Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit

ECON 5140 – Applied Econometrics – Spring 2023

Homework #5

Due date:  Thursday Mar 2, by 11:59 pm

· You may complete this assignment individually or with a group; in either case every student must turn in their own copy.

· If you miss the due date, you may submit your assignment within the next 24 hours, although there is a penalty for late submission

· Please upload your assignment answers into Canvas.  Your answers can be in Word, PDF, or Excel.

· Show work for partial credit.

1.  In the analysis of MA test scores and student-teacher ratios (etc.) in Chapter 9, the MA population differs somewhat from the CA population -- incomes are on average higher, classes are smaller, the % of English-first language students is much higher, etc. Given that the econometric results are similar between MA and CA, do population differences like these increase or decrease external validity? Briefly explain your answer.

2.  Using only the observations on individuals 120, 126, and 150 (24 observations total) in WagePanel.xlsx,

a) Estimate a regression of lwage (natural log of wage) on experience & marriage using LSDV, reporting the usual regression output (coefficients, standard errors, R^2, etc.). You can do this in excel or matlab.

b) Estimate a regression of lwage (natural log of wage) on experience & marriage using de-meaning, (coefficients, standard errors, R^2, etc.). Do you get the same coefficients on experience and marriage? Do this in excel.

c) Estimate a fixed-effects regression of lwage on experience & marriage using the matlab function panFE. Do you get the same coefficients on experience and marriage?

3.  Taking differences across time periods (subtracting each variable at an earlier time period from its value at a later time period) is another way of removing fixed effects. Using the data in MURDER, answer the questions below (the variable mrdrte is the murder rate, the number of murders per 100,000 people; exec is the total number of prisoners executed for the current and prior two years; unem is the state unemployment rate.). You can do this in any software you choose.

a) How many states executed at least one prisoner in 1991, 1992, or 1993?

b) Which state had the most executions?

c) Using the years 1990 and 1993, do a pooled regression of mrdrte on d93, exec, and unem (that is, use all of the 1990 and 1993 data and run OLS).

i) Does the coefficient on exec make sense?

ii) What might cause this result?

d) Using the differences from 1990 to 1993 only (for a total of 51 observations), estimate the equation

by OLS and report the results (e.g. the first observation for  is the murder rate in Alabama in 1993 minus the murder rate in Alabama in 1990).  Now, does capital punishment appear to have a deterrent effect?