Regression Analysis of Time Series

From Wikipedia, the free encyclopedia

RATS
Developed by	Estima
Latest release	7.0 / 2007
OS	Cross-platform
Genre	econometrics software
License	Proprietary
Website	RATS

RATS, an abbreviation of Regression Analysis of Time Series is a statistical package for time series analysis and econometrics.

History:

The forerunner of RATS was a FORTRAN program called SPECTRE, written by Christopher Sims, a professor of economics. SPECTRE was designed to overcome some limitations of existing software that affected Sims' research in the 1970s, by providing spectral analysis and also the ability to run long unrestricted distributed lags. The program was then expanded by Tom Doan, then of the Federal Reserve Bank of Minneapolis, who went on to found the consulting firm that owns and distributes RATS software. In its early incarnations, RATS was designed primarily for time series analysis, but as it evolved, it acquired other capabilities. With the advent of personal computers in 1984, RATS went from being a specialty mainframe program to an econometrics package sold to a much broader market.

Features:

RATS is a powerful program, which can perform a range of econometric and statistical operations. The following is a list of the major procedures in econometrics and time series analysis that can be implemented in RATS:

Linear regression, including stepwise. Regressions with heteroscedasticity and serial-correlation correction.
Non-linear least squares.
Two-stage least squares, three-stage least squares, and seemingly unrelated regressions.
Non-linear systems estimation.
Generalized Method of Moments.
Maximum likelihood estimation.
Simultaneous equation systems, large econometric models.
ARIMA (autoregressive, integrated moving average) and transfer function models.
Spectral analysis.
Kalman filter and State Space models.
Neural networks.
Regressions with discrete dependent variables, such as logistic regressions.
ARCH and GARCH models.
Vector autoregressions.

All these methods can be used in order to forecast, as well as to conduct data analysis. In addition, RATS can handle cross-sectional and panel data.

RATS can import data in any number of ways. It has menu-driven data wizards for reading in data, but more typically, data will be imported from other sources using a command line. For instance, RATS can read data from Excel files or flat text files, as well as other formats. It can handle virtually any data frequency, including daily, weekly, intra-day, and panel data.

RATS has extensive graphics capabilities. It can generate high-resolution time series graphs, high-resolution X-Y scatter plots, dual-scale graphs, and can export graphs to many formats, including PostScript and Windows Metafile.

Mode of Operation:

RATS can be run interactively, or in batch jobs. In the interactive mode, the programmer writes individual lines of text, which the program will then execute. In the batch mode, the programmer writes a complete program, which is then loaded into RATS and run as a single job. New users often prefer the interactive mode, while experienced users will often prefer to run batch jobs. After an interactive session, the code can be saved, and converted to a batch format. One advantage of RATS, as opposed to automated forecasting software, is that it is an actual programming language, which enables the user to design custom models, and change specifications.

Website:

The company that owns RATS software was incorporated as Estima, Inc., and is located in Evanston, IL. The company’s website is http://www.estima.com

Comparison with Other Software:

One advantage of the RATS program is that it is inexpensive, compared to larger programs such as SAS. RATS has many of the same capabilities as SAS in both time series analysis and other advanced statistical methods. . The two programs differ more in the details than in capabilities. SAS has routines for automated State Space estimation. RATS can be programmed to estimate State Space models, or regression models with time-varying coefficients. In this respect, RATS is actually more flexible. Similarly, SAS has an entire routine for estimating and forecasting with Unobserved Components Models. In RATS, estimation of this type would require extensive programming. Nevertheless, in general, the capabilities of RATS are comparable to SAS/ETS and SAS/STAT, but at a much lower price.