This page lists most of the changes made to RATS from Version 4.0 through 5.11.

See RATS: Latest Release for details on the current release, RATS Updates for information about updating your RATS software or RATS Information and Features for general information on RATS.

Version 5.11

Improvements made in 5.11 included:

  • Many new functions, including tools for manipulating equations, regressor lists, instrument lists, and parameter sets, plus functions for handling polynomials and some additional string operations.

  • A FIXED command which provides an easy way to define fixed-value variables and arrays in PROCEDURES

  • An NPREG instruction for non-parametric regressions

  • An RREG instruction for "robust" regressions, including LAD and MAD

  • A GSET command for setting the values of series of arrays

  • Improvements to DLM allowing for diffuse initial states

  • An RLS command for doing recursive least squares

  • The ability to call other executable programs from within WinRATS (this capability was already present in the UNIX version)

  • A DBOX command for creating user-defined dialog boxes.

  • New options/features on more than a dozen existing instructions

  • Several bug fixes

See the October, 2003 RATSletter for additional information.

Version 5.10

Version 5.10 was a major upgrade, offering significant advances both in technical capabilities and in user-friendliness. For full details, please see the July, 2003 RATSletter.

Here are the highlights of 5.10:

  • Callable Functions: You can now create your own user-defined functions. These are similar to RATS procedures, but unlike procedures, they can be called in expressions, such as in a FRML statement. This represents the single most important step in expanding the scope of optimization problems that RATS can handle since the introduction of the MAXIMIZE command years ago.
  • Wizards! New menu- and dialog-box-driven wizards greatly simplify many common tasks, including specifying a date scheme and reading in data, generating basic statistics, and performing univariate regressions and testing restrictions. We've also incorporated much of the functionality of RATSDATA into RATS.
  • Report generator: The new REPORT command can simplify and fool-proof the process of extracting numerical results into reports for publication. This can be a huge time-saver for users publishing results.
  • More Monte Carlo/Bootstrapping Support: New functions and expanded documentation (in PDF format) extend the capabilities for doing simulations and bootstrapping
  • GMM additions: We've added additional options for lag window estimators for handling serial correlation in covariance matrix calculations and GMM weight matrix computations

Version 5.04

Version 5.04 added a "Function Wizard", accessed via a toolbar icon, that simplifies the process of identifying and using the more than 200 functions included in RATS. It also allows you to define a "Procedure Library" that is loaded automatically at startup or when you clear the memory. The GRTEXT instruction has been significantly updated to support including line breaks in text, automatically positioning text in any corner of the graph, and more. The GRAPH instruction also gained a new KEY option, and several new functions were added.

Finally, RATS now displays an error "trace back" indicating the line number of any error that occurs in a compiled section—this is particularly helpful for users doing extensive programming with procedures or loops.

See the April, 2003 RATSLetter for further details.

Version 5.03

Version 5.03 was a fairly significant update, providing a number of improvements and new features. Full details are provided in the November, 2002 RATSLetter. We've summarized the changes below:

The most significant changes in 5.03 were some major improvements to the non-linear estimation algorithms. The program is now better able to deal with especially "difficult" situations, reports more information when you use the TRACE option or when it fails to reach convergence, and also does a better job of determining convergence or non-convergence in challenging cases.

Version 5.03 also added several improvements to the interactive mode interface, including new toolbar buttons for setting the "Input" and "Output" window flats, and for clearing the memory. The status bar at the bottom of the screen now displays the current line number and column number of the cursor position, and other information.

The RATS language also gained several useful new options: A new option for computing optimal GMM weights when doing instrumental variables with LINREG and NLLS; an option for saving the derivatives computed by MAXIMIZE; an option on CVMODEL for determining the precise form of the maximand; and an option on all instructions supporting BFGS estimation that allows the user to input an initial estimate for the inverse of the Hessian matrix.

Finally, the second printing of the RATS 5 User's Guide and Reference Manuals appeared with the debut of RATS 5.03, and included details on the new features added in 5.02 and 5.03. The PDF versions of the manuals were similarly updated.

Version 5.02

This was a minor update, available on all platforms. It included several bug fixes, as well as several useful new features, including an option for adding additional (non-ARIMA) regressors to a BOXJENK estimation, an option on DLM allowing for exogenous shifts in the state equation, and several new functions. See the July, 2002 RATSLetter for details.

Version 5.01

For the Windows and DOS platforms, Version 5.01 was a bug-fix update to 5.0, with no new functionality. See the list of Known Bugs for more information. For Macintosh and UNIX platforms, Version 5.01 was the first release of Version 5.0 available for those platforms (the Mac and UNIX versions of 5.01 included all the bug fixes introduced in the PC 5.01 versions).

Version 5.0

The release of Version 5 in 2000 marked the most significant upgrade to the program in eight years. Since that time, we have released several incremental upgrades offering a variety of improvements and new features. For more details on RATS 5.0, please see the September 2000 edition of the RATSletter.

New Instructions

RATS Version 5 offers five new instructions:

  • CVMODEL estimates structural VAR?s.
  • ECT adds error correction effects to a VAR.
  • DLM works with state space models, employing Kalman filtering and smoothing.
  • PFORM creates panel series from other sources
  • PREGRESS performs fixed and random effects estimation.
  • DENSITY computes empirical density functions.

New Functions

We've added more than fifty new functions, including new probability and distribution functions, matrix concatenation/extraction functions, tools for creating series on the fly, and more.

Non-Linear Estimation

Improvements include:

  • the ability to use arrays as non-linear parameters.
  • a new PARMSET variable type that makes it easy to combine several sets of non-linear parameters, allowing models to be created in pieces and combined only when estimated.
  • the ability to impose equality and inequality constraints.
  • a new "genetic" estimation method which allows a much broader search of the parameter space.
  • the ability to define vectors of formulas, including the option of creating them inside a loop.

Vector Autoregressions and Systems of Equations

In addition to the new CVMODEL and ECT instructions, several "convenience" features have been added. For example, it is now much easier to save VAR residuals, forecasts, and impulses responses, using the new "RESULTS" option. Also, the SYSTEM instruction now includes a MODEL option that allows you to define a VAR as a model, which makes setting up instructions like IMPULSE and FORECAST much easier.

Probability and Random Numbers

New functions have been added for computing probability distributions and their inverses, as well as factorials and binomial coefficients. Additional documentation covers techniques like Gibbs sampling and approximate randomization in detail.

Revised Documentation

The RATS documentation has been reformatted into two standard 7"x9" books, and will be included in electronic form (Adobe PDF files) with each copy of RATS. The manual features more examples, more technical details, and offers expanded/new coverage of many topics, including Structural (identified, Bernanke-Sims) VAR?s, Blanchard-Quah decomposition, ARCH and GARCH models, hazard models, state-space models, simulations and bootstrapping, and non-linear estimation.

Changes to Existing Instructions

See RATS 5 Instruction Details for a detailed instruction-by-instruction listing of new options, features, and other improvements included in Version 5.

Version 4.35 (March, 1999)

Version 4.35 is a Macintosh-only release, which introduced a native PowerMac version of RATS, and incorporated all of the changes and new features introduced with versions 4.1, 4.2, and 4.3 on other platforms, as described below.

Version 4.30 (March, 1997)

Version 4.3 added the following new features:

  • New instructions for building and testing neural net models.
  • Built-in instruction for solving linear programming and quadratic programming problems.
  • New function for solving systems of linear equations.
  • For UNIX systems, it added the ability to view graphs directly on the screen (already available for desktop versions).

Version 4.20 (April, 1995)

Version 4.2 was a major update that included the introduction of the first versions of RATS written for the Windows operating system. WinRATS is available in both a 16-bit version and a faster 32-bit (Win32s) version. Other improvements in RATS 4.2 included:

  • Several new instructions for writing interactive programs, including USERMENU for user-defined pull-down menus, INFOBOX and MESSAGEBOX for displaying informational dialog boxes, and SELECT for displaying pop-up selection lists.
  • Many enhancements to the graphics capabilities, including:
    • New GRTEXT instruction for placing text anywhere inside a graph.
    • Simplified handling of two-scale graphs.
    • Additional labeling options for SPGRAPH's.
    • The ability to export graphs to formats such as PostScript, PIC, and HPGL directly from RATS.
  • Several new functions:
    • %INVNORMAL and %INVCHISQR for inverse Normal and inverse Chi-square distributions.
    • %MILLS and %DMILLS for computing Mills' ratio and its derivative.
    • %XDIAG, %XROW, and %XCOL for extracting the diagonal or a specific row or column from an array.
    • %LEFT, %RIGHT, and %MID extract characters from LABEL variables.
  • Support for reading directly from Excel files and the DRI Basic Economics (formerly Citibase) database.

Version 4.10 (March, 1994)

Improvements in Version 4.10 included:

  • Several enhancements to the interface of the DOS versions, including mouse support, the ability to keep several graphs in memory at one time, the option of "buffering" output to improve performance, and more.
  • Added additional data compaction options and support for reading/writing data in the DIF and DBF formats.
  • Two functions: %LOGDENSITY(A,B), which returns -.5(log|A|+B'inv(A)B), and %RANGAMMA(x) which returns a random draw from a gamma distribution with shape parameter x.

Version 4.0 (April, 1992)

RATS Version 4.0 represented the most comprehensive update in the history of RATS. The program was entirely rewritten in the C programming language, many improvements were made, new features were added, and the RATS User's Manual benefited from extensive revisions. The following is a partial list of the changes in Version 4.0:

  • Significant speed improvements over Version 3.1.
  • New statistical features, including:
    • Non-linear systems estimation.
    • BOXJENK supports arbitrary lag structures (e.g., you can include terms for lags 1 and 3 without including a term for lag 2).
    • The instruction FIND uses the simplex method to maximize/minimize any expression computed by any sequence of RATS instructions.
    • Simplex method added to MAXIMIZE.
    • MAXIMIZE can also handle a much wider range of models, including multivariate likelihood models.
  • The RATS editor allows you to work with several "windows" at one time. For example, you can enter instructions in one window and have the output sent to a different window.
  • Much more helpful error messages.
  • Improved formatting of output.
  • Simplified handling of scalar and matrix calculations
  • Simplified system of data types makes programming easier, and allows "compound" data types, such as vectors of series.
  • Variables names can now be up to 16 characters in length, rather than 8 characters.
  • Scatter plots done as high-resolution graphics rather than character-based plots.
  • Added support for C-style comment blocks using /* and */ notation.
  • CALENDAR instruction allows for much wider range of frequencies, including multiple observations per day.