Economics 314 Research Assignment
J.R. Carter
Fall 2003
Your assignment is to undertake an econometric study based on single-equation multiple regression analysis. The study will consist of selecting a research question with one or more testable hypotheses, formulating an econometric model, collecting data, estimating the model using the EViews software, conducting appropriate tests, interpreting the results, and communicating the study by means of a carefully written research paper.
For background preparation, you might want to skim Jeffrey M. Wooldridge, Introductory Econometrics (South-Western College Publishing, 2000), Chap. 19, "Carrying Out an Empirical Project," a photocopy of which has been placed on reserve. The chapter provides useful suggestions about conducting an econometrics project and includes some possible topics. Also browse through Chap. 19 of your text by Hill et al. However, remember that the requirements for your paper are set forth here, not in Woolridge or your text.
Your first task is to select a suitable research question. This is the most important and perhaps difficult hurdle ahead of you. To structure the assignment, and also to keep you out of trouble, you are required to link your project explicitly to at least one econometric study already published in a scholarly source. Hence, selecting a topic entails locating one or more published articles upon which you will base your study. You can do this in one of two ways, roughly speaking:
(1) Choose a general topic of interest to you, and then search for a suitable article on that topic. You might already have a general topic in mind. If not, you might look through your current and past economics texts. Once you have a general topic, use the periodicals indexes and computer searches in Dinand to narrow your focus. The Social Science Citation Index is particularly good if you know how to use it. Don't hesitate to ask a reference librarian for help. From the indexes or searches, compile a list of relevant articles in scholarly journals (not popular magazines); locate the articles; identify and skim through those that involve econometric analysis and look promising; pay attention to the bibliographies for leads on earlier published work; and hope that something grabs you.
(2) Alternatively, don't worry about a general topic; just search directly for a suitable article. This approach is less systematic, but it can work well because for most of you it will be the process rather than the specific topic that dominates your interest and motivation in this assignment. Begin by browsing through scholarly journals in the current periodicals room and downstairs in the stacks. Skim table of contents; flip pages of articles looking for tables of regression results. Again, follow leads until you find an article that will work. Also, you might be able to abbreviate this search process by noting articles cited in your textbooks or in the Wooldridge chapter.
Some periodicals that have been used for this assignment in
previous years include American
Economist, American Journal of Economics and Sociology, American Political
Science Review, Economic Journal, Journal of Economics, Journal of Labor
Research, Journal of Money, Credit, and Banking, Public Choice, Quarterly
Review of Economics and Business, Review of Social Economy, and Southern Economic Journal. Others that might be particularly suited
include Applied Economics, Atlantic
Economic Journal, Eastern Economic Journal, Journal of Economics and Business, and Quarterly Journal of Business and
Economics. Most of these are what
would be called middle-level journals.
Articles in top journals, like American
Economic Review, Review of Economics and Statistics, Quarterly Journal of
Economics, and Journal of Political Economy, are usually (but not always)
too advanced to be helpful in this assignment.
When you begin to focus on one or several articles, keep the
following considerations in mind.
Econometric studies are carried out at all levels of sophistication and
complexity and are based on a wide array of data sources. Be realistic.
Look for studies that are within your reach in terms of the modeling and
econometric analysis. Older studies tend
to be simpler and hence more suitable.
Therefore, don't hesitate to look at articles published during the 1970s
and early 1980s. Remember that the
assignment calls for a single-equation model as distinct from a
simultaneous-equations model. This means
that your selected article should use ordinary least squares, not maximum
likelihood, two-stage least squares, or instrumental variables estimation. Concerning data, you will need at least forty
to fifty observations in order to use the various large sample properties and
methods introduced in the course. Give
strong preference to cross-sectional data (e.g., across 50
There may actually be several key articles, but suppose for sake of discussion that you have identified one article that will guide your research. At this point your simplest option is to undertake what is called a replication study: attempt to use the same model and data and see if the published results replicate. A more challenging option, and hence one likely to earn a higher grade, is to try to improve the previous research by taking it a step or two farther. Think about possibilities like: test the model on a larger or different cross-section or time series; conduct F-tests for structural stability; use a different functional form; allow for an important omitted variable or a better measure of a key variable; test and correct for heteroskedasticity or autocorrelation; etc. Again, be realistic. Resist the temptation to try to do everything. What you do, do well. In footnotes or the conclusion of your paper, you can always discuss what other improvements could be pursued in future research.
Next do the econometrics. If you have done your work in the course, this should be the easy part of the project, although granted it can be time consuming if you are using a lot of data. When you get your results, think about what they mean. Look at the signs and magnitudes of the estimated parameters; pay attention to and use correctly the relevant regression statistics. Correctly conduct your formal hypothesis test(s). Explicitly connect your regression results with your research question and hypotheses.
Present your research in a carefully written paper. Assume you are writing for your classmates who, like you, have already learned the basic methods and terminology of econometrics. Use language understandable to both you and your reader. If your topic involves specialized terminology, define the terms for your reader.
The paper should be typed and double spaced. I have no length requirements. Use your discretion. I will just remind you that a long paper is not necessarily a good paper. An excellent paper might be only eight or twelve pages of text plus endnotes, references, and tables. Keep the organization tight. Try to imitate the tone and organizational style of the published articles you have referenced. In the scholarly world, do as the scholars do. For example, break your paper down into sections. Most papers will be organized roughly as follows: introduction, including a clear statement of the research question, motivation, and perhaps review of previous research; model, variables, data sources; results and analysis; and conclusion.
You are required to submit the paper with the following format: title page, text, references, endnotes, and tables, plus in an appendix a copy of your data and final EVIEWS output. Note that you are required to present your results in tables which you construct. These are distinct from the EVIEWS printout in the appendix, which is there for me, not your imaginary reader. Also, you must use the author-and-year pattern of citing references. This means that citations are by last name and date of publication and are usually placed within the text rather than in endnotes. Endnotes are used for elaboration and discussion tangential to the flow of the text. The format for references must follow that found in the American Economic Review, a current issue of which can be found in the current periodicals room. Note again that this format is a requirement.
Your paper will be graded based on its strict conformity to these directions and on the quality of the writing and content. Follow directions. Refer back to these instructions frequently. I put a premium on good writing and professional appearence. I suggest that you use a reciprocal buddy system of proofreading. When you leave Holy Cross, you will find that if you don't show respect for your written product, no one else will. Plagiarism is forbidden (see "Academic Honesty," Holy Cross Catalog).
This assignment has two due dates. By no later than Tuesday, November 4, and preferably earlier, you must seek formal
approval of your topic from me. This
entails providing me with a one-page statement of your research question and
plan plus a photocopy of the published paper(s) upon which your project will be
based. If you have kept me informed all
along, no problem should arise here.
However, failure to seek approval will be penalized one grade level
(e.g., A, A minus, B plus, etc.) per week of lateness. If I do not approve your project, a second
proposal is due within one week of the first; failure to seek a second approval
will likewise be penalized for lateness.
Your completed paper is due no later than
You are at liberty to show me a preliminary draft of your paper. Assuming you give sufficient lead time before the due date, I will provide feedback on the paper's organization and content. You are also permitted to use the Writing Workshop for assistance.
You may use some or all of your paper to satisfy a requirement in another course or in an honors program. However, any deviation from the stipulations of this assignment requires prior approval from me.