Department of Economics

Economics 143 (Cameron) - Applied Regression Analysis


DUE: Tuesday, November 24, 1998


One of the best ways to "make real" a course in quantitative methods is to practice thinking about how you could put the skills you are learning to use--in a context that actually interests you personally. Since there is no time in Economics 143 to require the execution of an entire term paper, we will ask you to do just part of the task. Sometimes this is the hardest (although not the most time-consuming) part of a research project.

For this assignment, you will assemble a research PROPOSAL. For ideas, you might resort to the subject matter of some of the field courses you have taken in economics. Or, you may draw on your own work experience. I would prefer an Economics topic, and I myself am most interested in energy/environmental economics topics. However, anything related to economic current events would certainly be admissible (check a newspaper or the Daily Bruin), or even a completely original off-the-wall topic.

To keep you from getting carried away, and to encourage succinctness, the text of your submission should be no longer than roughly 1000 words (the equivalent of about 4 double-spaced, typewritten pages). Your proposal should be prepared in paragraph form and should be arranged roughly into the following sub-sections:

On the Cover Page:

Informative and to the point. Avoid gimmicky titles until you are famous.

Never propose a research idea without staking a claim to the intellectual property rights!

This single paragraph is the "advertisement" for your proposed project. It is a challenge, but you should be able to condense your proposal into a maximum of 150 words that summarize the gist of your proposal. Upon reading the abstract, people should know what your project will try to do, and how you will be doing it. The audience for this abstract should be considered to be other economists with at least your level of familiarity with economics and econometrics.
On four subsequent pages:
What is the main question that your econometric model will be especially designed to answer? Who will be interested in the answer, and why (i.e. what is the constituency for this research)? However, beware of the appearance of advocacy. Do not let it appear that the results of your study are a foregone conclusion in your mind. While you may have very strong suspicions that the world works in a particular way, do not convey the impression that you merely are out to prove your opinion. Instead of saying "this research will show that A has a negative effect on B," use language such as "this research will determine1 the effect of A on B." In motivating the reader's interest in your proposal, you can certainly point out that "Leading advocates in this area (or some other group) believe that A has a negative effect on B. Thus, quantifying this relationship will be important to resolving this important controversy."

What is your dependent variable (be very specific)? What sort of data will you require? (Be sure to list some of the variables, and be ABSOLUTELY SURE to indicate what would constitute an observation. Remember that ALL of your variables should be available for each of your units of observation.) Do you imagine that these data are readily available? Or, could the necessary data conceivably be assembled? Are there any variables which would have to be proxied, and by what? (The answers to data availability questions can sometimes be obtained from the Public Affairs division at URL, or from the ISSR Data Archive, if you really get serious.)
Presuming that the data you need are in fact available, what sort of econometric model would you explore first? Is there any economic theory to guide your specifications? What are your expectations about the signs of the coefficients, ex ante? What would be the interpretations of the more interesting coefficients? What specific hypotheses would you plan to test? If any important variables are fundamentally unmeasurable, how might they be correlated with things that are included in your model, and how might this affect the results? Be sure to offer a prototype specification explicitly (in terms of:
with the DEPVAR and VAR1, etc. variables tailored to the acronyms you propose for the relevant variables in your own study. For example,
You specification may be a simple linear one, like that shown, or something more complicated, with quadratic terms or interactions or log-log form or anything else that you think might be appropriate to your study. Use informative names for your variables, rather than just X1, X2,... or VAR1, VAR2,....
One thing to keep in mind: be sure that your explanatory and dependent variables actually vary across the observations in your data set. Any explanatory variable that is the same for every observation will be absorbed by the intercept term. You won't be able to estimate a slope on that variable.
Once your model is calibrated using the actual data, how could it be used to predict what will happen, or what might have happened, if circumstances are (or were) different. Are there any socially or politically interesting scenarios your model could be used to simulate? Are there "policy" implications to be drawn from your calibrated model? Researchers have to be aware that simply estimating a model (while it may be elegant) is sometimes not very useful. Using the estimated model as a tool for policy evaluation makes the exercise worth something to society.
If everything works out as planned, what do we stand to learn from your proposed project that we didn't know before? What will your model be able to do for society? One would hope that these new insights are worth as much to society as the opportunity cost of doing the research. Can you persuade me (the "granting agency", or "your boss") that I should "fund" your proposed research project?
On a separate final page:
If applicable, the full citations for any books or journal, magazine, or newspaper articles you referred to in coming up with your idea. (In the text of your proposal, it is standard to refer to these citations simply as Author (date), e.g. Smith (1992), or Smith (1992, p. 27).) Your proposed research is more likely to be funded (approved, sponsored, graded favorably,...) if there is evidence that you have substantial familiarity with work that has already been done in this and related area. Evaluators want to be assured that you have some existing "expertise" in the area behind a research proposal.

The body of the proposal can be arranged into sub-sections with headings similar to those indicated above, but the prose should flow smoothly. Each section should consist of a few paragraphs which, collectively, address at least the questions noted in these guidelines. Remember, a paragraph that runs on for more than half a page is probably too long. Each paragraph should cover one idea or issue.

Most students who make it as far as Economics 143 know how to prepare a "research document," but just in case, here are a couple of reminders: Avoid cute plastic binders, colored paper or other disguises for your proposal. No extra points will be awarded for fancy fonts. The content of your proposal is all that matters. Its presentation should be professional. The abstract should be single-spaced, and the references may be single-spaced if there are a lot of them. The body of the proposal must be double-spaced (not space-and-a-half). Fasten with a single staple in the upper left hand corner. SUBMIT TWO COPIES; RETAIN A BACKUP COPY IN YOUR FILES.
TONE (again)
Academic research should be objective, not subjective. You are not out to convince your audience that something is true. Instead, you are examining the facts. It may turn out that some worthy cause or policy is supported by your research, or perhaps it is not. Try to convert any passion for your topic into meticulous attention to what might go wrong with your specification (e.g. omitted variables bias) that could cause your data set to imply either appealing or disappointing results (from your point of view). But do not show your hand. Any reader of your research proposal should not be able to guess which "side" of an issue you are on.
In economics, obviously, lots of empirical studies on all sorts of different topics have already been done. If you truly cannot dream up an idea on your own, or if you are wondering if your proposed economic model has already been thoroughly researched, you can easily explore the existing literature. Perhaps a model similar to someone else's could be tested on a different set of data, or the model could be elaborated or changed in some way. If you choose to take this route, identify clearly the citation(s) for the source of your idea and proceed.
One of the greatest additions to the economist's tool-kit has been the "EconLit" CD-ROM database containing information concerning published research in the field. This lets you search a wide range of economics articles in journals over the last 20 years for papers on the subject(s) that you list. Try it out. It is no harder to use than ORION at the library, and it is a amazing boon to research. If you overlook crucial citations in the literature, it can doom your research proposal's chances for funding. From UCLA, EconLit can be most easily accessed via the UCLA Library Web Site. You may also want to use MELVYL and check the Current Contents database for interesting titles.
  1. Economics 147A,B usually requires a term project. Perhaps your efforts on this problem set will lead to a viable topic for your 147A project.
  2. If you find yourself becoming truly intrigued by a serious empirical research proposal, and you will not be taking 147A next quarter, you are invited to talk over the prospect of doing a 199 project with me during the winter quarter. I can only manage two or three 199s, but if I am also interested in your topic and it appears to be tractable, we should talk it over.
Your grade pertaining to the research proposal will consist of two parts. First, you will receive a grade out of 20 on your own proposal. Second, you will receive a grade out of 5 for your brief "referee report" on someone else's proposal. Being able to critically evaluate the proposed (or completed) econometric work of someone else is a necessary skill. This will be your first attempt.

Updated: 8:47 AM 10/6/98
Prepared by: Trudy Ann Cameron