Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Library: times

acf computes the autocorrelation function for time series
acfplot plots the autocorrelation function of a time series.
annarchtest This macro calculates either the Lagrange Multiplier (LM) form or the T-Rsquare (TR2) form of a test for conditional heteroskedasticity based on Artificial Neural Networks. The first argument of the function is the vector of residuals, the second optional argument is the order of the test, the third optional argument is the number of hidden units of the Neural Network. The second optional argument is either a vector or a scalar. If no second argument is provided, the default orders are 2, 3, 4, and 5. The third argument may be either a vector, or a scalar. If both second and third arguments are vectors, the test will be calculated for all combinations of orders and hidden units. If no third argument is provided, the number of hidden units by default is set to 3. The fourth argument is the form of the test. This argument is a string of characters, which can be either "LM" or "TR2". The default fourth argument is "LM", i.e., the Lagrange Multiplier form. The macro returns in the first column the order of the test, in the second column the number of hidden units, in the third column the value of the test, in the fourth column the 95% critical value of the null hypothesis for that order, and in the fifth column the P-value of the test.
annlintest This macro calculates the neural network test for neglected nonlinearity proposed by Lee, White and Granger (1993). This statistic is evaluated from uncentered squared multiple correlation of an auxiliary regression in which we regress the residuals of a linear regression on the regressors of this regression and the principal components of a nonlinear transformation of the regressors. The first argument of the macro is the series y. The second argument is either a set of regressors X if the series is regressed on X, or the number of lags if the series is regressed on its lagged realizations. The macro adds automatically a constant if the constant term is missing in X. If the series is regressed on its past realizations, then the second argument, i.e., the number of lags, may be a vector. In that case, the corresponding linear models and statistics for neglected nonlinearity are computed. The third optional argument is the number of hidden units of the neural network, which should be greater than or equal to 3. The fourth optional argument is the number of principal components used in the auxiliary regression. The number of principal components should be less than the number of corresponding hidden units. The default third argument is the vector (10,20), the default fourth argument is the vector (2,3). If the series is regressed on a set of exogeneous variables X, the macro returns the number of principal components used in the auxiliary regression, the value of the test, the 95% critical value for the null hypothesis of the test, and the P value of the test. If the series is regressed on its past realizations, the number of lagged explanatory variables is also displayed.
archest estimates a GARCH process with mean zero by QMLE
archtest This macro calculates either the Lagrange Multiplier (LM) form or the R squared (TR2) form of Engle's ARCH test. The first argument of the function is the vector of residuals, the second optional argument is the lag order of the test. This ; second argument may be either a scalar, or a vector. In the later case, the test is evaluated for all the order components of the vector. If this second optional argument is not provided, the default lag orders are 2, 3, 4, and 5. The third optional argument is the form of the test. This argument is a string, which can be either "LM" or "TR2". The default third argument is "LM", i,e., the LM form. The macro returns in the first column the order of the test, in the second column the value of the test, in the third column the 95% critical value of the null hypothesis for that order, and in the fourth column the P-value of the test.
armacls estimates an autoregressive moving average process with mean zero by conditional least squares
armalik estimates an ARMA(1,1) process with mean zero by maximum likelihood using the innovation algorithm
buildz computes lag vectors of y (help procedure for VARs)
corrint computes the correlation integral for time series
fracbrown the result*normal(2*p*nu+1) gives a fractional brownian motion with respect to alpha=2*H ( H = Hurst coefficent)
genarch generates a GARCH process with Gaussian innovations
genarma generates an autoregressive moving average process with mean zero
genbil generates a bilinear process x(t)=sum phi(i)x(t-i) + sum theta(j)e(t-j) + sum sum gamma(i,j)x(t-i)e(t-j)
genexpar generates an exponential AR process
gentar generates a Threshold AR process
gph Estimation of the degree of long memory of a time series by using a log-periodogram regression
gpplot returns the Grassberger-Procaccia plot for time series
hurst estimates the Hurst coefficent of a process using the R/S statistics
jump detects jump points in the time series. Optional parameter alpha controls the the sensitivity of the procedure. Recomended values are 0.5 ... 4.0. The default value is 2.0. The output vector j, which has the same length as data, indicates the detected jumps. NaN values are not allowed.
kem Calculates estimates of mu, F, Q and R in a state-space model using EM-algorithm. The state-space model is assumed to be in the following form:

y_t = H x_t + v_t

x_t = F x_t-1 + w_t

x_0 ~ (mu,Sig), v_t ~ (0,Q), w_t ~ (0,R)

Parameters Sig and H are assumed known.

kemitor Calculates observations of a given state-space model. The state-space model is assumed to be in the following form:

y_t = H x_t + ErrY_t

x_t = F x_t-1 + ErrX_t

x_0 = mu

kfilter Calculates a filtered time serie (uni- or multivariate) using the Kalman filter equations. The state-space model is assumed to be in the following form:

y_t = H x_t + v_t

x_t = F x_t-1 + w_t

x_0 ~ (mu,Sig), v_t ~ (0,Q), w_t ~ (0,R)

All parameters are assumed known.

kpss Calculation of the KPSS statistics for I(0) against long-memory alternatives. We consider the two tests, denoted by KPSS_mu and KPSS_t, and respectively based on a regression on a constant mu, and on a constant and a time trend t. As in the KPSS paper, we consider for the autocorrelation consistent variance estimator the truncation lags denoted by L0, L4, and L12. The quantlet returns the value of the truncation lag, the type of the test, i.e., const or trend, the estimated statistic, and the critical value for a 95 percent confidence interval for I(0). If the value of the test exceeds the critical value, a star symbol * is displayed after the test statistic.

ksmoother Calculates a smoothed time serie (uni- or multivariate) using the Kalman smoother equations. The state-space model is assumed to be in the following form:

y_t = H x_t + v_t

x_t = F x_t-1 + w_t

x_0 ~ (mu,Sig), v_t ~ (0,Q), w_t ~ (0,R)

All parameters are assumed known.

lo Calculation of the Lo statistic for long-range dependence. The first argument of the quantlet is the series, the second optional argument is the vector of truncation lags of the autocorrelation consistent variance estimator. If the second optional argument is missing, the vector of truncation lags is set to m = 5, 10, 25, 50. The quantlet returns the estimated statistic with its corresponding order. If the estimated statistic is outside the interval (0.809, 1.862), which is the 95 percent confidence interval for no long-memory, a star symbol * is displayed in the third column. The other critical values are in Lo's paper.

lobrob Semiparametric test for I(0) of a time series against fractional alternatives, i.e., long-memory and antipersistence. The test is semiparametric in the sense that it does not depend on a specific parametric form of the spectrum in the neighborhood of the zero frequency. The first argument of the function is the series. The second optional argument is the vector of bandwidth, i.e., the parameter specifying the number of harmonic frequencies around zero to be considered. By default, the macro uses the automatic bandwidth given in Lobato and Robinson. If the user provides his own vector of bandwidths, then the function returns the value of the test for each component of the bandwidth vector. If the value of the test is in the lower tail of the standard normal distribution, the null hypothesis of I(0) is rejected against the alternative that the series displays long-memory. If the value of the test is in the upper tail of the standard normal distribution, the null hypothesis I(0) is rejected against the alternative that the series is antipersistent.
neweywest Calculation of the Newey and West Heteroskedastic and Autocorrelation Consistent estimator of the variance. The first argument of the quantlet is the series, the second optional argument is the vector of truncation lags of the autocorrelation consistent variance estimator. If the second optional argument is missing, the vector of truncation lags is set to m = 5, 10, 25, 50.

pacf computes the partial autocorrelation function
pacfplot plots the partial autocorrelation function of a time series.
parzen calculates the parzen window
pgram computes and plots the raw (log) periodogram of a time series
roblm Semiparametric average periodogram estimator of the degree of long memory of a time series. The first argument of the macro is the series, the second optional argument is a strictly positive constant q, which is also strictly less than one. The third optional argument is the bandwidth vector m. By default q is set to 0.5 and the bandwidth vector is equal to m = n/4, n/8, n/16. If q and m contain several elements, the estimator is evaluated for all the combinations of q and m. The quantlet returns in the first column the estimated degree of long-memory, in the second column the number of frequencies considered, in the third column the value of q.
robwhittle Semiparametric Gaussian estimator of the degree of long memory of a time series, based on the Whittle estimator. The first argument is the series, the second argument is the vector of bandwidths, i.e., the number of frequencies after zero that are considered. By default, the bandwidth vector m = n/4, n/8, n/16, where n is the sample size. This quantlet displays the estimated parameter d, with the number of frequencies considered.
rvlm Calculation of the rescaled variance test for I(0) against long-memory alternatives. The statistic is the centered kpss statistic based on the deviation from the mean. The limit distribution of this statistic is a Brownian bridge whose distribution is related to the distribution of the Kolmogorov statistic. This statistic can also be used for detecting long-memory in ARCH models. The first argument of the quantlet is the series, the second optional argument is the vector of truncation lags for the spectral based autocorrelation consistent estimator of the variance. If this optional argument is not provided, the default vector of truncation lags used by Kwiatkowski, Phillips, Schmidt and Shin is used. The quantlet returns the order of the truncation lag, the rescaled variance statistic, with the 95% critical value.
spec estimates and plots the spectral density of a time series
tdiff a difference operator for time series allow multiple differences and seaonal difference
timeplot plots a time series in multiple windows with user-specified maximum length per window
timesmain loads the libraries needed for the macros in times
timestest executes some tests for the macros defined in times.lib
varest computes a VAR estimate
varfc computes one step forecast for a VAR-model

Keywords - Function groups - @ A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

(C) MD*TECH Method and Data Technologies, 28.6.1999