Chapter_1_Intro.pptx. Introductory econometric book
- 2. Learning Objectives • Define and describe the basics of econometrics • Describe how to do an econometric study
- 3. Step 1: What Do We Want to Do? • The first step in doing econometrics is to define the purpose of the modeling. It is easy to skip this step, but doing so means your analysis is unlikely to be useful. • Your purpose should be concrete and concise. • Often, if you can state your purpose in the form of a question, you will see whether you have defined it adequately.
- 4. Example: Do Good Teachers Produce Better Student Outcomes? https://www.aeaweb.org/articles?id=10.1257/aer.104.9.2593
- 5. YES! Good Teachers Raise Future Income Replacing an average teacher with a teacher in top 5% would increase students’ earnings later in life by 2.8%. • The average 12 year old in the United States can expect lifetime earnings of $522,000. • 2.8% earnings bump is worth about $14,500 per student. • Multiply that by 20 kids per classroom and an excellent teacher is really valuable.
- 6. Example: Does the Law of Demand Hold for Electricity? https://www.aeaweb.org/articles?id=10.1257/aer.104.4.1417
- 7. YES! Law of Demand Holds for Electricity – but only if people know how much they’re using
- 8. Example: Is it Possible to Forecast Stock Returns? https://academic.oup.com/rfs/article/21/4/1455/1565737
- 9. No! Stock Market Prediction Models are Unreliable? • Negative means that prediction model is worse than historical mean. • IS means “in-sample”. OOS means “out-of-sample”.
- 10. Step 2: Formulate Your Research Design and Specify the Econometric Model • Take your abstract objective from step 1 and convert it into an econometric model with data that can answer your questions. • Requires some economic theory, common sense and a little cleverness. • Making a good choice about which data to collect and use determines whether you will be able to meet your objective.
- 11. Step 2 Example: Do Good Teachers Produce Better Student Outcomes? • Who is a good teacher? • Can’t just measure test scores. Some teachers may get high test scores because they are easy graders or because smart students take their class. • Answer: a good teacher raises test scores more than average • With this “value added” measure of teacher quality, Chetty, Friedman and Rockoff can estimate using econometrics whether students who took classes from high VA teachers earn more later in life.
- 12. Step 2 Example: Does the Law of Demand Hold for Electricity? • Everyone wants to use their AC on hot days • Electricity is more expensive on hot days because demand is high on those days • To test law of demand, you could give some people a high price and others a low price on the same day • Jessoe and Rapson convinced an electric utility to let them raise prices for a random subset of customers • Then they estimated using econometrics how consumers responded to the high prices
- 13. Step 2 Example: Is it Possible to Forecast Stock Returns? • Everyone is a stock market expert after the fact • (or a bitcoin market expert) • Example: historically, the stock market did better in January. • Does this mean that the stock market will do better in future Januarys? • Hint: no. • Welch and Goyal build econometric models of stock returns using historical data and then see whether those models predict future returns.
- 14. Step 3: Apply Statistical Theory The Power of Statistics: Statistics lets us make statements of probability about an unobserved population using data from a single sample. Econometrics is challenging because it integrates all three fields—economics, mathematics, and statistics—into one.
- 15. How Does Poverty Affect Student Performance in California Schools? • To answer this question, we first need to measure “student performance” and “poverty.” • Academic performance index (API) • The California state government uses it to measure school performance • Constructed from students’ scores on state-wide standardized tests • Ranges from 200 to 1000. • We can use the API as our measure of average student performance at each California school • We do not know the household incomes for each student at each school • But U.S. National School Lunch Program provides free lunches to school children from low-income households • We know the share of students eligible for free lunches at each school • We can use free lunch eligibility (FLE) as an indicator of the share of students from households with incomes below the poverty line at each school.
- 16. Table 1.1 Academic Performance Index (API) and Free Lunch Eligibility (FLE) at 20 Randomly Chosen California Elementary Schools in 2013 (from Total Population of 5,765 Schools) School County API FLE Joe A. Gonsalves Elementary Los Angeles 960 16 Old River Elementary Kern 849 0 Sierra Vista Elementary Kern 722 96 West Portal Elementary San Francisco 914 44 Isabelle Jackson Elementary Sacramento 754 83 Rio Vista Elementary Orange 796 90 Poplar Avenue Elementary Butte 802 80 Cloverly Elementary Los Angeles 903 46 Creative Arts Charter San Francisco 844 33 Carolyn A. Clark Elementary Santa Clara 963 6 Raymond Elementary Orange 824 69 Fernangeles Elementary Los Angeles 730 100 Rainbow Ridge Elementary Riverside 826 90 Cyrus J. Morris Elementary Los Angeles 882 29 Benjamin Franklin Elementary Riverside 882 36 Salvador Elementary Napa 736 65 Bowers Elementary Santa Clara 788 59 Vintage Parkway Elementary Contra Costa 830 54 Balboa Magnet Elementary Los Angeles 981 22 Selby Lane Elementary San Mateo 730 80 Mean 835.80 54.90 Std. Deviation 81.14 31.00 Source: California Department of Education; http://www.cde.ca.gov/ta/ac/ap/apidatafiles.asp
- 17. Figure 1.1 2013 API in 20 California elementary schools. Academic performance varies widely across these 20 schools. Mean API for this sample = 835.80. Repeat for many random samples, and on average you would get μ, the mean API of all schools. But μ almost certainly does not equal 835.80! (BYAM: It’s 813.70. But we don’t know that yet!)
- 18. Let Free School Lunch Eligibility “Proxy” for Poverty in School Districts Figure 1.5. It sure looks like API decreases with FLE. What we can learn from FLE to help predict API? Begin by estimating a linear model in which FLE is the only variable explaining API; that is: i i i e X b b Y 1 0 API of school i FLE of school i Error, how much we miss by if we use FLE to predict API for school i Mean=836
- 19. What We Learned • Econometrics is about making sense of economic data. • Three steps to conducting econometric analysis. 1. State the purpose of the analysis. 2. Formulate the research design and specify the econometric model. 3. Apply statistical theory.