What Next

There is almost no limit to what you can do with RATS. It is only a question of learning how. You may need more information on the instructions we’ve already covered. Or, you may be interested in other types of analysis. With that in mind, we would like to offer some suggestions on how to find out what you need to know.

Yes, there are a lot of instructions. Don’t worry! RATS is a very powerful and general program, and you may only need to use a fraction of its capabilities. Some suggestions for learning about RATS:

•First, look through the sections on Graphics and Dealing with Data, two things you will likely be using almost any time you work with the program.

•Take a look at the Instructions section to get a feel for how it is organized and how details on each instruction are presented. Look up some instructions you’ve already seen, like LINREG. When you see something new in an example, use the Reference or the help to find out exactly what is going on.

•The Statistics and Algorithms pages list a wide variety of types of analysis, mainly grouped into categories, such as GARCH Models and Vector Autoregressions.

•If you want to use a search engine, it usually helps to include “RATS” as one of the search keywords. If there’s a RATS program that fits with the rest of the description, that will usually be at the top of the search list.

•There are two main lists of examples on the help. “Examples” are the main set that we use to demonstrate techniques. “Paper Replications” are more specifically for replications of specific papers.

However, we can’t emphasize enough that even if you find an example program that does exactly what you (think you) want, do not skip over the most important step which is to Check Your Data! In empirical work it’s often the case that it takes longer to do something quickly than to do something carefully. We’ve had users waste countless hours trying to fit and interpret existing models adapted to their data set. The changes to the program were fine—it was the data set that was wrong. Make sure that you run the original program with the original data, and see whether your data have at least (somewhat) similar properties to the original.