Reference no: EM131949176

1 Workflow

The project is usually done by following these steps: (i) posing an empirical question; (ii) review relevant literature; (iii) collect and clean data; (iv) run econometric analysis; (v) write your term paper. Provided that you work as a group, the division of labor is up to you. This section focuses the first four steps.

I cannot emphasize enough the importance of literature review.

Students tend to skip this step. Literature review is useful at least for three reasons. (i) It helps to argue the importance of your topic and how much interesting is your project. An "interesting" project is new relative to the literature.

Without literature review, you cannot argue that. (ii) The published paper will usually mention their data source. It helps you to find your data. (iii) The literature will discuss their econometric models and potential issues. For example, "endogeneity" is a common issue in empirical studies. You may find some good instrumental variables by reading the literature.

I have demonstrated how to search for literature using google scholar in class.

1.1 Question

You need to find your own question, not me. The topics can be any interesting empirical question in economics. To give you some examples, how does unemployment insurance affect labor supply in the U.S. (Krueger and Meyer, 2002);

how does family income affect child achievement (Dahl and Lochner, 2012); how does mortgage credit expansion affect the price of housing (Favara and Imbs, 2015). Keep in mind, your question needs to be specific. For example, "I am going to study education in the term paper" is not a true question. "I am going to study the return to education in earnings" is more specific and better. Even better, you could say "I am going to study the difference of the return to education between male and female workers". To find a topic, you might be interested in browsing some journal articles, such as the papers published on the American Economic Review. The Economist and the Financial Times could also inspire you, however, unlike the papers on the American Economic Review, their data sources are usually not clear. The appendix of Chapter 19 of our textbook (Jeffrey M. Wooldridge: Introductory econometrics: A modern approach, sixth edition) lists 33 sample empirical projects, which might inspire you.

1.2 Literature review
You paper, even short, should contain a review of relevant literature. You can find relevant literature by searching the keywords of your question in googlescholar (not Google itself). Time is limited, focus on the papers that were mostly cited taking account of their years of being published.

When you cite others' paper, please follow strictly the citation styles of The Chicago Manual of Style. If you do not what is Chicago style (not jazz), google it or check out this wikipedia page. This guideline used the Chicago style. Pay attention to the citations I made in Section 1.1 and References list at the end of this note.

1.3 Data collection
We will walk through several useful public datasets. The datasets include the Current Population Survey (CPS), the British Household Panel Survey (BHPS), the Panel Study of Income Dynamics (PSID), the Labor Force Survey (UK) et al.

Most recently published papers in the American Economic Review and other journals also provide the datasets used. If you are interested in the topics of particular papers, you might consider using their data at least as a starting point. If you read the paper carefully, you will find the exact source of their data in many cases.

After collecting/downloading data, it is important to become familiar with the data. For example, how were missing data coded in your sample. You should also be aware of the nature of the variables. Which are dummy variables, ordinal variables? What are the units of measurement of the variables? When you have time series or cross-section data in multiple years, questions like "were they seasonally adjusted", "are they weakly dependent", "have the income been CPI adjusted" are critical for doing right econometric analysis.

1.4 Econometric analysis
You determine the appropriate econometrics models and methods for the question. Keep in mind that you have to justify the selected model and method. For example, if you decided to use multiple linear regression, you should discuss why are the key OLS assumptions satisfied for your model. If you used robust variance/inference, you need to tell us why.

Good empirical papers usually contain sensitivity analysis. This means that after estimating your original model, you modify it in ways that seem reasonable and do the estimation again. The goal is to check if your major conclusion is sensitive to the specification of the original model.

2 Writing a paper
There is no limit for the length of your manuscript. A typical structure of paper would be (i) title; (ii) introduction; (iii) conceptual (or theoretical) framework for analyzing your question; (iv) econometric models and estimation methods; (v) data description; (vi) results; (vii) conclusion. If you have read a few paper in the literature, you know the structure.

The title page is unnecessary. Just provide title, authors' information, abstract, keywords. You do not need to include any graphic design, such as the logo of the university, - it is useless.

Introduction typically states the main objectives of the work and why they are important. Usually, people summarize the key findings to attract readers' attention. For term paper, I recommend summarizing the findings. Otherwise, I feel forced to read the paper, which is the mood you want me to avert. Literature review is also done within the introduction.

Conceptual (or theoretical) framework could be formal economics theory or intuitive arguments to answer your question.

Econometric models and estimation methods are different. For example, if I study the return to education in earnings, the model is to express earnings as a function of explanatory variables, e.g.

ln wage = α0 + ρss + β0x + β1x2 + u, (1)

where ?? is schooling level, and ?? is work experience. Estimation methods address how to estimate the unknown coefficients (α0, ρss, β0, β1) in the model. Of course, you also need to address possible issues in the estimation, i.e. argue why are the assumptions involved in the estimation method satisfied. If not, you need to provide remedies, which give rise to another estimation method. For example, because ability was not included in eq. (1) and ability is likely to correlated with s (school level) and x (work experience), the OLS assumption E(u|s, x) = 0 is not likely to hold. A remedy could be find proxy variable for abilityor use panel data methods. Keep the assumptions behind estimation method in mind! You should carefully describe your data. It is informative to present a table of summary statistics, such as min and max, mean and sd of each variable.

Table 1 gives you an example, which is part of Table 19.2 of the textbook.

Table 1: Summary Statistics
Variable Mean Standard Deviation Minimum Maximum

prate .869        .167                        .023                        1

Mrate .746                        .844                        .011                        5

You report your results in beautifully crafted tables not copy-and-paste fromsoftware like R or STATA. The table should always include at least the Rsquared and the number of observations for each equation. Always, state the dependent variable clearly. Table 2 gives an example, which is part of Table 19.3 of the textbook. The most important task to interpret your results. Do the coefficients have the expected sign? What is the magnitude of the effect?

Table 2: OLS Results: Dependent Variables: Participation Rate

Independent Variables                                 (1)                          (2)                          (3)

mrate                                                  .156                        .239                        .218

(.012)                     (.042)                    (.342)

mrate2                                                -                           -.087                     -.096

(.043)                      (.073)

Observations                                     3,784                     3,784                     3,784

R-squared                                           .143                        .152                        .162

Attachment:- Template.rar