1. Price of the house.

n=14
X1
Has a pool 1
No has a pool 0

a.1
Intercept. When there is no pool, no living room and no fire place, no square feet, no
rooms and no bathrooms, the expected value of the price of the house will be
39.0571 thousand US$.

a.2
Differential slope of having or not a living room (X2). -21.344 thousands of US$ is
the difference effect between having or not a living room on average, keeping
constant having or not having a pool, having or not having a fire place, the number
of square feet, the number of rooms and the number of bathrooms, being higher for
not having a living room.
Slope + units; the difference effect; between reference category y la dummy; on
average; keeping constant; being higher.

*Higher the reference category when the slope is negative. Higher the dummy when
the slope is positive.

a.3
Partial slope of the number of rooms. When we increase in 1 room the house, the
expected price of the house will decrease in 7.0455 thousand US$ on average,
keeping constant having or not having a pool, a fire place and a living room; and the
number of square feet and the number of bathrooms.

a.4
R2 = 0,9115 = 91,15%
The variations in having or not having a pool, having or not having a living room,
having or not having a fire place, the number of square feet, the number of rooms
and the number of bathrooms explain 91.15% of the variations in the price of the
house on average. It is a high value because it is closer to 1, therefore, it is a good
model.
The variations in all the IV explain R2% of the variations in the DV, on average.

b.
It has sense because we are applying ceteris paribus. Ceteris paribus means
keeping constant the rest of the independent variables when we analyze the slope
of each one. Therefore, in the three cases they have a negative slope because if we
maintain the number of square feet and we add a living room or an extra room or an
extra bathroom, we are reducing the area of other rooms, kitchen or bathroom just
because of adding what we have mentioned.
c. 1
SIGNIFICANCE ANALYSIS.
For analyzing the significance of having or not having a pool for explaining the price
of the house, we can use the p-value method in which we RH0 and therefore, this
independent variable would explain the price of housing, when the p-value of this
independent variable is lower than alpha. there are some cases:
For alpha =10% → p-value < 10%
For alpha = 5 % → p-value < 5%
For alpha = 1% → p-value < 1%

c.2
alpha = 10%
1. Formulation of the hypothesis
H0: B1 = 0 → having or not having a pool doesn’t affect the price of the house
H1: B1 ≠ 0 → having or not having a pool affects the price of the house
2. Calculate the t*
Df = 14 – 7 = 7
T*7, 10% = 1,895
3. Calculate the confidence interval.
b-t*xse(b) = 53,1958 – 1,895x22.0635 = 11.39
b+t*xse(b) = 53,1958 + 1,895x22.0635 = 95
Confidence interval = [11.39, 95]
4. Decision rule.
AH0: inside the interval
RH0: outside the interval
0 is not inside the confidence interval → RH0
5. Conclusion.
We conclude on a 90% of confidence, that we RH0, which means that having or not
having a pool is key to explain the price of the house.

c.3
alpha = 1%
1. Formulation of the hypothesis
H0: B1 = 0 → having or not having a pool doesn’t affect the price of the house
H1: B1 ≠ 0 → having or not having a pool affects the price of the house

2. Calculate the t-value

3. Calculate the t*.

4. Decision rule.
2.41 < 3.499 → AH0

5. Conclusion.
We conclude on a 99% of confidence, that we AH0, which means that having or not
having a pool is not key to explain the price of the house.

d.
x4
1. Formulation of the hypothesis.
H0: B4 = 1.6 → one additional square feet doesn’t increase the price of housing in
more than 1.6 thousands US$
H1: B4 > 1.6 → one additional square feet increases the price of housing in more
than 1.6 thousands US$

2. Calculate the t-value.

3. Calculate the t*
Df = 7
T* 7, 5% = 1.895

4. Decision rule
-4.83 < 1.895 → AH0

5. Conclusion.
We conclude on a 95% of confidence that we AH0 which means that one additional
square feet doesn’t increase the price of housing in more than 1.6 thousands US$.

e.
1. formulation of the hypothesis.
H0: B4 = 1.3 → one additional square feet doesn’t increase the price of housing in
less than 1.3 thousands US$
H1: B4 < 1.3 → one additional square feet increases the price of housing in less than
1.3 thousands US$

1. Calculate the t-value.

2. Calculate the t*
Df = 7
T* 7, 5% = 1.895

3. Decision rule
-3.83 < -1.895 → RH0

4. Conclusion.
We conclude on a 95% of confidence that we RH0 which means that one additional
square feet increases the price of housing in less than 1.3 thousands US$.
f. GLOBAL SIGNIFICANCE ANALYSIS.
1. Formulation of the hypothesis.
H0: B1 = B2 = B3 = B4 = B5 = B6 = 0 → all the IV don’t affect globally the price of the
house
H1: B1 ≠ B2 ≠ B3 ≠ B4 ≠ B5 ≠ B6 ≠ 0 → all the IV affect globally the price of the house

2. Calculate the F-value.

3. Calculate the F*.

4. Decision rule
F-value > F* → RH0
12.02 > 7.1914 → RH0

5. Conclusion
We conclude on a 99% of confidence, that we RH=, which means that all the IV
affect globally the price of the house.

g.
in order to improve this model, we need to look at the |t-values| of each independent
variable. If one of them is lower than 1, we can improve the model by substracting
from the model that independent variable. In this case, we have 5 independent
variables with an |t-value|<1, but we start substracting the one with the lowest |t-
value| because we need to discard them one by one. The number of bathrooms
variable is the one with the lowest |t-value| and it will be the first one discarded.
Once we discard this variable, we expect that the adjusted R^2 increases and, after
that, we will repeat the process in order to see if we can improve again the model by
discarding another variable with an |t-value|<1.

2a.
We choose the model B because since it is a log-linear model and the opther is a
linear model, the only able to explain the elasticity of the final demand is the log-log
model, therefore the model B.

2b.
Log (Final Demand) = 1.554 + 0.998* ln(Real GDP) – 0.333*ln(Real energy price).

2c.
Intercept. When the ln (real GDP) and the ln(real energy price) is 0$ (which is
meaningless because ln(0) da error), the expected value of the ln of the final
demand will be 1.554$.
Partial slope of the real GDP. When we increase in one % the real GDP, the value
of the final demand will increase in 0.998% on average, keeping constant the real
energy price.
Partial slope of the real energy price. When we increase in 1% the real energy
price, the value of the final demand will decrease 0.333% on average, keeping
constant the real GDP