Varying Cointegration,
BBCTest Test of unit root against SETAR alternative
Description
Test of unit root against a stationnary three regime SETAR alternative
Usage
BBCTest(
x,
m,
series,
testStat = c("LR", "Wald", "LM"),
trim = 0.1,
grid = c("minPerc", "minObs")
)
BBCTest 13
Arguments
x time series
m Number of lags under the alternative
testStat Type of test statistic to use
trim trimming parameter indicating the minimal percentage of observations in each
regime
grid Whether a minimal number of percentage or observations should be imposed.
See details
Details
TODO
Value
A object of class "BBC2004Test" containing:

-The value of the sup Test
-The version of test used (either Wald, LM or LR).
Author(s)
Matthieu Stigler
References
Bec, Ben Salem and Carrasco (2004) Tests for Unit-Root versus Threshold Specification With an
Application to the Purchasing Power Parity Relationship, Journal of Business and Economic Statis-
tics; 22:4
See Also
setarTest for a test with stationarity as a null.
Examples
BBCTest(lynx, m=3, test="Wald", grid="minPerc")

14 coefB
charac_root Characteristic roots of the AR coefficients
Description
Computes the (inverse) characteristic roots of the auto-regressive coefficients. To be stationary, the
values should be outside the unit circle. The function here returns the modulus of the roots.
Usage
charac_root(object, ...)
## S3 method for class 'nlar'

charac_root(object, ...)
Arguments
object object of class nlar
... currently unused
Details
Computes the roots of the polynomial (1, -phi) using function polyroot
Value
a data.frame, with the modulus of the roots. For models with multiple regimes (setar, lstar, star),
one column per regime.
Examples
mod.ar <- linear(lh, m = 5, include = "const")
mod.setar <- setar(lh, m = 5, include = "const")
charac_root(mod.ar)
charac_root(mod.setar)
coefB Extract cointegration parameters A, B and PI
Description
Extract parameters in VECM: adjustment coefficients A, cointegrating coefficients B , or the com-
posite matrix PI
coefB 15
Usage
coefB(object, ...)
## S3 method for class 'VECM'

coefB(object, ...)
## S3 method for class 'ca.jo'

coefB(object, r = 1, normalize = TRUE, ...)
coefA(object, ...)

coefA(object, ...)
## S3 method for class 'ca.jo'

coefA(object, r = 1, normalize = TRUE, ...)
coefPI(object, ...)
Arguments
object An object of class VECM, ca.jo
r The cointegrating rank
normalize Whether to normalize the A/B coefficients. See details
... Further arguments passed to methods
Details
The functions extract the parameters from a VECM with K variables and rank r:
A Adjustment coefficients, of dim K × r
B Cointegrating coefficients, of dim K × r
0
Pi Matrix Π = AB , of dim K × K
Coefficients are extracted from a VECM in package tsDyn, or from a VECM obtained in package
urca from ca.jo or cajorls.
Note that by default, the A and B coefficients returned are normalized (see below). This is the case
for results obtained from VECM/lineVar and cajorls, while for ca.jo, the user has the choice (but
normalize=TRUE by default), in which case the rank r is also to be specified. The normalization
is the Phillips triangular representation, as suggested by Johansen (1995, p. 72), standardising the
first r × r coefficients to Ir :
0 0
B Bnorm = B(c B)−1 with c = (Ir , 0K−r,r )
0
A Anorm = AB c
Finally, note that the function also apply to objects obtained from tests of class ca.jo.test (from
blrtest etc...). Care should be taken however, since the normalization might override the restric-
tions imposed.
16 delta
Value
A matrix containing the coefficients
Author(s)
Matthieu Stigler
References
Johansen, Soren, (1995), Likelihood-Based Inference in Cointegrated Vector Autoregressive Mod-
els, Oxford University Press
Examples
data(barry)
vecm <- VECM(barry, lag=1, estim="ML")
vecm_r2 <- VECM(barry, lag=1, estim="ML", r=2)
## extract coefficients:
coefA(vecm)
coefB(vecm)
coefPI(vecm)
coefB(vecm_r2)
coefPI(vecm_r2)
## Beta-Restricted VECM:
beta_vecm2 <- coefB(vecm_r2)
beta_vecm2[3,2] <- 0.02
vecm_r2_rest <- VECM(barry, lag=1, estim="ML", r=2, beta=beta_vecm2)
round(coefB(vecm_r2_rest),5)
## Package vars/urca
if(require(urca)){
vecm_ur <- ca.jo(barry, K=2)
coefB(vecm_ur)
coefB(vecm_ur,r=2)
coefB(cajorls(vecm_ur, r=2))
all.equal(coefB(vecm), coefB(vecm_ur), check.attributes=FALSE)
all.equal(coefB(vecm_r2), coefB(vecm_ur, r=2), check.attributes=FALSE)
}
delta delta test of conditional independence
Description
delta statistic of conditional independence and associated bootstrap test
delta 17
Usage
delta(x, m, d = 1, eps)
delta.test(
x,
m = 2:3,
d = 1,
eps = seq(0.5 * sd(x), 2 * sd(x), length = 4),
B = 49
)
Arguments
x time series
m vector of embedding dimensions
d time delay
eps vector of length scales
B number of bootstrap replications
Details
delta statistic of conditional independence and associated bootstrap test. For details, see Man-
zan(2003).
Value
delta returns the computed delta statistic. delta.test returns the bootstrap based 1-sided p-value.
Warning
Results are sensible to the choice of the window eps. So, try the test for a grid of m and eps values.
Also, be aware of the course of dimensionality: m can’t be too high for relatively small time series.
See references for further details.
Author(s)
References
Sebastiano Manzan, Essays in Nonlinear Economic Dynamics, Thela Thesis (2003)
See Also
BDS marginal independence test: bds.test in package tseries
Teraesvirta’s neural network test for nonlinearity: terasvirta.test in package tseries
delta test for nonlinearity: delta.lin.test
18 delta.lin
Examples
delta(log10(lynx), m=3, eps=sd(log10(lynx)))
delta.lin delta test of linearity
Description
delta test of linearity based on conditional mutual information
Usage
delta.lin(x, m, d = 1)
delta.lin.test(
x,
m = 2:3,
d = 1,
eps = seq(0.5 * sd(x), 2 * sd(x), length = 4),
B = 49
)
Arguments
x time series
m vector of embedding dimensions
d time delay
eps vector of length scales
B number of bootstrap replications
Details
delta test of linearity based on conditional mutual information
Value
delta.lin returns the parametrically estimated delta statistic for the given time series (assuming
linearity). delta.lin.test returns the bootstrap based 1-sided p-value. The test statistic is the
difference between the parametric and nonparametric delta estimators.
Author(s)
fevd.nlVar 19
References
Sebastiano Manzan, Essays in Nonlinear Economic Dynamics, Thela Thesis (2003)
Examples
delta.lin(log10(lynx), m=3)
fevd.nlVar Forecast Error Variance Decomposition
Description
Use the fevd function from package vars to compute the forecast error variance decomposition of
a VAR(p) or VECM for n.ahead steps.
Usage
## S3 method for class 'nlVar'
fevd(x, n.ahead = 10, ...)
Arguments
x Object of class ‘VAR’ generated by lineVar(), or an object of class ‘VECM’ gen-
erated by VECM()
n.ahead Integer specifying the number of steps.
... Currently not used.
Details
The function converts the VAR or VECM computed by package tsDyn into an object of class
‘vec2var’, on which then the fevd method is applied. For details, see package vars.
Value
A list with class attribute ‘varfevd’ of length K holding the forecast error variances as matrices.
Author(s)
Bernhard Pfaff
References
Hamilton, J. (1994), Time Series Analysis, Princeton University Press, Princeton.
Lutkepohl, H. (2006), New Introduction to Multiple Time Series Analysis, Springer, New York.
20 fitted.nlVar
See Also
plot for the plot method. lineVar, VECM for the models.
Examples
data(zeroyld)
mod_vecm <- VECM(zeroyld, lag = 2)
fevd(mod_vecm, n.ahead = 5)
fitted.nlVar fitted method for objects of class nlVar, i.e. VAR and VECM models.
Description
Returns the fitted values of the model, either as computed in the model, or back to the original series
level.
Usage
fitted(object, level = c("model", "original"), ...)
Arguments
object An object of class ‘nlVar’; generated by VECM or lineVar.
level How to return the fitted values. See below.
Details
In case of a VAR in differences, in ADF specification, or a VECM, the fitted values are actually
in differences. With the option level="original", the function returns the series in the original
level.
For VAR in levels, the two arguments are evidently the same and hence it is not taken into account,
returning a warning.
Value
A matrix.
Author(s)
Matthieu Stigler
getTh 21
Examples
## estimate models
data(barry)
ve <- VECM(barry, lag=2)

va <- lineVar(barry, lag=1)
va_diff <- lineVar(barry, lag=1, I="diff")
va_ADF <- lineVar(barry, lag=1, I="ADF")
## get fitted values:

tail(fitted(ve))
tail(fitted(ve, level="original"))
tail(fitted(va))
tail(fitted(object=va, level="original"))
tail(fitted(va_diff))
tail(fitted(object=va_diff, level="original"))
tail(fitted(va_ADF))
tail(fitted(object=va_ADF, level="original"))
getTh Extract threshold(s) coefficient
Description
Extract threshold coefficient(s)
Usage
getTh(object, ...)
getTh(object, ...)
Arguments
object object of class setar, summary.setar, nlVar
... additional arguments to getTh
Value
Threshold value.
22 GIRF
Author(s)
Matthieu Stigler
Examples
set<-setar(lynx, m=3)
getTh(set)
getTh(summary(set))
GIRF Generalized Impulse response Function (GIRF)
Description
Generates a GIRF for multiple innovations and histories
Usage
GIRF(object, n.ahead, seed = NULL, ...)

GIRF(
object,
n.ahead = 10,
seed = NULL,
n.hist = 20,
n.shock = 20,
R = 10,
hist_li = NULL,
shock_li = NULL,
...
)

GIRF(
object,
n.ahead = 10,
seed = NULL,
n.hist = 20,
n.shock = 20,
R = 10,
hist_li = NULL,
shock_li = NULL,
...
)
GIRF 23

GIRF(
object,
n.ahead = 10,
seed = NULL,
n.hist = 20,
n.shock = 20,
R = 10,
hist_li = NULL,
shock_li = NULL,
...
)
## S3 method for class 'GIRF_df'

plot(
x,
plot_type = c("density", "line"),
n.ahead = c(1, 5, 10),
var = unique(x$var)[1],
n_simu = c(1, 2),
...
)
Arguments
object An object of class linear, setar or nlVar (TVAR, TVECM)

n.ahead The number of steps ahead to compute
seed optional, the seed for the random numbers
n.hist The number of past histories to consider. Should be high, ideally size of data
(minus lags).
n.shock The number of actual shocks to consider
R the number of draws to use for the n.ahead innovations
hist_li optional, a list of histories (each of same length as lags in the model). If not
provided, n.hist histories will be randomly drawn for the original series.
shock_li optional, a list of innovations. If not provided, n.shock shocks will be randomly
drawn from the estimated residuals.
x output of girf
plot_type plot: density (for each n.ahead), or line (for multipe n_simu)?
var plot: which variable to plot?
n_simu line plot: which simulation to plot?
... Further arguments passed to specific methods.
24 GIRF
Details
In a nonlinear model, the Impulse response Function (IRF) is not time-, scale- and sign-invariant as
in linear models. To address this, Koop et al (1996) introduced the Generalized Impulse response
Function (GIRF):
GIRF (h, δ, ωt−1 ) = E[yt+h |t = δ, t+h = 0, ωt−1 ] − E[yt+h |t = 0, t+h = 0, ωt−1 ]
It is the difference between two conditional expectations, one containing the shock of interest, the
second one averaging it out. The averaging-out is done by comparing against random innovations
(unlike the IRF, which compares against innovation 0), The parameter R corresponds to the number
of times this is done.
The GIRF as defined here depends on the particular shock, as well as on the history. Koop et
al (1996) suggest to draw multiple combinations of histories and innovations. This is done with
arguments n.hist and n.shock (or, alternatively, provide one of, or both, hist_li and hist_li as
list of histories and shocks).
The output is a data-frame containing the two average paths for each combinations of shocks and
histories.
Value
A data-frame, with:
n_simu: Id for the simulation (total number is n.hist times n.shock)

hist, shock History and shock used in the nth simulation
n.ahead: The forecasting horizon. Note the shocks happens at time 0
var: The variable (on which the shock happens, corresponds hence to the response argument in
irf)
sim_1, sim_2 The average (over R times) simulation with the specific shock (sim_1) or with ran-
dom shocks (sim_2).
girf The difference between sim_1 and sim_2
Author(s)
Matthieu Stigler
References
Koop, G, Pesaran, M. H. & Potter, S. M. (1996) Impulse response analysis in nonlinear multivariate
models. Journal of Econometrics, 74, 119-147
See Also
irf.nlVar for the IRF, for linear models, or in case of non-linear models, for each regime.
IIPUs 25
Examples
## simulate a SETAR for the example. Higher regime more persistent (AR coef 0.5 instead of 0.2)
set <- setar.sim(B = c(0.5, 0.2, 0.5, 0.5), lag = 1, Thresh = 0.5, n = 500)
set_estim <- setar(set, m = 1)
## regime-specific IRF:
plot(irf(set_estim, regime = "L", boot = FALSE))
plot(irf(set_estim, regime = "H", boot = FALSE))
## GIRF
girf_out <- GIRF(set_estim, n.hist = 10, n.shock = 10) # smaller number for example only
## the GIRF shows a very fast convergence (the shock at n.ahead = 4 is already very close to 0)
plot(girf_out, n.ahead = 1:4)
## investigate a few specific GIRFS:
plot(girf_out, plot_type = "line", n_simu = 1:5)
IIPUs US monthly industrial production from Hansen (1999)
Description
This data, used as example in Hansen (1999), contains the US monthly industrial production.
Usage
IIPUs
Format
A monthly time series of class ts starting in January 1960 and ending in September 1997. Note that
the series ends at 1997 and not 1998 as in the paper of Hansen, even if the data was taken from hi
site and the graph is exactly the same.
Source
Hansen (1999) Testing for linearity, Journal of Economic Surveys, Volume 13, Number 5, Decem-
ber 1999 , pp. 551-576(26) available at: http://www.ssc.wisc.edu/~bhansen/papers/cv.htm
Examples
data(IIPUs)
end(IIPUs) #not same date as in the paper
## Figure 2 in paper (p. 555)

plot(IIPUs)
26 IIPUs
## Table 1 in paper (p. 556)

ar_1 <- linear(IIPUs, m=16)
coef(summary(ar_1))[, 1:2]
deviance(ar_1)

set_1 <- setar(IIPUs, m=16, thDelay=5, trim=0.1)
## tsDyn finds another threshold, with a better SSR:
getTh(set_1)
deviance(set_1)
## estimate with Hansen threshold:

set_1_han <- setar(IIPUs, m=16, thDelay=5, trim=0.1, th = 0.23)
deviance(set_1_han)
set_1_CO <- coef(summary(set_1_han))[, 1:2]
cbind(set_1_CO[1:17,], set_1_CO[18:34,])

set_2 <- setar(IIPUs, m=16, thDelay=5, trim=0.1, nthresh =2)
getTh(set_2)
deviance(set_2)
##most of the results agree, except constant in the low regime which has opposed signs!
set_2_CO <- coef(summary(set_2))[, 1:2]
cbind(set_2_CO[1:17,], set_1_CO[18:34,])
#this is obviously a error in Hansen, see:

XX<-embed(IIPUs, 17)
Y<-XX[,1]
X<-XX[,-1]
dummyDown<-ifelse(X[,6]<= -2.5, 1,0)
sum(dummyDown)
M<-cbind(1*dummyDown,X*dummyDown )
lm(Y~M-1)
## setar test (takes long to compute, even with small nboot)

## Not run:
test_1 <- setarTest(IIPUs, m=16, thDelay=5, nboot=10)
#because of the discrepency. test1vs2 does not correspond, test 1vs3 conforms
test_1$Ftests ## compare with Table 2 (F12) and Table 4 (F13, F23)
summary(test_1)
setarTest(IIPUs, m=16, thDelay=5, nboot=10, test="2vs3")

#test 2vs3 is also different of the version in the article (27)
## End(Not run)
## results from the test is stored in: setarTest_IIPUs_results

data(setarTest_IIPUs_results)
## Table 5 and 6
test_1vs <- setarTest_IIPUs_results$test_1
irf.linear 27
test_1vs
## Table 7
test_2vs <- setarTest_IIPUs_results$test_2
test_2vs
plot(test_2vs)
irf.linear Impulse response function
Description
Compute the impulse response coefficients (IRF) of a VAR(p) (or transformed VECM to VAR(p))
for n.ahead steps. For TVECM and TVAR model, argument regime offers regime-specific IRF.
Usage
irf(
x,
impulse = NULL,
response = NULL,
n.ahead = 10,
ortho = TRUE,
cumulative = FALSE,
boot = TRUE,
ci = 0.95,
runs = 100,
seed = NULL,
...
)

irf(
x,
impulse = NULL,
response = NULL,
n.ahead = 10,
ortho = TRUE,
cumulative = FALSE,
boot = TRUE,
ci = 0.95,
runs = 100,
seed = NULL,
regime = c("L", "M", "H"),
...
)
28 irf.linear
## S3 method for class 'ar'

irf(
x,
impulse = NULL,
response = NULL,
n.ahead = 10,
ortho = TRUE,
cumulative = FALSE,
boot = TRUE,
ci = 0.95,
runs = 100,
seed = NULL,
...
)
## S3 method for class 'VAR'

irf(
x,
impulse = NULL,
response = NULL,
n.ahead = 10,
ortho = TRUE,
cumulative = FALSE,
boot = TRUE,
ci = 0.95,
runs = 100,
seed = NULL,
...
)

irf(
x,
impulse = NULL,
response = NULL,
n.ahead = 10,
ortho = TRUE,
cumulative = FALSE,
boot = TRUE,
ci = 0.95,
runs = 100,
seed = NULL,
...
)
## S3 method for class 'TVAR'

irf(
irf.linear 29
x,
impulse = NULL,
response = NULL,
n.ahead = 10,
ortho = TRUE,
cumulative = FALSE,
boot = TRUE,
ci = 0.95,
runs = 100,
seed = NULL,
regime = c("L", "M", "H"),
...
)
## S3 method for class 'TVECM'

irf(
x,
impulse = NULL,
response = NULL,
n.ahead = 10,
ortho = TRUE,
cumulative = FALSE,
boot = TRUE,
ci = 0.95,
runs = 100,
seed = NULL,
regime = c("L", "M", "H"),
...
)
Arguments
x Object of class ‘VAR’; generated by lineVar(), or object of class ‘VECM’; gen-

erated byVECM().
impulse, response
Not used!
n.ahead Integer specifying the steps.
ortho Logical, if TRUE (the default) the orthogonalised impulse response coefficients
are computed .
cumulative Logical, if TRUE the cumulated impulse response coefficients are computed. The
default value is false.
boot, ci, runs, seed
Arguments for the bootstrap, see irf.varest
regime For a setar model, which regime (L, M or H) to produce IRF for?
30 isLinear
Value
A list of class ‘varirf’ see irf.varest
Author(s)
Matthieu Stigler
See Also
plot for the plot method. lineVar, VECM for the models. Hamilton, J. (1994), Time Series Analysis,
Princeton University Press, Princeton.
Lutkepohl, H. (2006), New Introduction to Multiple Time Series Analysis, Springer, New York.
Examples
data(barry)
## For VAR
mod_var <- lineVar(barry, lag = 2)
irf(mod_var, impulse = "dolcan", response = c("dolcan", "cpiUSA", "cpiCAN"), boot =
FALSE)
## For VECM
mod_VECM <- VECM(barry, lag = 2, estim="ML", r=2)
irf(mod_VECM, impulse = "dolcan", response = c("dolcan", "cpiUSA", "cpiCAN"), boot =
FALSE)
isLinear isLinear
Description
Generic NLAR linearity test
Usage
isLinear(object, ...)
Arguments
object fitted time series model
... arguments to and from other methods
Author(s)
A. F. Di Narzo
KapShinTest 31
KapShinTest Test of unit root against SETAR alternative with
Description
Test of unit root against a stationnary 3 regime SETAR alternative with random walk in the inner
regime
Usage
KapShinTest(x, m=1, series, include = c("none","const", "trend", "both"),
c=3, delta=0.5, points=NULL,minObsMid=10,
trick=c("for", "apply", "mapply"), trace=FALSE)
Arguments
x time series
m Number of lags under the alternative
include Whether data should be raw, de-meaned or de-meaned and de-trended
c Argument for the grid search. See details
delta Argument for the grid search. See details
points Points for the grid search. See details
minObsMid Minimal number of observations in the inner regime
trick type of internal function used
trace should additional infos be printed? (logical)
Details
This function is currently spurious.
Value
A object of class KapShin2006Test containing:
statistic The three (SupW, AvgW, ExpW) test statistics computed

case Whether the data was transformed, corresponds to input argument include
series The name of the series
Author(s)
Matthieu Stigler
32 lags.select
See Also
BBCTest for a similar test. setarTest for a test with stationarity as a null.
Examples
KapShinTest(lynx, m=1, trace=FALSE, include="none", points=10)
lags.select Selection of the lag with Information criterion.
Description
Selection of the cointegrating rank and the lags with Information criterion (AIC, BIC).
Usage
lags.select(
data,
lag.max = 10,
include = c("const", "trend", "none", "both"),
fitMeasure = c("SSR", "LL"),
sameSample = TRUE
)
Arguments
data multivariate time series.
lag.max Maximum number of lags to investigate.
include Type of deterministic regressors to include.
fitMeasure Whether the AIC/BIC should be based on the full likelihood, or just the SSR.
See explanations in logLik.VECM.
sameSample Logical. Whether the data should be shortened so that the AIC/BIC are esti-
mated on the same sample. Default to TRUE.
Details
This function selects the lag according to AIC, BIC and Hannan-Quinn.
Value
An object of class rank.select, with ‘print’ and ‘summary methods’, containing among other the
matrices of AIC/BIC/HQ.
Author(s)
Matthieu Stigler
LINEAR 33
See Also
rank.select, the underlying function, to estimate the rank also.
VARselect in package vars, does basically the same.
Examples
data(barry)
#
rk_sel <- lags.select(barry)
rk_sel
summary(rk_sel)
LINEAR Linear AutoRegressive models
Description
AR(m) model
Usage
linear(x, m, d=1, steps=d, series, include = c( "const", "trend","none", "both"),
type=c("level", "diff", "ADF"), warn_root=TRUE)
Arguments
x time series
include Type of deterministic regressors to include
type Whether the variable is taken is level, difference or a mix (diff y= y-1, diff lags)
as in the ADF test
warn_root Whether to check (and warn) for roots outside the unit circle?
Details
AR(m) model:
xt+s = φ0 + φ1 xt + φ2 xt−d + · · · + φm xt−(m−1)d + t+s

34 lineVar
Value
A nlar object, linear subclass.
Author(s)
See Also
nlar for fitting this and other models to time series data
Examples
#fit an AR(2) model
mod.linear <- linear(log(lynx), m=2)
mod.linear
summary(mod.linear)
lineVar Multivariate linear models: VAR and VECM
Description
Estimate either a VAR or a VECM.
Usage
lineVar(
data,
lag,
r = 1,
model = c("VAR", "VECM"),
I = c("level", "diff", "ADF"),
beta = NULL,
estim = c("2OLS", "ML"),
LRinclude = c("none", "const", "trend", "both"),
exogen = NULL
)
Arguments
data multivariate time series (first row being first=oldest value)
lag Number of lags to include in each regime
r Number of cointegrating relationships
lineVar 35
model Model to estimate. Either a VAR or a VECM

I For VAR only: whether in the VAR the variables are to be taken in levels (orig-
inal series) or in difference, or similarly to the univariate ADF case.
beta for VECM only: user-specified cointegrating value. If NULL, will be estimated
using the estimator specified in estim
estim Type of estimator for the VECM: ’2OLS’ for the two-step approach or ’ML’ for
Johansen MLE
LRinclude Possibility to include in the long-run relationship and the ECT a trend, a, con-
stant, etc. Can also be a matrix with exogeneous regressors
exogen Inclusion of exogenous variables (first row being first=oldest value). Is either of
same size than data (then automatically cut) or than end-sample.
Details
This function provides basic functionalities for VAR and VECM models. More comprehensive
functions are in package vars. A few differences appear in the VECM estimation:
Engle-Granger estimator The Engle-Granger estimator is available

Presentation Results are printed in a different ways, using a matrix form
lateX export The matrix of coefficients can be exported to latex, with or without standard-values
and significance stars
Two estimators are available: the Engle-Granger two-steps approach (2OLS) or the Johansen (ML).
For the 2OLS, deterministic regressors (or external variables if LRinclude is of class numeric) can
be added for the estimation of the cointegrating value and for the ECT. This is only working when
the beta value is not pre-specified.
The argument beta is only for VECM, look at the specific help page for more details.
Value
Fitted model data
Author(s)
Matthieu Stigler
See Also
VECM which is just a wrapper for lineVar(..., model="VECM"). Methods predict.VAR, VARrep,
regime, irf and toLatex.
TVAR and TVECM for the corresponding threshold models. linear for the univariate AR model.
36 llar
Examples
data(zeroyld)
#Fit a VAR
VAR <- lineVar(zeroyld, lag=1)
VAR
summary(VAR)
#compare results with package vars:

if(require(vars)) {
a<-VAR(zeroyld, p=1)
coef_vars <- t(sapply(coef(a), function(x) x[c(3,1,2),1]))
all.equal(coef(VAR),coef_vars, check.attributes=FALSE)
}
###VECM
VECM.EG <- lineVar(zeroyld, lag=2, model="VECM")
VECM.EG
summary(VECM.EG)
VECM.ML <- lineVar(zeroyld, lag=2, model="VECM", estim="ML")

VECM.ML
summary(VECM.ML)
###Check Johansen MLE

myVECM <- lineVar(zeroyld, lag=1, include="const", model="VECM", estim="ML")
summary(myVECM, digits=7)
#comparing with vars package
if(require(vars)){
a<-ca.jo(zeroyld, spec="trans")
summary(a)
#same answer also!
}
##export to Latex
toLatex(VECM.EG)
toLatex(summary(VECM.EG))
options("show.signif.stars"=FALSE)
toLatex(summary(VECM.EG), parenthese="Pvalue")
options("show.signif.stars"=TRUE)
llar Locally linear model

llar 37
Description
Casdagli test of nonlinearity via locally linear forecasts
Usage
llar(x, m, d = 1, steps = d, series, eps.min = sd(x)/2,
eps.max = diff(range(x)), neps = 30, trace = 0)
llar.predict(x, m, d=1, steps=d, series, n.ahead=1,

eps=stop("you must specify a window value"),
onvoid=c("fail","enlarge"), r = 20, trace=1)
llar.fitted(x, m, d=1, steps=d, series, eps, trace=0)
Arguments
x time series
n.ahead n. of steps ahead to forecast
eps.min, eps.max
min and max neighbourhood size
neps number of neighbourhood levels along which iterate
eps neighbourhood size
onvoid what to do in case of an isolated point: stop or enlarge neighbourhood size by
an r%
r if an isolated point is found, enlarge neighbourhood window by r%
trace tracing level: 0, 1 or more than 1 for llar, 0 or 1 for llar.forecast
Details
llar does the Casdagli test of non-linearity. Given the embedding state-space (of dimension m and
time delay d) obtained from time series series, for a sequence of distance values eps, the relative
error made by forecasting time series values with a linear autoregressive model estimated on points
closer than eps is computed. If minimum error is reached at relatively small length scales, a global
linear model may be inappropriate (using current embedding parameters). This was suggested by
Casdagli(1991) as a test for non-linearity.
llar.predict tries to extend the given time series by n.ahead points by iteratively fitting locally
(in the embedding space of dimension m and time delay d) a linear model. If the spatial neighbour-
hood window is too small, your time series last point would be probably isolated. You can ask to
automatically enlarge the window eps by a factor of r% sequentially, until enough neighbours are
found for fitting the linear model.
llar.fitted gives out-of-sample fitted values from locally linear models.
38 llar
Value
llar gives an object of class ’llar’. I.e., a list of components:
RMSE vector of relative errors

eps vector of neighbourhood sizes (in the same order of RMSE)
frac vector of fractions of the time series used for RMSE computation
avfound vector of average number of neighbours for each point in the time series which
can be plotted using the plot method, and transformed to a regular data.frame
with the as.data.frame function.
Function llar.forecast gives the vector of n steps ahead locally linear iterated forecasts.
Function llar.fitted gives out-of-sample fitted values from locally linear models.
Warning
For long time series, this can be slow, especially for relatively big neighbourhood sizes.
Note
The C implementation was re-adapted from that in the TISEAN package ("ll-ar" routine, see refer-
ences). However, here the euclidean norm is used, in place of the max-norm.
Author(s)
References
M. Casdagli, Chaos and deterministic versus stochastic nonlinear modelling, J. Roy. Stat. Soc. 54,
303 (1991)
Hegger, R., Kantz, H., Schreiber, T., Practical implementation of nonlinear time series methods:
The TISEAN package; CHAOS 9, 413-435 (1999)
Examples
res <- llar(log(lynx), m=3, neps=7)
plot(res)
x.new <- llar.predict(log(lynx),n.ahead=20, m=3, eps=1, onvoid="enlarge", r=5)

lag.plot(x.new, labels=FALSE)
x.fitted <- llar.fitted(log(lynx), m=3, eps=1)

lag.plot(x.fitted, labels=FALSE)
logLik.nlVar 39
logLik.nlVar Extract Log-Likelihood
Description
Log-Likelihood method for VAR and VECM models.
Usage
logLik(object, ...)

logLik(object, r, ...)
Arguments
object object of class VAR computed by lineVar, or class VECM computed by VECM.
... additional arguments to logLik.
r The cointegrating rank. By default the rank specified in the call to VECM, but can
be set differently by user.
Details
For a VAR, the Log-Likelihood is computed as in Luetkepohl (2006) equ. 3.4.5 (p. 89) and Juselius
(2006) p. 56:
T h i
X 0 0 0
LL = −(T K/2) log(2π) − (T /2) log |Σ| − (1/2) (yt − A xt ) Σ−1 (yt − A xt )
Where Σ is the Variance matrix of residuals, and xt is the matrix stacking the regressors (lags and
deterministic).
However, we use a computationally simpler version:
LL = −(T K/2) log(2π) − (T /2) log |Σ| − (T K/2)
See Juselius (2006), p. 57.

(Note that Hamilton (1994) 11.1.10, p. 293 gives +(T /2) log |Σ−1 |, which is the same as −(T /2) log |Σ|).
For VECM, the Log-Likelihood is computed in two different ways, depending on whether the VECM
was estimated with ML (Johansen) or 2OLS (Engle and Granger).
When the model is estimated with ML, the LL is computed as in Hamilton (1994) 20.2.10 (p. 637):
r
X
LL = −(T K/2) log(2π) − (T K/2) − (T /2) log |Σ̂U U | − (T /2) log(1 − λ̂i )
i=1
40 LSTAR
Where ΣU U is the variance matrix of residuals from the first auxiliary regression, i.e. regressing ∆yt
on a constant and lags, ∆yt−1 , . . . , ∆yt−p . λi are the eigenvalues from the Σ−1 −1
V V ΣV U ΣU U Σ U V ,
see 20.2.9 in Hamilton (1994).
When the model is estimated with 2OLS, the LL is computed as:
LL = log |Σ|
Where Σ is the variance matrix of residuals from the the VECM model. There is hence no corre-
spondence between the LL from the VECM computed with 2OLS or ML.
Value
Log-Likelihood value.
Author(s)
Matthieu Stigler
References
Hamilton (1994) Time Series Analysis, Princeton University Press
Juselius (2006) The Cointegrated VAR model: methodology and Applications, Oxford Univesity
Press
Luetkepohl (2006) New Introduction to Multiple Time Series Analysis, Springer
Examples
data(zeroyld)
#Fit a VAR
VAR <- lineVar(zeroyld, lag=1)
logLik(VAR)
#'#Fit a VECM
vecm <- VECM(zeroyld, lag=1, r=1, estim="ML")
logLik(vecm)
LSTAR Logistic Smooth Transition AutoRegressive model
Description
Logistic Smooth Transition AutoRegressive model.
LSTAR 41
Usage
lstar(x, m, d=1, steps=d, series, mL, mH, mTh, thDelay,
thVar, th, gamma, trace=TRUE, include = c("const", "trend","none", "both"),
control=list(), starting.control=list())
Arguments
x time series
mL autoregressive order for ’low’ regime (default: m). Must be <=m
mH autoregressive order for ’high’ regime (default: m). Must be <=m
thDelay ’time delay’ for the threshold variable (as multiple of embedding time delay d)
mTh coefficients for the lagged time series, to obtain the threshold variable
thVar external threshold variable
th, gamma starting values for coefficients in the LSTAR model. If missing, a grid search is
performed
control further arguments to be passed as control list to optim
starting.control
further arguments for the grid search (dimension, bounds). See details below.
Details
xt+s = (φ1,0 +φ1,1 xt +φ1,2 xt−d +· · ·+φ1,mL xt−(mL−1)d )G(zt , th, γ)+(φ2,0 +φ2,1 xt +φ2,2 xt−d +· · ·+φ2,mH xt−(mH−1)d )
with z the threshold variable, and G the logistic function, computed as plogis(q, location = th,
scale = 1/gamma), so see plogis documentation for details on the logistic function formulation
and parameters meanings. The threshold variable can alternatively be specified by:
mTh z[t] = x[t]mT h[1] + x[t − d]mT h[2] + · · · + x[t − (m − 1)d]mT h[m]
thDelay z[t] = x[t − thDelay ∗ d]
thVar z[t] = thV ar[t]
Note that if starting values for phi1 and phi2 are provided, isn’t necessary to specify mL and mH.
Further, the user has to specify only one parameter between mTh, thDelay and thVar for indicating
the threshold variable.
Estimation of the transition parameters th and gamma, as well as the regression parameters phi1
and phi2 , is done using concentrated least squares, as suggested in Leybourne et al. (1996).
Given th and gamma, the model is linear, so regression coefficients can be obtained as usual by
OLS. So the nonlinear numerical search needs only to be done for th and gamma; the regression
parameters are then recovered by OLS again from the optimal th and gamma.
42 LSTAR
For the nonlinear estimation of the parameters th and gamma, the program uses the optim func-
tion, with optimization method BFGS using the analytical gradient. For the estimation of standard
values, optim is re-run using the complete Least Squares objective function, and the standard errors
are obtained by inverting the hessian. You can pass further arguments to optim directly with the
control list argument. For instance, the option maxit maybe useful when there are convergence
issues (see examples).
Starting parameters are obtained doing a simple two-dimensional grid-search over th and gamma.
Parameters of the grid (interval for the values, dimension of the grid) can be passed to starting.control.
nTh The number of threshold values (th ) in the grid. Defaults to 200
nGamma The number of smoothing values (gamma) in the grid. Defaults to 40
trim The minimal percentage of observations in each regime. Defaults to 10% (possible threshold
values are between the 0.1 and 0.9 quantile)
gammaInt The lower and higher smoothing values of the grid. Defaults to c(1,40)
thInt The lower and higher threshold values of the grid. When not specified (default, i.e NA), the
interval are the trim quantiles above.
Value
An object of class nlar, subclass lstar, i.e. a list with fitted model informations.
Author(s)
References
Non-linear time series models in empirical finance, Philip Hans Franses and Dick van Dijk, Cam-
bridge: Cambridge University Press (2000).
Non-Linear Time Series: A Dynamical Systems Approach, Tong, H., Oxford: Oxford University
Press (1990).
Leybourne, S., Newbold, P., Vougas, D. (1998) Unit roots and smooth transitions, Journal of Time
Series Analysis, 19: 83-97
See Also
plot.lstar for details on plots produced for this model from the plot generic.
Examples
#fit a LSTAR model. Note 'maxit': slow convergence
mod.lstar <- lstar(log10(lynx), m=2, mTh=c(0,1), control=list(maxit=3000))
mod.lstar
#fit a LSTAR model without a constant in both regimes.

mod.lstar2 <- lstar(log10(lynx), m=1, include="none")
mod.lstar2
m.unrate 43
#Note in example below that the initial grid search seems to be to narrow.
# Extend it, and evaluate more values (slow!):
controls <- list(gammaInt=c(1,2000), nGamma=50)
mod.lstar3 <- lstar(log10(lynx), m=1, include="none", starting.control=controls)
mod.lstar3
# a few methods for lstar:

summary(mod.lstar)
residuals(mod.lstar)
AIC(mod.lstar)
BIC(mod.lstar)
plot(mod.lstar)
predict(mod.lstar, n.ahead=5)
m.unrate Monthly US unemployment
Description
Monthly US civilian unemployment rate from 1948 through 2004. Used in the book financial time
series for Tsay (2005, chapter 4)
Usage
data(m.unrate)
Format
zoo object
Source
‘https://faculty.chicagobooth.edu/ruey-s-tsay/research/analysis-of-financial-time-series-2nd-edition
References
Ruey Tsay (2005) Analysis of Financial Time Series, 2nd ed. (Wiley, ch. 4)
44 MakeThSpec
MakeThSpec Specification of the threshold search
Description
This optional function allows the user to set different restrictions for the threshold grid search in
function selectSETAR.
Usage
MakeThSpec(
ngrid = c("All", "Half", "Third", "Quarter"),
exact = NULL,
int = c("from", "to"),
around = "val",
...
)
Arguments
ngrid The number of values to search for
exact The user give an exact threshold value
int The user gives an interval to search inside
around The user gives an point to search around
Details
This function is just to check the inputs for the specification of the grid search. If not provided, the
search will be in the biggest interval (ngrid ="All") between the minimum and maximum values.
The user can reduce it by giving setting "Half" (only every two points is taken) and so on, or setting
a number.
The search can also be made around a point, or between two points. When between a point, the
argument ngrid is still used, whereas for around, a value of 30 is taken as default value if ngrid is
not specified by user.
Value
The input values are given as output after checking for consistency (only one of exact/int/around
should be given).
Author(s)
Matthieu Stigler
MAPE 45
See Also
selectSETAR
Examples
sun<-(sqrt(sunspot.year+1)-1)*2
selectSETAR(sun, m=3, th=MakeThSpec(exact=10.40967),criterion="SSR", d=1, thDelay=0:2,
plot=FALSE, nthresh=1)
#when pre-sepcified value does not correspond, function will search nearest value
selectSETAR(sun, m=3, th=MakeThSpec(exact=10.4),criterion="SSR", d=1, thDelay=0:2,
#search around:
selectSETAR(sun, m=3, th=MakeThSpec(around=10.40967, ngrid=20),criterion="SSR", d=1, thDelay=0:2,
#search in an interval
selectSETAR(sun, m=3, th=MakeThSpec(int=c(10, 11), ngrid=20),criterion="SSR", d=1, thDelay=0:2,
#reduce size of the grid:
selectSETAR(sun, m=3, th=MakeThSpec(ngrid="Half"),criterion="SSR", d=1, thDelay=0:2,
# 2 thresholds:
selectSETAR(sun, m=3, th=MakeThSpec(ngrid="Half"),criterion="SSR", d=1, thDelay=0:2,
MAPE Mean Absolute Percent Error
Description
Generic function to compute the Mean Absolute Percent Error of a fitted model.
Usage
MAPE(object, ...)
MAPE(object, ...)
Arguments
object object of class nlar.fit
... additional arguments to MAPE
46 mse
Value
Computed Mean Absolute Percent Error for the fitted model.
Author(s)
mse Mean Square Error
Description
Generic function to compute the Mean Squared Error of a fitted model.
Usage
mse(object, ...)
mse(object, ...)
Arguments
object object of class nlar.fit

... additional arguments to mse
Value
Computed MSE for the fitted model.
Author(s)

nlar-methods 47
nlar-methods NLAR methods
Description
Generic ‘nlar’ methods. Method ‘nlar’ is described in a separate page: nlar
Usage
coef(object, ...)

fitted(object, ...)

residuals(object, initVal = TRUE, timeAttr = TRUE, ...)

deviance(object, ...)

mse(object, ...)

AIC(object, k = 2, ...)

BIC(object, ...)

MAPE(object, ...)

summary(object, ...)

plot(x, ask = interactive(), ...)

toLatex(object, digits, label, ...)
Arguments
... further arguments to be passed to and from other methods
initVal Whether to return NA for initial values in residuals/predicitons etc
timeAttr Whether the time attributes should be returned in the output
48 nlar-methods
k numeric, the penalty per parameter to be used for AIC/BIC; the default k = 2 is
the classical AIC
x, object fitted ‘nlar’ object
ask graphical option. See par
digits For print method, see printCoefmat.
label LaTeX label passed to the equation
Details
MAPE Mean Absolute Percent Error

mse Mean Square Error
plot Diagnostic plots
Author(s)
See Also
availableModels for listing all currently available models.
Examples
x <- log10(lynx)
mod.setar <- setar(x, m=2, thDelay=1, th=3.25)
mod.setar
AIC(mod.setar)
mse(mod.setar)
MAPE(mod.setar)
coef(mod.setar)
summary(mod.setar)
e <- residuals(mod.setar)
e <- e[!is.na(e)]
plot(e)
acf(e)
plot(x)
lines(fitted(mod.setar), lty=2)
legend(x=1910, y=3.9,lty=c(1,2), legend=c("observed","fitted"))
plot(mod.setar)
NNET 49
NNET Neural Network nonlinear autoregressive model
Description
Neural Network nonlinear autoregressive model.
Usage
nnetTs(x, m, d = 1, steps = d, series, size,
control = list(trace = FALSE))
Arguments
x time series
size number of hidden units in the neural network
control control list to be passed to nnet::nnet optimizer
Details
Neural network model with 1 hidden layer and linear output:
D
X m
X
xt+s = β0 + βj g(γ0j + γij xt−(i−1)d )
j=1 i=1
Model is estimated using the nnet function in nnet package. Optimization is done via the BFGS
method of optim. Note that for this model, no additional model-specific summary and plot methods
are made available from this package.
Value
An object of class nlar, subclass nnetTs, i.e. a list with mostly nnet::nnet internal structures.
Author(s)
References
Press (1990).
Chaos: A Statistical Perspective, Chan, K., Tong, H., New York: Springer-Verlag (2001).
50 plot methods
Examples
#fit a Neural Network model
mod.nnet <- nnetTs(log(lynx), m=2, size=3)
mod.nnet
plot methods Plotting methods for SETAR and LSTAR subclasses
Description
Plotting methods ‘setar’ and ‘lstar’ subclasses
Usage
plot(x, ask=interactive(), legend=FALSE, regSwStart, regSwStop, ...)
plot(x, ask=interactive(), legend=FALSE, regSwStart, regSwStop, ...)
Arguments
x fitted ‘setar’ or ‘lstar’ object
ask graphical option. See par
legend Should a legend be plotted? (logical)
regSwStart, regSwStop
optional starting and stopping time indices for regime switching plot
... further arguments to be passed to and from other methods
Details
These plot methods produce a plot which gives to you an idea of the behaviour of the fitted model.
Firstly, if embedding dimension is, say, m, m scatterplots are produced. On the x axis you have
the lagged time series values. On the y axis the ‘response’ time series values. Observed points are
represented with different colors-symbols depending on the level of the threshold variable. Specifi-
cally, for the setar model, black means ‘low regime’, red means ‘high regime’. For the lstar model,
where the self-threshold variable is continuous, threshold values are grouped in 5 different zones
with the same number of points in each. Note that if more than 300 points are to be plotted, they all
share the same symbol, and regimes can be distinguished only by color. If you want, by specifying
legend=TRUE a legend is added at the upper-left corner of each scatterplot. To each scatterplot, a
dashed line is superposed, which links subsequent fitted values.
Finally, a new time series plot is produced, with lines segments coloured depending on the regime
(colors meanings are the same of those in the preceding scatterplots). Optionally, you can specify a
starting and ending time indices, for zooming on a particular segment of the time series.
plot_ECT 51
Author(s)
See Also
setar, lstar
nlar-methods for other generic available methods for this kind of objects.
Examples
##
##See 'setar' examples
##
plot_ECT Plot the Error Correct Term (ECT) response
Description
This plot shows how variables in a (T)VECM respond to deviations from the long-term equilibrium
Usage
plot_ECT(x, add.legend = TRUE, legend.location = "topright", ...)
Arguments
x object of class VECM or TVECM
add.legend logical. Whether to add a legend?
legend.location
character. Location of the legend, see legend
... arguments passed to the initial plot call, see plot
Value
a plot, and invisibly the underlying data.frame, containing the ECT and the response for each vari-
able
Examples
data(zeroyld)
vec_l1 <- VECM(zeroyldMeta[, c("long.run", "short.run")], lag =1)
tvec_l1 <- TVECM(zeroyldMeta[, c("long.run", "short.run")], lag =1,
plot = FALSE, trace = FALSE, th1 = list(exact = -1.263))
plot_ECT(vec_l1)
plot_ECT(tvec_l1, legend.location = "bottomright")
52 predict.nlar
predict.nlar Predict method for objects of class ‘nlar’.
Description
Forecasting a non-linear model object of general class ‘nlar’, including ‘setar’ and ‘star’.
Usage
predict(
object,
newdata,
n.ahead = 1,
type = c("naive", "MC", "bootstrap", "block-bootstrap"),
nboot = 100,
ci = 0.95,
block.size = 3,
boot1Zero = TRUE,
seed = NULL,
...
)
Arguments
object An object of class ‘nlar’; generated by setar() or lstar().
newdata Optional. A new data frame to predict from.
n.ahead An integer specifying the number of forecast steps.
type Type of forecasting method used. See details.
nboot The number of replications for type MC or bootstrap.
ci The forecast confidence interval (available only with types MC and bootstrap).
block.size The block size when the block-bootstrap is used.
boot1Zero Whether the first innovation for MC/bootstrap should be set to zero.
seed optional, the seed for the random generation.
... Further arguments passed to the internal ‘oneStep’ function. Mainly argument
‘thVar’ if an external threshold variable was provided
Details
The forecasts are obtained recursively from the estimated model. Given that the models are non-
linear, ignoring the residuals in the 2- and more steps ahead forecasts leads to biased forecasts
(so-called naive). Different resampling methods, averaging n.boot times over future residuals, are
available:
naive No residuals
predict.nlar 53
MC Monte-Carlo method, where residuals are taken from a normal distribution, with a standard
deviation equal to the residuals sd.
bootstrap Residuals are resampled from the empirical residuals from the model.
block-bootstrap Same as bootstrap, but residuals are resampled in block, with size block.size
The MC and bootstrap methods correspond to equations 3.90 and 3.91 of Franses and van Dijk
(2000, p. 121). The bootstrap/MC is initiated either from the first forecast, n.ahead=1 (set with
boot1zero to TRUE), or from the second only.
When the forecast method is based on resampling, forecast intervals are available. These are ob-
tained simply as empirical ci quantiles of the resampled forecasts (cf Method 2 in Franses and van
Dijk, 2000, p. 122).
Value
A ‘ts’ object, or, in the case of MC/bootstrap, a list containing the prediction (pred) and the forecast
standard errors (se).
Author(s)
Matthieu Stigler
References
See Also
The model fitting functions setar, lstar.
A more sophisticated predict function, allowing to do sub-sample rolling predictions: predict_rolling.
Examples
x.train <- window(log10(lynx), end = 1924)

x.test <- window(log10(lynx), start = 1925)
### Use different forecasting methods:

mod.set <- setar(x.train, m=2, thDelay=0)
pred_setar_naive <- predict(mod.set, n.ahead=10)
pred_setar_boot <- predict(mod.set, n.ahead=10, type="bootstrap", n.boot=200)
pred_setar_Bboot <- predict(mod.set, n.ahead=10, type="block-bootstrap", n.boot=200)
pred_setar_MC <- predict(mod.set, n.ahead=10, type="bootstrap", n.boot=200)
## Plot to compare results:

pred_range <- range(pred_setar_naive, pred_setar_boot$pred, pred_setar_MC$pred, na.rm=TRUE)
plot(x.test, ylim=pred_range, main="Comparison of forecasts methods from same SETAR")
lines(pred_setar_naive, lty=2, col=2)
54 predict.TVAR
lines(pred_setar_boot$pred, lty=3, col=3)

lines(pred_setar_Bboot$pred, lty=4, col=4)
lines(pred_setar_MC$pred, lty=5, col=5)
legLabels <- c("Observed", "Naive F", "Bootstrap F","Block-Bootstrap F", "MC F")
legend("bottomleft", leg=legLabels, lty=1:5, col=1:5)
predict.TVAR Predict method for objects of class ‘VAR’, ‘VECM’ or ‘TVAR’
Description
Forecasting the level of a series estimated by ‘VAR’ / ‘VECM’ or ‘TVAR’
Usage
## S3 method for class 'TVAR'
predict(object, newdata, n.ahead = 5, newdataTrendStart, ...)

predict(object, newdata, n.ahead = 5, newdataTrendStart, exoPred = NULL, ...)
Arguments
object An object of class ‘VAR’, ‘VECM’ or ‘TVAR’
newdata Optional. A new data frame to predict from. This should contain lags of the
level of the original series. See Details.
newdataTrendStart
If ‘newdata’ is provided by the user, and the estimated model includes a trend,
this argument specifies where the trend should start
... Arguments passed to the unexported ‘VAR.gen’ or ‘TVAR.gen’ function
exoPred vector/matrix of predictions for the exogeneous variable(s) (with ‘n.ahead’ rows).
Only for ‘VAR’/‘VECM’, not for ‘TVAR’.
Details
The forecasts are obtained recursively, and are for the levels of the series.
When providing newdata, newdata has to be ordered chronologically, so that the first row/element
is the earliest value.
For VECM, the forecasts are obtained by transforming the VECM to a VAR (using function VARrep).
Note that a VECM(lag=p) corresponds to a VAR(lag=p+1), so that if the user provides newdata for
a VECM(lag=p), newdata should actually contain p+1 rows.
predict_rolling 55
Value
A matrix of predicted values.
Author(s)
Matthieu Stigler
See Also
lineVar and VECM. VARrep
A more sophisticated predict function, allowing to do sub-sample rolling predictions: predict_rolling.
Examples
data(barry)
barry_in <- head(barry, -5)
barry_out <- tail(barry, 5)
mod_vecm <- VECM(barry_in, lag=2)

mod_var <- lineVar(barry_in, lag=3)
mod_tvar <- TVAR(barry_in, lag=3, nthresh=1, thDelay=1)
pred_vecm <- predict(mod_vecm)

pred_var <- predict(mod_var)
pred_tvar <- predict(mod_tvar)
## compare forecasts on a plot

n <- 30
plot(1:n, tail(barry[,1], n), type="l", xlim=c(0,n))
lines((n-5+1):n, pred_var[,1], lty=2, col=2)
lines((n-5+1):n, pred_vecm[,1], lty=2, col=3)
lines((n-5+1):n, pred_tvar[,1], lty=2, col=4)
legend("bottomright", lty=c(1,2,2,2), col=1:4, legend=c("true", "var", "vecm", "tvar"))
## example for newdata:

all.equal(predict(mod_vecm), predict(mod_vecm, newdata=barry[c(317, 318, 319),]))
predict_rolling Rolling forecasts
Description
Forecasts a VAR or VECM by discarding a part of the sample, and recursively generating a series
of updated forecasts.
56 predict_rolling
Usage
predict_rolling(object, ...)

predict_rolling(object, nroll = 10, n.ahead = 1, refit.every, newdata, ...)
Arguments
object A linear object of class ‘nlVar’; generated by lineVar or VECM.

nroll The number of rolling forecasts
refit.every Determines every how many periods the model is re-estimated.
newdata In case the given model is already estimated on the sub-sample, the out-of-
sample data can be provided with this argument. Note it should contain ob-
servations to predict the first values, that are also contained in the in-sample.
Details
This function allows to check the out-of sample forecasting accuracy by estimating the model on a
sub-sample of the original, then making nroll forecasts of horizon n.ahead, each time by updating
the sample. In other words, with a given model estimated on 100 observations, the function will
estimate it on say 90 first obs (nroll=10), generate a say 1 step-ahead (n.ahead=1) from obs 90,
then using true value 91 for to predict value at 92, etc, till the full sample is used.
Unlike with usual predict() methods, specifiying n.ahead=2 will not generate a 1 step-ahead and
a 2 step-ahead forecasts, but only nroll 2 step-ahead forecasts.
Note that while the forecasts are updated with new values, the model estimation is (by default) not
updated. This can however be done with the argument fit.every, specifiying at which frequency
the model should be re-estimated. By setting it to 1 for example, each time a new observation is
taken, the model is reestimated. This is similar to the ugarchroll in package rugarch.
Value
A matrix containing the forecasts.
Author(s)
Matthieu Stigler
See Also
predict.nlar for the standard predict function.

rank.select 57
Examples
data(barry)
## model estimated on full sample:

mod_vec <- VECM(barry, lag=2)
## generate 10 1-step-ahead forecasts:

preds_roll <- predict_rolling(mod_vec, nroll=10)
## plot the results:

plot(window(barry[,"dolcan"],start=1998), type="l", ylab="barry: dolcan")
preds_roll_ts <- ts(preds_roll$pred, start=time(barry)[nrow(barry)-10], freq=12)
lines(preds_roll_ts[,"dolcan"], col=2, lty=2)
legend("bottomright", lty=c(1,2), col=1:2, leg=c("True", "Fitted"))
title("Comparison of true and rolling 1-ahead forecasts\n")
rank.select Selection of the cointegrating rank with Information criterion.
Description
Selection of the cointegrating rank and the lags with Information criterion (AIC, BIC).
Usage
rank.select(
data,
lag.max = 10,
r.max = ncol(data) - 1,
fitMeasure = c("SSR", "LL"),
sameSample = TRUE,
returnModels = FALSE
)
## S3 method for class 'rank.select'

print(x, ...)

as.data.frame(x, ...)

summary(object, ...)
58 rank.select
Arguments
data multivariate time series.
lag.max Maximum number of lags to investigate.
r.max Maximum rank to investigate.
include Type of deterministic regressors to include. See VECM or lineVar.
fitMeasure Whether the AIC/BIC should be based on the full likelihood, or just the SSR.
See explanations in logLik.VECM.
sameSample Logical. Whether the data should be shortened so that the AIC/BIC are esti-
mated on the same sample. Default to TRUE.
returnModels Logical, default to FALSE. Whether the output should also contain the list of
each model computed.
x The output from rank.select for the print method.
... Unused.
object The output from rank.select for the summary method.
Details
This function estimates the AIC, BIC and Hannan-Quinn for each rank (up to lags.max) and lags
(up to lags.max). This method has been shown to be useful to select simultaneously the rank and
the lags, see references.
Value
An object of class ‘rank.select’, with ‘print’ and ‘summary methods’, containing among other the
matrices of AIC/BIC/HQ, the Likelihood, and best ranks according to each criterion.
Author(s)
Matthieu Stigler
References
- Aznar A and Salvador M (2002). Selecting The Rank Of The Cointegration Space And The Form
Of The Intercept Using An Information Criterion. Econometric Theory, *18*(04), pp. 926-947.
<URL: http://ideas.repec.org/a/cup/etheor/v18y2002i04p926-947_18.html>.
-Cheng X and Phillips PCB (2009). Semiparametric cointegrating rank selection. Econometrics
Journal , *12*(s1), pp. S83-S104. <URL: http://ideas.repec.org/a/ect/emjrnl/v12y2009is1ps83-
s104.html>.
- Gonzalo J and Pitarakis J (1998). Specification via model selection in vector error correction mod-
els. Economics Letters, *60*(3), pp. 321 - 328. ISSN 0165-1765, <URL: http://dx.doi.org/DOI:
10.1016/S0165-1765(98)00129-3>.
- Kapetanios G (2004). The Asymptotic Distribution Of The Cointegration Rank Estimator Un-
der The Akaike Information Criterion. Econometric Theory, *20*(04), pp. 735-742. <URL:
http://ideas.repec.org/a/cup/etheor/v20y2004i04p735-742_20.html>.
rank.test 59
- Wang Z and Bessler DA (2005). A Monte Carlo Study On The Selection Of Cointegrating Rank
Using Information Criteria. Econometric Theory, *21*(03), pp. 593-620. <URL: http://ideas.repec.org/a/cup/etheor/v21y200
620_05.html>.
See Also
VECM for estimating a VECM. rank.test (or ca.jo in package urca) for the classical Johansen
cointegration test.
Examples
data(barry)
#
rk_sel <- rank.select(barry)
rk_sel
summary(rk_sel)
rank.test Test of the cointegrating rank
Description
Maximum-likelihood test of the cointegrating rank.
Usage
rank.test(vecm, type = c("eigen", "trace"), r_null, cval = 0.05)
## S3 method for class 'rank.test'

print(x, ...)
## S3 method for class 'rank.test'

summary(object, digits = max(1, getOption("digits") - 3), ...)
Arguments
vecm ‘VECM’ object computed with the function VECM.
type Type of test, either ’trace’ or ’eigenvalue’. See details below.
r_null Rank to test specifically.
cval Critical value level for the automatic test.
x The output from rank.test for the print method.
... Unused.
object The output from rank.test for the summary method.
digits The number of digits to use in format.pval
60 rank.test
Details
This function computes the two maximum-likelihood tests for the cointegration rank from Johansen
(1996). Tests are:
trace Test the hypothesis of rank ‘h’ against rank ‘K’, i.e. against the alternative that the system is
stationary.
eigenvalue Test the hypothesis of rank ‘h’ against rank ‘h+1’.
The test works for five specifications of the deterministic terms as in Doornik et al (1998), to be
specified in the previous call to VECM:
H_ql Unrestricted constant and trend: use include="both"
H_l Unrestricted constant and restricted trend: use include="const"and LRinclude="trend"
H_lc Unrestricted constant and no trend: use include="const"
H_c Restricted constant and no trend: use LRinclude="const"
H_z No constant nor trend: use include="none"
Two testing procedures can be used:
Specific test By specifying a value for ‘r_null’. The ‘pval’ value returned gives the speciifc p-
value.
Automatic test If not value is specified for ‘r_null’, the function makes a simple automatic test: re-
turns the rank (slot ‘r’) of the first test not rejected (level specified by arg cval) as recommend
i.a. in Doornik et al (1998, p. 544).
A full table with both test statistics ad their respective p-values is given in the summary method.
P-values are obtained from the gamma approximation from Doornik (1998, 1999). Small sample
values adjusted for the sample site are also available in the summary method. Note that the ‘effective
sample size’ for the these values is different from output in gretl for example.
Value
An object of class ‘rank.test’, with ‘print’ and ‘summary methods’.
Comparison with urca

While ca.jo in package urca and rank.test both implement Johansen tests, there are a few dif-
ferences:
• rank.test gives p-values, while ca.jo gives only critical values.

• rank.test allows for five different specifications of deterministic terms (see above), ca.jo
for only three.
• ca.jo allows for seasonal and exogenous regressors, which is not available in rank.test.
• The lag is specified differently: K from ca.jo corresponds to lag+1 in rank.test.
Author(s)
Matthieu Stigler
rank.test 61
References
- Doornik, J. A. (1998) Approximations to the Asymptotic Distributions of Cointegration Tests,
Journal of Economic Surveys, 12, 573-93
- Doornik, J. A. (1999) Erratum [Approximations to the Asymptotic Distribution of Cointegration
Tests], Journal of Economic Surveys, 13, i
- Doornik, Hendry and Nielsen (1998) Inference in Cointegrating Models: UK M1 Revisited, Jour-
nal of Economic Surveys, 12, 533-72
- Johansen, S. (1996) Likelihood-based inference in cointegrated Vector Autoregressive Models,
Oxford University Press
See Also
VECM for estimating a VECM. rank.select to estimate the rank based on information criteria.
ca.jo in package urca for another implementation of Johansen cointegration test (see section
‘Comparison with urca’ for more infos).
Examples
data(barry)
## estimate the VECM with Johansen!

ve <- VECM(barry, lag=1, estim="ML")
## specific test:
ve_test_spec <- rank.test(ve, r_null=1)
ve_test_spec_tr <- rank.test(ve, r_null=1, type="trace")
ve_test_spec
ve_test_spec_tr
## No specific test: automatic method

ve_test_unspec <- rank.test(ve)
ve_test_unspec_tr <- rank.test(ve, type="trace")
ve_test_unspec
ve_test_unspec_tr
## summary method: output will be same for all types/test procedure:

summary(ve_test_unspec_tr)
## The function works for many specification of the VECM(), try:

rank.test(VECM(barry, lag=3, estim="ML"))
rank.test(VECM(barry, lag=3, include="both",estim="ML"))
rank.test(VECM(barry, lag=3, LRinclude="const",estim="ML"))
## Note that the tests are simple likelihood ratio, and hence can be obtained also manually:
-2*(logLik(ve, r=1)-logLik(ve, r=2)) # eigen test, 1 against 2
-2*(logLik(ve, r=1)-logLik(ve, r=3)) # eigen test, 1 against 3
62 regime
regime Extract a variable showing the regime
Description
This function allows to extract the indicator variable specifying the regime in which the process is
at time t.
Usage
regime(object, initVal = TRUE, timeAttr = TRUE, series = NULL, ...)

regime(object, initVal = TRUE, timeAttr = TRUE, series, discretize = TRUE, ...)
Arguments
object object of class setar or nlVar
initVal Logical. Whether the NA initial values should be returned. Default to TRUE.
timeAttr Logical. Whether the time attributes should be returned. Default to TRUE.
series Optional. A numeric vector to classify according to the model.
... additional arguments to regime
discretize logical (default TRUE) whether the series are discretized to {1,2}, or whether
regime probabilities are returned.
Value
Time series of same attributes as input to setar.
Author(s)
Matthieu Stigler
Examples
set<-setar(lynx, m=3)
regime(set)
regime(set, time=FALSE, initVal=FALSE)
plot(regime(set))
resample_vec 63
resample_vec Resampling schemes
Description
Bootstrap a vector according to multiple resampling schemes: resampling, block resampling, Wild
bootstrap.
Usage
resample_vec(
x,
boot.scheme = c("resample", "resample_block", "wild1", "wild2", "check"),
seed = NULL,
block.size = 2
)
Arguments
x A numeric vector
boot.scheme The type of resampling scheme used, see Details
seed the seed used, see set.seed
block.size for the resample_block scheme, the size of the blocks.
Details
This function offers various bootstrap/resampling schemes:
resample Resampling with replacement

resample_block Resampling contiguous observations (blocks) with replacement. Use argument
block.size
wild1 Wild bootstrap: do not resample, but add a N(0,1) distribution to each value
wild12 Wild bootstrap: same, but add instead -1 or 1.
resVar Residual variance
Description
Extracts the global and regime-dependent variance of the residuals
Usage
resVar(x, adj=c("OLS", "ML"))
64 resVar
Arguments
x setar object
adj Degrees of freedom adjustment for the variance
Details
The degree of freedom adjustment in the formula for the variance is the number of parameters when
adj="OLS" or zero when adj="ML".
Value
A vector containing:
Total The residual variance of the full sample

L, (M), H The residual variance of the lower (L), middle (if two thresholds) (M) and higher
(H) regimes
Author(s)
Matthieu Stigler
References
Press (1990).
Examples
#Lynx model as in Tong (1980, p. 387)

mod.setar <- setar(log10(lynx), mL=7,mH=2, thDelay=1, th=3.116)
summary(mod.setar)
#coefficients are same for lower regime but differ for higer
resVar(mod.setar, adj="ML")
#variance or the residuals is same for lower regime but differ for higer regime and hence for total
#Lynx model as in Tong (1980, p. 405)

mod.setar2 <- setar(log10(lynx), mL=1,mM=7,mH=2, thDelay=1, nthresh=2,th=c(2.373, 3.154))
round(coefficients(mod.setar2),3)
resVar(mod.setar2, adj="ML")
selectHyperParms 65
selectHyperParms Automatic selection of model hyper-parameters
Description
Automatic selection of model hyper-parameters
Usage
selectLSTAR(x, m, d=1, steps=d, mL = 1:m, mH = 1:m, thDelay=0:(m-1),
fast=TRUE, trace=FALSE)
selectNNET(x, m, d=1, steps=d, size=1:(m+1), maxit=1e3, trace=FALSE)
Arguments
x time series
m, d, steps embedding parameters. For their meanings, see help about nlar
mL,mH Vector of ‘low’ and ‘high’ regimes autoregressive orders
thDelay Vector of ‘threshold delay’ values
size Vector of numbers of hidden units in the nnet model
maxit Max. number of iterations for each model estimation
fast For LSTAR selection, whether a fast algorithm using starting values fro previous
models should be used
trace Logical. Whether informations from each model should be returned.
Details
Functions for automatic selection of LSTAR and NNET models hyper parameters. An exhaustive
search over all possible combinations of values of specified hyper-parameters is performed. Em-
bedding parameters m,d,steps are kept fixed.
Selection criterion is the usual AIC.
For the LSTAR model, two methods are offered:
fast=FALSE Each model is run separately, each time using the full grid search for starting values.
fast=TRUE Only the first model is run with a full grid search, while the subsequent use the first
model results for their starting values.
Value
A data-frame, with columns giving hyper-parameter values and the computed AIC for each row
(only the best 10s are returned)
Author(s)
66 selectSETAR
Examples
llynx <- log10(lynx)
selectLSTAR(llynx, m=2)
selectNNET(llynx, m=3, size=1:5)
selectSETAR Automatic selection of SETAR hyper-parameters
Description
Automatic selection of SETAR hyper-parameters
Usage
selectSETAR(x, m, d=1, steps=d, series, mL, mH,mM, thDelay=0, mTh, thVar,
th=MakeThSpec(), trace=TRUE, include = c("const", "trend","none", "both"),
common=c("none", "include","lags", "both"), model=c("TAR", "MTAR"),
ML=seq_len(mL), MH=seq_len(mH), MM=seq_len(mM), nthresh=1, trim=0.15,
criterion = c("pooled-AIC", "AIC","BIC", "SSR"), plot=TRUE,
max.iter=2, type=c("level", "diff", "ADF"), same.lags=FALSE,
restriction=c("none","OuterSymAll","OuterSymTh"), hpc=c("none", "foreach") )
Arguments
x time series
m, d, steps embedding parameters. For their meanings, see help about nlar
mL, mH,mM autoregressive order for ‘low’ (mL) ‘middle’ (mM, only useful if nthresh=2) and
‘high’ (mH) regime (default values: m). Must be <=m. Alternatively, you can
specify ML
thDelay Vector of possible ‘threshold delay’ values to check for
th Different specifications of the grid search, to pre-specify a value or set the num-
ber of points to search. SeeMakeThSpec
common Indicates which elements are common to all regimes: no, only the include
variables, the lags or both
model Currently not implemented
ML,MM,MH vector of lags for order for ‘low’ (ML) ‘middle’ (MM, only useful if nthresh=2)
and ‘high’ (MH)regime. Max must be <=m
nthresh Number of threshold of the model
selectSETAR 67

regime. Default to 0.15
criterion Model selection criterion
plot Should a plot showing the criterion values be printed? (logical)
max.iter Number of iterations for the algorithm
as in the ADF test
same.lags Logical. When AIC or pooled-AIC is used and arg m is given, should it search
for same number of lags in each regime (TRUE) or allow for different (FALSE)
lags in each regime. Different lags involves more computation
restriction Restriction on the threshold. OuterSymAll will take a symmetric threshold and
symmetric coefficients for outer regimes. OuterSymTh currently unavailable
hpc Possibility to run the bootstrap on parallel core. See details
Details
Routine for automatic selection of SETAR models hyper parameters.
An exhaustive search over all possible combinations of values of specified hyper-parameters is
performed. Thus the threshold delay, the number of lags in each regime and the threshold value are
computed.
Embedding parameters d,steps are kept fixed.
Possible criteria are the usual SSR, AIC and a pooled AIC formula: AIC(lowregimemodel) +
AIC(highregimemodel). The default criterion is the pooled AIC formula. SSR criterion can’t be
used to compare models with different lags.
When two thresholds(nthresh=2) have to be computed, the search for the second is made condi-
tional on results for first threshold as suggested in Gonzalo and Pittarakis (2002). Refinements can
be obtained by using max.iter (first threshold being re-estimated based on the second one). If SSR
is used, the number of lags in the inner regime is either the same if only arg m was given, otherwise
it has to be pre-specified. Criterion AIC can be used to determine the number of lags in the nner
regime, whereas pooled-AIC is currently not implemented for nthresh=2.
By default, all threshold values excluding the upper and lower trim of the threshold values are
taken as potential threshold. restriction can be made with arg th. See function MakeThSpec.
With the argument hpc, the heavy grid search can be run on parallel cores, thus alleviating the time
of computation. Preliminary results indicate however that the length of the series must be very
considerable in order that the parallel code becomes advantageous. To use it, the user needs simply
to choose a package (among doMC, doMPI, doSNOW or doRedis) and register the backend. See
the vignette for more details.
Value
An object of class selectSETAR (print and plot methods) with:
res A data-frame, with columns giving hyper-parameter values and the computed
AIC for each row (only the best 10/5s are returned)
res2 Same as res, returned if nthresh=2 otherwise set to NULL
68 SETAR
bests estimated hyper-parameters

th, firstBests, bests2th, ML, MM, MH
estimated parameters, from first and conditional search
criterion, nthresh,same.lags
returns args given by user
allTh all threshold values and corresponding criterion from first search
Author(s)
Antonio, Fabio Di Narzo and Stigler, Matthieu
References
Gonzalo, J. & Pitarakis, J. (2002) Estimation and model selection based inference in single and
multiple threshold models, Journal of Econometrics, 110, 319 - 352
See Also
selectLSTAR, selectNNET, MakeThSpec
Examples
llynx <- log10(lynx)
selectSETAR(llynx, m=2)
#Suggested model is the following:
setar(llynx, m=2, thDelay=1, th=3.4)
SETAR Self Threshold Autoregressive model
Description
Self Exciting Threshold AutoRegressive model.
Usage
setar(x, m, d=1, steps=d, series, mL, mM, mH, thDelay=0, mTh, thVar, th, trace=FALSE,
nested=FALSE, include = c( "const", "trend","none", "both"),
common=c("none", "include","lags", "both"), model=c("TAR", "MTAR"), ML=seq_len(mL),
MM=seq_len(mM), MH=seq_len(mH),nthresh=1,trim=0.15, type=c("level", "diff", "ADF"),
restriction=c("none","OuterSymAll","OuterSymTh") )
SETAR 69
Arguments
x time series
mL,mM, mH autoregressive order for ‘low’ (mL) ‘middle’ (mM, only useful if nthresh=2) and
‘high’ (mH)regime (default values: m). Must be <=m. Alternatively, you can
specify ML
th threshold value (if missing, a search over a reasonable grid is tried)
common Indicates which elements are common to all regimes: no, only the include
variables, the lags or both
ML,MM,MH vector of lags for order for ‘low’ (ML) ‘middle’ (MM, only useful if nthresh=2)
and ‘high’ (MH)regime. Max must be <=m
model Whether the threshold variable is taken in levels (TAR) or differences (MTAR)
nthresh Number of threshold of the model
regime. Default to 0.15
as in the ADF test
restriction Restriction on the threshold. OuterSymAll will take a symmetric threshold and
symmetric coefficients for outer regimes. OuterSymTh currently unavailable
nested Whether is this a nested call? (useful for correcting final model df)
Details
Self Exciting Threshold AutoRegressive model.
Xt+s = xt+s = (φ1,0 +φ1,1 xt +φ1,2 xt−d +· · ·+φ1,mL xt−(mL−1)d )I(zt ≤ th)+(φ2,0 +φ2,1 xt +φ2,2 xt−d +· · ·+φ2,mH xt−(m
with z the threshold variable. The threshold variable can alternatively be specified by (in that order):
thDelay z[t] = x[t - thDelay*d ]
mTh z[t] = x[t] mTh[1] + x[t-d] mTh[2] + ... + x[t-(m-1)d] mTh[m]
thVar z[t] = thVar[t]
For fixed th and threshold variable, the model is linear, so phi1 and phi2 estimation can be done
directly by CLS (Conditional Least Squares). Standard errors for phi1 and phi2 coefficients pro-
vided by the summary method for this model are taken from the linear regression theory, and are to
be considered asymptotical.
70 setar.sim
Value
An object of class nlar, subclass setar
Author(s)
References
Press (1990).
See Also
plot.setar for details on plots produced for this model from the plot generic.
Examples
#fit a SETAR model, with threshold as suggested in Tong(1990, p 377)
mod.setar <- setar(log10(lynx), m=2, thDelay=1, th=3.25)
mod.setar
summary(mod.setar)
## example in Tsay (2005)

data(m.unrate)
setar(diff(m.unrate), ML=c(2,3,4,12), MH=c(2,4,12), th=0.1, include="none")
setar.sim Simulation and bootstrap of Threshold Autoregressive model (SETAR)
Description
Simulate or bootstrap a Threshold AR (SETAR) The setar.sim function allows to simulate a
SETAR model from scratch. The most important argument is the B argument, which should be a
one row matrix, with first the constant/trend arguments, then the slope coefficients, and this for each
regime (lower, middle, high). Other arguments such as lag, nthresh indicate the dimension of this
matrix. As an example, a SETAR with 2 lags, 1 threshold, a constant, would have coefficient in
the order: c(const_L, phi_1_L, phi_2_L, const_H, phi_1_H, phi_2_H) where L is for Lower
regime, H for Higher.
setar.boot on the other side resample/bootstraps an existing setar output. It uses a recursive
approach, reconstructing the series. Residuals from the original model are resampled using different
bootstrap schemes, see resample_vec.
setar.sim 71
Usage
setar.boot(
setarObject,
seed = NULL,
...
)
linear.boot(
linearObject,
seed = NULL,
...
)
setar.sim(
B,
n = 200,
lag = 1,
nthresh = 1,
Thresh,
starting = NULL,
innov = rnorm(n),
...
)
linear.sim(
B,
n = 200,
lag = 1,
starting = NULL,
innov = rnorm(n),
...
)
Arguments
setarObject Bootstrap: the setar object to resample data from.
boot.scheme Bootstrap: which resampling scheme to use for the residuals. See resample_vec.
seed Bootstrap: seed used in the resampling
... additional arguments for the unexported setar.gen.
linearObject Bootstrap: the linear object to resample data from.
B Simulation: vector of coefficients to simulate from.
n Simulation: Number of observations to simulate.
72 setarTest
Thresh, nthresh, lag, include

Simulation: parameters for the SETAR to simulate. See setar for their descrip-
tion.
starting Simulation: Starting values (same length as lag)
innov Simulation: time series of innovations/residuals.
Value
a list with the simulated/bootstrapped data and the parameter matrix used.
Author(s)
Matthieu Stigler
See Also
SETAR to estimate a SETAR, arima.sim to simulate an ARMA.
Examples
##Simulation of a TAR with 1 threshold

TvarMat <- c(2.9,-0.4,-0.1,-1.5, 0.2,0.3)
sim<-setar.sim(B=TvarMat,lag=2, type="simul", nthresh=1, Thresh=2, starting=c(2.8,2.2))
mean(ifelse(sim>2,1,0)) #approximation of values over the threshold
#check the result

selectSETAR(sim, m=2)
##Bootstrap a TAR with two threshold (three regimes)

sun <- (sqrt(sunspot.year+1)-1)*2
sun_est <- setar(sun, nthresh=2, m=2)
sun_est_boot <- setar.boot(sun_est)
head(sun_est_boot)
##Check the bootstrap: with no resampling, is it the same series?

sun_est_boot <- setar.boot(sun_est, boot.scheme = "check")
all.equal(as.numeric(sun), sun_est_boot)
setarTest Test of linearity against threshold (SETAR)
Description
Test of linearity against threshold of Hansen (1999) with bootstrap distribution
setarTest 73
Usage
setarTest(
x,
m,
thDelay = 0,
trim = 0.1,
nboot = 10,
test = c("1vs", "2vs3"),
hpc = c("none", "foreach"),
seed = NULL
)
Arguments
x time series
m, thDelay lag and ’time delay’ for the threshold variable
regime
include Type of deterministic regressors to include: none, a constant, a trend, or constant
and trend (both).
nboot number of bootstrap replications
test whether to test AR against SETAR, or SETAR(1 reg) against SETAR(2 reg)
hpc Possibility to run the bootstrap on parallel core. See details in
boot.scheme Type of resampling scheme to use for the residuals. See resample_vec.
seed Seed used in the bootstrap resampling TVECM.HStest
Details
Estimation of the first threshold parameter is made with CLS, a conditional search with one iteration
is made for the second threshold. The Ftest comparing the residual sum of squares (SSR) of each
model is computed.
Fij = T ((Si − Sj )/Sj )
where Si is the SSR of the model with i regimes (and so i-1 thresholds).
Three test are available. The both first can be seen as linearity test, whereas the third can be seen
as a specification test: once the 1vs2 or/and 1vs3 rejected the linearity and henceforth accepted the
presence of a threshold, is a model with one or two thresholds preferable?
Test 1vs2: Linear AR versus 1 threshold TAR
Test 1vs3: Linear AR versus 2 thresholds TAR
Test 2vs3: 1 threshold TAR versus 2 thresholds TAR
The two first tests are computed together and available with test="1vs". The third test is available
with test="2vs3".
74 setarTest
The homoskedastic bootstrap distribution is based on resampling the residuals from H0 model (ar
for test 1vs, and setar(1) for test 2vs3), estimating the threshold parameter and then computing the
Ftest, so it involves many computations and is pretty slow.
Value
A object of class "Hansen99Test" containing:
SSRs The residual Sum of squares of model AR, 1 threshold TAR and 2 thresholds
TAR
Ftests The Ftest statistic for the test
PvalBoot The bootstrap p-values for the test selected
CriticalValBoot
The critical values for the test selected
Ftestboot All the F-test computed
firstBests, secBests
The thresholds for the original series, obtained from search for 1 thresh (firstBests)
and conditional search for 2 thresh (secBests)
nboot,m The number of bootstrap replications (nboot), the lags used (m)
Author(s)
Matthieu Stigler
References
See Also
TVAR.LRtest for the multivariate version. SETAR for estimation of the model.
Examples
#Data used by Hansen

sun <- (sqrt(sunspot.year + 1) - 1) * 2
#Test 1vs2 and 1vs3

#setarTest(sun, m=11, thDelay=0:1, nboot=5,trim=0.1, test="1vs")
setarTest_IIPUs_results 75
setarTest_IIPUs_results
Results from the setarTest, applied on Hansen (1999) data
Description
Saved objects
Usage
setarTest_IIPUs_results
Format
A list containing output from two tests
See Also
Example in IIPUs
Examples
library(tsDyn)
data(IIPUs)
B <- 1000
## Not run:
test_1 <- setarTest(IIPUs, m=16, thDelay=5, nboot=B)
test_2 <- setarTest(IIPUs, m=16, thDelay=5, nboot=B, test = "2vs3")
## End(Not run)
sigmoid sigmoid functions
Description
Some sigmoid functions. See R sources for their definition
Usage
sigmoid(x)
dsigmoid(x)
Arguments
x numeric vector
76 STAR
Author(s)
J. L. Aznarte
STAR STAR model
Description
STAR model fitting with automatic selection of the number of regimes based on LM tests.
Usage
star(x, m=2, noRegimes, d = 1, steps = d, series, rob = FALSE,
mTh, thDelay, thVar, sig=0.05, trace=TRUE, control=list(), ...)
Arguments
x time series
noRegimes max number of regimes
rob perform robust test (not implemented)
sig significance level for the tests to select the number of regimes.
control further arguments to be passed as control list to optim
trace should additional infos be printed out?
Details
The function star implements the iterative building strategy described in [1] to identify and estimate
Smooth Transition Autoregressive models.
[1] T. Terasvirta, "Specification, estimation and evaluation of smooth transition autoregressive mod-
els", J. Am. Stat. Assoc. 89 (1994): 208-218.
Value
star returns an object of class nlar, subclass star, i.e. a list with informations about the fitted
model.
toLatex.setar 77
Author(s)
J. L. Aznarte M.
See Also
addRegime
Examples
mod.star <- star(log10(lynx), mTh=c(0,1), control=list(maxit=3000))
mod.star
addRegime(mod.star)
toLatex.setar Latex representation of fitted setar models
Description
Produce LaTeX output of the SETAR model.
Usage
toLatex(object, digits = 3, label, ...)
Arguments
object fitted setar model (using nlar)
digits options to be passed to format for formatting numbers
label LaTeX label passed to the equation
... Not used
Author(s)
See Also
setar, nlar-methods
Examples
mod.setar <- setar(log10(lynx), m=2, thDelay=1, th=3.25)

toLatex(mod.setar)
78 TVAR
TVAR Multivariate Threshold Vector Autoregressive model
Description
Estimate a multivariate Threshold VAR (TVAR), either using lags as transition variable (default),
or specifying an external variable.
Usage
TVAR(
data,
lag,
model = c("TAR", "MTAR"),
commonInter = FALSE,
nthresh = 1,
thDelay = 1,
mTh = 1,
thVar,
trim = 0.1,
ngrid,
gamma = NULL,
around,
plot = FALSE,
dummyToBothRegimes = TRUE,
trace = TRUE,
trick = "for",
max.iter = 2
)
Arguments
data time series
model Whether the transition variable is taken in levels (TAR) or difference (MTAR)
commonInter Whether the deterministic regressors are regime specific (commonInter=FALSE)
or not.
nthresh Number of thresholds
PLEASE NOTE that the notation is currently different to univariate models in
tsDyn. The left side variable is taken at time t, and not t+1 as in univariate cases.
mTh combination of variables with same lag order for the transition variable. Either
a single value (indicating which variable to take) or a combination
TVAR 79
thVar Optional. External transition variable.

regime
ngrid number of elements of the grid, especially for nthresh=3
gamma prespecified threshold values
around The grid search is restricted to ngrid values around this point. Especially useful
for nthresh=3.
plot Whether a plot showing the results of the grid search should be printed
dummyToBothRegimes
Whether the dummy in the one threshold model is applied to each regime or not.
trace should additional infos be printed out?
trick type of R function called: for or mapply
max.iter Number of iterations for the algorithm
Details
For fixed th and threshold variable, the model is linear, so estimation can be done directly by CLS
(Conditional Least Squares). The search of the parameters values is made upon a grid of potential
values. So it is pretty slow.
nthresh=1: estimation of one threshold model (two regimes) upon a grid of ngrid values (default to
ALL) possible thresholds and delays values.
nthresh=2: estimation of two thresholds model (three regimes) Conditional on the threshold found
in model where nthresh=1, the second threshold is searched. When both are found, a second grid
search is made with 30 values around each threshold.
nthresh=3: DOES NOT estimate a 3 thresholds model, but a 2 thresholds model with a whole grid
over the thresholds parameters (so is really slow) with a given delay, is there rather to check the
consistency of the method nthresh=2
Value
An object of class TVAR, with standard methods.
Author(s)
Matthieu Stigler
References
Lo and Zivot (2001) "Threshold Cointegration and Nonlinear Adjustment to the Law of One Price,"
Macroeconomic Dynamics, Cambridge University Press, vol. 5(4), pages 533-76, September.
See Also
lineVar for the linear VAR/VECM, TVAR.LRtest to test for TVAR, TVAR.sim to simulate/bootstrap
a TVAR.
80 TVAR.LRtest
Examples
data(zeroyld)
tv <- TVAR(zeroyld, lag=2, nthresh=2, thDelay=1, trim=0.1, mTh=1, plot=FALSE)
print(tv)
summary(tv)
# a few useful methods:

plot(tv)
predict(tv)
c(AIC(tv), BIC(tv), logLik(tv))
TVAR.LRtest Test of linearity
Description
Multivariate extension of the linearity against threshold test from Hansen (1999) with bootstrap
distribution
Usage
TVAR.LRtest(
data,
lag = 1,
trend = TRUE,
series,
thDelay = 1:m,
mTh = 1,
thVar,
nboot = 10,
plot = FALSE,
trim = 0.1,
test = c("1vs", "2vs3"),
model = c("TAR", "MTAR"),
trace = FALSE,
check = FALSE
)
Arguments
data multivariate time series
trend whether a trend should be added
TVAR.LRtest 81
series name of the series

thVar external transition variable
nboot Number of bootstrap replications
plot Whether a plot showing the results of the grid search should be printed
regime
test Type of usual and alternative hypothesis. See details
model Whether the threshold variable is taken in level (TAR) or difference (MTAR)
hpc Possibility to run the bootstrap on parallel core. See details in TVECM.HStest
check Possibility to check the function by no sampling: the test value should be the
same as in the original data
Details
This test is just the multivariate extension proposed by Lo and Zivot of the linearity test of Hansen
(1999). As in univariate case, estimation of the first threshold parameter is made with CLS, for the
second threshold a conditional search with one iteration is made. Instead of a Ftest comparing the
SSR for the univariate case, a Likelihood Ratio (LR) test comparing the covariance matrix of each
model is computed.
LRij = T (ln(det Σ̂i ) − ln(det Σ̂j ))
where Σ̂i is the estimated covariance matrix of the model with i regimes (and so i-1 thresholds).
Three test are available. The both first can be seen as linearity test, whereas the third can be seen
as a specification test: once the 1vs2 or/and 1vs3 rejected the linearity and henceforth accepted the
presence of a threshold, is a model with one or two thresholds preferable?
Test 1vs2: Linear VAR versus 1 threshold TVAR
Test 1vs3: Linear VAR versus 2 threshold2 TVAR
Test 2vs3: 1 threshold TAR versus 2 threshold2 TAR
The both first are computed together and available with test="1vs". The third test is available with
test="2vs3".
The homoskedastik bootstrap distribution is based on resampling the residuals from H0 model,
estimating the threshold parameter and then computing the Ftest, so it involves many computations
and is pretty slow.
82 TVAR.sim
Value
A list containing:
-The values of each LR test
-The bootstrap Pvalues and critical values for the test selected
Author(s)
Matthieu Stigler
References
Lo and Zivot (2001) "Threshold Cointegration and Nonlinear Adjustment to the Law of One Price,"
Macroeconomic Dynamics, Cambridge University Press, vol. 5(4), pages 533-76, September.
See Also
setarTest for the univariate version. OlsTVAR for estimation of the model.
Examples
data(zeroyld)
data<-zeroyld
TVAR.LRtest(data, lag=2, mTh=1,thDelay=1:2, nboot=3, plot=FALSE, trim=0.1, test="1vs")
TVAR.sim Simulation of a multivariate Threshold Autoregressive model (TVAR)
Description
Simulate a multivariate Threshold VAR (TVAR)
Usage
TVAR.sim(
B,
Thresh,
nthresh = 1,
n = 200,
lag = 1,
thDelay = 1,
mTh = 1,
TVAR.sim 83
starting = NULL,
innov = rmnorm(n, varcov = varcov),
varcov = diag(1, nrow(B)),
show.parMat = FALSE,
...
)
TVAR.boot(
TVARobject,
innov,
seed,
boot.scheme = c("resample", "wild1", "wild2", "check"),
...
)
Arguments
B Matrix of coefficients to simulate

Thresh The threshold value(s). Vector of length nthresh
nthresh number of threshold (see details)
n Number of observations to create when type="simul"
include Type of deterministic regressors to include. NOT WORKING PROPERLY
CURRENTLY if not const
starting Starting values (matrix of dimension lag x k). If not given, set to zero.
innov Innovations used for simulation. Should be matrix of dim n x k. By default
multivariate normal. For the bootstrap case TVAR.boot, residuals are resampled
if argument is missing.
varcov Variance-covariance matrix for the innovations. By default identity matrix.
show.parMat Logical. Should the parameter matrix be shown? Useful to understand how to
give right input
... Further arguments passed to the underlying (un-exported) TVAR.gen function
TVARobject Object of class TVAR generated by function TVAR
seed Optional. Seed for the random resampling function.
boot.scheme The bootstrap scheme.
84 TVAR.sim
Details
This function offers the possibility to generate series following a TVAR.
By giving a matrix of coefficients, on can only simulate a VAR (nthresh=0) or TVAR (nthresh=1
or 2). One can have a specification with constant (default), trend, both or none (see arg include).
Order in parameters is include/lags (VECM) and include/lags/include/lags for TVECM, hence, a
matrix for a TVECM with 3 regimes, a const and a 2 lags would have 2 lines and 2*(1+4) columns.
The innovations can be given by the user (a matrix of dim nxk, here n does not include the starting
values!), by default it uses a multivariate normal distribution, with covariance matrix specified by
varcov. The starting values (of dim lags x k) can be given. The user should take care for their
choice, since it is not sure that the simulated values will cross the threshold even once.
The matrix ‘B’ has to be in the form: constant, trend, lags, then repeated if many regimes. In case
of uncertainty, using who.parMat=TRUE will print the matrix as interpreted by the function, helping
the user to feed the right input.
For the bootstrap, the function resamples data from a given TVAR model generated by TVAR, re-
turning the resampled data. A residual recursive bootstrap is used, where one uses either a simple
resampling, or the Wild bootstrap, either with a normal distribution (wild1) or inverting the sign
randomly (wild2)
Value
A matrix with the simulated/bootstrapped series.
Author(s)
Matthieu Stigler
See Also
TVAR to estimate the TVAR. Similar TVECM.sim and TVECM.boot for TVECM, VAR.sim and VAR.boot
for VAR models estimated with lineVar.
Examples
## TVAR.sim: Simulation of a TVAR with 1 threshold

B <- rbind(c(0.11928245, 1.00880447, -0.009974585, -0.089316, 0.95425564, 0.02592617),
c(0.25283578, 0.09182279, 0.914763741, -0.0530613, 0.02248586, 0.94309347))
colnames(B) <- paste(rep(c("Const", "Lag_1_var1", "Lag_1_var2"), 2), c("Low", "High"), sep="_")
sim <- TVAR.sim(B=B,nthresh=1,n=500, mTh=1, Thresh=5, starting=matrix(c(5.2, 5.5), nrow=1))
#estimate the new serie

TVAR(sim, lag=1, dummyToBothRegimes=TRUE)
TVECM 85
## TVAR.boot: Bootstrap a TVAR with two threshold (three regimes)

data(zeroyld)
serie <- zeroyld
mod <- TVAR(data=serie,lag=1, nthresh=1)
TVAR.boot(mod)
TVECM Threshold Vector Error Correction model (VECM)
Description
Estimate a Threshold Vector Error Correction model (VECM)
Usage
TVECM(
data,
lag = 1,
nthresh = 1,
trim = 0.05,
ngridBeta = 50,
ngridTh = 50,
plot = TRUE,
th1 = list(exact = NULL, int = c("from", "to"), around = "val"),
th2 = list(exact = NULL, int = c("from", "to"), around = "val"),
beta = list(exact = NULL, int = c("from", "to"), around = c("val", "by")),
restr = c("none", "equal", "signOp"),
common = c("All", "only_ECT"),
dummyToBothRegimes = TRUE,
beta0 = 0,
methodMapply = FALSE,
trace = TRUE
)
Arguments
data time series
nthresh number of threshold (see details)
regime
ngridBeta number of elements to search for the cointegrating value
ngridTh number of elements to search for the threshold value
plot Whether the grid with the SSR of each threshold should be plotted.
86 TVECM
th1 different possibilities to pre-specify an exact value, an interval or a central point

for the search of the threshold (or first threshold if nthresh=2)
th2 different possibilities to pre-specify an exact value or a central point for the
search of the second threshold (used only if nthresh=2)
beta different possibilities to pre-specify an exact value, an interval or a central point
for the search of the cointegrating value
restr Currently not available
common Whether the regime-specific dynamics are only for the ECT or for the ECT and
the lags
dummyToBothRegimes
Whether the dummy in the one threshold model is applied to each regime or not.
beta0 Additional regressors to include in the cointegrating relation
methodMapply only for programming. Is to make the choice between a for loop or mapply
implementation
Details
For fixed threshold and cointegrating vector, the model is linear, so estimation of the regression
parameters can be done directly by CLS (Conditional Least Squares). The search of the threshold
and cointegrating parameters values which minimize the residual sum of squares (SSR) is made on
a grid of potential values. For specification of the grids, see below.
The function can estimate one as well as two thresholds:
nthresh=1: estimation of one threshold model (two regimes) upon a grid of ngridTh values (default
to ALL) possible thresholds and delays values.
nthresh=2: estimation of two thresholds model (three regimes). Conditional on the threshold found
in model where nthresh=1, the second threshold is searched. When both are found, a second
grid search is made with 30 values around each threshold.
The model can be either with a threshold effect on all variables ("All") or only on the error correction
term (ECT) (argument "only ECT"). In the second case, the value for the middle threshold is taken
a null, as in Balke and Fomby (1997).
The grid for the threshold parameters can be set in different ways, through the argument th1 , th2
and beta:
exact: Pre-specified value (for beta: cointegrating vector will be c(1,-beta))

int: Specify an interval (of length ngridTh ) in which to search.
around: Specify to take ngridTh points around the value given.
The default is to do an interval search. Interval bounds for the threshold interval are simply the trim
and 1-trim percents of the sorted error correction term. For the cointegrating parameter, bounds of
the interval are obtained from the (OLS) confidence interval of the linear cointegration case. It is
often found however that this interval is too tight. It is hence recommended to inspect the plot of
the grid search.
TVECM.HStest 87
Value
Fitted model data
Author(s)
Matthieu Stigler
References
Hansen, B. and Seo, B. (2002), Testing for two-regime threshold cointegration in vector error-
correction models, Journal of Econometrics, 110, pages 293 - 318
Seo, M. H. (2009) Estimation of non linear error-correction models, Working paper
See Also
VECM for the linear VECM, TVAR for the threshold VAR, TVECM.SeoTest to test for TVECM,
TVECM.sim to simulate/bootstrap a TVECM.
Examples
data(zeroyld)
##Estimate a TVECM (we use here minimal grid, it should be usually much bigger!)
tvec <- TVECM(zeroyld, nthresh=2,lag=1, ngridBeta=20, ngridTh=30, plot=TRUE,trim=0.05, common="All")
print(tvec)
summary(tvec)
#Obtain diverse infos:

AIC(tvec)
BIC(tvec)
res.tvec<-residuals(tvec)
#export the equations as Latex:

toLatex(tvec)
TVECM.HStest Test of linear cointegration vs threshold cointegration
Description
Tests the null of linear cointegration against threshold cointegration following Hansen and Seo
(2002). Fixed regressor and residual bootstrap are available.
88 TVECM.HStest
Usage
TVECM.HStest(data, lag=1, ngridTh=300, trim=0.05,
nboot=100, fixed.beta=NULL, intercept=TRUE,
boot.type=c("FixedReg", "ResBoot"),
hpc=c("none", "foreach"))
Arguments
data Time series
ngridTh Number of threshold grid points to evaluate the optimal threshold.
trim Trimming parameter indicating the minimal percentage of observations in each
regime
fixed.beta Numeric. User pre-specified cointegrating value, as in VECM (i.e. cointegrating
vector will be c(1, -beta.fixed)). When NULL (default), the value is estimated
from the linear VECM.
intercept Logical. Whether an intercept has to be included in the VECM
boot.type Character. Type of bootstrap simulation (only if nboot>0)
hpc Possibility to run the bootstrap on parallel core. See details
Details
This test follows the implementation done by Hansen and Seo (2002). The cointegrating value is
estimated from the linear VECM. Then, conditional on this value, the LM test is run for a range of
different threshold values. The maximum of those LM test values is reported.
Two bootstrap are available: a fixed regressor, as well as a usual residual bootstrap (using the
function TVECM.sim).
Available methods are print(), summary() and plot().
With the argument hpc, the burdensome bootstrap replication can be run on parallel cores, thus
alleviating the time of computation. The user needs simply to choose a package (among doMC,
doMPI, doSNOW or doRedis) and register the backend. See the vignette for more details.
Value
A list containing diverse values:
stat The sup-LM statistic.

values The whole LM values.
PvalBoot The bootstrap p-value
CriticalValBoot
The bootstrap critical values
allBoots The boot sup-LM values
args Some user given args (nboot, boot.type)
TVECM.HStest 89
Reproducibility
Comparison with original paper is made difficult as values of the test are not shown in the paper,
only their critical values, which depend on random bootstrap.
Comparison is done with the GAUSS code available on the page of Bruce Hansen. Running tar_ci,
we have the same sup-LM value when lags=1 and lags=2, a higher value with lag=3. When the test
is run with pre-specified beta values, we have different results, sometimes higher but also smaller
sup-LM value.
Author(s)
Matthieu Stigler
References
See Also
zeroyld: data used in the original paper of Hansen and Seo.
TVECM.SeoTest: a similar test, but with null hypothesis of no-cointegration.
TVECM for estimating a TVECM, TVECM.sim for simulating/bootstrap a TVECM,
Examples
#Use original data from paper:
data(zeroyld)
dataPaper<-zeroyld
# Test: nboot, number of bootstrap replications, should be high
## Not run:
test1<-TVECM.HStest(dataPaper, lag=1, intercept=TRUE, nboot=1000)
## End(Not run)
#we use here for the example a much smaller number of bootstrap:
test1<-TVECM.HStest(dataPaper, lag=1, intercept=TRUE, nboot=10)
test1
summary(test1)
plot(test1)
#can have only specific plots:

plot(test1, which="LM values")
plot(test1, which="Density")
## Run the function in parallel:

## Not run:
#we show here the use with package doMC
library(doMC)
registerDoMC(2) #Number of cores
90 TVECM.SeoTest
test1<-TVECM.HStest(dataPaper, lag=1, intercept=TRUE, nboot=1000, hpc="foreach")
## End(Not run)
TVECM.SeoTest No cointegration vs threshold cointegration test
Description
Test the null of no cointegration against threshold cointegration with bootstrap distribution of Seo
(2006)
Usage
TVECM.SeoTest(
data,
lag,
beta,
trim = 0.1,
nboot,
plot = FALSE,
check = FALSE
)
Arguments
data time series
beta Pre-specified cointegrating value (i.e. cointegrating vector will be c(1, -beta))
regime
plot Whether a grid with the SSR of each threshold should be printed
hpc Possibility to run the bootstrap on parallel core. See details in TVECM.HStest
check Possibility to check the function by no sampling: the test value should be the
same as in the original data
Details
For this test, the cointegrating value has to be specified by the user.
The model used is one where the threshold effect concerns only the cointegrating vector, and only
in the outer regimes.
TVECM.SeoTest 91
Due to the presence of parameters unidentified under the null hypothesis, the test employed is a
Sup-Wald test, that means that for each combination of the thresholds, a Wald Test is computed and
the supremum of all tests is taken. For each bootstrap replication, this approach is taken, so that the
test is really slow.
Value
A list containing diverse informations:

Estimated threshold parameters and usual slope parameters.
Value of the test.
Critical and Pvalue from bootstrap distribution.
Author(s)
Matthieu Stigler
References
Seo, Myunghwan, 2006. "Bootstrap testing for the null of no cointegration in a threshold vector
error correction model," Journal of Econometrics, vol. 127(1), pages 129-150, September.
See Also
TVECM for estimating a TVECM, TVECM.sim for simulating/bootstrap a TVECM,
Examples
# As the function takes long long time to be executed, we show in in don't run environement
## Not run:
data(zeroyld)
#can be useful to check whether the bootstrap is working:

#without sampling, results of boot should be same as original
#this is indeed not always the case duye to floating point algorithm
TVECM.SeoTest(zeroyld,lag=2, beta=1, trim=0.1,nboot=2, plot=FALSE,check=TRUE)
#then run the function:

TVECM.SeoTest(zeroyld,lag=2, beta=1, trim=0.1,nboot=100, plot=FALSE,check=FALSE)
## End(Not run)
92 TVECM.sim
TVECM.sim Simulation and bootstrap a VECM or bivariate TVECM
Description
Estimate or bootstraps a multivariate Threshold VAR
Usage
TVECM.sim(
B,
n = 200,
lag = 1,
beta,
nthresh = 1,
Thresh,
starting = NULL,
innov = rmnorm(n, varcov = diag(1, nrow(B))),
returnStarting = FALSE,
...
)
VECM.sim(
B,
n = 200,
lag = 1,
beta,
starting = NULL,
innov = rmnorm(n, varcov = diag(1, nrow(B))),
...
)
VECM.boot(
object,
seed = NULL,
...
)
TVECM.boot(
object,
TVECM.sim 93
seed = NULL,
...
)
Arguments
B Simulation: Matrix of coefficients to simulate
n Simulation: Number of observations to simulate.
beta The cointegrating value
Thresh, nthresh, lag, include
Simulation: parameters for the VECM/TVECM to simulate. See TVECM for their
description.
starting Simulation: Starting values (same length as lag = 1)
innov Simulation: time series of innovations/residuals.
show.parMat Logical. Whether to show how the parameter matrix B is interpreted.
returnStarting Logical. Whether to return the starting values.
... additional arguments for the unexported TVECM.gen.
object Object computed by function TVECM or linear VECM
boot.scheme Bootstrap: which resampling scheme to use for the residuals. See resample_vec.
seed Bootstrap: seed used in the resampling
Details
This function offers the possibility to generate series following a VECM/TVECM from two ap-
proaches: bootstrap or simulation. VECM.sim is just a wrapper for TVECM.sim.
When the argument matrix is given, on can only simulate a VECM (nthresh=0) or TVECM
(nthresh=1 or 2). One can have a specification with constant ("const"), "trend", "both" or
"none" (see argument include). Order for the parameters is ECT/include/lags for VECM and
ECT1/include1/lags1/ECT2/include2/lags2 for TVECM. To be sure that once is using it correctly,
setting show.parMat = TRUE will show the matrix of parameters together with their values and
names.
The argument beta is the cointegrating value on the right side of the long-run relationship, and
hence the function use the vector (1,-beta). The innov argument specifies the innovations. It should
be given as a matrix of dim nxk, (here n does not include the starting values!), by default it uses a
multivariate normal distribution, with covariance matrix specified by varcov.
The starting values (of dim lags x k) can be given through argument starting. The user should
take care for their choice, since it is not sure that the simulated values will cross the threshold even
once. Notice that only one cointegrating value is allowed. User interested in simulating a VECM
with more cointegrating values should do use the VAR representation and use TVAR.sim.
The second possibility is to bootstrap series. This is done on a object generated by TVECM (or VECM).
A simple residual bootstrap is done, or one can simulate a series with the same parameter matrix
and with normal distributed residuals (with variance pre-specified), corresponding to Monte-carlo
simulations.
One can alternatively give only the series, and then the function will call internally TVECM.
94 TVECM.sim
Value
A matrix with the simulated/bootstrapped series.
Author(s)
Matthieu Stigler
See Also
VECM or TVECM to estimate the VECM or TVECM. Similar TVAR.sim and TVAR.boot for TVAR,
VAR.sim and VAR.boot for VAR models estimated with lineVar models.
Examples
###reproduce example in Enders (2004, 2 edition) p. 350,

# (similar example in Enders (2010, 3 edition) 301-302).
if(require(mnormt)){
#see that the full "VAR" coefficient matrix is:
A <- matrix(c(-0.2, 0.2, 0.2, -0.2), byrow=TRUE, ncol=2)
# but this is not the input of VECM.sim. You should decompose into the a and b matrix:
a<-matrix(c(-0.2, 0.2), ncol=1)
b<-matrix(c(1,-1), nrow=1)
# so that:
a%*%b
# The a matrix is the input under argument B, while the b matrix is under argument beta:
# (the other zeros in B are for the not-specified lags)
innov<-rmnorm(100, varcov=diag(2))
Bvecm <- rbind(c(-0.2, 0,0), c(0.2, 0,0))
vecm1 <- VECM.sim(B=Bvecm, beta=1,n=100, lag=1,include="none", innov=innov)
ECT <- vecm1[,1]-vecm1[,2]
#add an intercept as in panel B

Bvecm2 <- rbind(c(-0.2, 0.1,0,0), c(0.2,0.4, 0,0))
vecm2 <- VECM.sim(B=Bvecm2, n=100,beta=1, lag=1,include="const", innov=innov)
par(mfrow=c(2,1))
plot(vecm1[,1], type="l", main="Panel a: no drift or intercept", ylab="", xlab="")
lines(vecm1[,2], lty=2)
plot(vecm2[,1], type="l", main="Panel b: drift terms (0.1)", ylab="", xlab="")
lines(vecm2[,2], lty=2)
}
##Bootstrap a TVAR with 1 threshold (two regimes)
data(zeroyld)
TVECMobject <- TVECM(zeroyld, nthresh=1, lag=1, ngridBeta=20, ngridTh=20, plot=FALSE, trace = FALSE)
TVECM.boot(TVECMobject)
UsUnemp 95
##Check the bootstrap: do we get original series, when not resampling residuals?
TVECM.boot.check <- TVECM.boot(TVECMobject, boot.scheme = "check")
all.equal(as.data.frame(TVECM.boot.check), zeroyld)
UsUnemp US unemployment series used in Caner and Hansen (2001)
Description
This data, used as example in Caner and Hansen (2001), contains the monthly US adult male un-
employment from 1956 to 1999.
Usage
data(UsUnemp)
Format
A monthly time series of class ts starting in January 1956 and ending in August 1999.
Source
Caner and Hansen,Threshold autoregression with a unit root Econometrica, 2001, 69, 1555-1596
available at: http://www.ssc.wisc.edu/~bhansen/papers/cv.htm
VAR.sim Simulate or bootstrap a VAR model
Description
Allow to either simulate from scratch (by providing coefficients) or bootstrap from an estimated
VAR model,
Usage
VAR.sim(
B,
n = 200,
lag = 1,
starting = NULL,
innov = rmnorm(n, varcov = varcov),
varcov = diag(1, nrow(B)),
96 VAR.sim
...
)
VAR.boot(
VARobject,
boot.scheme = c("resample", "wild1", "wild2", "check"),
seed,
...
)
Arguments
B Matrix of coefficients.
n Number of observations to simulate
lag Number of lags of the VAR to simulate
include Type of deterministic regressors to include in the VAR to simulate
starting Starting values (matrix of dimension lag x k) for the VAR to simulate. If not
given, set to zero.
innov Innovations used for in the VAR to simulate. Should be matrix of dim n x k. By
default multivariate normal.
varcov Variance-covariance matrix for the innovations. By default identity matrix.
show.parMat Logical. Should the parameter matrix be shown? Useful to understand how to
give right input
returnStarting Whether starting values are returned. Default to FALSE
... Further arguments passed to the underlying (un-exported) VAR.gen function
VARobject Object of class VAR generated by function lineVar
boot.scheme The bootstrap scheme. See details.
seed Optional. Seed for the random resampling function.
Details
For the bootstrap, the function resamples data from a given VAR model generated by lineVar,
returning the resampled data. A residual recursive bootstrap is used, where one uses either a simple
resampling, or the Wild bootstrap, either with a normal distribution (wild1) or inverting the sign
randomly (wild2)
Value
A matrix with the resampled series.
Author(s)
Matthieu Stigler
VARrep 97
See Also
lineVar to estimate the VAR. Similar TVECM.sim and TVECM.boot for TVECM, TVAR.sim and
TVAR.boot for TVAR models.
Examples
## VAR.sim: simulate VAR as in Enders 2004, p 268

B1<-matrix(c(0.7, 0.2, 0.2, 0.7), 2)
var1 <- VAR.sim(B=B1, n=100, include="none")
ts.plot(var1, type="l", col=c(1,2))
B2<-rbind(c(0.5, 0.5, 0.5), c(0, 0.5, 0.5))

varcov<-matrix(c(1,0.2, 0.3, 1),2)
var2 <- VAR.sim(B=B2, n=100, include="const", varcov=varcov)
ts.plot(var2, type="l", col=c(1,2))
## VAR.boot: Bootstrap a VAR

data(zeroyld)
mod <- lineVar(data=zeroyld,lag=1)
VAR.boot(mod)
VARrep VAR representation
Description
Show the VAR representation of a VECM
Usage
VARrep(object, ...)

VARrep(object, ...)

VARrep(object, ...)
Arguments
object An object of class ‘VECM’ created by VECM, or of class ‘VAR’ created by
lineVar
... Currently not used
98 VECM
Value
A matrix containing the parameters of the VECM under their VAR representation.
Author(s)
Matthieu Stigler
References
Examples
data(barry)
# VECM model:
mod_vecm <- VECM(barry, lag=2, estim="ML")
VARrep(mod_vecm)
# VAR model:
mod_var <- lineVar(barry, lag=2, I="diff")
VARrep(mod_var)
VECM Estimation of Vector error correction model (VECM)
Description
Estimate a VECM by either Engle-Granger (2OLS) or Johansen (MLE) method.
Usage
VECM(
data,
lag,
r = 1,
beta = NULL,
estim = c("2OLS", "ML"),
LRinclude = c("none", "const", "trend", "both"),
exogen = NULL
)
VECM 99
Arguments
data multivariate time series (first row being first=oldest value)
lag Number of lags (in the VECM representation, see Details)
r Number of cointegrating relationships
beta for VECM only: user-specified cointegrating values, the cointegrating vector
will be taken as: (1, -beta) If NULL, will be estimated using the estimator
specified in estim
estim Type of estimator: 2OLS for the two-step approach or ML for Johansen MLE
LRinclude Type of deterministic regressors to include in the long-term relationship. Can
also be a matrix with exogeneous regressors (2OLS only).
exogen Inclusion of exogenous variables (first row being first=oldest value). Is either of
same size than data (then automatically cut) or than end-sample.
Details
This function is just a wrapper for the lineVar, with model="VECM".
More comprehensive functions for VECM are in package vars. Differences with that package are:
Engle-Granger estimator The Engle-Granger estimator is available

Presentation Results are printed in a different ways, using a matrix form
lateX export The matrix of coefficients can be exported to latex, with or without standard-values
and significance stars
Prediction The predict method contains a newdata argument allowing to compute rolling fore-
casts.
Two estimators are available: the Engle-Granger two step approach (2OLS) or the Johansen (ML).
For the 2OLS, deterministic regressors (or external variables if LRinclude is of class numeric) can
be added for the estimation of the cointegrating value and for the ECT. This is only working when
the beta value is not pre-specified.
The arg beta is the cointegrating value, the cointegrating vector will be taken as: (1, -beta).
Note that the lag specification corresponds to the lags in the VECM representation, not in the VAR
(as is done in package vars or software GRETL). Basically, a VAR with 2 lags corresponds here to
a VECM with 1 lag. The lag can be set to 0, although some methods (irf, fevd) won’t work for this
case.
0
#’The arg beta allows to specify constrained cointegrating values, leading to ECT = β Xt−1 . It
should be specified as a K × r matrix. In case of r = 1, can also be specified as a vector. Note that
the vector should be normalised, with the first value to 1, and the next values showing the opposite
sign in the long-run relationship −β. In case the vector has K − 1 values, this is what lineVar is
doing, setting (1, −β). Note finally one should provide values for all the coefficients (eventually
except for special case of r=1 and k-1), if you want to provide only part of the parameters, and let
the others be estimated, look at the functions in package urca.
The eigenvector matrix β is normalised using the Phillips triangular representation, see Hamilton
(1994, p. 576) and Juselius (2006, p. 216), see coefA for more details.
100 VECM
Value
An object of class VECM (and higher classes VAR and nlVar) with methods:
Usual methods: Print, summary, residuals, fitted, vcov
Fit criteria: AIC, BIC, MAPE, mse, logLik (the latter only for models estimated with MLE)
Prediction: predict and predict_rolling
coef extraction: Extract cointegrating/adjustment coefficients, coefA, coefB coefPI
VAR/VECM methods: Impulse response function (irf.VECM) and forecast error variance decom-
position (fevd)
LaTeX: toLatex
Author(s)
Matthieu Stigler
References
Juselius (2006) The Cointegrated VAR model, Oxford University Press
See Also
coefA, coefB and coefPI to extract the relevant parameter matrices.
lineVar TVAR and TVECM for the corresponding threshold models. linear for the univariate AR
model.
Examples
data(zeroyld)
data<-zeroyld
#Fit a VECM with Engle-Granger 2OLS estimator:

vecm.eg<-VECM(zeroyld, lag=2)
#Fit a VECM with Johansen MLE estimator:

vecm.jo<-VECM(zeroyld, lag=2, estim="ML")
#compare results with package vars:

if(require(vars)) {
data(finland)
#check long coint values
all.equal(VECM(finland, lag=2, estim="ML", r=2)$model.specific$beta,
cajorls(ca.jo(finland, K=3, spec="transitory"), r=2) $beta, check.attributes=FALSE)
# check OLS parameters
all.equal(t(coefficients(VECM(finland, lag=2, estim="ML", r=2))),
coefficients(cajorls(ca.jo(finland, K=3, spec="transitory"), r=2)$rlm), check.attributes=FALSE)
}
VECM_symbolic 101
##export to Latex
toLatex(vecm.eg)
toLatex(summary(vecm.eg))
options("show.signif.stars"=FALSE)
toLatex(summary(vecm.eg), parenthese="Pvalue")
options("show.signif.stars"=TRUE)
VECM_symbolic Virtual VECM model
Description
Pedagogical tool to create a symbolic VECM model, i.e. just for representation purpose.
Usage
VECM_symbolic(
alpha,
beta,
lags,
inc,
include = c("none", "const", "trend", "both")
)
Arguments
alpha Matrix of alpha speed adjustment coefficients.
beta Matrix of alpha, cointegrating coefficients.
lags Matrix containing the lags coefficients.
inc Matrix containing the include (see following arg.) coefficients.
include Character indicating the type of deterministic term included, if any.
Value
An object of class ‘VECM’, without however any data.
Examples
a<-matrix(c(-0.4, 0.1), ncol=1)

b<-matrix(c(1, -2), ncol=2)
102 zeroyld
# VECM_symb(alpha=a, beta=t(b))
d<- VECM_symbolic(alpha=a, beta=t(b))
VARrep(d)
d<- VECM_symbolic(alpha=a, beta=t(b), lags=matrix(0, ncol=2, nrow=2))
VARrep(d)
LagMat <- matrix(c(0.1, 0.3, 0.1, 0.2), ncol=2, nrow=2)
incMat <- matrix(c(0.5, 0.1), ncol=1)
d3<- VECM_symbolic(alpha=a, beta=t(b), lags=LagMat, inc=incMat, include="const")
VARrep(d3)
zeroyld zeroyld time series
Description
U.S. Term Structure Data, 1951-1991. Dataset used by Hansen and Seo (2002). The data contains
the 12 month short rate and 120 month long rate.
Usage
zeroyld
zeroyldMeta
Format
zeroyld contains two variables, while zeryldMeta contains also Year and Month columns.
zeroyld is a data frame with 482 observations and 2 variables:
short.run numeric Short term, 12 month

long.run numeric Long term, 120 month
Source
The data can be downloaded from: http://www.ssc.wisc.edu/~bhansen/progs/joe_02r.zip.
The authors themselves took the data from the webpage of Huston McCulloch: https://www.asc.
ohio-state.edu/mcculloch.2/ts/mcckwon/mccull.htm
See Also
TVECM.HStest: Hansen and Seo test.

TVECM for estimating a TVECM.
zeroyld 103
Examples
data(zeroyldMeta)
plot(zeroyldMeta$Date, zeroyldMeta$short.run, type = "l", xlab = "Date", ylab ="Rate")
lines(zeroyldMeta$Date, zeroyldMeta$long.run, lty = 2)
legend("topleft", lty = c(1, 2), legend = c("Short rate: 12 months", "Long rate: 120 months"))
Index
∗ Testing for linearity delta.lin, 18

UsUnemp, 95 getTh, 21
∗ VAR isLinear, 30
VARrep, 97 lags.select, 32
∗ VECM LINEAR, 33
rank.select, 57 lineVar, 34
VARrep, 97 llar, 36
VECM_symbolic, 101 logLik.nlVar, 39
∗ bootstrap LSTAR, 40
TVAR.sim, 82 MAPE, 45
VAR.sim, 95 mse, 46
∗ cointegration nlar-methods, 47
rank.select, 57 NNET, 49
VARrep, 97 plot methods, 50
VECM_symbolic, 101 rank.select, 57
∗ datasets rank.test, 59
barry, 12 regime, 62
IIPUs, 25 resVar, 63
m.unrate, 43 selectHyperParms, 65
selectSETAR, 66
setarTest_IIPUs_results, 75
SETAR, 68
UsUnemp, 95
setar.sim, 70
zeroyld, 102
setarTest, 72
∗ predict
sigmoid, 75
predict_rolling, 55
STAR, 76
∗ regression
toLatex.setar, 77
fevd.nlVar, 19
tsDyn-package, 3
fitted.nlVar, 20
TVAR, 78
irf.linear, 27 TVAR.LRtest, 80
predict.nlar, 52 TVAR.sim, 82
predict.TVAR, 54 TVECM, 85
∗ ts TVECM.HStest, 87
aar, 4 TVECM.SeoTest, 90
accuracy_stat, 5 TVECM.sim, 92
addRegime, 6 VAR.sim, 95
autopairs, 8 VARrep, 97
autotriples, 9 VECM, 98
autotriples.rgl, 10 VECM_symbolic, 101
availableModels, 11
delta, 16 AAR (aar), 4
104
INDEX 105
aar, 4 GIRF, 22
accuracy, 5
accuracy_stat, 5 hnorm, 9, 10
addRegime, 6, 77
AIC.nlar (nlar-methods), 47 IIPUs, 25, 75
ar_mean, 7 irf, 35
irf.ar (irf.linear), 27
arima.sim, 72
irf.linear, 27
as.data.frame, 38
irf.nlVar, 24
as.data.frame.llar (llar), 36
irf.nlVar (irf.linear), 27
as.data.frame.rank.select
irf.setar (irf.linear), 27
(rank.select), 57
irf.TVAR (irf.linear), 27
autopairs, 4, 8
irf.TVECM (irf.linear), 27
autotriples, 4, 9, 11
irf.VAR (irf.linear), 27
autotriples.rgl, 4, 10
irf.varest, 29, 30
availableModels, 4, 11, 48
irf.VECM, 100
barry, 12 irf.VECM (irf.linear), 27
BBCTest, 12, 32 isLinear, 30
bds.test, 17 KapShinTest, 31
BIC.nlar (nlar-methods), 47
blrtest, 15 lags.select, 32
legend, 51
ca.jo, 15, 59–61 LINEAR, 33
cajorls, 15 linear, 7, 23, 35, 71, 100
charac_root, 14 linear (LINEAR), 33
coef.nlar (nlar-methods), 47 linear.boot (setar.sim), 70
coefA, 99, 100 linear.sim (setar.sim), 70
coefA (coefB), 14 lineVar, 15, 20, 30, 34, 39, 55, 56, 58, 79, 84,
coefB, 14, 100 94, 96, 97, 99, 100
coefPI, 100 llar, 4, 36
coefPI (coefB), 14 logLik, 100
logLik.nlVar, 39
d2sigmoid (sigmoid), 75 logLik.VAR (logLik.nlVar), 39
delta, 4, 16 logLik.VECM, 32, 58
delta.lin, 4, 18 logLik.VECM (logLik.nlVar), 39
delta.lin.test, 17 LSTAR, 40
deviance.nlar (nlar-methods), 47 lstar, 7, 51, 53
dsigmoid (sigmoid), 75 lstar (LSTAR), 40
fevd, 19, 100 m.unrate, 43
fevd.nlVar, 19 MakeThSpec, 44, 66–68
fitted (fitted.nlVar), 20 makeThSpec (MakeThSpec), 44
fitted.nlar (nlar-methods), 47 MAPE, 45, 100
fitted.nlVar, 20 MAPE.nlar (nlar-methods), 47
format, 77 mse, 46, 100
format.pval, 59 mse.nlar (nlar-methods), 47
gam, 4 nlar, 14, 34, 47, 65, 66, 77

getTh, 21 nlar-methods, 47, 51
106 INDEX
NNET, 49 selectNNET, 68
nnet, 49 selectNNET (selectHyperParms), 65
nnetTs (NNET), 49 selectSETAR, 44, 45, 66
selectSetar (selectSETAR), 66
OlsTVAR, 82 selectsetar (selectSETAR), 66
OlsTVAR (TVAR), 78 set.seed, 63
optim, 41, 42, 49, 76 SETAR, 68, 72, 74
setar, 7, 23, 51, 53, 71, 72, 77
par, 48, 50 setar (SETAR), 68
plogis, 41 setar.boot (setar.sim), 70
plot, 20, 30, 51 setar.sim, 70
plot methods, 50 setarTest, 13, 32, 72, 82
plot-methods (plot methods), 50 setartest (setarTest), 72
plot.aar (aar), 4 setarTest_IIPUs_results, 75
plot.GIRF_df (GIRF), 22 sigmoid, 75
plot.llar (llar), 36 sm, 10
plot.lstar, 42 sm.autoregression, 10
plot.lstar (plot methods), 50 sm.density, 9
plot.nlar (nlar-methods), 47 sm.regression, 9, 10
plot.setar, 70 sm.ts.pdf, 9
plot.setar (plot methods), 50 STAR, 76
plot_ECT, 51 star, 7
polyroot, 14 star (STAR), 76
predict (predict.nlar), 52 summary.aar (aar), 4
predict.nlar, 52, 56 summary.linear (LINEAR), 33
predict.TVAR, 54 summary.nlar (nlar-methods), 47
predict.VAR, 35 summary.rank.select (rank.select), 57
predict.VAR (predict.TVAR), 54 summary.rank.test (rank.test), 59
predict.VECM (predict.TVAR), 54 summary.setar (SETAR), 68
predict_rolling, 53, 55, 55, 100
print.aar (aar), 4 terasvirta.test, 17
print.linear (LINEAR), 33 toLatex, 35
print.llar (llar), 36 toLatex.nlar (nlar-methods), 47
print.rank.select (rank.select), 57 toLatex.setar, 77
print.rank.test (rank.test), 59 tsDyn (tsDyn-package), 3
print.summary.linear (LINEAR), 33 tsDyn-package, 3
printCoefmat, 48 TVAR, 23, 35, 78, 83, 84, 87, 94, 97, 100
TVAR.boot, 94, 97
rank.select, 32, 33, 57, 61 TVAR.boot (TVAR.sim), 82
rank.test, 59, 59 TVAR.LRtest, 74, 79, 80
regime, 35, 62 TVAR.sim, 79, 82, 93, 94, 97
resample_vec, 63, 70, 71, 73, 93 TVECM, 23, 35, 51, 84, 85, 89, 91, 93, 94, 97,
residuals.nlar (nlar-methods), 47 100, 102
resVar, 63 TVECM.boot, 84, 97
rgl, 10 TVECM.boot (TVECM.sim), 92
TVECM.HStest, 73, 81, 87, 90, 102
selectHyperParms, 65 TVECM.HStest (TVECM.HStest), 87
selectLSTAR, 68 TVECM.SeoTest, 87, 89, 90
selectLSTAR (selectHyperParms), 65 TVECM.sim, 84, 87–89, 91, 92, 93, 97
INDEX 107
ugarchroll, 56
UsUnemp, 95
VAR.boot, 84, 94
VAR.boot (VAR.sim), 95
VAR.sim, 84, 94, 95
VARrep, 35, 54, 55, 97
VARselect, 33
VECM, 15, 20, 30, 35, 39, 51, 55, 56, 58–61, 87,
88, 93, 94, 97, 98
VECM.boot (TVECM.sim), 92
VECM.sim (TVECM.sim), 92
VECM_symbolic, 101
zeroyld, 89, 102

zeroyldMeta (zeroyld), 102

Ts Dyn

Uploaded by

Ts Dyn

Uploaded by

Package ‘tsDyn’

tsDyn-package Getting started with the tsDyn package

aar Additive nonlinear autoregressive model

accuracy_stat Forecasting accuracy measures.

## S3 method for class 'pred_roll'

mod_var_pred <-predict_rolling(object=mod_var, nroll=10, n.ahead=1:3)

addRegime addRegime test

ar_mean Long-term mean of an AR(p) process

Computes the long term mean of an AR process

## S3 method for class 'linear'

## S3 method for class 'setar'

## S3 method for class 'lstar'

object an object of class linear, setar or lstar

No constant nor trend The LT mean is 0

autopairs Bivariate time series plots

autotriples Trivariate time series plots

autotriples.rgl Interactive trivariate time series plots

availableModels Available models

dolcan Exchange rate US/Can dollar.

BBCTest Test of unit root against SETAR alternative

A object of class "BBC2004Test" containing:

setarTest for a test with stationarity as a null.

BBCTest(lynx, m=3, test="Wald", grid="minPerc")

charac_root Characteristic roots of the AR coefficients

## S3 method for class 'nlar'

coefB Extract cointegration parameters A, B and PI

## S3 method for class 'VECM'

## S3 method for class 'ca.jo'

## S3 method for class 'VECM'

## S3 method for class 'ca.jo'

delta delta test of conditional independence

delta(log10(lynx), m=3, eps=sd(log10(lynx)))

delta.lin delta test of linearity

fevd.nlVar Forecast Error Variance Decomposition

ve <- VECM(barry, lag=2)

## get fitted values:

getTh Extract threshold(s) coefficient

GIRF Generalized Impulse response Function (GIRF)

## S3 method for class 'setar'

## S3 method for class 'linear'

## S3 method for class 'nlVar'

## S3 method for class 'GIRF_df'

object An object of class linear, setar or nlVar (TVAR, TVECM)

n_simu: Id for the simulation (total number is n.hist times n.shock)

IIPUs US monthly industrial production from Hansen (1999)

## Figure 2 in paper (p. 555)

## Table 1 in paper (p. 556)

## Table 2 in paper (p. 559)

## estimate with Hansen threshold:

## Table 4 in paper (p. 561)

#this is obviously a error in Hansen, see:

## setar test (takes long to compute, even with small nboot)

setarTest(IIPUs, m=16, thDelay=5, nboot=10, test="2vs3")

## results from the test is stored in: setarTest_IIPUs_results

irf.linear Impulse response function

## S3 method for class 'setar'

## S3 method for class 'ar'

## S3 method for class 'VAR'

## S3 method for class 'VECM'

## S3 method for class 'TVAR'

## S3 method for class 'TVECM'

x Object of class ‘VAR’; generated by lineVar(), or object of class ‘VECM’; gen-

KapShinTest Test of unit root against SETAR alternative with

statistic The three (SupW, AvgW, ExpW) test statistics computed

lags.select Selection of the lag with Information criterion.

LINEAR Linear AutoRegressive models

xt+s = φ0 + φ1 xt + φ2 xt−d + · · · + φm xt−(m−1)d + t+s

lineVar Multivariate linear models: VAR and VECM

model Model to estimate. Either a VAR or a VECM

Engle-Granger estimator The Engle-Granger estimator is available

Fitted model data

#compare results with package vars:

VECM.ML <- lineVar(zeroyld, lag=2, model="VECM", estim="ML")

###Check Johansen MLE

llar Locally linear model

llar.predict(x, m, d=1, steps=d, series, n.ahead=1,

xt+s = φ0 + φ1 xt + φ2 xt−d + · · · + φm xt−(m−1)d + t+s