RATS (Regression Analysis of Time Series) is a fast, efficient, and comprehensive time series analysis and econometrics software package. RATS makes simple tasks easy to accomplish, while its command-driven interface and extensive programmability also make it a very flexible and powerful tool for more complex jobs.

State Space Methods in RATS

This paper provides an introduction and overview to working with state space models in RATS.

RATS Standard 8.3

RATS provides an interactive editor that allows you to quickly experiment with different models or procedures, without having to repeat earlier steps each time, with the batch mode automatically able to read and execute instructions from one or more input files, and saves the output to disk.

New in version 8.3

With RATS version 8.3, several important new tools have been added to make you more productive. It is now using a public domain “programmer’s editor” called Scintilla. Now using a proprietary text editor since WinRATS was introduced, as the text editor supplied with Windows at the time was limited to 32K bytes, which could be completely filled with one set of impulse responses on a reasonably large model. Unfortunately, that editor was orphaned several years ago, and wasn’t portable to the Macintosh or unix.
Scintilla is portable to all three platforms and is specifically designed as a programmer’s editor, which is what is needed for RATS. One very useful feature is a “bracket matcher”, which shows the (apparent) matches for ( ) and { } pairs (showing the mates in green), to help fix parenthesis nesting errors. If one of these characters has no obvious partner, it is shown in red as you cursor over it. We’ve also added the ability to “mark” a location in a text file so you can easily return to it.
A big productivity aid is the new Find in Files operation. Among standard, textbook and paper replication examples, we have well over 1000 running examples. With Find in Files, you can search across these for particular text, like MV=DCC to find examples which do dcc garch models. Both Find in Files, and the standard Edit-Find operations also now support “regular expressions”, a standard set of codings for text patterns. For instance, a search pattern of sur(.*model= will locate all instances of a SUR instruction with the MODEL option, with .* matching any number of characters. A simple sur(model= will only find examples where MODEL is the first option listed.
RATS 8.3 now uses multiple “threads” with the editor interface in one and the statistical engine in another. This makes it possible for the editor part to still be responsive while calculations are going on. It also slightly improves performance, particularly on computers with multiple cores, since it can give over one entire core to the numerical calculations while running the interface (and anything else you’re doing) on another.
The 64-bit executable (on WinRATS Pro) now runs about 15-20% faster than the 32-bit on the same computer due to improvements in compiler technology. We still provide both executables, since some of the added data formats aren’t available for the 64-bit at this point. The install process for Win-
RATS Pro has been simplified, with the downloadable version now including both 32 and 64 bit.
The following changes have been made to existing instructions:

  • BOXJENK has a MEANEQ option to define the equation for the mean (only) in a RegARIMA model

  • SSTATS adds a START option similar to those on SET and MAXIMIZE. It also will now ignore entries which evaluate to NA so you don’t have to use a SMPL option to skip over them.

  • THEIL now has a FORECASTS option which allows you to input a set of forecasts generated by some method other than the standard techniques available, while still being able to let THEIL do the forecast error calculations.

  • The STACKED bar graph styles on GRAPH will now allow a mix of positive and negative values at a given time period, stacking away from zero in each direction.

  • The VTICKS and HTICKS options on the graphics instructions will now accept a VECTOR of values at which you want tick marks, allowing you to override the automatic choices made by RATS.

8.3 adds several new functions:

  • %MODELLARGESTROOT(model) returns the absolute value of the largest root of the companion matrix of a model. This is helpful in doing Monte Carlo simulations on a var, as you can easily reject draws that are unstable.

  • %STRESCAPE(s,chars,escape) returns a string with any character from chars “escaped” by prefixing with it the character or string escape.

  • %STRFIND(s,f) looks for the string f in s, returning the position if found and zero otherwise

  • %STRTRIM(s) returns a copy of a string with leading and trailing quotes and spaces removed

In addition, quite a few existing functions have been altered to allow them to apply to each element of a vector or matrix rather than just to single real values: COS, SIN, TAN, %PLUS and %MINUS. %VALID, when applied to a matrix will return 1 only if no element of the matrix is missing.

Statistical Methods

Estimation Techniques

  • Multiple regressions including stepwise

  • Regression with autoregressive errors

  • Heteroscedasticity/serial-correlation correction, including Newey-West

  • Non-linear least squares

  • Two-stage least squares for linear, non-linear, and autocorrelated models

  • Seemingly unrelated regressions and three-stage least squares

  • Non-linear systems estimation

  • Generalized Method of Moments

  • Maximum likelihood estimation

  • Constrained optimization

  • Extensive built-in hypothesis testing capabilities.

  • Pre-written procedures for a huge variety of other tests, including unit-root, stability, and much more

  • Limited and discrete dependent variable models: logit, probit, censored/truncated (Tobit), count models

  • Panel data support, including fixed and random effects estimators

  • Non-parametric regressions

  • Kernel density estimation

  • Robust estimation

  • Recursive least squares

  • State-space models, including Kalman filtering and smoothing, simulations, and optimal control models

  • Neural network models

  • Linear and quadratic programming

  • Dynamic Stochastic General Equilibrium (DSGE) models

Time Series Procedures

  • Easy to specify lags and leads for time-series model estimation and analysis

  • ARIMA and ARMAX models including multiplicative seasonal models; support for arbitrary lag structures

  • Transfer function/intervention models

  • Error correction models

  • Kalman filter

  • Spectral analysis


  • Time series models

  • Regression models

  • Exponential smoothing

  • Static or dynamic forecasts

  • Simultaneous equation models (unlimited number of equations)

  • Simulations with random or user-supplied shocks

  • Forecast performance statistics, including Theil U statistics

Vector Autoregressions (VARs)

  • Unmatched support for VAR models

  • Error Correction models

  • Structural VARs. Choice of factorizations, including estimating a factor matrix from a covariance matrix model

  • Impulse responses, with Monte Carlo and Importance Sampling techniques for standard error bands.

  • Forecasting

  • Variance decomposition

  • Historical decomposition

  • Extensive hypothesis testing tools

  • CATS 2.0 add-on provides industry-leading cointegration analysis

ARCH and GARCH Models

  • Univariate and multivariate, including BEKK, diagonal, CC, DCC, and Vech multivariate models

  • Support for GARCH-in-mean models

  • Additional exogenous variables in mean and/or variance equations

  • Normal, t and GED distributions

  • Exponential and Asymmetric models

  • Robust standard errors 

Working With Data

Data Entry

  • Menu-driven Data Wizards for reading in data

  • Reads and writes Excel® files (including Excel 2007), text files, Stata®, Eviews®, Matlab®, Haver databases, and other formats

  • Pro version supports SQL/ODBC access, online access to the FRED® database, CRSP® data, and more

  • On-screen data viewer and editor, with point-and-click graphing and statistics tools

  • Can handle virtually any data frequency, including daily, weekly, intra-day, and panel data

  • Can automatically compact or expand data to different frequencies

  • RATS data file format is fast and easy, supports all frequencies, and allows you to store series of different frequencies on the same file

Data Transformations

  • Flexible transformations with algebraic formulas

  • Easy to create trend series, seasonal, and time period dummies

  • Extensive filtering operations, including Hodrick-Prescott, Henderson, Spencer, and custom filters

  • Supports regular, seasonal, and fractional differencing


  • Time series graphics

  • X-Y scatter plots

  • Dual-scale graphs

  • Box plots

  • Contour graphs

  • Ability to arrange multiple graphs on a single page

  • Copy-and-paste graphs into other applications

  • Export graphs to many formats, including PostScript and Windows Metafile

  • User can customize attributes such as line thickness, colors and grayscale levels, and fill patterns 


Interactive Mode Environment

  • Text-editor based

  • Point-and-click “wizards” for many tasks, greatly enhancing ease-of-use

  • Saved programs can be re-run with just a few mouse clicks

  • Designed so that you can reproduce results, output, and graphs easily and accurately (a critical but often overlooked requirement for producing reliable, publication-quality results)

  • True multiple window support. Simultaneously view your input commands and output, spreadsheet-style “report” windows, graphs, and more


  • Extensive looping capabilities and support for applying operations to lists of variables make it possible to automate many repetitive tasks

  • You can write procedures, which can perform complex tasks with a single instruction, and write your own callable functions.

  • A library of procedures written by RATS users from around the world is available free of charge on our web site

  • A variety of interface-related instructions allow you to create your own drop-down menus, custom dialog boxes, and more

Report Capabilities

  • Strong focus on making it easy to get results easily and accurately into documents or other applications.

  • Tables of output can be viewed in Report Windws, for easy exporting or copying-and-pasting into other applications.

  • Powerful report-generation features for constructing and exporting your own tables of information.

  • Easy control over displayed precision in output

  • TeX: Support for exporting TeX tables 

RATS Professional 8.3

The Professional versions of RATS adds the following features not found in the Standard version:

  • Support for reading databases via ODBC/SQL

  • Census Bureau X12-ARIMA seasonal adjustment routine

  • Support for FAME data files (for Windows and unix/linux)

  • Support for CRSP data files

  • Online access to the FRED database

Back to Top