Running Your Own Regression Project
This course requires student or students to form a team (Max. two students)
to carry out an empirical project.
General guidelines
For more detail descriptions of running you own regression project,
Please read
Chapter 3 and Chapter 11 of "Using
Econometrics: A Practical Guide", by Studenmund (2006)
Pearson-Addison-Wesley,
pp.66-79, and pp.390-413.
1. Selecting a topic: You have studied many economic theories about the behavior of economic agents and the relationships among economic variables. Which economic phenomenon is related to your daily life? So ask yourself now which of the theoretical relationship you studied would be worth empirically estimating and which bodies of the theories can be put to empirical tests.
A systematic way to approach the problem of choosing a specific topic is to make effective use of a classification system adopted by the Journal of Economics Literature (JEL), which is a quarterly publication that presents a classified list of books and journal articles published in the preceding quarter. Check it in the library.
2. Review of literature: The next step after choosing a topic and formulating a research question is to find out what other researchers have done on that topic. This bibliographic search is crucial because you will not only learn how models have been formulated and estimated, but also what the data sources are.
3. Formulating a general model: Based on the literature review you should construct a general formulation of you own model or following the previous research models. The initial model will be stated in boards terms, and will identify the dependent and independent variables for which you would like to get data.
Whether cross-section or time-series data are appropriate for your stated objectives should also be decided at this stage. If you goal is to explain what makes the values of the dependent variable change over time, then the relevant data will be time series. If, on the other hand, you wish to investigate why different groups behave differently at a given point of time, then cross-section data are called for and the dummy variables might need to be applied for.
prepare a write-up explaining why you believe the independent variables you have chosen are likely to affect the dependent variable (s). Describe the hypotheses you plan to test and the expected nature of the effect of independent variables. In particular, discuss the expected signs of regression coefficients, whether linear or nonlinear might be present, what kinds of interactions among independent variables you should look for, and so on.
4. Collecting the data: Collect the necessary data and organize them into a form (i.e., Excel or other spreadsheet formatted, etc) that computers can process, and finally enter them on the computer software (i.e., Eviews, or SAS, etc) for further analysis. We have seen that the higher the degree of freedom, the better the precision of an estimated and the greater the power of tests of hypotheses. Increasing the degree of freedom means having more observations relative to the number of independent variables. A rule of thumb is to have at least 30 degree of freedom.
5. Empirical results and conclusions: State what you observe in terms of the original hypotheses and expectations. If you found unexpected results, present some rationalization for them. Interpret the relationships of the economic variables and give the implications. Last, provide some concluding remarks regarding your study and put it in perspective with other studies.
¡@
Or read Chapter 19 of "Introductory
Econometrics: A modern Approach-- Carrying out an empirical project"
by Jeffery M. Wooldridge(2000) , South-Western, pp. 616-642.
Plagiarism is a serious mistake of academic ethics which can have serious consequences.)
Return to Dr. Bill Hung's "ECON2170 Applied Econometrics" Main Page
Return to Hong Kong Baptist University HomePage
Comments to billhung@hkbu.edu.hk