Homework 2 Spring 2012

Economics 140A Prof Heather Royer

Due Wednesday, April 18th in class by 2:05pm. Label clearly your TA on the front of your homework. You may hand in the homework either typed or written. To receive full credit, also hand in the STATA log le.

Written Problem 1
A professor decides to run an experiment to measure the eect of time pressure on nal exam scores. He gives each of the 400 students in his course the same nal exam, but some students have 90 minutes to complete the exam while others have 120 minutes. Each student is randomly assigned one of the examination times based on the ip of a coin. Let yi denote the number of points scored on the exam by the ith student (yi ranges between 0 and 100), let xi denote the amount of time that the student has to complete the exam (xi =90 or 120). Suppose we estimate the regression model yi = 0 + 1 xi + i . a Explain what the term
i

represents. Why will dierent students have dierent values of

i?

b Explain why E( i |xi ) = 0 for this regression model. c Are the other assumptions discussed in class satised? Explain. d The estimated regression is yi = 49 + 0.24xi . (a) Compute the estimated regressions prediction for the average score of students given 90 minutes to complete the exam; 120 minutes; and 150 minutes. (b) Compute the estimated gain in score for a student who is given an additional 10 minutes on the exam.

Written Problem 2
In An Economic Theory of Suicide, published in the Journal of Political Economy, Jan/Feb 1974, Hamermesh and Soss state, When unemployment rises, individuals expectations of future incomes (and utilities) are revised downward. Holding real income of the employed constant, an increased number of people will believe future prospects to have diminished and will commit suicide. As a crude test of their hypothesis, we can estimate st = 0 + 1 ut + t where st is the number of suicides per 100,000 population in t, ut is the unemployment rate in year t, and t is a random error term satisfying the usual assumptions. The relevant data are:

t 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978

st 10.7 11.1 11.5 11.7 12.0 12.0 12.1 12.7 12.5 13.3 12.5

ut 3.6 3.5 4.9 5.9 5.6 4.9 5.6 8.5 7.7 7.0 6.0

1. Construct a scatter diagram with suicide rates on the vertical axis and the unemployment rate on the horizontal axis. 2. Compute the ordinary least squares estimate of the parameters 0 and 1 by hand. 3. Calculate the predicted suicide rates for each year. Calculate the estimated error for each year. 4. What is the error variance for the regression? Calculate by hand. 5. What is the standard error of 0 and 1 , respectively? Calculate by hand. 6. What is the R-squared of the regression? Calculate by hand.

Written Problem 3
Suppose I want to know the eect of income on happiness. Suppose I run two regressions: (a) one with income measured in thousands of dollars (e.g., 10,000 dollars would be expressed as 10) and (b) one with income measured in dollars (e.g., 10,000 dollars would be expressed as 10,000 dollars). Call the OLS estimates of the intercept and slope from (a) 0 and 1 . Call the OLS estimates of the intercept and slope from (b) 0 and 1 . 1. Express 0 in terms of 0 . 2. Express 1 in terms of 1 .

Written Problem 4
Let kids denote the number of children ever born to a woman, and let educ denote years of education for the woman. A simple model relating fertility to years of education is kids = 0 + 1 educ + where is the unobserved error.

1. What kinds of factors are contained in ? Are these likely to be correlated with level of education? 2. If the factors contained in 0 and 1 ? are correlated with x, what does this imply about our OLS estimators

3. Will a simple regression analysis uncover the ceteris paribus eect of education on fertility? Explain.

Written Problem 5
1. A linear regression yields 1 = 0. Show that R2 = 0. 2. A linear regression yields R2 = 0. Does this imply that 1 = 0?

Computer Exercise 1
The dataset bwght.csv is a comma-delimited le containing the data on births to women in the United States. Two variables of interest are the dependent variable, infant birth weight in ounces (bwght), and an explanatory variable, average number of cigarettes the mother smoked per day during pregnancy (cigs). To read the dataset into STATA, you will need to use the command insheet using bwght.csv ; note you may need to use the command cd to the change the directory to the location of where you have bwght.csv. 1. Suppose you run a regression of birth weight on cigarette consumption. What are the estimates of 0 and 1 ? What are the estimates of var(0 ) and var(1 )? 2. What is the R-squared for the regression? 3. What is the predicted birth weight when cigs=0? What about when cigs=10 (one pack per day)? Comment on the dierence. 4. Does this simple regression necessarily capture a causal relationship between the childs birth weight and the mothers smoking habits? Explain. 5. To predict a birth weight of 135 ounces, what would cigs have to be? Comment. 6. The proportion of women in the sample who do not smoke while pregnant is about 0.85. Does this help reconcile your nding from the previous part of this problem?