Procedures

Many of the capabilities of RATS are made available through the use of procedures. Procedures are text files of RATS instructions which function much like a built-in instruction, but with greater flexibility. Consider, for instance, a Dickey-Fuller test. This is just a simple linear regression with a specific set of variables. A augmented Dickey-Fuller test with lags chosen by some means requires a set of linear regressions. There is no great advantage to having this "hard-coded" into a RATS instruction (like LINREG itself) since the calculation is quite straightforward. The advantage of providing the D-F test through a procedure is that it's easy to modify either by you, or by us.

The RATS distribution includes over 300 procedures, with more created all the time. Because procedures are so important to the use of RATS, it's important for the program to be able to find the most current versions. RATS maintains a set of search directories. This is generally set up properly when you first install the program, but you can change the directory, or add additional ones for your own procedures.

If you execute a procedure with (say)

@dfunit lgdp

RATS will automatically search for a file whose name matches that of the procedure, with a .SRC extension, so in this case dfunit.src. The search path(s) are set in the program's preferences in the Procedures Directories field on the Directories tab. If you have more than one directory listed, it will start with the first, then, if it's not found, move to the next. By putting your own directory above the standard one, you can use an altered copy before the distribution one.

Executing a Procedure

The basic syntax for executing a procedure is:

@procedure_name( options ) parameters

# < supplementary cards > (if needed)

Everything is the same as a standard RATS instruction except two things:

•The procedure name is preceded by @.

•You can’t abbreviate the procedure name.

The following uses the procedure @DFUNIT (Dickey–Fuller Unit root test) to do unit root tests on the series TBILL using several sets of options:

@dfunit(det=constant) tbill

@dfunit(det=constant,maxlags=6,method=gtos) tbill

@dfunit(det=constant,maxlags=6,method=aic) tbill

Loading the Procedure

To use a procedure or function stored on a separate file, you need to have RATS execute the instructions that define the procedure. In most cases, RATS can do this automatically, by searching for a file with a .src extension whose name matches the name of the procedure. Otherwise, you can compile procedures stored on a file using the SOURCE instruction:

source name of file with PROCEDURE or FUNCTION

If RATS can’t load a procedure:

•Check to make sure you did a full installation, including all of the files supplied with RATS. "Installing RATS" for details.

•Check the “Procedure Directory” setting on the “Directories” tab in the preferences. In order for RATS to find procedure files automatically, this should point to the directory containing your procedures. See "Where to Find It" for Windows or Macintosh.

•If you know where the procedure file is installed on your computer, include a SOURCE instruction (with a complete path and filename) prior to the procedure call to source in the procedure.

•You may be running an out-of-date version of RATS that did not include the procedure in question. If you do Help—Update Procedures..., RATS will list all procedures that have been added or changed since your version was released and that can be used on what you are running. You can do Help—About RATS to check which version you are using. See "Updates" or email sales@estima.com for information on updating. If you are using a network license at a company or university, your institution may already have the most recent version available—check with your system support staff.

•If all else fails, you should be able to download the procedure using the “Procedure Browser” on the “Resources” section of our website.

Procedure Listing

The most important of the procedures are listed in this section alphabetically with (for most) links to a description of their use. For other procedures (or even for the ones shown here), you can open the .SRC file and read the comments to see how they work.

@ABLAGS	Generates Arellano-Bond set of instruments
@ACF	Performs autocorrelation analysis on a series
@ACF2PACF	Computes autocorrelations from partial autocorrelations
@ADFAUTOSELECT	Selects optimal lag length to be used for an ADF test
@ADTEST	Anderson-Darling test for normality
@AGFRACTD	Andrews-Guggenberger estimate of fractional difference
@APBREAKTEST	Andrews-Ploberger Structural Break Test
@APGRADIENTTEST	Andrews-Ploberger Structural Break Test for GARCH/Maximum Likelihood
@ARAUTOLAGS	Computes Information Criteria for AR models using Yule-Walker or Burg
@ARCHTEST	Tests a series for ARCH effects
@ARMADLM	Sets up a DLM based upon an ARMA model
@ARMASPECTRUM	Graphs the spectral density for an input ARMA model
@BAING	Estimates factors in a factor model using Bai-Ng formulas
@BAIPERRON	Bai-Perron Test for Multiple Structural Changes
@BAYESTST	Bayesian Unit Root test
@BDINDTESTS	Battery of independence tests
@BDSTEST	Brock-Decher-Scheinkman test for i.i.d.
%BETAPARMS	Computes parameters required for beta distribution
@BICORRTEST	Hinich bi-correlations test for autocorrelation
@BJAUTOFIT	Automated ARIMA model selection
@BJDIFF	Aids in selection of differencing operations
@BJEST	Estimates an ARIMA model
@BJFORE	Estimates and forecasts an ARIMA model
@BJIDENT	Box-Jenkins identification tool
@BJTHEIL	Computes Theil U statistics for an ARIMA model
@BJTRANS	Aids in selection of preliminary transformation
@BKFILTER	Band pass filter using Baxter-King method
@BNDECOMP	Beveridge-Nelson decomposition
@BPPANELTESTS	Does Breusch-Pagan (and related) tests for random effects
@BQDODRAWS	Does Monte Carlo draws from a VAR with BQ factorization
@BREITUNG	Breitung test for unit roots in panel data
@BRYBOSCHAN	Bry-Boschan business cycle dating (Pagan-Harding for quarterly data)
@BSOPTION	Black-Scholes option pricing procedure
@BVARBUILDPRIOR	Builds up a prior for a linear equation one element at a time.
@BVARBUILDPRIORMN	Builds up a prior for a Bayesian VAR
@CANCORR	Computes canonical correlations for two sets of series
@CFEAT	Identifying turning points and cyclical phases of a series
@CFFILTER	Band pass filter using Christiano-Fitzgerald method
@CHOWDENNING	Chow-Denning multiple variance ratio test
@CHOWLIN	Distributes a series to a higher frequency using related series (obsolete: use @DISAGGREGATE instead)
@CLASSICALDECOMP	Decomposes a series into trend, seasonal, irregular
@CONDITION	Conditional forecasting procedure
@CORRADO	Corrado non-parametric event test
@CORRINTEGRAL	Computes a correlation integral for a series
@CROSSCORR	Computes and graphs cross correlations of two series
@CROSSPEC	Computes and graphs phase and coherence
@CSERIESSYMM	Complex series symmetrizer
@CUMPDGM	Durbin’s Cumulated Periodogram for serial correlation
@CUSUMTESTS	Computes and displays CUSUM and CUSUMQ tests
@CVSTABTEST	Stability tests on a covariance matrix of series
@CXLOGDENSITY	Computes Whittle likelihood using complex matrices
@CXLOGDENSITYCV	Computes concentrated multivariate Whittle likelihood using complex matrices
@DENTON	Distributes a series to a higher frequency using proportional Denton method (obsolete: use @DISAGGREGATE instead)
@DFUNIT	Dickey-Fuller unit root test
@DISAGGREGATE	General disaggregation (interpolation/distribution) procedure
@DISTRIB	Distribution from one frequency to a higher frequency
@DIVISIA	Computes a Divisia index
@DLMIRF	Impulse Reponse Function from a State-Space model
@DMARIANO	Diebold-Mariano Forecast Comparison Test
@DURBINLEVINSON	Computes Autoregressive Representations using Durbin-Levinson recursion
@EBA	Extreme Bounds Analysis, from Granger and Uhlig
@EGCRTVAL	Computes "exact" critical values for Dickey-Fuller unit root and Engle-Granger cointegration tests.
@EGTEST	Engle-Granger test for Cointegration
@EGTESTRESIDS	Engle-Granger test for Cointegration on 1st stage residuals
@ELFCALC	Computes empirical likelihood for a set of moment conditions
@ENDERSGRANGER	Enders/Granger threshold unit root tests
@ENDERSSIKLOS	Enders-Siklos test for cointegration with threshold effect
@EQNTOACF	Creates an ACF from an ARMA equation
@ERSTEST	Elliott-Rothenberg-Stock unit root tests
@EXACTINVERSE	Computes exact (limit) inverse with “infinite” components
@FLUX	General Nyblom fluctuations test
@FM	Estimates cointegrating vectors using Fully Modified Least Squares
@FORCEDFACTOR	Factors covariance matrix with specific vector column/row
@GAIN	Computes and graphs the gain and phase of a pair of series
%GAMMAPARMS	Computes parameters required for gamma distribution
@GARCHFORE	Univariate GARCH forecasting procedure
@GAUSSHERMITE	Generates weights and grid points for Gauss-Hermite numerical integration
@GEDDRAW	Generates draws for a generalized error distribution
@GENCOMBOS	Generate all combinations of a set of integers
@GLSDETREND	Local to unity GLS detrending routine
@GMAUTOFIT	Automated ARIMA model selection (seasonal models)
@GNEWBOLD	Granger-Newbold forecast comparison test
@GPH	Geweke-Porter-Hudak estimate of fractional differencing
@GRAPHMATRIX	Graphs a RECTANGULAR array of series on separate graphs
@GREGORYHANSEN	Gregory-Hansen test for cointegration with breaks
@GRIDSERIES	Generates a series with an equally spaced grid
@HADRI	Hadri test for unit roots in panel data
@HALTON	Generates Halton sequences
@HANNANRISSANEN	Estimates an ARIMA model using the Hannan-Rissanen algorithm
@HEGY	Performs a HEGY seasonal unit root test for quarterly data
@HILLGEV	Estimates tail index for a distribution using Hill’s method
@HINICHTEST	Hinich bispectrum test for linearity and Gaussianity
@HISTOGRAM	Histogram procedure
@HJBOUNDS	Computes Hansen-Jagannathan bounds for a set of returns
@HPFILTER	Hodrick-Prescott filter (obsolete)
@HTUNIT	Harris-Tzavalis unit root test for panel data
@HURST	Computes a Hurst exponent
@ICSS	Inclan-Tiao test for breaks in variance
@IMHOF	Computes CDF for quadratic form in Normal(0,1) variables. Obsolete—use functions %QFORMPDF or %QFORMDPDF functions.
@IMPACTSIGNFLIP	Corrects signs of columns in a factor matrix
@INTERPOL	Interpolation from one frequency to a higher one
%INVCHISQRPARMS	Computes parameters required for inverse chi-squared distribution
%INVGAMMAPARMS	Computes parameters required for inverse gamma distribution
@IPSHIN	Im, Pesaran and Shin panel unit root test
@IRFRESTRICT	Used to build up a set of zero-valued restrictions for impulse responses impact and higher steps
@JOHMLE	Johansen ML Cointegration analysis
@KPSS	KPSS (Kwiatowski, Phillips, Schmidt, and Shin) stationarity test
@KSCPOSTDRAW	Draws from posterior density needed in stochastic volatility model
@LAGPOLYROOTS	Creates table of the roots of a lag polynomial
@LEVINLIN	Levin-Lin-Chu test for unit roots in panel data
@LIML	Limited information maximum likelihood estimation
@LOCALDLM	Creates matrices for local (level or trend) DLM
@LOCALDLMINIT	Calculates initial guesses for component variances in a local level/trend
@LOCALTREND	Local smoothing regression
@LOGMVSKEWT	Function for log density of multivariate skew-t distribution
%LOGNORMALPARMS	Computes parameters required for log normal distribution
%LOGSKEWGEDDENSITY	Function for log density of skew-GED distribution
%LOGSKEWGEDGARCH	Function for log density of skew-GED distribution for use with GARCH with DENSITY option.
%LOGSKEWTDENSITY	Function for log density of skew-t distribution
%LOGSKEWTGARCH	Function for log density of skew-t distribution for use with GARCH with DENSITY option.
@LPUNIT	Lumsdaine-Papell unit root test with structural breaks
@LSDVC	Estimates a dynamic FE model with correction for bias
@LSUNIT	Lee-Strazicich unit root tests with one or more structural breaks
@MAAUTOLAGS	Computes Information Criteria for MA models using innovations algorithm
@MACKINNONCV	Computes Mackinnon’s Critical values for DF and EG tests
@MANNWHITNEY	Performs Mann-Whitney test for comparison of samples
@MARKOV	Functions supporting Markov Chain Models (obsolete use @MSSETUP instead)
@MATPEEK	Extracting/inserting information from/into rectangular arrays (obsolete: use %XSUBMAT and %PSUBMAT functions instead)
@MCFEVDTABLE	Organizes tables of FEVD’s with confidence bands
@MCGRAPHIRF	Organizes graphs of IRF’s with confidence bands
@MCLEODLI	Performs a McLeod-Li test for 2nd order dependence
@MCMCPOSTPROC	Calculates sample statistics from MCMC realizations
@MCPROCESSIRF	Organizes error bands for IRF’s based upon MC results
@MCVARDODRAWS	Does Monte Carlo draws from a VAR to generate IRF’s
@MEANGROUP	Mean group estimator for panel data
@MEPLOT	Does Mean Excess Return plots
@MESA	Computes and graphs a spectrum using Maximum Entropy Method
@MHEGY	Implements the monthly version of the “HEGY” tests
@MIXED	Mixed estimation of a single equation
@MIXVAR	Mixed estimation of an equation with a Bayesian prior
@MONTEVAR	Monte Carlo Integration of VAR Impulse Response confidence bands
@MSEMSETUPSTD	Markov switching procedures for EM estimation
@MSREGRESSION	Markov switching linear regression procedures
@MSSETUP	Markov switching general support procedures
@MSSYSREGRESSION	Markov switching linear systems regression procedures
@MSVARSETUP	Markov switching VAR setup procedures
@MULTIPLEBREAKS	Multiple structural change analysis per Bai-Perron
@MVARCHTEST	Multivariate test for ARCH
@MVBNDECOMP	Multivariate Beveridge-Nelson decomposition via a VAR
@MVGARCHFORE	Multivariate GARCH forecasting
@MVGARCHTOVECH	Extracts a VECH representation from GARCH estimates
@MVIDENT	Creates a Tiao-Box cross correlation matrix
@MVJB	Multivariate Jarque-Bera normality test
@MVQSTAT	Hosking’s Multivariate Q statistic
@NBERCYCLES	Generates dummies based upon NBER cycle dates
@OLSHODRICK	Computes Hodrick standard errors
@PACF2AR	Generates coefficients for an AR from input covariances
@PANCOINT	Panel data unit root/cointegration testing procedure (Pedroni tests)
@PANELDOLS	Panel data group mean DOLS
@PANELFM	Panel data group mean FMOLS
@PANELTHRESH	Does analysis of up to two threshold breaks in a fixed effects panel model
@PDL	Polynomial Distributed Lags regression
@PDLREG	Polynomial Distributed Lags regression
@PERRONBREAKS	Compute various unit root tests with breaks
@PERRONNGMTESTS	Compute various Perron-Ng “M” unit root tests
@PERRONRODRIGUEZ	Perron-Rodriguez unit root test allowing for break at unknown date
@PERSIST	Sum of coefficients of a MA representation for a series
@PHILLIPSHANNAN	Phillips-Hannan Efficient estimator for multivariate regressions
@POLYMULT	Multiplies lag polynomial coefficients. Obsolete: use %POLYMULT function instead.
@POTEST	Phillips-Ouliaris-Hansen test for Cointegration
@POTESTRESIDS	Phillips-Ouliaris-Hansen test for Cointegration on 1st stage residuals
@PPUNIT	Phillips-Perron Unit Root test
@PRINCOMP	Extracts principal components
@PRINFACTORS	Principal components-based factor analysis
@PRJCONDITIONAL	Predicted probabilities for conditional logit model
@PRJMULTINOMIAL	Predicted probabilities for multinomial logit model
@PRJPOISSON	Predictions and marginal effects for Poisson count model
@QPLOT	Graphs a Q plot against a hypothesized distribution
@QUARTIMAX	Does factor rotation using quartimax criterion
%RANGRID	Random draw from a distribution approximated across grid of points
@RANMIXTURE	Random draws from a mixture of Normals
@RANNORMALTRUNC	Random draws from a truncated Normal (procedure). Obsolete: use %RANTRUNCATE function instead.
@REGACTFIT	Regression post-processing, fancy graph of actual/fitted/resids
@REGANOVA	Displays an analysis of variance table from most recent regression
@REGCONFIDENCE	Displays a table of confidence intervals from most recent regression
@REGCORRS	Computes and graphs autocorrelations from residuals
@REGCRITS	Computes information criteria for most recent regression
@REGEXACTDW	Computes the exact significance level for the Durbin-Watson
@REGHBREAK	Performs structural break test with bootstrapped p-values
@REGPARTCORR	Computes partial correlations between the regressors and dep var.
@REGPCSE	Panel-corrected standard error calculation
@REGRECURSIVE	Regression post-processing, computes recursive resids, does tests (largely obsolete—use RLS instruction instead).
@REGRESET	Performs Ramsey RESET test on regression
@REGTOTEX	Creates a TeX equation from the most recent regression
@REGTREE	Performs a CART (Classification and Regression Trees) analysis
@REGWHITENNTEST	Performs White neural network test on regression
@REGWHITETEST	Performs White heteroscedasticity test on regression
@REGWUTEST	Performs Wu (or Durbin-Wu-Hausman) specification test on regression
@RGSE	Semiparametric fractional differencing parameter estimation
@ROBUSTLMTEST	Robust LM test for orthogonality of residuals and input series
@ROLLREG	Computes rolling regressions for least squares
@RRGQTEST	Computes a Goldfeld-Quandt test on recursive residuals
@RSSTATISTIC	R/S Statistic (classical or Lo’s modified)
@RUNTEST	Computes a run test for a two-state series
@SEASONALDLM	Creates the matrices for the seasonal component of a DLM
@SHORTANDLONG	Factor covariance matrix with short and long run restrictions
@SPECFORE	Forecasting using spectral techniques
@SPECTRUM	Computes/graphs spectral density
@SPLOM	Produces an NxN matrix of SCATTER plots
@SPUNIT	Computes various “Schmidt-Phillips” tests (TAU) for a unit roots
@SSMSPECTRUM	Multivariate spectral density of a state space model
@STABTEST	Perfoms Hansen’s stability test for OLS
@STAMPDIAGS	Performs a standard battery of specification tests for a state space model
@STARTEST	Test for linearity vs. LSTAR or ESTAR
@STEPPROBIT	Backwards stepwise reduction of a probit model
@STOCKWAT	Stock-Watson and Dickey-Fuller Unit Root Tests
@STRUCTRESIDS	Computes structural residuals from standard residuals
@SURGIBBSSETUP	Sets up Gibbs sampler for SUR model
@SWAMY	Computes a GLS matrix weighted estimator for a panel data set
@SWDOLS	Estimates cointegrating vectors using dynamic OLS
@SWTRENDS	Tests cointegration rank using common trends analysis
@TAR	Estimates a threshold autoregression, tests for threshold effect
@THRESHTEST	Hansen’s Test for Threshold Break
@TLOOKUP	Provides a procedure for doing table lookups
@TSAYNLTEST	Tsay test for neglected non-linearities
@TSAYTEST	Tsay arranged regression test for threshold autoregression (TAR)
@TSECCTEST	Tse test for constant correlation in MV-GARCH model
@TVARSET	Time-varying VAR setup routine
@UFOREERRORS	Forecast errors for a univariate model
@UHLIGFUNCS	Computes criteria for Uhlig sign-restricted shocks
%UNIFORMPARMS	Computes required parameters for uniform distribution
@UNIQUEVALUES	Extracts unique values from a series
@VARBOOTSETUP	Sets up a parallel system for bootstrapping a VAR
@VARCALC	Does a direct calculation of a simple OLS VAR
@VARFPE	Minimum FPE representation for the equations in a VAR
@VARFROMDLM	Computes a state space representation to its implied VAR
@VARIMAX	Does factor rotation using varimax criterion
@VARIRF	Organizes graphs of Impulse responses for an estimated VAR
@VARIRFDELTA	Computes the covariance matrix of an IRF using the delta method
@VARLAGMD	Computes the sums of the VAR lag coefficients (obsolete, use %MODELLAGSUMS function)
@VARLAGSELECT	Selects lag length for a VAR model
@VARMADLM	Routines for analyzing a VARMA using DLM
@VARSPECTRUM	Multivariate spectral density of a Vector Autoregression
@VRATIO	Variance ratio random walk test procedure
@WESTCHOTEST	Heteroscedasticity-robust serial correlation test
@WHITTLETEST	Implements Whittle test for independence of state sequences
@YULELAGS	Computes Information Criteria for AR models using Yule-Walker (obsolete: use @ARAutoLags instead).
@YULEVAR	Estimates a VAR on stationary data using Yule-Walker Equations
@ZIVOT	Zivot-Andrews Unit Root Test