Assignment:
In this homework, we work on a paper by Goldberg (2016). The paper and its data is available for UMD students and on ELMS (PS03-Goldberg2016-Labor supply in rural malawi.pdf and PS03-Goldberg2016.dta)
The assignment has two parts. In part 1 you have to read the paper and answer some simple questions.
In part two, you need to use Stata. You can practice by writing commands in command window to answer questions. Butyou need to write a do file that contains all the codes you write and then generate the log file. You must upload this file (HW1.doc) with answers as well as the log fileon ELMS.
You will be assigned to groups. One copy per group should be uploaded on ELMS. But all members of a group should have the answers. I may ask members to send me their answers and log files.
Part 1:
Read the paper and answer the following questions:
[Hint: read the abstract very carefully, read the introduction, data, and conclusion, and skim the rest of the paper. Look at graphs and tables. Forget about most technical parts.]
a. What are the main questions the paper asks?
b. Explain how the author collect the data to answer the question. Explain the setting and unit of observations. What are the variable of interest and the main explanatory variable?
c. What are the main findings of the paper? What is the main explanation?
Part 2:
Now work on the data set and answer the following questions. There are many variables in data set. We will work with some of them.
The main variable of interest is the proportion of people in each village at each week who accepted the offer: vil_labor. The main explanatory variable is the wage: wage.
To answer these questions, you need to write codes in Stata. The codes you need are the ones we used in previous Problem Solving Session instructions. If there is a need for any new code, it is given below.
1. Describe your variable of interest (i.e., find average, standard deviation, minimum and maximum. Also draw a histogram of it and comment of the shape of the distribution. Report the numbers and paste the histogram below).
2. Describe the main explanatory variable (the same type of descriptions mentioned above.) Use the command tab with this variable to see how many observationsare in each wage. Comment. Does the shape of the wage histogram confirm the information you have on wage? (You can use the options in hist command to generate a histogram that represents the design of the experiment. Try it)
3. We would like to find the mean and standard deviation of acceptance rate for the 20 observations with the lowest wage, 20 observations with highest wage, and 20 observations in the middle of wage distribution. To do so we use the following command to sort the data from lowest wage to highest wage. Then we summarize the acceptance rate in the first, the middle and the last 20 observations. The code for sorting and for the first 20 observations are as follows. Comment on the differences in acceptance rates based on wage groups.
4. Find the correlation coefficient between acceptance rate and wage. Does the coefficient confirm your answer for the previous question?
5. Now regress wage on acceptance rate. Carefully interpret the results including the coefficient, statistical significance using t and p-value, and R-squared.
6. Calculate the predicted values and draw a graph that contains the observations, predicted values, and the line through predicted values. (Use practice 5 in problem solving session 2 as your guide)
7. The coefficients for wage seems to be very small. Generate a new variable that is 1000 times smaller than wage (wage/1000) and use it instead of wage in the regression with only wage as explanatory variable. Compare these results with the results in question 5. What has changed and what has remained constant? Why?
8. Finally, use the log value of wage (lnwage) as explanatory variable instead of wage. Interpret the results and compare them with the results in question 5.
9. Use equation in the paragraph just below equation 2 in page 140 to calculate the elasticity. (A challenge: can you write codes to calculate he elasticity? Hint: you need to save the numbers from doing regression and summarizing variables and ask Stata to do the math)
Readings:
Kwacha Gonna Do? Experimental Evidence about Labor Supply in Rural Malawi
By Jessica Goldberg