You are on page 1of 33

BUS 525

MANGERIAL ECONOMICS

Data Analysis

SUBMITTED BY:

MD. SHAHRIAZ SIRAJ

ID: 2115350660

SECTION: 05

SUBMITTED TO:

Professor Dr. K. M. Zahidul Islam

DATE OF SUBMISSION:

NOVEMBER 20, 2021


ANSWER - PDF PROBLEMS

Answer to the Problem No: 4-14


Data on income (in thousands of dollars), education (years), experience (years) and age
(age) for twenty people are shown here:

Person Income Age Education Job Experience


1 5 29 2 9
2 9.7 36 4 18
3 28.4 41 8 21
4 8.8 30 8 12
5 21 34 8 14
6 26.6 36 10 16
7 25.4 61 12 16
8 23.1 29 12 9
9 22.5 54 12 18
10 19.5 30 12 5
11 21.7 28 12 7
12 24.8 29 13 9
13 30.1 35 14 12
14 24.8 59 14 17
15 28.5 65 15 19
16 26 30 15 6
17 38.9 40 16 17
18 22.1 23 16 1
19 33.1 58 17 10
20 48.3 60 21 17

A) Regression Analysis: Income versus Age

Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 1 455.4 455.39 6.00 0.025
Age 1 455.4 455.39 6.00 0.025
Error 18 1365.0 75.84
Lack-of-Fit 13 830.5 63.89 0.60 0.790
Pure Error 5 534.5 106.90
Total 19 1820.4

Model Summary

S R-sq R-sq(adj) R-sq(pred)


8.70836 25.02% 20.85% 7.27%

Page | 1
Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant 9.93 6.22 1.60 0.128
Age 0.359 0.146 2.45 0.025 1.00

Regression Equation

Income = 9.93 + 0.359 Age

Fits and Diagnostics for Unusual Observations

Std
Obs Income Fit Resid Resid
20 48.30 31.47 16.83 2.11 R

R Large residual

Explanation:

H0 ( Null hypothesis): Age has no impact on Income.

Ha ( Alternative Hypothesis) : Age has an impact on Income.

Here, the equation is

Income = 9.93 + 0.359 Age

Coefficient determination = Adj R^2 = 20.85%: Approximately 21% variation in the


dependent variable (Income) is explained by the variations of the independent variable
(Age). Approximately 79% variation is unexplained.

t-statistics: Here, calculated T-value of Age = 2.45

Degrees of freedom = n – k -1 = 20 -1 – 1 = 18. Here, n = 20; k = 1 [Age] with a significance


level of 5%.

So, Critical Value = 2.101

Therefore, T-Value > Critical Value (2.45>2.101) So, we reject the null hypothesis and accept
the alternative hypothesis.

Thus, we can state that, Age has a statistically significant positive impact on Income.

>> Now here, the sign of the age co-efficient is (+) positive. In general, we expect that
income will increase with age. Here our equation shows that age and income have a positive

Page | 2
relationship. Which means that this regression model supports our assumption that income
will increase with increase in age.

B) Regression Analysis: Income versus Age, Education, Job Experience

Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 3 1591.37 530.46 37.05 0.000
Age 1 53.14 53.14 3.71 0.072
Education 1 1124.00 1124.00 78.51 0.000
Job Experience 1 236.87 236.87 16.55 0.001
Error 16 229.06 14.32
Total 19 1820.43

Model Summary

S R-sq R-sq(adj) R-sq(pred)


3.78368 87.42% 85.06% 78.66%

Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant -7.06 3.37 -2.10 0.052
Age -0.211 0.110 -1.93 0.072 2.98
Education 2.245 0.253 8.86 0.000 1.71
Job Experience 1.024 0.252 4.07 0.001 2.51

Regression Equation

Income = -7.06 - 0.211 Age + 2.245 Education + 1.024 Job Experience

Fits and Diagnostics for Unusual Observations

Obs Income Fit Resid Std Resid


4 8.80 16.85 -8.05 -2.25 R

R Large residual

Explanation: For Age,

H0 ( Null hypothesis): Age has no impact on Income.

Ha ( Alternative Hypothesis) : Age has an impact on Income.

Here, the equation is

Income = -7.06 - 0.211 Age + 2.245 Education + 1.024 Job Experience

Page | 3
Coefficient determination = Adj R^2 = 85.06%: Approximately 85% variation in the
dependent variable (Income) is explained by the variations of the independent variables
(Age, Education and Job Experience). Approximately 15% variation is unexplained.

t-statistics: Here, calculated T-value of Age = 1.93 (since we always take absolute value of
T)

Degrees of freedom = n – k -1 = 20 -3 – 1 = 16. Here, n = 20; k = 3 [Age, Education, Job


Experience] with a significance level of 5%.

So, Critical Value = 2.12

Therefore, T-Value < Critical Value (1.93<2.12) So, we accept the null hypothesis and reject
the alternative hypothesis. So, according to this estimation, age has no significance on
income.

Again, for Education,

H0 ( Null hypothesis): Education has no impact on Income.

Ha ( Alternative Hypothesis) : Education has an impact on Income.

Here, the equation is

Income = -7.06 - 0.211 Age + 2.245 Education + 1.024 Job Experience

Coefficient determination = Adj R^2 = 85.06%: Approximately 85% variation in the


dependent variable (Income) is explained by the variations of the independent variables
(Age, Education and Job Experience). Approximately 15% variation is unexplained.

t-statistics: Here, calculated T-value of Education = 8.86

Degrees of freedom = n – k -1 = 20 -3 – 1 = 16. Here, n = 20; k = 3 [Age, Education, Job


Experience] with a significance level of 5%.

So, Critical Value = 2.12

Page | 4
Therefore, T-Value > Critical Value (8.86>2.12) So, we reject the null hypothesis and accept
the alternative hypothesis. So, education has statistically significant positive impact on
income.

Again, for Job experience,

H0 ( Null hypothesis): Job experience has no impact on Income.

Ha ( Alternative Hypothesis) : Job experience has an impact on Income.

Here, the equation is

Income = -7.06 - 0.211 Age + 2.245 Education + 1.024 Job Experience

Coefficient determination = Adj R^2 = 85.06%: Approximately 85% variation in the


dependent variable (Income) is explained by the variations of the independent variables
(Age, Education and Job Experience). Approximately 15% variation is unexplained.

t-statistics: Here, calculated T-value of Job Experience = 4.07

Degrees of freedom = n – k -1 = 20 -3 – 1 = 16. Here, n = 20; k = 3 [Age, Education, Job


Experience] with a significance level of 5%.

So, Critical Value = 2.12

Here, T-Value > Critical Value (4.07>2.12) So, we reject the null hypothesis and accept the
alternative hypothesis. So, job experience has statistically significant positive impact on
income.

Therefore, according to this regression model, age does not have a significant impact on
income but education and job experience do have statistically significant positive impact on
income.

The results of part B are a bit different compared to the results of part A due to the
difference in the impact of Age on Income. We found a statistically significant impact of Age
on Income in part A whereas in Part B our results show the opposite. I believe this has
happened due to the difference in the number of variables used in the two regression
models. Due to the high number of variables in part B, age seemed to have no impact on

Page | 5
age. However, when we analyzed the impact of only the one variable age on income, we
found a statistically significant impact of age on income.

Page | 6
C) Forecasted Income if Age = 45 years, Education = 14 years, Job Experience = 10 years;

Here, our Regression Equation:


Income = -7.06 - 0.211 AGE + 2.245 EDU + 1.024 EXP
= - 7.06 – 0.211*45 + 2.245*14 + 1.024*10
= -7.06 – 9.495 + 31.43 + 10.24
= 25.115

Therefore, forecasted Income = $25,115 (since income is shown in thousands of dollars)

Page | 7
Answer to the Problem No: 4-15

Data on grade point average and IQ for 12 high school students:

Grade Point Average IQ


2.1 116
2.2 129
3.1 123
2.3 121
3.4 131
2.9 134
2.9 126
2.7 122
2.1 114
1.7 109
3.3 132
3.5 140

A) Regression Analysis: Grade Point Average versus IQ

Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 1 2.701 2.7014 23.38 0.001
IQ 1 2.701 2.7014 23.38 0.001
Error 10 1.155 0.1155
Total 11 3.857

Model Summary

S R-sq R-sq(adj) R-sq(pred)


0.339893 70.04% 67.05% 61.68%

Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant -4.17 1.42 -2.94 0.015
IQ 0.0550 0.0114 4.84 0.001 1.00

Regression Equation

Grade Point Average = -4.17 + 0.0550 IQ

Fits and Diagnostics for Unusual Observations

Grade
Point
Obs Average Fit Resid Std Resid
2 2.200 2.917 -0.717 -2.23 R

R Large residual

Page | 8
Explanation:

H0 ( Null hypothesis): IQ has no impact on Grade Point Average.

Ha ( Alternative Hypothesis) : IQ has an impact on Grade Point Average.

Here, the equation is

Grade Point Average = -4.17 + 0.0550 IQ

Coefficient determination = Adj R^2 = 67.05%: Approximately 67% variation in the


dependent variable (Grade Point Average) is explained by the variations of the independent
variable (IQ). Approximately 33% variation is unexplained.

t-statistics: Here, calculated T-value of IQ = 4.84


Degrees of freedom = n – k -1 = 12 -1 – 1 = 10. Here, n = 12; k = 1 [IQ] with a significance
level of 5%.

So, Critical Value = 2.179

Therefore, T-Value > Critical Value (2.45>2.179) So, we reject the null hypothesis and accept
the alternative hypothesis.

According to the results it can be stated that, IQ has a statistically significant positive impact
on grade point average.

My presumption was that IQ should have an impact on grade point average. Grade point
average is generally high for people with a high IQ. So, the results of this regression model
are consistent with my prior expectations that IQ does have a significant impact on GPA.
Grade Point Average tends to be high in people with a higher IQ score and low for people
with a low IQ score.

Page | 9
B) Forecasted grade point average if IQ = 120

Here our regression equation:

Grade Point Average = -4.17 + 0.0550 IQ


= - 4.17 + 0.0550 * 120
= - 4.17 + 6.6
= 2.43

Forecasted grade point average if IQ = 150

Here our regression equation:

Grade Point Average = -4.17 + 0.0550 IQ


= - 4.17 + 0.0550 * 150
= - 4.17 + 8.25
= 4.08

Here, I have more confidence in the first forecast with the IQ of 120. In the second forecast
with the IQ of 150 the estimated GPA is 4.08 which is generally not possible since GPA is
calculated on a scale of 4. I believe the reason behind this is that the IQ score of 120 is closer
to the data set and 150 is an outlier in this data set.

Thus, I believe the first forecast where GPA is forecasted to be 2.43 for a person with an IQ
of 120, is more acceptable and hence I have more confidence in the first forecast.

Page | 10
Answer to the Problem No: 4-16

Data on electric power consumption (in billions of kilowatt-hours), GNP (in billions of
dollars), and electricity prices (in cents per kilowatt-hour) for the period 1969 – 1983 are
given below:

Year Consumption Pre-consumption GNP Price


1969 407.9 367.7 944 2.09
1970 447.8 407.9 992.7 2.1
1971 479.1 447.8 1077.6 2.19
1972 511.4 479.1 1185.9 2.29
1973 554.2 511.4 1326.4 2.38
1974 555 554.2 1434.2 2.83
1975 586.1 555 1594.2 3.21
1976 613.1 586.1 1718 3.45
1977 652.3 613.1 1918.3 3.78
1978 679.2 652.3 2163.9 4.03
1979 696 679.2 2417.8 4.43
1980 734.4 696 2631.7 5.12
1981 730.5 734.4 2957.8 5.8
1982 732.7 730.5 3069.3 6.44
1983 750.9 732.7 3304.8 6.83

A) Regression Analysis: Consumption versus Pre-consumption, GNP, Price

Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 3 177631 59210.4 333.86 0.000
Pre-consumption 1 6078 6077.9 34.27 0.000
GNP 1 175 174.8 0.99 0.342
Price 1 197 197.0 1.11 0.315
Error 11 1951 177.4
Total 14 179582

Model Summary

S R-sq R-sq(adj) R-sq(pred)


13.3174 98.91% 98.62% 98.10%

Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant 96.1 43.4 2.22 0.049
Pre-consumption 0.834 0.142 5.85 0.000 23.77
GNP 0.0649 0.0654 0.99 0.342 215.21
Price -25.9 24.5 -1.05 0.315 123.71

Regression Equation

Page | 11
Consumption = 96.1 + 0.834 Pre-consumption + 0.0649 GNP - 25.9 Price

Fits and Diagnostics for Unusual Observations

Obs Consumption Fit Resid Std Resid


6 555.00 578.32 -23.32 -2.07 R

R Large residual

Explanation: For Pre-consumption,

H0 ( Null hypothesis): Pre-consumption has no impact on Consumption.

Ha ( Alternative Hypothesis) : Pre-consumption has an impact on Consumption.

Here, the equation is

Consumption = 96.1 + 0.834 Pre-consumption + 0.0649 GNP - 25.9 Price

Coefficient determination = Adj R^2 = 98.62%: Approximately 99% variation in the


dependent variable (Consumption) is explained by the variations of the independent
variables (Pre-consumption , GNP and Price). Approximately 1% variation is unexplained.

t-statistics: Here, calculated T-value of Pre-consumption =5.85


Degrees of freedom = n – k -1 = 15 -3 – 1 = 11. Here, n = 15; k = 3 [Pre-consumption, GNP,
Price] with a significance level of 5%.

So, Critical Value = 2.201

Therefore, T-Value > Critical Value (5.85>2.201) So, we reject the null hypothesis and accept
the alternative hypothesis.

Thus, we can state that, Pre-consumption has a statistically significant positive impact on
consumption.

For GNP,

H0 ( Null hypothesis): GNP has no impact on Consumption.

Ha ( Alternative Hypothesis) : GNP has an impact on Consumption.

Here, the equation is

Page | 12
Consumption = 96.1 + 0.834 Pre-consumption + 0.0649 GNP - 25.9 Price

Coefficient determination = Adj R^2 = 98.62%: Approximately 99% variation in the


dependent variable (Consumption) is explained by the variations of the independent
variables (Pre-consumption , GNP and Price). Approximately 1% variation is unexplained.
t-statistics: Here, calculated T-value of GNP = 0.99
Degrees of freedom = n – k -1 = 15 -3 – 1 = 11. Here, n = 15; k = 3 [Pre-consumption, GNP,
Price] with a significance level of 5%.

So, Critical Value = 2.201

Therefore, T-Value < Critical Value (0.99<2.201) So, we accept the null hypothesis and reject
the alternative hypothesis.

Thus, we can state that, GNP does not have a statistically significant impact on consumption.

For Price,

H0 ( Null hypothesis): Price has no impact on Consumption.

Ha ( Alternative Hypothesis) : Price has an impact on Consumption.

Here, the equation is

Consumption = 96.1 + 0.834 Pre-consumption + 0.0649 GNP - 25.9 Price

Coefficient determination = Adj R^2 = 98.62%: Approximately 99% variation in the


dependent variable (Consumption) is explained by the variations of the independent
variables (Pre-consumption , GNP and Price). Approximately 1% variation is unexplained.
t-statistics: Here, calculated T-value of Price = 1.05
Degrees of freedom = n – k -1 = 15 -3 – 1 = 11. Here, n = 15; k = 3 [Pre-consumption, GNP,
Price] with a significance level of 5%.

So, Critical Value = 2.201

Therefore, T-Value < Critical Value (1.05<2.201) So, we accept the null hypothesis and reject
the alternative hypothesis.

Page | 13
Thus, we can state that, Price does not have a statistically significant impact on
consumption.

As per economic theory consumption increases when GNP rises and decreases when price
rises. So, this means our regression model is not consistent with the economic theory.
However, the signs of the estimated coefficients are consistent with economic theory
because in our equation, pre-consumption has a positive sign, GNP has a positive sign and
price has a negative sign.

As per our regression model, only pre-consumption has a significant impact on consumption
at a significance level of 0.05. Neither of the other two variables, GNP and Price, have any
significant impact on consumption at a significance level of 0.05.

B) Here our Regression Equation:

Consumption = 96.1 + 0.834 Pre-consumption + 0.0649 GNP - 25.9 Price

Forecasted consumption for 1984 if GNP in 1984 is 3661.3 million and the price of electricity
is 7.16 cent per kilowatt will be:

Consumption = 96.1 + 0.834 (750.9) + 0.0649 (3661.3) - 25.9 (7.16)

= 96.1 + 626.251 + 237.618 – 185.444

= 774.525

Therefore, as per our estimating equation our forecasted electricity consumption in 1984 is
774.525 billion kilowatt-hours.

Page | 14
Answer to the Problem No: 4-17

Consumption of hamburgers (thousands of burgers per week) in 12 different cities is shown


here. Price of hamburgers, income per capita (in $1000), and prices of hot dogs for cities are
also shown:

City Hamburger Consumption Hamburger Price Income Hot Dog Price


1 50 1.5 12 1.8
2 80 1.35 14.2 1.55
3 95 1.25 15 1.45
4 105 1.2 16 1.35
5 70 1.4 13.8 1.6
6 85 1.3 14.3 1.5
7 55 1.5 13.3 1.7
8 60 1.45 13.3 1.7
9 75 1.35 13.7 1.6
10 90 1.25 14.5 1.5
11 100 1.2 15.2 1.35
12 65 1.45 13.6 1.65

A) In order to estimate hamburger consumption as a multiplicative function of the price of


hamburgers, income and hot dog price we need four new columns using LN function:
City LN HC LN HP LN INC LN HDP
1 3.912 0.405 2.485 0.588
2 4.382 0.300 2.653 0.438
3 4.554 0.223 2.708 0.372
4 4.654 0.182 2.773 0.300
5 4.248 0.336 2.625 0.470
6 4.443 0.262 2.660 0.405
7 4.007 0.405 2.588 0.531
8 4.094 0.372 2.588 0.531
9 4.317 0.300 2.617 0.470
10 4.500 0.223 2.674 0.405
11 4.605 0.182 2.721 0.300
12 4.174 0.372 2.610 0.501

Page | 15
Regression Analysis: LN HC versus LN HP, LN INC, LN HDP

Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 3 0.628924 0.209641 151.44 0.000
LN HP 1 0.020570 0.020570 14.86 0.005
LN INC 1 0.007797 0.007797 5.63 0.045
LN HDP 1 0.000525 0.000525 0.38 0.555
Error 8 0.011075 0.001384
Total 11 0.639999

Model Summary

S R-sq R-sq(adj) R-sq(pred)


0.0372068 98.27% 97.62% 95.62%

Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant 0.97 1.83 0.53 0.610
LN HP -2.210 0.573 -3.85 0.005 17.72
LN INC 1.434 0.604 2.37 0.045 16.28
LN HDP 0.500 0.811 0.62 0.555 42.75

Regression Equation

LN HC = 0.97 - 2.210 LN HP + 1.434 LN INC + 0.500 LN HDP

Explanation: For HP,

H0 ( Null hypothesis): Hamburger Price has no impact on Hamburger Consumption.

Ha ( Alternative Hypothesis) : Hamburger Price has an impact on Hamburger Consumption.

Here, the equation is

LN HC = 0.97 - 2.210 LN HP + 1.434 LN INC + 0.500 LN HDP

Coefficient determination = Adj R^2 = 97.62%: Approximately 98% variation in the


dependent variable (HC) is explained by the variations of the independent variables
(Hamburger Price, Income and Hot Dog Price). Approximately 2% variation is unexplained.
t-statistics: Here, calculated T-value of HP = 3.85
Degrees of freedom = n – k -1 = 12 -3 – 1 = 8. Here, n = 12; k = 3 [HP, INC and HDP] with a
significance level of 5%.

So, Critical Value = 2.179

Page | 16
Therefore, T-Value > Critical Value (3.85>2.179) So, we reject the null hypothesis and accept
the alternative hypothesis.

Thus, we can state that, Hamburger Price has a statistically significant impact on Hamburger
Consumption.

For INC,

H0 ( Null hypothesis): Income has no impact on Hamburger Consumption.

Ha ( Alternative Hypothesis) : Income has an impact on Hamburger Consumption.

Here, the equation is

LN HC = 0.97 - 2.210 LN HP + 1.434 LN INC + 0.500 LN HDP

Coefficient determination = Adj R^2 = 97.62%: Approximately 98% variation in the


dependent variable (HC) is explained by the variations of the independent variables
(Hamburger Price, Income and Hot Dog Price). Approximately 2% variation is unexplained.
t-statistics: Here, calculated T-value of INC = 2.37
Degrees of freedom = n – k -1 = 12 -3 – 1 = 8. Here, n = 12; k = 3 [HP, INC and HDP] with a
significance level of 5%.

So, Critical Value = 2.179

Therefore, T-Value > Critical Value (2.37>2.179) So, we reject the null hypothesis and accept
the alternative hypothesis.

Thus, we can state that, Income has a statistically significant positive impact on Hamburger
Consumption.

For HDP,

H0 ( Null hypothesis): Hot Dog Price has no impact on Hamburger Consumption.

Ha ( Alternative Hypothesis) : Hot Dog Price has an impact on Hamburger Consumption.

Here, the equation is

LN HC = 0.97 - 2.210 LN HP + 1.434 LN INC + 0.500 LN HDP

Page | 17
Coefficient determination = Adj R^2 = 97.62%: Approximately 98% variation in the
dependent variable (HC) is explained by the variations of the independent variables
(Hamburger Price, Income and Hot Dog Price). Approximately 2% variation is unexplained.
t-statistics: Here, calculated T-value of HDP = 0.62
Degrees of freedom = n – k -1 = 12 -3 – 1 = 8. Here, n = 12; k = 3 [HP, INC and HDP] with a
significance level of 5%.

So, Critical Value = 2.179

Therefore, T-Value < Critical Value (0.62<2.179) So, we accept the null hypothesis and reject
the alternative hypothesis.

Thus, we can state that, Hot Dog Price does not have a statistically significant impact on
Hamburger Consumption.

>>Therefore, at a significance level of 5% Hamburger Price and Income have a statistically


significant income on Hamburger Consumption but Hot Dog Price does not have a
statistically significant impact on Hamburger Consumption. This means that a rise or decline
in hot dog price will not impact the consumption of hamburger much. Therefore, HP and
INC are the two coefficients that are significant at the 0.05 level.

B) From the estimates of our regression model in A,


Regression Equation is
LN HC = 0.97 - 2.210 LN HP + 1.434 LN INC + 0.500 LN HDP
From the signs of the coefficients, it can be seen that Income has a positive (+) sign, Hot Dog
Price has a positive (+) sign and Hamburger Price has a negative (-) sign. This means
Hamburger Consumption will increase when Income and Hot Dog Price will increase but
Hamburger Consumption will decrease when Hamburger Price will increase meaning
Income Elasticity and Cross Price Elasticity is positive and Hamburger Price Elasticity in
Negative. This is consistent with the economic theory.

Page | 18
According to economic theory, In case of normal goods,
 If product price increases, consumption will decrease and if product price decreases,
consumption will increase. (Negative Elasticity)
 If income increases, consumption will increase and if income decreases,
consumption will degrease. (Positive Elasticity)
 If price of substitute product increases, consumption will increase and if price of
substitute product decreases, consumption will decrease. (Positive Elasticity)

However, as per our estimations, on 5% significance level, we found that only Hamburger
Price and Income have significant impact on Hamburger consumption. Hot Dog can be
considered a substitute product of Hamburger. But according to our estimations, Hot Dog
price does not have any impact on Hamburger consumption.

The cross elasticity of demand theory states that demand for one good increases, when
price of a substitute good increases. This means, price of Hot Dog should have an impact on
Consumption of Hamburger. Therefore, estimations of our regression model on cross
elasticity are not consistent with economic theory.

Page | 19
ANSWER - EXCEL PROBLEMS

ADEXP VS SALES

ADEXP SALES
2000 15000
2000 30000
5000 30000
3000 25000
9000 55000
8000 45000
7000 60000

Fitted Line Plot


SALES = 11573 + 4.972 ADEXP

60000 S 8782.64
R-Sq 76.5%
R-Sq(adj) 71.8%

50000

40000
SALES

30000

20000

10000
2000 3000 4000 5000 6000 7000 8000 9000
ADEXP

Regression Analysis: SALES versus ADEXP

Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 1 1257182986 1257182986 16.30 0.010
ADEXP 1 1257182986 1257182986 16.30 0.010
Error 5 385674157 77134831
Lack-of-Fit 4 273174157 68293539 0.61 0.731
Pure Error 1 112500000 112500000
Total 6 1642857143

Model Summary

S R-sq R-sq(adj) R-sq(pred)


8782.64 76.52% 71.83% 56.97%

Page | 20
Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant 11573 7151 1.62 0.166
ADEXP 4.97 1.23 4.04 0.010 1.00

Regression Equation

SALES = 11573 + 4.97 ADEXP

Explanation:

H0 ( Null hypothesis): Advertising Expense has no impact on Sales.

Ha ( Alternative Hypothesis) : Advertising Expense has an impact on Sales.

Here, the equation is

SALES = 11573 + 4.97 ADEXP

Coefficient determination = Adj R^2 = 71.83%: Approximately 72% variation in the


dependent variable (Sales) is explained by the variations of the independent variable
(Advertising Expense). Approximately 28% variation is unexplained.
T-test: Here, calculated T-value of ADEXP = 4.04
Degrees of freedom = n – k -1 = 7 - 1 – 1 = 5. Here, n = 7; k = 1 [ADEXP] with a significance
level of 5%.

So, Critical Value = 2.571

Therefore, T-Value > Critical Value (4.04>2.571) So, we reject the null hypothesis and accept
the alternative hypothesis.

P-Value Test: Here, our calculated P- value of ADEXP is 0.010 which is less than 0.05. Hence,
we reject the null hypothesis and accept the alternative hypothesis.

Thus, we can state that, Advertising Expense has a statistically significant positive impact on
Sales.

INSURANCE COSTS VS CLAIMS & PREMIUMS

Page | 21
Data Table:

COSTS CLAIMS PREMIUMS


10197 23.9 319.04
9270 19.5 255
9665 16.7 225.51
8705 16.3 208.95
8888 15.2 200.13
8631 15.6 196.22
8330 16 191.8
8654 14.4 191.46
8516 13 190.78
7738 14.1 189.01
7881 15.3 181.42
7723 14.4 179.74
8702 13.2 177.92
9077 10.7 176.65
9873 10.6 171.38
7842 14.7 168.11
7717 11.9 160.97
7798 12.2 151.02
8783 11.1 129.84
9803 9.7 126.34

Regression Analysis: COSTS versus CLAIMS, PREMIUMS

Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 2 4184066 2092033 4.85 0.022
CLAIMS 1 2142693 2142693 4.97 0.040
PREMIUMS 1 3541112 3541112 8.21 0.011
Error 17 7332278 431310
Total 19 11516345

Model Summary

S R-sq R-sq(adj) R-sq(pred)


656.742 36.33% 28.84% 4.50%

Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant 7406 696 10.64 0.000
CLAIMS -281 126 -2.23 0.040 7.63
PREMIUMS 28.19 9.84 2.87 0.011 7.63

Regression Equation

COSTS = 7406 - 281 CLAIMS + 28.19 PREMIUMS

Page | 22
Fits and Diagnostics for Unusual Observations

Std
Obs COSTS Fit Resid Resid
1 10197 9672 525 1.18 X
20 9803 8237 1566 2.61 R

R Large residual
X Unusual X

Explanation: For Claims,

H0 ( Null hypothesis): Claims have no impact on Insurance cost.

Ha ( Alternative Hypothesis) : Claims have an impact on Insurance cost.

Here, the equation is

COSTS = 7406 - 281 CLAIMS + 28.19 PREMIUMS


Coefficient determination = Adj R^2 = 28.84%: Approximately 29% variation in the
dependent variable (Insurance Costs) is explained by the variations of the independent
variables (Claims and Premiums). Approximately 71% variation is unexplained.
T-test: Here, calculated T-value of Claims = 2.23
Degrees of freedom = n – k -1 = 20 - 2 – 1 = 17. Here, n = 20; k = 2 [Claims, Premiums] with a
significance level of 5%.

So, Critical Value = 2.110

Therefore, T-Value > Critical Value (2.23>2.110) So, we reject the null hypothesis and accept
the alternative hypothesis.

P-Value Test: Here, our calculated P- value of claims is 0.040 which is less than 0.05. Hence,
we reject the null hypothesis and accept the alternative hypothesis.

Thus, we can state that, Claims have a statistically significant positive impact on Insurance
costs.

For Premiums,

H0 ( Null hypothesis): Premiums have no impact on Insurance cost.

Ha ( Alternative Hypothesis) : Premiums have an impact on Insurance cost.

Page | 23
Here, the equation is

COSTS = 7406 - 281 CLAIMS + 28.19 PREMIUMS


Coefficient determination = Adj R^2 = 28.84%: Approximately 29% variation in the
dependent variable (Insurance Costs) is explained by the variations of the independent
variables (Claims and Premiums). Approximately 71% variation is unexplained.
T-test: Here, calculated T-value of Premiums = 2.87
Degrees of freedom = n – k -1 = 20 - 2 – 1 = 17. Here, n = 20; k = 2 [Claims, Premiums] with a
significance level of 5%.

So, Critical Value = 2.110

Therefore, T-Value > Critical Value (2.87>2.110) So, we reject the null hypothesis and accept
the alternative hypothesis.

P-Value Test: Here, our calculated P- value of Premiums is 0.011 which is less than 0.05.
Hence, we reject the null hypothesis and accept the alternative hypothesis.

Thus, we can state that, Premiums have a statistically significant positive impact on
Insurance costs.

Therefore, as per our estimation, Claims and Premiums both of the variables have a
statistically significant impact on Insurance Costs.

Page | 24
ESTIMATION OF DEMAND AS A FUNCTION OF PRICE

Data Table:

TIME Qd Price
1 44 10
2 40 9
3 42 11
4 46 12
5 48 11
6 52 12
7 54 13
8 58 13
9 56 14
10 60 15

Fitted Line Plot


Qd = 7.600 + 3.533 Price
S 2.86065
60 R-Sq 85.1%
R-Sq(adj) 83.3%

55
Qd

50

45

40

9 10 11 12 13 14 15
Price

Regression Analysis: Qd versus Price

Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 1 374.53 374.533 45.77 0.000
Price 1 374.53 374.533 45.77 0.000
Error 8 65.47 8.183
Lack-of-Fit 5 21.47 4.293 0.29 0.890
Pure Error 3 44.00 14.667
Total 9 440.00

Page | 25
Model Summary

S R-sq R-sq(adj) R-sq(pred)


2.86065 85.12% 83.26% 80.20%

Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant 7.60 6.33 1.20 0.264
Price 3.533 0.522 6.77 0.000 1.00

Regression Equation

Qd = 7.60 + 3.533 Price

Explanation:

H0 ( Null hypothesis): Price has no impact on Demand.

Ha ( Alternative Hypothesis) : Price has an impact on Demand.

Here, the equation is

Qd = 7.60 + 3.533 Price
Coefficient determination = Adj R^2 = 83.26%: Approximately 83% variation in the
dependent variable (Price) is explained by the variations of the independent variable
(Demand). Approximately 17% variation is unexplained.
T-Test: Here, calculated T-value of Price = 6.77
Degrees of freedom = n – k -1 = 10 - 1 – 1 = 8. Here, n = 10; k = 1 [Demand] with a
significance level of 5%.

So, Critical Value = 2.306

Therefore, T-Value > Critical Value (6.77>2.306) So, we reject the null hypothesis and accept
the alternative hypothesis.

P-Value Test: Here, our calculated P- value of Price is 0.00 which is less than 0.05. Hence, we
reject the null hypothesis and accept the alternative hypothesis.

Thus, we can state that, Price has a statistically significant impact on demand.

Page | 26
>>What would happen to the demand if price is Taka 20?
Ans: Here our equation:
Qd = 7.60 + 3.533 Price
By putting a price of 20 in our equation we get:

Qd = 7.60 + 3.533 (20)

=> Qd = 7.60 + 70.66

=> Qd = 78.26

Therefore, if price is 20 taka, then demand would be 78.26.

Page | 27
FORECASTING – MR. X’S INCOME FOR THE NEXT FIVE YEARS

Data Table:

Year Income ($) B P


1994 6036 85.1 20.4
1995 6113 87.8 20.2
1996 6271 88.9 21.3
1997 6378 94.5 19.9
1998 6727 99.9 18
1999 7027 99.5 19.9
2000 7280 104.2 22.2
2001 7513 106.5 22.3
2002 7728 109.7 23.4
2003 7891 110.8 26.2
2004 8134 113.7 27.1
2005 8322 113 29
2006 8562 116 33.5
2007 9042 108.7 42.8
2008 8867 115.4 35.6
2009 8944 118.9 32.2
2010 9175 127.4 33.7
2011 9381 123.5 34.4
2012 9735 117.9 48.5
2013 9829 105.4 66.1
2014 9722 103.2 62.4
2015 9769 104.2 58.6
2016 9725 103.7 56.7
2017 9930 105.7 55.5
2018 10419 105.5 57.3
2019 10625 106.5 53.7
2020 10905 107.3 52.6

Trend Analysis for Income ($)

Page | 28
Data Income ($)
Length 27
NMissing 0

Fitted Trend Equation

Yt = 5963.4 + 182.64×t

Accuracy Measures

MAPE 2.1
MAD 178.3
MSD 47493.7

Forecasts

Period Forecast
2021 11077.3
2022 11260.0
2023 11442.6
2024 11625.2
2025 11807.9

Trend Analysis Plot for Income ($)

Trend Analysis Plot for Income ($)


Linear Trend Model
Yt = 5963.4 + 182.64×t
12000 Variable
Actual
Fits
11000
Forecasts

Accuracy Measures
10000 MAPE 2.1
Income ($)

MAD 178.3
MSD 47493.7
9000

8000

7000

6000

1994 1999 2004 2009 2014 2019 2024


Year

Regression Analysis: Income ($) versus B, P

Page | 29
Analysis of Variance

Source DF Adj SS Adj MS F-Value P-Value


Regression 2 52301325 26150662 173.36 0.000
B 1 8111769 8111769 53.78 0.000
P 1 34549096 34549096 229.04 0.000
Error 24 3620260 150844
Total 26 55921584

Model Summary

S R-sq R-sq(adj) R-sq(pred)


388.387 93.53% 92.99% 91.77%

Coefficients

Term Coef SE Coef T-Value P-Value VIF


Constant -231 803 -0.29 0.776
B 56.22 7.67 7.33 0.000 1.05
P 74.71 4.94 15.13 0.000 1.05

Regression Equation

Income ($) = -231 + 56.22 B + 74.71 P

Fits and Diagnostics for Unusual Observations

Income
Obs ($) Fit Resid Std Resid
20 9829 10632 -803 -2.29 R
26 10625 9768 857 2.31 R
27 10905 9731 1174 3.15 R

R Large residual

Here our Regression Equation:

Income ($) = -231 + 56.22 B + 74.71 P

Forecasted Income if B = 110 and P = 55,


Income ($) = -231 + 56.22 (110) + 74.71 (55)
= - 231 + 6184.2 + 4109.05
= 10,062.25
Therefore, as per our regression equation our forecasted income is $10,062.25 (Ans.)

Page | 30
FORECASTING DEMAND FOR THE MONTH OF JANUARY & FEBRUARY

Data Table:

Month Time Q
Jan 1 46
Feb 2 56
Mar 3 72
Apr 4 67
May 5 77
Jun 6 66
Jul 7 69
Aug 8 79
Sep 9 88
Oct 10 91
Nov 11 94
Dec 12 104
Jan 13 103.636
Feb 14 107.927

Trend Analysis for Q

Data Q
Length 12
NMissing 0

Fitted Trend Equation

Yt = 47.86 + 4.290×t

Accuracy Measures

MAPE 6.5651
MAD 4.5565
MSD 33.3505

Forecasts

Period Forecast
Jan 103.636
Feb 107.927

Page | 31
Trend Analysis Plot for Q

Trend Analysis Plot for Q


Linear Trend Model
Yt = 47.86 + 4.290×t
110 Variable
Actual
100 Fits
Forecasts

90 Accuracy Measures
MAPE 6.5651
MAD 4.5565
80
MSD 33.3505
Q

70

60

50

40
Jan Mar May Jul Sep Nov Jan
Month

Therefore, forecasted demand for the month of January and February is as follows:
Period Forecast
Jan 103.636
Feb 107.927

-----X-----

Page | 32

You might also like