代做Econ 320 Spring 2024 Problem Set 6 - Instrumental Variables and 2SLS帮做R程序
- 首页 >> Database作业Econ 320 Spring 2024
Problem Set 6 - Instrumental Variables and 2SLS
The due date for this assignment is April 14 at 11:59 p.m EST.
1. (40 points) There are many reasons why instrumental variables estimates can differ from OLS. In this question, we will develop a simple model of how the coefficients differ, which depends on an interpretation of instrumental variables under certain conditions to justify a Local Average Treatment Effect (LATE) estimate, as discussed in the class.
The model: Suppose people in this area have 10 years to live, and they can either go to school or work during this time. They can choose 3 different levels of schooling:
- They don’t attend school and work for 10 years.
- They go to school for the first year and then work for the remaining 9 years.
- They can go to school for 2 years (the first two) and then work for the remaining 8 years.
If they go to school, they get a better job and earn more per year according to the following rule:
- 0 year of schooling → Yearly salary of $2.
- 1 year of schooling → Yearly salary of $4.
- 2 years of schooling → Yearly salary of $5.
(a) (3 points) In terms of yearly salary, what is the return to education of going to 1 year of school (as opposed to 0)? What is the return to 2 years (as opposed to 1)? At which level of schooling is the return highest?
(b) (3 points) What is the socially optimal level of education for everyone (in terms of creating the biggest pot of available resources for society)?
(c) (4 points) What are the lifetime earnings of people who don’t go to any school, people who go for 1 year, and people who go for 2 years? Assume that there is no discounting and the only cost of school is that you don’t earn wages for every year you are in school. If everyone faces this same problem, how many years of schooling will everyone get?
(d) (5 points) Now suppose that going to school costs $1 per year, and you still don’t earn a wage while in school. What are the net lifetime earnings associated with each choice now? Still assuming no discounting, does this cost change the number of years of education people choose?
(e) (10 points) Suppose that school still costs $1 per year, and there are two types of people, patient and impatient people. Patients have a discount rate of 0, so they value their lifetime earnings exactly as in part (d). But impatient people have a discount rate of .6, meaning that after the first year, they deflate future earnings by a term equal to 1/(1.6x ) where x is the number of years after the first. What is the lifetime value of each schooling choice for the impatient people? What level of education will each type of people choose?
(f) (5 points) Finally, suppose that going to school costs now costs $2 per year, and there are impatient and patient people as before. Will either type of person choose a different amount of schooling than they did when school cost $1 as in part (e)? If so, which type of person changes their amount of education?
(g) (10 points) OLS calculates the correlation between years of education and wages for ev- eryone in the sample. IV calculates the return only for people affected by the instrument (“compliers”). If we have an instrument that is the distance from where you live to the nearest school (e.g., Z = 1 if you live near a school, Z = 0 if you live far from a school), such an instrument would mainly affect the upfront cost of going to school instead of working during school age. Suppose we were to have patient and impatient people in the population with discounting similar to those discussed above. Would you expect the slope coefficient of an IV regression to be larger or smaller than the slope coefficient of an OLS regression? Why? Use the model from above to justify your answer.
2. (30 points) A bank offers a week-long management training program to all its loan officers at the end of the second year of their job. Participation in the program is voluntary. The bank is interested in knowing whether participation in the program makes it more likely for a loan officer to be promoted to branch manager. You have data on all the bank’s loan officers, their participation in the program, and whether they have been promoted between years three and five of being in their job. You run a regression for the promotion decision on a constant and a dummy for participation in the training program.
(a) (15 points) Suppose the bank encourages participation in the training program by send- ing a personal letter from the CEO to selected employees who have been recommended by their supervisor as showing particular potential for a position in management. Would it be useful to use the CEO letter as an instrument for participation in the training pro- gram? Explain why or why not.
(b) (15 points)Suppose the bank encourages participation in the training program by send- ing a personal letter from the CEO to a randomly chosen set of employees. Would it be useful to use this CEO letter as an instrument for participation in the training program? Explain why or why not.
3. (30 points) You are interested in estimating the impact of a household head’s level of educa- tion on the food security status of his or her household. (Roughly stated, the issue of food security concerns whether a person or household can count on having access to a sufficient number of calories) You have data from a 2008 household survey run in Senegal. The survey is nationally representative and includes over 20,000 households and the following variables:
variable Description |
|
foodSecure |
food security index (higher values indicates higher food security) |
educHd |
years of education of household head |
rural |
dummy if household is situated in a rural area; zero otherwise |
employHd |
dummy if household head is currently employed |
members |
number of individuals living in the household |
kids |
number of children in the household |
nMale |
number of men in the household |
nFem |
number of women in the household |
(a) (10 points) Once you looked at this data, you are tempted to use OLS. However, a colleague suggests that if you do so, your estimate of the effect of household head’s level of education will suffers from omitted variable bias since you do not have a measure of household income. Describe the circumstances under which your colleague would be correct. Give an example and be specific about the nature of the bias that you would expect.
(b) (10 points) Your colleague suggests using the education level of the household head’s father as an instrumental variable for household head’s level of education. What as- sumptions does such an instrument need to satisfy? Do you believe that this instrument is valid? Explain, being as specific as possible.
(c) (10 points) Suppose the instrument is valid. Describe how you would get the causal effect of an additional years of education on food security index, using all available information you have on the data. Be as specific as possible. (Hint: describe the two steps of the two-stage least squares regression)