Lecture No.

10
 Analysis of Method of Least Square

 Detrending

 Analyzing the seasonal variations.

 Analysis of Ratio-to-Moving-Average Method

 Analysis of Ratio-to-Trend Method

 Analysis of Link-Relative Method

 This method is most frequently used for the computation of seasonal index
numbers.
 The first step is to eliminate the trend component. To this end, divide the
original observations for each month or quarter by the corresponding 12-
month or 4-quarter centered moving average and express the result as a
percentage, i.e. compute the seasonal relative for each month or quarter.
 It is worthwhile to note that each monthly or quarterly value is assumed to
consist of the product of the effects of T, C, S and I components, and each
moving average is a measure of the combined effect of trend and cyclical
components, i.e. T  C.
 Dividing the original data by the corresponding moving averages and then
multiplying by 100, an estimate of the effects of seasonal and irregular
components combined is obtained; i.e.
 original data TCSI
100  100  SI (Seasonal relatives)
moving average TC
 The next step is to remove the effects of the irregular component in order
to obtain seasonal indices. To achieve this end, arrange the seasonal
relatives by months or quarters and find the monthly or quarterly averages,
using either the mean or the median.
 If mean is to be used, compute a modified mean by discarding the
unusually large or small seasonal relative under each month or quarter so
that the average is not distorted.
 If these monthly or quarterly averages do not average to 100, then adjust
them by multiplying each median or modifies mean estimate of seasonal
index by the correction factor that will make the average of all the indices
equal 100.
Compute the seasonal indices for the four quarters by the method of ratio-to-
moving averages.

Year Quarters

I II III IV

1949 105 77 68 95

1950 107 83 74 106

1951 117 99 86 112

Year Y- 4-quarter 4-quarter 4-quarter Seasonal
Values moving centered centered moving relatives
totals moving averages
total
1949-I 105
II 77
345
III 68 347 692 86.5 78.6
IV 95 353 700 87.5 108.6
359 712 89.0 120.2
1950-I 107
370 729 91.1 91.1
II 83 380 750 93.8 78.9
III 74 396 776 97.0 109.3
408 804 100.5 116.4
IV 106
414 822 102.8 96.3
1951-I 117
II 98
III 86
IV 112
 Seasonal Indices

Quarters Total
Year I II III IV
1949 78.6 108.6
1950 120.2 91.1 78.9 109.3
1951 116.4 96.3
Total 236.6 187.4 157.5 217.9
Mean 118.30 93.70 78.75 108.95 399.7
S.I
 In this method, the trend values are obtained for each time period by fitting
a least-squares trend line either to the observed time series data or to the
annual averages.
 The rest of the computational procedure is the same as that of the ratio-to-
moving-average method as the seasonal index computed by it includes
cyclical and irregular variations.
Compute seasonal indices for the four quarters by the ratio-to-trend method
for the following data.

Year Quarters

I II III IV
1961 122 125 118 117

1962 119 114 114 109

1963 105 99 93 89

1964 86 80 83 84
Year Y-Values X XY Trend
X2
Yt  103.56  1.60 X

1961-I 122 -15 225 -1830 127.56

II 125 -13 169 -1625 124.36
III 118 -11 121 -1298 121.16
IV 117 -9 81 -1053 117.96
1962-I 119 -7 49 -833 114.76
II 114 -5 25 -570 111.56
III 114 -3 9 -342 108.36
IV 109 -1 1 -109 105.16
1963-I 105 1 1 105 101.96
II 99 3 9 297 98.76
III 93 5 25 465 95.56
IV 89 7 49 623 92.36
1964-I 86 9 81 774 89.16
II 80 11 121 880 85.96
III 83 13 169 1079 82.76
IV 84 15 225 1260 79.56
The Equation of the straight line:

Yt  a  bX
Two Normal Equations:

a
 Y

1657
 103.56
n 16

b
 XY 2177
  1.60
 X 1360
2

Yt  103.56  1.60 X
 Seasonal Indices

Quarters Total
Year
I II III IV
1961 95.64 100.51 97.39 99.19
1962 103.69 102.19 105.20 103.65
1963 102.98 100.24 97.32 96.36
1964 96.46 93.07 100.29 105.58
Total 398.77 302.94 295.00 404.78
Mean 99.69 100.98 98.33 101.20 400.20

S.I 99.64 100.93 98.28 101.15 400

 Compute the link-relatives by expressing each monthly or quarterly value
as a percentage of the preceding monthly or quarterly value.
 Arrange the link-relatives by months or quarters and find an appropriate
average of these relatives for each month or quarter.
 Convert the average (median or mean) relatives into a series of chain
relatives by setting the value of January or the first quarter as 100, and
carrying the process to include the first unit of the next period.
 A discrepancy due to trend increment positive an negative) exists between
the chain relative for the first January or quarter and that for the next
period. Adjust the chain relatives for the trend component by subtracting
one-twelfth of the discrepancy, from the value of February, two-twelfth
from the value of March and so on or by one-fourth of the discrepancy
from the relative of second quarter, two-fourth from the third quarter
relative and three fourth from the fourth quarter relative.
 To obtain the seasonal indices, reduce the adjusted chain relatives to the
same level as January or the first quarter by multiplying each of the
adjusted chain relatives by the correction factor that will make the average
of the indices equal 100.
Obtain the seasonal indices for the following data by using the link-relative
method.

Year Quarters

I II III IV

1948 71 89 106 78

1949 71 90 108 79

1950 73 91 111 81

1951 76 97 122 89
 Link-Relatives

Year Quarters

I II III IV

1948 125.4 119.1 73.6

1949 91.0 126.8 120.0 73.1

1950 92.4 124.7 122.0 73.0

1951 93.8 127.6 125.8 73.0

Median 92.4 126.1 121.0 73.1

 Chain relatives

Quarter I II III IV I
Chain Relative 100 126.1 152.6 111.6 103.1

The chain relative for the first quarter works out to be 103.1 which as a matter
of fact, ought to have been 100. This increase of 3.1 is due to the trend
component present in the data. An adjustment for the trend therefore become
necessary. Since the difference is positive so we subtract one-fourth of this
form the second quarter, two-fourth from the third quarter and three-fourth
from the fourth quarter.
 Seasonal Indices

Quarter I II III IV Total

Adjusted Chain 100 125.32 151.05 109.28 485.65
Relative
Seasonal Index 82.4 103.2 124.4 90.0 400.0