ECON 41: Data Analysis and Econometrics Fall 2014 Prof. William A. Sundstrom Lucas 216G [email protected] 408-554-6892 Office hours: Monday 1:30-3:30 p.m. Overview Most interesting ideas in economics and other social sciences make claims about relationships between variables. For example, one interesting and influential claim is that a person can earn a higher income if she attains more education. Economic theory offers various reasons to expect this to be the case: education is an investment in skills that are valued in the job market, and more education may also signal to employers that the person has persistence and smarts. But is it true? It’s immediately obvious that to answer this we need to look at some evidence regarding the direction of the relationship between earnings and education. (Preview: It’s positive!) Is that enough? No: The economic claim is not just that earnings and education are correlated, but that more education causes increased earnings. Thus it is a claim about a causal relationship. Finding correlations is easy, but establishing causality is not. In many natural sciences, causality is put to the test using controlled experiments in the lab or randomized trials in the field. Although economics has become more open to experimental methods, experiments are often impractical or impossible, and we must then rely on drawing inferences from observational data collected from surveys of individuals, households, businesses, governments, and other organizations. An important question is whether, and if so how, observational data may be analyzed to determine causality in a way that approximates the experimental ideal. In addition to knowing whether a causal relationship exists, we need to know how big or important the effect is. For example, compared with a high-school diploma, can you expect your Santa Clara degree to increase your annual income by 5 percent, 20 percent, 50 percent, or 100 percent? Economists are interested in the answer, and I’m guessing you are too. Econometrics is the application of data analysis and statistics to these kinds of questions. In this course you will learn the basic conceptual foundations and tools of econometrics by applying them to case studies with real-world data. The key statistical technique used through most of the course is multiple linear regression. Regression is a powerful tool for estimating the impact of one or more hypothesized causal variables on an outcome of interest, while controlling for other factors. It is also widely used in forecasting, and has many applications beyond the field of economics. Learning objectives After successfully completing this course you should be able to: • • • • •

 

calculate and interpret descriptive statistics and graphs commonly used in economics; estimate and interpret linear regression models with one or more regressors, using actual data and employing appropriate statistical tests; estimate and interpret regression models that take account of qualitative effects, interactions, and other nonlinearities; assess internal and external validity of studies that use regression to estimate causal effects from observational data; use regression to forecast; 1  

•

appreciate the power and the limitations of data analysis in the social sciences and the importance of using it wisely and honestly.

Prerequisites ECON 1 and 2; MATH 12 or 31; MATH 8 or OMIS 40 or equivalent. This course is open only to Economics majors or by permission of instructor. To take this course you MUST simultaneously enroll in the 2-unit lab course, ECON 42. In ECON 42 you will learn and apply R, a powerful open-source statistical software package, to the hands-on analysis of data. Topics and assignments are closely coordinated across the two courses. ECON 41 culminates in a significant data project that will apply the techniques learned in both courses to analyze an interesting economic or social issue using real-world data. I will assume that you learned something in your previous stats course! We will review key statistical concepts as we go along, within the context of the applications. However, if you are rusty, you should definitely take extra time to review the core material in chapters 1–4 of OpenIntro Statistics and chapters 2–3 of the Stock and Watson textbook. You may also find the videos for chapter 4 of OpenIntro Statistics helpful. Required readings, software, and hardware • • • • • •

OpenIntro Statistics (free on-line: http://www.openintro.org/stat/ ) Stock and Watson, Introduction to Econometrics (see p. 3 below for purchase options) Guide to R for SCU Economics Students (free on Camino site) Camino (Canvas): There is just one Camino site for all sections of ECON 41 and 42. o Some additional short required and optional readings may be posted to Camino. Software: R and RStudio, both available free online (details to be provided). Hardware: o i>clicker: Please bring it to all class meetings. o Access to a computer for the project.

Assignments and grading Assignments and grading weights are as follows: • • •

In-class assignments and clicker questions: 15% Exams (3 total): 60% Data analysis project: 25%

In-class assignments and clicker questions are intended to develop and test your understanding of the material and provide an incentive to be prepared for class. The three exams receive equal weight (20% each); the first two will be held during the scheduled class time, the third during the scheduled final exam period. They will be a mix of multiple choice, short-answer, and problems. You will be permitted to bring a single “cheat sheet” to each exam. The data analysis project will apply skills you learn in this course and the lab course, ECON 42. It will be due in two chunks. Additional graded data analysis assignments using R will be assigned weekly in ECON 42. You course grade will be based on a weighted average of the above assignments, out of 100 points. The following grading scale is guaranteed, but I may curve if averages are lower than expected: A A-

 

93 90

B+ B B-

87 83 80

C+ C C-

77 73 70

2  

D F

60 <60

Class time Please come to class prepared: Read the assigned material in advance, and work the suggested practice problems. Solutions to odd-numbered textbook problems are available online. During class, feel free to raise questions at any time—I don’t mind being interrupted. Use of laptops, phones, and/or other electronic devices during ECON 41 class meetings is NOT permitted, unless you require such a device for an approved disability accommodation (see below); so please put them away. I will distribute an outline of the day’s topics at the beginning of each class. Please note that key announcements, assignments, and due dates will be listed in a box at the bottom of the outline. If you miss class, make sure to get a copy of the handout from the Camino site so you can keep up. Academic integrity You must submit your own work on all assignments. You may confer with other students taking this course on the data analysis project, but the final result should be your own effort. Collaboration or consultation with other students or anyone else on exams is not permitted. Cheating will result in failing the course and will be reported to the Office of Student Life. If you are uncertain about what counts as cheating, just ask me! Accommodations for disability To request academic accommodations for a disability, students must contact Disabilities Resources located on the second floor of Benson. Phone numbers are (408) 554-4111; TTY (408) 554-5445. Students must register and provide documentation of a disability to Disabilities Resources prior to receiving academic accommodations. Please let me know if the class is not working for you, for any reason. Textbook options The Stock and Watson Introduction to Econometrics (3rd ed.) textbook can be obtained in various formats. The bookstore has hardbound copies, which can be purchased new or used, or rented. The book can also be purchased or rented online, for example from Amazon or other sellers. Shop around: the Amazon rental price may be lower than our bookstore’s. The publisher, Pearson, also offers access to an eText digital version for about $100 (for 180 days). Be aware that the end-of chapter problems may be different in the e-text, so make sure you look at the correct on-line solutions!

 

3  

Schedule of topics and reading assignments Please complete all assigned readings in advance of class. Schedule is subject to adjustment. OIS refers to OpenIntro Statistics (http://www.openintro.org/stat/ ). SW refers to Stock and Watson, Introduction to Econometrics.

 

Date

Topic

Chapter

Sept. 23

Introduction; data and data sources

Sept. 25

Looking at data: descriptive statistics and graphs

OIS 1

Sept. 30

Fitting a line using simple regression

OIS 7.0-7.3, SW 4.1-4.3

Oct. 2

Random variables and sampling

SW 2

Oct. 7

Confidence intervals and hypothesis tests

SW 3

Oct. 9

The regression model

SW 4.4-4.6

Oct. 14

Regression hypothesis tests and confidence intervals

SW 5.1-5.2

Oct. 16

Exam #1

Oct. 21

Regression statistics, cont.; heteroskedasticity

SW 5.3-5.4

Oct. 23

Omitted variable bias; multiple regression

SW 6.1-6.3

Oct. 28

Multiple regression fit, multicollinearity

SW 6.4-6.8

Oct. 30

Hypothesis tests and confidence intervals for multiple regression

SW 7.1–7.3

Nov. 4

Multiple regression, cont.; model specification; conditional mean independence

SW 7.5-7.7, Appendix 7.2

Nov. 6

Exam #2

Nov. 11

Nonlinear regressions: polynomials and logs

SW 8.1-8.2

Nov. 13

Nonlinear regressions: dummy variables, interaction terms Data analysis project part 1 due

SW 8.3-8.5

Nov. 18

Internal and external validity of regression studies

SW 9.1-9.3

Nov. 20

Assessing external validity

SW 9.4-9.5

Nov. 25, 27

No Class: Happy Thanksgiving!

Dec. 2

Regression with a binary dependent (Y) variable

Dec. 4

Catch-up and review

Dec. 9, 11

Exam #3: scheduled final exam periods (2 hours): 8:30 a.m. section: Tues., Dec. 9, 9:10 a.m. 2:00 p.m. section: Thurs., Dec. 11, 1:30 p.m.

Dec. 11

Data analysis project final report due before midnight

4  

SW 11.1-11.2