ECON 2P91: Business Econometrics with Applications
Fall 2009
Assignment 3 (Due Date: Monday November 30th 2009) -
SOLUTIONS
In assignments1 and 2, you investigated the relationship between residential housing
price and lot size by constructing hedonic price models for the City of Prince George,
British Columbia, Canada. In this assignment, you will refine the previous models
with a focus on the following questions. First, are there seasonal differences in
residential housing prices? Second, is the relationship between lot size and price
linear? As a by-product, you will examine the use of econometric models for
estimating the means of variables and the associated confidence intervals.
For this assignment, you will employ the data for 1011 residential houses sold in
Prince George during the period from the first quarter of 2001 to the first quarter of
2002. As in the previous assignments, LOT denotes lot size (acres), PRICE is the
residential housing price (thousands of dollars). We have added a binary variable
D
(
1
=
D
if the house was sold in Spring/Summer and
0
=
D
otherwise).
ID
is the
housing identification variable. The relevant data are provided in the Excel data file
princegeorge_housing3.xls
, which is available on Sakai.
1.
Obtain the descriptive statistics of PRICE and interpret the values of the
coefficient of skewness and the coefficient of excess kurtosis.
Summary Statistics, using the observations 1 - 1011
for the variable 'price' (1011 valid observations)
Mean
119.91
Median
117.00
Minimum
22.500
Maximum
385.00
Standard deviation
44.169
C.V.
0.36835
Skewness
1.1046
Ex. kurtosis
3.8509
Coefficient of skewness (1.1046) is positive indicating positive skewness i.e. long tail
is in the positive direction.
Excess kurtosis (3.8509) is positive, meaning that the coefficient of kurtosis is greater
than 3, indicating that the distribution has a higher peak than the normal
distribution. Note that excess kurtosis is equal to coefficient of kurtosis minus 3. In
this case, the coefficient of kurtosis must be 3.8507+3=6.8507.
1

This
** preview**
has intentionally

**sections.**

*blurred***to view the full version.**

*Sign up*2.
Use the mean and standard deviation you obtained in part (1) to compute (by
hand) a 95 percent confidence interval for the average price.
95 percent confidence interval =
)
1011
/
169
.
44
(
96
.
1
91
.
119
)
/
(
96
.
1
±
=
±
n
s
x
x
=
723
.
2
91
.
119
±
i.e. 117.187 to 122.633.
3.
Run a regression of PRICE on a constant and compare the estimate of the
constant with the estimate of the mean you obtained in part (1). On the basis
of these results, what can you conclude about the effect of running a
regression of any variable on a constant?
Model 1: OLS estimates using the 1011 observations 1-1011
Dependent variable: price
VARIABLE
COEFFICIENT
STDERROR
T STAT
P-VALUE
const
119.911
1.38911
86.322
<0.00001 ***
Mean of dependent variable = 119.911
Standard deviation of dep. var. = 44.1686
Sum of squared residuals = 1.97037e+006
Standard error of residuals = 44.1686
Unadjusted R-squared = 0.00000
Adjusted R-squared = 0.00000

This is the end of the preview.
Sign up
to
access the rest of the document.