The following provides answers to some frequently asked questions regarding the CATS (Cointegration Analysis of Time Series) procedures.
FAQ updated on: September 25, 1998
Q1. Is CATS a stand-alone application?
Q3. When do I use the BATCH option?
Q4. What are the SC and HQ criteria produced by the residual analysis operation (p. 26)?
A1. No, CATS is a sophisticated set of procedures written in the RATS language. You use CATS by running a set up program in RATS which reads in your data and compiles and executes the procedure. CATS requires version 4.20 or later of RATS.
A2. This is indeed a bit confusing. At this point, CATS is actually waiting for you to type in the word "RETURN" in the input window, and then execute this line as a command (by hitting the Enter key or clicking on the "Run" icon). CATS is actually using the CNTRL(USER) command to temporarily give control back to the user, which allows you to scroll through the list model choices before having CATS do the tests. After you execute the RETURN command, CATS will prompt you for the number (1 through 5) of the first model you want to test. After you hit "OK", CATS will prompt you for the last model you want to test (again, 1 through 5). If, for example, you enter 2 and 5, respectively, CATS will then go on to test models 2, 3, 4, and 5.
A3. You only need to use the BATCH option if you are running CATS under RATS for UNIX, or if you are running it under RATS386 in batch mode--that is, if you start RATS and execute your setup program by typing something like: "RATS386 UK.CAT" at the DOS prompt. If you are using RATS386 in interactive mode, or using RATS for Windows, you should not use the BATCH option in your @CATS command.
A4. These are the Schwarz and Hannan-Quinn information criteria, respectively. Basic formulas are given on pages 72 and 73 of the CATS manual. See Information Criteria for some additional details provided by Henrik Hansen, one of the authors of CATS.
A5. The explanation of this in the manual isn't quite as clear as it could be, so we've provided the following table, which condenses information from equations 2.8 and 2.9 on page 5 of the CATS manual, the discussion of "Cases" 1 through 5 on pages 5-6, the available settings for the DETREND option listed on pages 13-14, and the critical-values tables from Appendix B. We hope this chart will help you choose a DETREND setting and refer you to the correct critical value table for the selected model.
| Case | Restriction(s) Imposed | DETREND Option | Appendix B Table |
|---|---|---|---|
| 1 | None | NA (not supported in CATS) | 5 |
| 2 | d2 = 0 (no quadradric trends in data) | DETREND=CIDRIFT | 4 |
| 3 | d1 = 0, d2 = 0 (above plus no trend in CR) | DETREND=DRIFT (default setting) | 3 |
| 4 | d1 = 0, d2 = 0, u2 = 0 (above plus no trend in data) | DETREND=CIMEAN | 2 |
| 5 | d1 = 0, d2 = 0, u1 = 0, u2 = 0 (no trends, intercepts) | DETREND=NONE | 1 |
Notes:
These terms refer to the mu and delta terms described in equation 2.8, 2.9 on page 5 of the CATS manual: