Economics and stata
PART A: Stata Application: Reset and Chow tests (50 marks)
Question 1: (40 marks) We will use the data set from Assignment 1 to perform a CHOW test to
determine whether or not there was any change in the relationship between crime, the lagged
clearance rate, the size of the local population and the overall unemployment rate after Mr. Rob
Knecht was hired as the new Chief of Police in Edmonton.
The test will be based on the following “log-log” models (NOTE: the data set includes the
current value of Clearance, but the models use the lagged value of this variable).
MODEL 1
ln(Crimet) = δo + δ1 ln(Clearancet-1) + δ2 ln(Popt) + δ3 ln(Unempt) + ut
MODEL 2
ln(Crimet) = βo + β1 ln(Clearancet-1) + β2 ln(Popt) + β3 ln(Unempt) + β4Newchieft +
β5 Newchieft * ln(Clearancet-1) + β6 Newchieft * ln(Popt) +
β7 Newchieft * ln(Unempt) + νt;
(a) Find out when Rob Knecht became chief of police for the City of Edmonton and create a dummy
variable, Newchief, that is equal to 1 for all observations after Rob Knecht became the chief of
police (i.e., do not include quarter during which he became police chief).
(b) Using the same sample period as in Assignment 1 (i.e., commencing in 2009 Q2), calculate
summary statistics (including correlations) for Crimet
, Clearancet-1, Popt
, Unempt and Newchieft
for
i. the full sample period;
ii. the sub-sample where Newchieft=0
iii. the sub-sample where Newchieft=1.
(c) Estimate the coefficients for Model 1 and Model 2.
(d) Test whether or not the elasticticity of crime with respect to unemp for the period during which
Chief Knecht has been in office is significantly different from zero.
(e) Perform RESET tests for both models
(f) Perform a CHOW test for differences across the two subsamples.
Complete Part A of the Answer Worksheet and Attach your Stata log file
PART B: Some Properties of Simple Regression Models with Dummy Variables (20 marks)
Consider the following model, where the data are drawn from two different groups (such as two
different time periods), Group A and Group B:
Yt = γ0 + γ1Dt + εt
i = 1, … T; where
Dt = 0 for the first TA observations in the data set (Group A)
1 for the last TB (= T-TA) observations in the data set (Group B).
(i) (2 marks)Find the expected value of Y for a random observation drawn from Group A
(ii) (2 marks)Find the expected value of Y for a random observation drawn from Group B
(iii) Use a Stata experiment where Yt = Crimet and Dt = Newchieft from Part A, to
determine which of the following statements are true for this data set. For each
statement clearly explain how your Stata experiment supports or refutes the statement.
(a) (4 marks) ∑ [(?? − ?̅)??
] = (????)/(?)
?
?=1
(b) (4 marks) ∑ [(?? − ?̅)??
] = ??(?̅
? − ?̅)
?
?=1
where ?̅
? is the average value of Y
for the subset of observations belonging to group B
(c) (2 marks) If we use OLS to estimate the model, the estimate of γ0 will be ?̅
?
(d) (2 marks) If we use OLS to estimate the model, the estimate of γ1 will be ?̅
? -?̅
?
(e) (4 marks) A “Chow” test, which in this case is simply the F-test for overall
significance, indicates that there are significant differences in the average value of
Y across the two time periods.
Attach your Stata log file to your answers to Part B.
PART C: Preliminary Data and Literature Review Work on Research Report (30 marks)
(1) Collect the data that you will need to estimate your model. The sample size should be at least
30 observations. Report the exact sources (4 marks) of your data (including web site addresses
and series numbers if applicable).
(2) (4 marks) Calculate the minimum, maximum, and average values for all of your variables and
the correlations of each of your explanatory variables with your dependent variable. Present
these values in a table using the general format from Assignment 1. Submit the corresponding
Stata logs for these calculations.
(3) (2 marks) Create a Stata plot with your dependent variable on the vertical axis and the main
explanatory variable corresponding to your main research question on the horizontal axis.
(4) (10 marks) Write a paragraph (at least 1/3 of a page single spaced or at least 2/3 of a page
double-spaced) describing the highlights of your results from (2) and (3)
(5) (10 marks) Write a paragraph (at least 1/3 of a page single spaced or at least 2/3 of a page
double-spaced) describing one or more major results or ideas from one of the articles that you
will be using for your literature review. Provide a complete reference for the article (See, for
example, the references on the class website.) Note: Wikipedia and other similar websites such
as Investopedia are not appropriate sources for your literature review.
NOTE: If you have changed topics, PART C for Assignment #2 will only be marked if you also
hand in a revised version of PART C from Assignment # 1.
Assignment #2 Answer Worksheet
Due Date: March 23, 2017 (in class)
Name: __________________ Student ID ______________
(A1) (2 marks) Rob Knecht became the Chief of Police for the City of Edmonton on
________________. Enter dates as Month/Day/Year (for example November 30,1999).
(A2) (8 marks) Summary Statistics
1. Full Sample: 2009 Q2 through 2016 Q3
crime lagged
clearance
pop unemp Newchief
Sample
Average
Minimum
Maximum
Standard
Deviation
Correlation
with crime
Sub-Sample 1: Observations from 2009 Q2 through 2016 Q3 where Newchief=0
crime lagged
clearance
pop unemp Newchief
Sample
Average
Minimum
Maximum
Standard
Deviation
Correlation
with crime
Sub-Sample 2: Observations from 2009 Q2 through 2016 Q3 where Newchief=1
crime lagged
clearance
pop unemp Newchief
Sample
Average
Minimum
Maximum
Standard
Deviation
Correlation
with crime
(A4) Complete the following tables (12 marks =6 marks per Model)
Model 2 Results
?̂0
[se(?̂0 )]
?̂1
[se(?̂1
)]
?̂2
[se(?̂2
)]
?̂3
[se(?̂3
)]
?̂4
[se(?̂4 )]
?̂5
[se(?̂5
)]
?̂6
[se(?̂6
)]
?̂7
[se(?̂7
)]
# of observations
R
2
F test of overall
significance
(p-value)
RESET
(p-value)
Model 1 Results
?̂0
[se(?̂0 )]
?1
̂
[se(?1
̂ )]
?̂2
[se(?2
̂ )]
?̂3
[se(?3
̂ )]
# of observations
R
2
F test of overall
significance
(p-value)
RESET
(p-value)
(A5) (6 marks) According to the RESET test results, the functional form for:
Model 1 appears to be adequate: _YES / NO___(circle one)
Model 2 appears to be adequate: _YES / NO___(circle one)
(A6) (10 marks) Construct a 95% Confidence interval for the elasticity of crime with respect to
the unemployment rate for the period after Chief Knecht took office. Interpret this interval from
the perspective of an economist (i.e., What do we learn about the elasticity from this C.I.?):
ANSWER:
(A7) (12 marks) Perform a CHOW test for whether or not the relationship is the same in both
subsamples. Include all of the following steps:
Indicate the null and alternative hypotheses: H0 and Ha ;
provide the appropriate critical value from your statistical tables;
provide the value of the test statistic from your Stata output;
indicate whether you have decided to reject or not reject your null hypothesis;
interpret your decision in the context of the model.
ANSWER: