Upload
others
View
19
Download
0
Embed Size (px)
Citation preview
FEEDBACK TUTORIAL LETTER
2ND SEMESTER 2020
ASSIGNMENT 1 & 2
Statistics for Economists 2B SFE612S
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
1
COURSE: STATISTICS FOR ECONOMISTS 2B
COURSE CODE: SFE612S
TUTORIAL LETTER: 2020
DATE: 04/02/ 2021
Dear Student
Congratulations on the successful completion of your first and second assignments for
semester 2 2020.
I am convinced the study guide gave you enough exposure to applications of Statistics for
Economists skills in daily financial transactions.
I have no doubts that working through the questions must have in no small way improved on
your statistical, analytical and other calculation skills. The attached memoranda are for you to
see the step by step methods of realising the final calculations and will also prepare you for
the comprehensive test.
Regards,
Dr. Dismas Ntirampeba
Tel. +26461 207 2808
Email: [email protected]
mailto:[email protected]
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
2
Assignment 1
Question 1 [50 marks]
1.1
1.1.1
Ho: 1 2 3
i.e. There is no difference in the three aspects of judgment
H1: , for at leat one pair ( , ),i j i j i j
i.e. At least two aspects of judgment affect the quality of judgement differently.
2
2..
1 1
22 2 2 436.5 = 17 18.5 ... 24.2
21
=317.34
a b
TOTij
i j
SSN
yy
23
2..
.1
22 2 2
1
1 1 1 436.8 = 119 142.8 175
7 7 7 21
=225.68
treati
ii
SSn N
yy
317.34 - 225.68
91.66
T TreatSSE SS SS
225.68112.84
1 2
treattreat
SSMS
k
91.665.0922
18
SSEMSE
N k
Test statistic:
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
3
112.8422.159
5.0922
treatMSFMSE
Critical value:
0.05,2,18 3.55F
Because the test statistic ( 22.159F ) is greater than the critical value
( 0.05,2,18 3.55F ) we reject the null hypothesis.
Conclusion
At 5% significance level, the data provide sufficient evidence to indicate the basis of
judgment affects the quality of judgment.
1.1.2
0 : i jH
1 : i jH , i j ( we compare all possible pairs)
Pair of means i and j would be declared significantly different if
1 3
2
1 1. . , n ni j n k
y y t MSE .
Test statistics: . .i jy y ,
corresponding critical values: 2
1 1, i jn nn k
LSD t MSE
Since 1 2 3 7n n n , LSD is constant.
0.051 3
2
1 1,36
1 17 7 =2.101 5.0922
=2.534
n nLSD t MSE
. .i jy y LSD
1. 2.y y 3.4 2.534
1. 3.y y 8 2.534
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
4
2. 3.y y 4.6 2.534
Since all values of . .i jy y are greater than 2.534LSD , we conclude that, at 5% level
of significance, all aspects of judgement are different and the best one is the direct
experience( i.e lower scores are obtained through this aspect of judgement).
1.2
0 1 2 6: ...H
There is no difference in students’ mean performances.
1 : for at least one pair ( , ) and i jH i j i j
At least two two sudents perform differently from others.
Now, we calculate the test statistics.
2
2..
1 1
22 2 2 18132 = 526 594 ... 420
18
=65798
a b
Tij
i j
SSN
yy
26
2..
.1
22 2 2 2 2 2
1
1 18132 = 1590 1770 1374 1680 1344 1308
3 18
=63250
studentsi
i
SSb N
yy
23
2..
de t .
1
22 2 2
1
1 18132 = 3012 3090 2964
6 18
=1348
Aptitu est partsj
j
SSa N
yy
de t 65798 - (63250 1348)
1200
T students aptitu est partsSSE SS SS SS
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
5
( ) 6325012650
1 6 1
Treat students
students
SSMS
a
1200
120-1 -1 10
EE
SSMS
a b
F-Statistic for treatment (student performance) comparison:
12650105.4167
120
studentsMSFMSE
F-critical:
0.05,5,10 3.33F
Because the test statistic 105.4167F is greater than the critical value 0.05,5,10 3.33F ,
we reject the null hypothesis.
Conclusion:
At 5% significance level, there is enough evidence that students perform differently in
three portions of the SAT
0 1 2 3:H
There is no difference in trouble give to students by the three portions of SAT.
1 : for at least one pair ( , ) and i jH i j i j
At least two two portions of SAT give different trouble to students.
( ) 1348674
1 3 1
block aptitude
aptitude
SSMS
b
F-Statistic for block (three portions of the SAT) comparison:
( ) 6745.617
120
block aptitudeMSF
MSE
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
6
F-critical:
0.05,2,10 4.10F
Because the test statistic 5.617F is greater than the critical value 0.05,2,10 4.10F , we
reject the null hypothesis.
Conclusion:
At 5% significance level, there is enough evidence to indicate that the three portions
of the SAT do not give the students the same trouble.
Question 2 [50 marks]
2.1
Temperature (oC) 22 17 24 26 28 29
Number of passengers 173 149 175 188 186 198
2.1.1 Fit an ordinary least square (OLS) simple linear regression model for predicting number
of passengers throughout a particular hour given the temperature at the beginning of the hour.
[7]
Let
Model is
2.1.2 Without using Pearson (product moment) correlation formula, compute coefficient of
determination and interpret it. [5]
Meaning: About 86.14% of total variation in number of passengers ( is explained (accounted
for) by temperature ( .
2.1.3 Compute the standard error of the estimate. [3]
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
7
[OR using ]
Then
2.1.4 Construct the 90% prediction interval (P.I.) for number of passengers in an hour with a
beginning temperature of 25oC. [5]
Note: [Rounding appears incorrect but it is correct since these are number of passengers; we
round LPL down and UPL up regardless of value of the tenth digit]
2.2
Yes, a linear equation seems to a be good fit for the transformed data
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
8
log log logy x
Let0log , log , ,and logY y b b X x . Then the above transformed equation can
be rewritten as
Y bo bX
We use the same computational procedure as in 2.1.2 to obtain the coefficient of the regression model.
0
0.3047 , 0.1498 , 0.727, 0.481
0.14980.492
0.3047
0.481 0.492 0.727 0.128
0.1283 0.492
X XY
XY
X
SS SS X Y
SSb
SS
b Y bX
Y X
Thus,
0.1283 log and 0.492b
0.12830.1283 log 10 1.344
2.3
2.3.1
1 2 3 4 5 6ˆ 18.859 16.202 0.175 11.526 13.580 5.311x x x x x x
2.3.2
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
9
2 1
1
6541.41121 =1
95980.35226
=0.992
SSEn p
AdjRSST
n
Interpretation: Accounting for degrees of freedom, about 99.2% of the total variation in
1x̂ is mutually explained by 2x , 3x ,…,and 6x ).
2.3.3
Test for overall adequacy of the fitted model at 5% level?
Hypotheses
against
Observed F-value
Critical F-value
Decision Rule
Conclusion
Since , we highly reject and conclude that overall fitted
model is significantly adequate at 5% level.
2.3.4
against
Since the p-values associated with the regression coefficients are all less than 0.05,
we reject the null hypotheses and concluded that all regression coefficient are
statistically different from zero.
Therefore, we would not consider eliminating any variable.
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
10
Total: 90
Assignment 2
Question 1[35 marks]
1.1 [10]
Step 1:
The null and alternative hypotheses are
1 2 3 4 5 6: 0.12, 0.29, 0.11, 0.10, 0.14, 0.24OH p p p p p p
1H : At least one of the proportions is different from stated proportion.
Step 2
The significance level is 0.01
Step 3: Computation of the test statistic
In step 3, we have to compute the test statistic using the formula 2
2cal
o
e
fn
f
From this formula, we notice that we must find the expected frequencies ef as they are not
given.
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
11
We recall that e if np
Thus, the expected frequency for the first category is given by1 519 0.12 62.28f ,
the expected frequency for the second category is given by 2 519 0.29 150.51f
the expected frequency for the third category is given by 3 519 0.11 57.09f ,
the expected frequency for the first category is given by 4 519 0.1 51.9f ,
the expected frequency for the fifth category is given by 5 519 0.14 72.66f ,
and the expected frequency sixth category is given by 5 519 0.24 125.6f
22
cal
o
e
fn
f
2 2 2 2 2 22 88 135 52 40 76 128 519
62.28 150.51 57.09 51.9 72.66 125.6
=15.659
cal
Step 4: Select the critical value: 2
, 1k
There are six age categories. Hence, 6k and the number of degrees of freedom
is 1 6 1 5k . The critical value is 2
0.01,5 15.08627 .
Step 5: Formulation of the rejection rule
We reject 0H for a value of statistic greater than2
0.01,5 15.08627 . In our case, 2 =15.609
cal
is greater than 2
0.01,5 15.08627 . Hence, 0H is rejected.
Step 6: conclusion
At 1% significance level, we conclude that the data provide enough evidence to indicate that the
the distribution of customer ages in the Snoop report does not agree with that of the sample
report.
1.2 [15]
H0: The performance and the size of a company are statistically independent.
H1: The performance and the size of a company are associated
Computation of expected values ( iE ) and test statistic (2D ):
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
12
i
Column total row totalE
Grandtotal
22 i
i
OD n
E
Company size
Performance(% price change)
40% Total
Small 66(77.0) 90(85) 28(29.2) 39(31.9)
223
Medium 24(29.3) 33(32.4)
17(11.1) 11(12.2)
85
Large 55(38.7) 37(42.7)
10(14.7) 10(16.0)
112
Total 145 160 55 60 420
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
13
22
2 2 266 90 10 = .... 420
77 85 16
=19.09
i
i
OD n
E
2(0.05)
6 12.592X
Conclusion:
Because 2 19.09D >
2(0.05)
6 12.592X we reject Ho.
Thus our financial analyst has demonstrated that there is significant relationship between the
performance of firm and its size.
1.3 [10]
Test Hypotheses
)
Since we are testing adequacy of a discrete uniform model, , the expected
counts are simply for all . Then the observed and
expected counts are as follows.
Score
1 62 50
2 45 50
3 63 50
4 32 50
5 47 50
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
14
6 51 50
Total
(Note: This table is optional and it has no mark at all. However, if a candidate only gives this
table without explanation just above then will still be awarded the same 2 marks above.)
Observed
Critical
Decision Rule
Decision Making
Conclusion
At 1% significance level, we conclude that the data do not provide sufficient evidence to infer
that the die is not balanced (fair).
Question 2 [44 marks]
2.1 [12]
Year Quarter rate 4_MA(8marks) Seasonal ratio (4 marks)
2006 1 0.561
2 0.702
3 0.8 0.6595 121.304
4 0.568 0.66575 85.31731
2007 1 0.575 0.67875 84.71455
2 0.738 0.691875 106.6667
3 0.868 0.698875 124.1996
4 0.605 0.70125 86.27451
2008 1 0.594 0.683875 86.85798
2 0.738 0.665875 110.8316
3 0.729 0.66875 109.0093
4 0.6 0.6685 89.75318
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
15
2009 1 0.622 0.674375 92.23355
2 0.708 0.688 102.907
3 0.806 0.697375 115.5763
4 0.632 0.718625 87.94573
2010 1 0.665 0.742875 89.51708
2 0.835 0.756 110.4497
3 0.873
4 0.67
2.2 [10]
Q1 Q2 Q3 Q4 Total
2006 121.304 85.31731
2007 84.71455 106.6667 124.1996 86.27451
2008 86.85798 110.8316 109.0093 89.75318
2009 92.23355 102.907 115.5763 87.94573
2010 89.51708 110.4497
Seasonal median
88.18753 108.5582 118.4401 87.11012 402.296
100
400 = 0.994
402.296
kAdjustment factor
Median seasonal index
Quarter Seasonal median(SM) Seasonal index=
Q1 88.18753 87.68422
Q2 108.5582 107.9386
Q3 118.4401 117.7642
Q4 87.11012 86.61296
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
16
2.3 [6]
Year Quarter rate SI De-seasonalised index
2006 1 0.561 87.68422 0.639796
2 0.702 107.9386 0.65037
3 0.8 117.7642 0.679324
4 0.568 86.61296 0.655791
2007 1 0.575 87.68422 0.655762
2 0.738 107.9386 0.683722
3 0.868 117.7642 0.737066
4 0.605 86.61296 0.69851
2008 1 0.594 87.68422 0.677431
2 0.738 107.9386 0.683722
3 0.729 117.7642 0.619034
4 0.6 86.61296 0.692737
2009 1 0.622 87.68422 0.709364
2 0.708 107.9386 0.655928
3 0.806 117.7642 0.684419
4 0.632 86.61296 0.729683
2010 1 0.665 87.68422 0.758403
2 0.835 107.9386 0.773588
3 0.873 117.7642 0.741312
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
17
4 0.67 86.61296 0.773556
Interpretation of the first de-seasonilised index (0.6398):
2006 quarter 1 occupancy rate would have been higher at 0.6398, instead of the actual sales of
0.561, had seasonal influences not been present.
2.4 [10]
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
18
Trend Line formula: abxy
2
6.9770.00263
2660
13.889int 0.6945
20
xySlope b
x
yy ercept a
n
Trend Line: 0.00263 0.6945y x
2.5 [2]
Quarter
1 0.561 -19 -10.659 361
2 0.702 -17 -11.934 289
3 0.8 -15 -12 225
4 0.568 -13 -7.384 169
1 0.575 -11 -6.325 121
2 0.738 -9 -6.642 81
3 0.868 -7 -6.076 49
4 0.605 -5 -3.025 25
1 0.594 -3 -1.782 9
2 0.738 -1 -0.738 1
3 0.729 1 0.729 1
4 0.6 3 1.8 9
1 0.622 5 3.11 25
2 0.708 7 4.956 49
3 0.806 9 7.254 81
4 0.632 11 6.952 121
1 0.665 13 8.645 169
2 0.835 15 12.525 225
3 0.873 17 14.841 289
4 0.67 19 12.73 361
Tot 13.889y 0x 6.977xy 2 2660x
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
19
Estimate the trend value of the time series for Quarter 3 in 2011. .
Quarter 3 in 2011: ˆ 0.00263(25 0.6945 0.76T y
2.6 [4]
The seasonally-adjusted trend estimate is calculated usingT S .
where the S is the seasonal index for the specific period (Q3)
Thus, 117.7642
0.76 0.895100
T S .
Question 3 [6 marks]
Year Quarter Sales
Exponentially smoothed sales
2007 1 18 18.000
2 22 18.400
3 27 19.260
4 31 20.434
2008 1 33 21.691
2 20 21.522
3 38 23.169
4 26 23.452
2009 1 25 23.607
2 36 24.846
3 44 26.762
4 29 26.986
Question 4[15 marks]
4.1.
Price relative for food= 1
0
45100% 100% 112.5%
40
p
p
This shows that since 1981, the price of food has increased by 12. 5%.
Price relative for accommodation = 1
0
40100% 100% 133.33%
30
p
p
This shows that the average price of accommodation has increased by 33.33% since 1981.
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
20
Price relative for petrol = 1
0
1.40100% 100% 107.69%
1.30
p
p
This shows that the average price of petrol has increased by 7.69% since 1981.
4.2
Quantity relative for food= 1
0
25100% 100% 83.33%
30
p
p
This shows that since 1981, the food consumed by typical family of four has decreased by
16.67%.
Quantity relative for accommodation = 1
0
5100% 100% 62.5%
8
p
p
This shows that the consumption of accommodation units has decreased by 37.5% since 1981.
Quantity relative for petrol = 1
0
100% 75 100% 80.00%p
p
This shows that the number of units of petrol consumed by a typical family of four has
decreased by 20% since 1981.
4.3-4.5
Quantities Prices
Item
1981
1982
1981
1982
Food 30 25 40 45 1350 1200 1125
Accommodation
8 5 30 40 320 240 200
Petrol 75 60 1.3 1.4 105 97.5 84
Entertainment
10 8 10 12 120 100 96
Total
5 1505
TUTORIAL LETTER MEMO
SEMESTER 2/2020
STATISTICS FOR ECONOMISTS 2B
SFE612S
21
4.3.
Laspeyres price index = 1 0
0 0
1895100% 100% 115.725%
1637.5
p q
p q
If quantities (number of units consumed) are held constant at 1981(base period) levels, the
price of the number of units consumed by a typical family has increased by, on average
15.725% since 1981.
[4]
4.4
Paasche’s quantity index= 1 1
1 0
100% 100% 79.42%p q
p q
If prices are held constant at current period levels (1982), the quantities (the number of units
consumed by a typical family) would have increased by 20.58%. [4]
4.5 [1]
unweighted composite (aggregate) price index
Total:100
Statistics for Economists 2B SFE612SSFE612S_2020_FTL MEMO_Final (003)