Almost every dissertation should include some theory. Whether this deserves its own section is a matter of preference, and depends on the topic: it is perfectly possible to introduce the models that you are referring to in your literature review, for example.
Theory is useful for two reasons:
- it can be used directly to analyse a problem
- it can be used to establish testable hypotheses (which can be addressed in an empirical section)
What do we mean by theory? Well, it is unlikely that examiners will be impressed simply by words: some formal analysis, either diagrams or (more probably) mathematical analysis will be needed. Of course, what theory you introduce depends very much on the topic.
Option 1: Here’s one I made earlier
Using a model developed elsewhere (either in a textbook, in a lecture or from an academic paper) can be appropriate and a time-saver. Of course, it goes without saying that you should reference where it came from carefully, signalling what you have taken and what is original work.
If you use a model developed elsewhere (as most students do) you need to do something interesting with it: use it to show a particular result, or to prove a particular relationship. Of course, what you do will depend on your topic – but it is rare that a verbatim copy of a model is exactly appropriate, and it hardly demonstrates to your examiner what you are doing.
It is important to be critical of any model that you use. What are its assumptions, and are they plausible? What happens to the relevant results if the assumptions vary? Does a model developed for a certain situation (a particular industry, for instance) apply in the situation (another industry) you are concerned with?
Option 2: Do it yourself
Your other option is developing a model yourself. This is not easy and can be time-consuming, and should not be entered into lightly. However, if done correctly it can be very impressive.
Setting up a model is only useful if it will address the question you are interested in. Most economic models start with some simple expressions (a utility function, for example) and analyse what happens when agents optimise subject to different constraints. They can also analyse the interaction between agents.
In general, if you don’t have a clear idea about what you want to do, developing a model might be time-consuming. Discuss any model that you plan on making with your supervisor – they might have a good idea of whether it is feasible or not.