REGRESSION ANALYSIS 337

correlation analysis. The sign of correlation coefficient indicates the nature (direct or inverse) of
relationship between two variables, while the absolute value of correlation coefficient indicates
the extent of relationship.
2. Correlation analysis determines an association between two variables x and y but not that they
have a cause-and-effect relationship. Regression analysis, in contrast to correlation, determines
the cause-and-effect relationship between x and y, that is, a change in the value of independent
variable x causes a corresponding change (effect) in the value of dependent variable y if all other
factors that affect y remain unchanged.
3. In linear regression analysis one variable is considered as dependent variable and other as
independent variable, while in correlation analysis both variables are considered to be
independent.
4. The coefficient of determination r2 indicates the proportion of total variance in the dependent variable that
is explained or accounted for by the variation in the independent variable. Since value of r2 is determined
from a sample, its value is subject to sampling error. Even if the value of r2 is high, the assumption
of a linear regression may be incorrect because it may represent a portion of the relationship that
actually is in the form of a curve.

10.2 ADVANTAGES OF REGRESSION ANALYSIS

The following are some important advantages of regression analysis:
1. Regression analysis helps in developing a regression equation by which the value of a dependent
variable can be estimated given a value of an independent variable.
2. Regression analysis helps to determine standard error of estimate to measure the variability or
spread of values of a dependent variable with respect to the regression line. Smaller the variance
and error of estimate, the closer the pair of values (x, y) fall about the regression line and better
the line fits the data, that is, a good estimate can be made of the value of variable y. When all the
points fall on the line, the standard error of estimate equals zero.
3. When the sample size is large (df ≥ 29), the interval estimation for predicting the value of a
dependent variable based on standard error of estimate is considered to be acceptable by changing
the values of either x or y. The magnitude of r2 remains the same regardless of the values of the
two variables.

Remarks
1. The relationship between the dependent variable y and independent variable x exists and is
linear. The average relationship between x and y can be described by a simple linear regression
equation y = a + bx + e, where e is the deviation of a particular value of y from its expected value
for a given value of independent variable x.
2. For every value of the independent variable x, there is an expected (or mean) value of the dependent
variable y and these values are normally distributed. The mean of these normally distributed
values fall on the line of regression.
3. The dependent variable y is a continuous random variable, whereas values of the independent
variable x are fixed values and are not random.
4. The sampling error associated with the expected value of the dependent variable y is assumed to
be an independent random variable distributed normally with mean zero and constant standard
deviation. The errors are not related with each other in successive observations.
5. The standard deviation and variance of expected values of the dependent variable y about the
regression line are constant for all values of the independent variable x within the range of the
sample data.
338 BUSINESS S T A T I S T I C S

6. The value of the dependent variable cannot be estimated for a value of an independent variable
lying outside the range of values in the sample data.

10.3 PARAMETERS OF SIMPLE LINEAR REGRESSION MODEL

The fundamental aim of regression analysis is to determine a regression equation (line) that makes
sense and fits the representative data such that the error of variance is as small as possible. This
implies that the regression equation should adequately be used for prediction. J. R. Stockton stated
that
• The device used for estimating the values of one variable from the value of the other consists of a line
through the points, drawn in such a manner as to represent the average relationship between the two
variables. Such a line is called line of regression.
The two variables x and y which are correlated can be expressed in terms of each other in the
form of straight line equations called regression equations. Such lines should be able to provide the best
fit of sample data to the population data. The algebraic expression of regression lines is written as:
• The regression equation of y on x
y = a + bx
is used for estimating the value of y for given values of x.
• Regression equation of x on y
x = c + dy
is used for estimating the value of x for given values of y.

Remarks
1. When variables x and y are correlated perfectly (either positive or negative) these lines coincide, that is, we
have only one line.
2. Higher the degree of correlation, nearer the two regression lines are to each other.
3. Lesser the degree of correlation, more the two regression lines are away from each other. That is, when r
= 0, the two lines are at right angle to each other.
4. Two linear regression lines intersect each other at the point of the average value of variables x and y.

10.3.1 Regression Coefficients

To estimate values of population parameter β0 and β1, under certain assumptions, the fitted or estimated
regression equation representing the straight line regression model is written as:
ŷ = a + bx
where ŷ = estimated average (mean) value of dependent variable y for a given value of independent
variable x.
a or b0 = y-intercept that represents average value of ŷ
b = slope of regression line that represents the expected change in the value of y for unit
change in the value of x

To determine the value of ŷ for a given value of x, this equation requires the determination of two
unknown constants a (intercept) and b (also called regression coefficient). Once these constants are
calculated, the regression line can be used to compute an estimated value of the dependent variable
y for a given value of independent variable x.
 REGRESSION ANALYSIS 339

The particular values of a and b define a specific linear relationship between x and y based on
sample data. The coefficient ‘a’ represents the level of fitted line (i.e., the distance of the line above or
below the origin) when x equals zero, whereas coefficient ‘b’ represents the slope of the line (a measure
of the change in the estimated value of y for a one-unit change in x).
The regression coefficient ‘b’ is also denoted as:
• byx (regression coefficient of y on x) in the regression line, y = a + bx
• bxy (regression coefficient of x on y) in the regression line, x = c + dy

Properties of regression coefficients

1. The correlation coefficient is the geometric mean of two regression coefficients, that is, r = byx × bxy .
2. If one regression coefficient is greater than one, then other regression coefficient must be less than one,
because the value of correlation coefficient r cannot exceed one. However, both the regression coefficients
may be less than one.
3. Both regression coefficients must have the same sign (either positive or negative). This property rules out
the case of opposite sign of two regression coefficients.
4. The correlation coefficient will have the same sign (either positive or negative) as that of the two regression
coefficients. For example, if by x = – 0.664 and bxy = – 0.234, then r = – 0.664 × 0.234 = – 0.394.
5. The arithmetic mean of regression coefficients bxy and byx is more than or equal to the correlation coefficient
r, that is, (by x + bx y ) / 2 ≥ r. For example, if byx = – 0.664 and bx y = – 0.234, then the arithmetic mean of
these two values is (– 0.664 – 0.234)/2 = – 0.449, and this value is more than the value of r = – 0.394.
6. Regression coefficients are independent of origin but not of scale.

10.4 METHODS TO DETERMINE REGRESSION COEFFICIENTS

Following are the methods to determine the parameters of a fitted regression equation:

10.4.1 Least Squares Normal Equations

Let ŷ = a + bx be the least squares line of y on x, where ŷ is the estimated average value of dependent
variable y. The line that minimizes the sum of squares of the deviations of the observed values of y
from those predicted is the best fitting line. Thus the sum of residuals for any least-square line is
minimum, where

L = Σ ( y − ˆy)2 = Σ {y – (a + bx)}2; a, b = constants

Differentiating L with respect to a and b and equating to zero, we have
∂L
= – 2 Σ{y – (a + bx)} = 0
∂a
∂S
= – 2 Σ{y – (a + bx)}x = 0
∂b
Solving these two equations, we get the same set of equations as equations (10.1)
Σ y = na + bΣ x (10.1)
Σ xy = aΣx + bΣ x2
340 BUSINESS S T A T I S T I C S

where n is the total number of pairs of values of x and y in a sample data. The equations (10.1) are
called normal equations with respect to the regression line of y on x. After solving these equations for
a and b, the values of a and b are substituted in the regression equation, y = a + bx.
Similarly if we have a least squares line x̂ = c + dy of x on y, where x̂ is the estimated mean value
of dependent variable x, then the normal equations will be
Σ x = nc + d Σ y
Σ xy = n Σ y + d Σ y2
These equations are solved in the same manner as described above for constants c and d. The values
of these constants are substituted to the regression equation x = c + dy.

Alternative method to calculate value of constants

Instead of using the algebraic method to calculate values of a and b, we may directly use the results of
the solutions of these normal equation.
The gradient ‘b’ (regression coefficient of y on x) and ‘d’ (regression coefficient of x on y) are
calculated as:
Sx y n n 1 n n
b = , where Sx y = ∑ ( xi − x ) ( yi − y ) = ∑ xi yi − ∑ xi ∑ yi
S xx i =1 i =1 n i =1 i =1

2
n 2 2
n 1  n 
Sxx = ∑ ( xi − x ) = ∑ xi −  ∑ xi 
i =1 i =1 n  i =1 

2
S yx n 2 n 1 n 
Syy = ∑ ( yi − y ) = ∑ yi −
2
and d = , where  ∑ y
S yy i =1 i =1 n  i =1 
Since the regression line passes through the point ( x , y ), the mean values of x and y and the
regression equations can be used to find the value of constants a and c as follows:
a = y − bx for regression equation of y on x
c = x – d y for regression equation of x on y
The calculated values of a, b and c, d are substituted in the regression line y = a + bx and
x = c + dy respectively to determine the exact relationship.
Example 10.1: Use least squares regression line to estimate the increase in sales revenue expected
from an increase of 7.5 per cent in advertising expenditure.

Firm Annual Percentage Increase Annual Percentage Increase

in Advertising Expenditure in Sales Revenue
A 1 1
B 3 2
C 4 2
D 6 4
E 8 6
F 9 8
G 11 8
H 14 9
 REGRESSION ANALYSIS 341

Solution: Assume sales revenue ( y) is dependent on advertising expenditure (x). Calculations for
regression line using following normal equations are shown in Table 10.1
Σ y = na + bΣ x and Σ x y = a Σ x + bΣ x2

Table 10.1 Calculation for Normal Equations

Sales Revenue Advertising x2 xy

y Expenditure, x
1 1 1 1
2 3 9 6
2 4 16 8
4 6 36 24
6 8 64 48
8 9 81 72
8 11 121 88
9 14 196 126
40 56 524 373

Approach 1 (Normal Equations):

Σ y = na + bΣ x or 40 = 8a + 56b
2
Σ xy = aΣ x + bΣ x or 373 = 56a + 524b
Solving these equations, we get a = 0.072 and b = 0.704. Substituting these values in the regression
equation
y = a + bx = 0.072 + 0.704x
For x = 7.5% or 0.075 increase in advertising expenditure, the estimated increase in sales revenue
will be y= 0.072 + 0.704 (0.075) = 0.1248 or 12.48%
Approach 2 (Short-cut method):
S xy 93
b = = = 0.704,
S xx 132

Σ xΣ y 40 × 56
where Sxy = Σ x y – = 373 – = 93
n 8

(Σ x)2 (56)2
Sxx = Σ x2 – = 524 – = 132
n 8
The intercept ‘a’ on the y-axis is calculated as:
40 56
a = y − bx = – 0.704 × = 5 – 0.704×7 = 0.072
8 8
Substituting the values of a = 0.072 and b = 0.704 in the regression equation, we get
y = a + bx = 0.072 + 0.704 x
For x = 0.075, we have y = 0.072 + 0.704 (0.075) = 0.1248 or 12.48%.
342 BUSINESS S T A T I S T I C S

Example 10.2: The owner of a small garment shop is hopeful that his sales are rising significantly
week by week. Treating the sales for the previous six weeks as a typical example of this rising trend,
he recorded them in Rs. 1000’s and analysed the results.
Week : 1 2 3 4 5 6
Sales : 2.69 2.62 2.80 2.70 2.75 2.81
Fit a linear regression equation to suggest to him the weekly rate at which his sales are rising and
use this equation to estimate expected sales for the 7th week.
Solution: Assume sales ( y) is dependent on weeks (x). Then the normal equations for regression
equation: y = a + bx are written as:
Σ y = na + bΣ x and Σxy = aΣ x + bΣx2
Calculations for sales during various weeks are shown in Table 10.2.

Table 10.2 Calculations of Normal Equations

Week (x) Sales ( y) x2 xy

1 2.69 1 2.69
2 2.62 4 5.24
3 2.80 9 8.40
4 2.70 16 10.80
5 2.75 25 13.75
6 2.81 36 16.86
21 16.37 91 57.74

The gradient ‘b’ is calculated as:

S xy 0.445 ΣxΣ y 21 × 16.37
b= = = 0.025; Sxy = Σ xy – = 57.74 – = 0.445
S xx 17.5 n 6

(∑ x)2 (21)2
Sxx = ∑ x 2 − = 91 – = 17.5
n 6
The intercept ‘a’ on the y-axis is calculated as
16.37 21
a = y − bx = − 0.025 ×
6 6
= 2.728 – 0.025×3.5 = 2.64
Substituting the values a = 2.64 and b = 0.025 in the regression equation, we have
y = a + bx = 2.64 + 0.025x
For x = 7, we have y = 2.64 + 0.025 (7) = 2.815
Hence the expected sales during the 7th week is likely to be Rs. 2.815 (in Rs. 1000’s).

10.4.2 Deviations Method

Calculations to least squares normal equations become lengthy and tedious when values of x and y are
large. Thus the following two methods may be used to reduce the computational time.
 REGRESSION ANALYSIS 343

(a) Deviations Taken from Actual Mean Values of x and y If deviations of actual values of variables x and
y are taken from their mean values x and y , then the regression equations can be written as:
• Regression equation of y on x • Regression equation of x on y
y – y = byx (x – x ) x – x = bx y (y – y )
where by x = regression coefficient of where bx y = regression coefficient
= y on x. where bxy = of x on y.
The value of byx can be calculated using The value of bx y can be calculated
the using the formula formula
Σ (x − x ) ( y − y ) Σ (x − x ) ( y − y )
by x = bx y =
Σ ( x − x )2 Σ ( y − y )2

(b) Deviations Taken from Assumed Mean Values for x and y If mean value of either x or y or both are in
fractions, then we must prefer to take deviations of actual values of variables x and y from their
assumed means.
• Regression equation of y on x • Regression equation of x on y
y – y = byx (x – x ) x – x = bxy (y – y )
n Σ dx dy − (Σ dx ) (Σ d y ) n Σ dx dy − (Σ dx )(Σ dy )
where by x = where bx y =
nΣ dx2 2
− (Σ dx ) n Σ dy2 − (Σ dy )2
n = number of observations n = number of observations
dx = x – A; A is assumed mean of x dx = x – A; A is assumed mean
dx = = of x
dy = y – B; B is assumed mean of y dy = y – B; B is assumed mean
of y
(c) Regression Coefficients in Terms of Correlation Coefficient If deviations are taken from actual mean
values, then the values of regression coefficients can be alternatively calculated as follows:
Σ (x − x ) ( y − y ) Σ (x − x ) ( y − y )
byx = 2
bxy =
Σ (x − x ) Σ ( y − y )2
Covariance(x, y) σy Covariance(x, y) σx
= = r⋅ = = r⋅
σ2x σx σ2y σy

Example 10.3: Compute the two regression coefficients using the value of actual mean value of X and
Y from the data given below and then work out the value of r.
X : 7 4 8 6 5
Y : 6 5 9 8 2 [GJU (Hisar), BBA, 2004]
Solution: Calculations for two regression coefficients are given below:
X Y x=X – X y=Y– Y xy x2 y2
7 6 +1 0 0 1 0
4 5 –2 –1 2 4 1
8 9 +2 +3 6 4 9
6 8 0 +2 0 0 4
5 2 –1 –4 4 1 16

30 30 Σx = 0 Σy = 0 Σxy = 12 Σx2 =10 Σy2 = 30

344 BUSINESS S T A T I S T I C S

ΣX 30 ΣY ΣY
X = = =6 and Y = = =6
n 5 n 5
Σ xy 12
Regression coefficient of X on Y is: bxy = = = 0.4
Σ y2 30

Σ xy 12
Regression coefficient of Y on X: byx = = = 1.2
Σ x2 10

Co-relation coefficient, r = bxy × byx = 1.2 × 0.4 = 0.48 = 0.69.

Example 10.4: Use following data to find out the two lines of regression and compute the Karl
Pearson’s coefficient of correlation.
Σ x = 250, Σ y = 300, Σ x y = 7900, Σ x2 = 6500, Σ y2 = 10000, n = 10
Solution: Calculate x , y , bxy and byx with the help of given information as follows:
Σx 250 Σy 300
x = = = 25; y = = = 30
n 10 n 10
nΣ x y − Σ xΣ y 10(7900) − (250) (300)
bxy = 2 2
= 2
= 0.4
n Σ y − (Σ y) 10(10000) − (300)
Let the regression line of x on y be:
x − x = bxy ( y − y )
x – 25 = 0.4( y – 30)
x = 0.4y – 12 + 25 = 0.4y + 13
n Σ x y − Σ xΣ y 10(7900) − (250) (300)
Also byx = 2 2 = = 1.6
n Σ x − (Σ x) 10(6500) − (250)2
Let the regression line of y on x be
y − y = byx ( x − x )
y – 30 = 1.6(x – 25)
y = 1.6x – 40 + 30 = 1.6x – 10

Hence, the coefficient of correlation is: r = bxy × byx = 0.4 × 1.6 = 0.8.

Example 10.5: A departmental store gives training to its salesman which is followed by a test. The
store is considering whether it should terminate the service of any salesman who does not do well in
the test. The following data gives the scores and sales made by nine salesmen during a certain period:
Test Scores : 14 19 24 21 26 22 15 20 19
Sales (‘00 Rs.) : 31 36 48 37 50 45 33 41 39
Calculate the correlation coefficient between test scores and the sales. Does it indicate that the
termination of services of low test scores is justified? If the firm wants a minimum sales volume of Rs.
3000, what is the minimum test score that will ensure continuation of service? Also estimate the most
probable sales volume of a sales making a score of 28. [Delhi Univ., B.Com(Hons), 1998, 2002]
Solution: Let test scores be x and sales be y. Calculation required to determine co-relation coeffi-
cient are shown below:
 REGRESSION ANALYSIS 345

X dx = X – 20 dx2 Y dy = Y – 40 dy2 dx dy

14 –6 36 31 –9 81 54
19 –1 1 36 –4 16 4
24 4 16 48 8 64 32
21 1 1 37 –3 9 –3
26 6 36 50 10 100 60
22 2 4 45 5 25 10
15 –5 25 33 –7 49 35
20 0 0 41 1 1 0
19 –1 1 39 –1 1 1
180 120 360 346 193

Σ x 180 Σ y 360
x = = = 20 and y = = = 40
n 9 n 9
Σxy 193
r= = = 0.9476
2
Σx × Σy2 120 × 346
The value of r shown that there is a high degree of correlation between test scores (x) and sales (y).
This implies that the persons having low test scores will not be able to make good sales hence termi-
nation is justified.
Regression equation of X on Y is:
Σ dx dy − (Σ dx ) (Σdy ) 193
X – X = bxy ( Y − Y ); where bxy = = = 0.557
Σd2y − (Σdy )2 346
x – 20 = 0.557 (y – 40)
x = 0.557 y – 22.28 + 20 = 0.557 Y – 2.28
When sales is Rs. 3000, i.e. y = 30, then x = 0.557 (30) – 2.28 = 14.43
Hence, test score = 14.43 when sales volume is Rs. 3000.
Regression equation of Y on X:
Σxy 193
y − y = byx ( x − x ) ; where byx = = =1.6
Σx 2 120
∴ y – 40 = 1.6 (x – 20)
y = 1.6x – 32 + 40 = 1.6x + 8
When test score is 28, i.e. x = 28, then y= 1.6(28) + 8 = 52.8
Thus sales volume = 52.8 × 100 = Rs. 5280.
Example 10.6: The following data relate to the scores obtained by 9 salesmen of a company in an
intelligence test and their weekly sales (in Rs. 1000’s)

Salesmen : A B C D E F G H I
Test scores : 50 60 50 60 80 50 80 40 70
Weekly sales : 30 60 40 50 60 30 70 50 60
(a) Obtain the regression equation of sales on intelligence test scores of the salesmen.
(b) If the intelligence test score of a salesman is 65, what would be his expected weekly sales.
[HP Univ., MCom, 1996]
Solution: Assume weekly sales (y) as dependent variable and test scores (x) as independent variable.
Calculations for the following regression equation are shown in Table 10.3.
y – y = by x (x – x )
346 BUSINESS S T A T I S T I C S

Table 10.3 Calculation for Regression Equation

Weekly dx = x – 60 dx2 Test dy = y – 50 dy2 dxdy

Sales, x Score, y
50 – 10 100 30 – 20 400 200
60 0 0 60 10 100 0
50 – 10 100 40 – 10 100 100
60 0 0 50 0 0 0
80 20 400 60 10 100 200
50 – 10 100 30 – 20 400 200
80 20 400 70 20 400 400
40 – 20 400 50 0 0 0
70 10 100 60 10 100 100
540 0 1600 450 0 1600 1200

Σx 540 Σy 450
(a) x = = = 60; y = = = 50
n 9 n 9

Σ dx dy − (Σ dx )(Σ d y ) 1200
byx = = = 0.75
Σ dx2 − (Σ dx )2
1600

Substituting values in the regression equation, we have

y – 50 = 0.75 (x – 60) or y = 5 + 0.75x
For test score x = 65 of salesman, we have
y = 5 + 0.75 (65) = 53.75
Hence we conclude that the weekly sales is expected to be Rs. 53.75 (in Rs. 1000’s) for a test score of 65.
Example 10.7: A company is introducing a job evaluation scheme in which all jobs are graded by
points for skill, responsibility, and so on. Monthly pay scales (Rs. in 1000’s) are then drawn up
according to the number of points allocated and other factors such as experience and local conditions.
To date the company has applied this scheme to 9 jobs.

Job : A B C D E F G H I
Points : 5 25 7 19 10 12 15 28 16
Pay (Rs.) : 3.0 5.0 3.25 6.5 5.5 5.6 6.0 7.2 6.1

(a) Find the least squares regression line for linking pay scales to points.
(b) Estimate the monthly pay for a job graded by 20 points.
Solution: Assume monthly pay (y) as the dependent variable and job grade points (x) as the independent
variable. Calculations for the following regression equation are shown in Table 10.4.
y – y = byx (x – x )
 REGRESSION ANALYSIS 347

Table 10.4 Calculations for Regression Equation

Grade dx = x – 15 dx2 Pay Scale, y dy = y – 5 dy2 dx dy

Points, x
5 – 10 100 3.0 – 2.0 4 20
25 10 100 5.0 ← B 0 0 0
7 –8 64 3.25 – 1.75 3.06 14
19 4 16 6.5 1.50 2.25 6
10 –5 25 5.5 0.50 0.25 – 2.5
12 –3 9 5.6 0.60 0.36 – 1.8
15 ← A 0 0 6.0 1.00 1.00 0
28 13 169 7.2 2.2 4.84 28.6
16 1 1 6.1 1.1 1.21 1.1
137 2 484 48.15 3.15 16.97 65.40

Σx 137 Σy 48.15
(a) x = = = 15.22; y = = = 5.35
n 9 n 9
Since mean values x and y are non-integer value, therefore deviations are taken from assumed mean
as shown in Table 10.4.
n Σ dx d y − (Σ dx ) (Σ d y ) 9 × 65.40 − 2 × 3.15 582.3
byx = = = = 0.133
n Σ dx2 − (Σ dx ) 2
9 × 484 − (2)2 4352
Substituting values in the regression equation, we have
y – y = byx (x – x ) or y – 5.35 = 0.133 (x – 15.22) = 3.326 + 0.133x
(b) For job grade point x=20, the estimated average pay scale is given by
y = 3.326 + 0.133x = 3.326 + 0.133 (20) = 5.986
Hence, likely monthly pay for a job with grade points 20 is Rs. 5986.

Example 10.8: Find two regression equations from the data given below:
x : 57 58 59 59 60 61 62 64
y : 77 78 75 78 82 82 79 81 [GJU (Hisar), BBA, 2005]
Solution: Calculations of two regression equations as shown below:

x y dx = x – 59 dY = y – 82 dx dy dx2 dy2

57 77 –2 –5 +10 4 25
58 78 –1 –4 +4 1 16
59 75 0 –7 0 0 49
59 78 0 –4 0 0 16
60 82 +1 0 0 1 0
61 82 +2 0 0 4 0
62 79 +3 –3 –9 9 9
64 81 +5 –1 –5 25 1

480 632 8 –24 0 44 116

348 BUSINESS S T A T I S T I C S

Σx 480 Σy 632
x = = = 60 and y = = = 79
n 8 n 8

nΣ dx dy − Σ dx Σ dy 8(0) − (8)(−24) 192

We know that bxy = = = = 0.55
N Σd2y − (Σd y ) 2
8(116) − (−24)2 352

nΣ dx d y − Σ dx Σ d y 8(0) − (8)(−24) 192

byx = = = = 0.67
N Σ2x − (Σ dx )2
8(44) − (8)2 288

Regression equation of x on y: x − x = bxy ( y − y )

x – 60 = 0.55( y − y )
x = 0.55y – 43.45 + 60 = 0.55y + 16.55
Regression equation of y on x: y − y = byx ( x − x )
y – 79 = 0.67 (x – 60)
y = 0.67 x – 40.2 + 79 = 0.67 x + 38.8.
Example 10.9: The following data give the ages and blood pressure of 10 women:
Age : 156 142 136 147 149 142 160 172 163 155
Blood pressure : 147 125 118 128 145 140 155 160 149 150
(a) Find the correlation coefficient between age and blood pressure.
(b) Determine the least squares regression equation of blood pressure on age.
(c) Estimate the blood pressure of a woman whose age is 45 years.

Solution: Assume blood pressure (y) as the dependent variable and age (x) as the independent variable.
Calculations for regression equation of blood pressure on age are shown in Table 10.5.
Table 10.5 Calculations for Regression Equation

Age, x dx = x – 49 dx2 Blood, y dy = y – 145 dy2 dx dy

56 7 49 147 2 4 14
42 –7 49 125 – 20 400 140
36 – 13 169 118 – 27 729 351
47 –2 4 128 – 17 289 34
49 ← A 0 0 145 ← B 0 0 0
42 –7 49 140 –5 25 35
60 11 121 155 10 100 110
72 23 529 160 15 225 345
63 14 196 149 4 16 56
55 6 36 150 5 25 30
522 32 1202 1417 – 33 1813 1115

(a) Coefficient of correlation between age and blood pressure is given by

n Σ dx dy − Σ dxΣ dy
r=
n Σ dx2 − (Σ dx )2 n Σ dy2 − (Σ dy )2
 REGRESSION ANALYSIS 349

10(1115) − (32) (− 33)

=
10(1202) − (32)2 10(1813) − (− 33)2

11150 + 1056 12206

= = = 0.892
12020 − 1024 18130 − 1089 13689

We may conclude that there is a high degree of positive correlation between age and blood pressure.
(b) The regression equation of blood pressure on age is given by
y – y = byx (x – x )

Σx 522 Σy 1417
x = = = 52.2; y = = = 141.7
n 10 n 10
n Σ dx dy − Σ dx Σ dy 10(1115) − 32(− 33) 12206
and byx = = = = 1.11
n Σ dx2 2
− (Σ dx ) 10(1202) − (32)2 10996

Substituting these values in the above equation, we have

y – 141.7 = 1.11 (x – 52.2) or y = 83.758+1.11x
This is the required regression equation of y on x.
(c) For a women whose age is 45, the estimated average blood pressure will be
y = 83.758+1.11(45) = 83.758+49.95 = 133.708
Hence, the likely blood pressure of a woman of 45 years is 134.

Example 10.10: The General Sales Manager of Kiran Enterprises—an enterprise dealing in the sale
of readymade men’s wear—is toying with the idea of increasing his sales to
Rs. 80,000. On checking the records of sales during the last 10 years, it was found that the annual
sale proceeds and advertisement expenditure were highly correlated to the extent of 0.8. It was
further noted that the annual average sale has been Rs. 45,000 and annual average advertisement
expenditure Rs. 30,000, with a variance of Rs. 1600 and Rs. 625 in advertisement expenditure
respectively.
In view of the above, how much expenditure on advertisement would you suggest the General
Sales Manager of the enterprise to incur to meet his target of sales?
Solution: Assume advertisement expenditure (y) as the dependent variable and sales (x) as the
independent variable. Then the regression equation advertisement expenditure on sales is given by
σy
(y – y ) = r (x − x )
σx
Given r = 0.8, σx = 40, σy = 25, x = 45,000, y = 30,000. Substituting these value in the above
equation, we have
25
(y – 30,000) = 0.8 (x – 45,000) = 0.5 (x – 45,000)
40
y = 30,000 + 0.5x – 22,500 = 7500 + 0.5x
When a sales target is fixed at x = 80,000, the estimated amount likely to the spent on advertisement
would be
y = 7500 + 0.5×80,000 = 7500 + 40,000 = Rs. 47,500
350 BUSINESS S T A T I S T I C S

Example 10.11: You are given the following information about advertising expenditure and sales:

Advertisement (x) Sales (y)

(Rs. in lakh) (Rs. in lakh)

Arithmetic mean, x 10 90
Standard deviation, σ 3 12

Correlation coefficient = 0.8

(a) Obtain the two regression equations.
(b) Find the likely sales when advertisement budget is Rs. 15 lakh.
(c) What should be the advertisement budget if the company wants to attain sales target of Rs. 120
lakh?
Solution: (a) Regression equation of x on y is given by
σx
x– x = r ( y − y)
σy
Given x = 10, r = 0.8, σx = 3, σy = 12, y = 90. Substituting these values in the above regression
equation, we have
3
x – 10 = 0.8 (y – 90) or x = – 8 + 0.2y
12
Regression equation of y on x is given by
σy
(y – y ) = r (x − x )
σx
12
y – 90 = 0.8 ( x − 10) or y = 58 + 3.2x
3
(b) Substituting x = 15 in regression equation of y on x. The likely average sales volume would be
y = 58 + 3.2 (15) = 58 + 48 = 106
Thus the likely sales for advertisement budget of Rs. 15 lakh is Rs. 106 lakh.
(c) Substituting y = 120 in the regression equation of x on y. The likely advertisement budget to
attain desired sales target of Rs. 120 lakh would be
x = – 8 + 0.2 y = – 8 + 0.2 (120) = 16
Hence, the likely advertisement budget of Rs. 16 lakh should be sufficient to attain the sales target
of Rs. 120 lakh.

Example 10.12: For 100 students of a class, the regression equation of marks in statistics (x) on marks
in economics (y) is: 3y – 5x + 180 = 0. If marks in economics is 50 and variance of marks in statistics
is (4/9) of variance of marks in economics, find mean marks in statistics and the coefficient of correlation
between them. [Delhi Univ., BCom (Hons), 2005]

Solution: Regression equation of x on y: 3y – 5x + 180 = 0

3 y 180
+ x=
5 5
3
Thus, regression coefficient of x on y, bxy =
5
 REGRESSION ANALYSIS 351

Also it is given that y = 50. To find x , put y = 50 in the regression equation we get
3(50) – 5x + 180 = 0 or x = 66, i.e. x = 66
4 2 2 σx 2
It is given that σ2x = σ y i.e. σ2x = σy or =
9 3 σy 3

3 σx 3
Since, bxy = or r . =
5 σy 5

r ( 23 ) = 53 or r =
3 3
×
5 2
=
9
10
= 0.9

Example 10.13: Given that

Regression euqation of y on x is: y = 20 + 0.4 x
Mean of x = 30
Correlation coefficient = 0.8
Find regression equation of x on y. [Delhi Univ., BCom, 2005]

Solution: Given, regression equation of y on x : y = 20 + 0.4x. This implies that byx = 0.4.
Also, x = 30, put x = 30 in the regression equation of y on x, we get
y = 20 + 0.4 (30) = 20 + 12 = 32
Thus y = 32
We know that, r2 = bxy × byx
(0.8)2 = bxy × 0.4 [since r = 0.8, byx = 0.4]
0.64
bxy = = 1.6
0.4
Regression equation of x on y is:
x − x = bxy ( y − y )
x – 30 = 1.6 ( y – 32) = 1.6 y – 51.2 + 30 = 1.6 y – 21.2.

Example 10.14: Given:

X Series Y Series

Mean 18 100
Standard deviation 14 20

Coefficient of correlation = 0.8

Find the most probable value of Y if X = 70 and most probable value of X if Y = 90.
[Delhi Univ., BCom (Hons), 2005]
Solution: Given: X = 18, Y = 100, σX = 14, σY = 20, and r = 0.8
r σX
Regression equation of X on Y: X−X = (Y − Y )
σY
(0.8)(14)
X – 18 = (Y − 100) = 0.56 (Y – 100)
20
X = 0.56Y – 56 + 18 = 0.56Y – 38
352 BUSINESS S T A T I S T I C S

If Y = 90, then X = 0.56 (90) – 38 = 12.4

r σY
Regression equation of Y on X : Y – Y = (X − X)
σX

(0.8)(20)
Y – 100 = ( X − 18) = 1.14 (X – 18)
14
Y = 1.14 X – 20.52 + 100 = 1.14 X + 79.48
If X = 70, then Y = 1.14 (70) + 79.48 = 159.28.

Example 10.15: Given x = 4y + 5 and y = kx + 4 are the lines of regression of x on y and y on x

respectively. If k is positive, prove that it cannot exceed 1/4. If k = 1/16, find the means of two variable
and correlation coefficient between them. [Delhi Univ., BCom, 2006]
Solution: Regression equation of x on y : x = 4y + 5. This implies that bxy = 4
Regression equation of y on x : y = kx + 4. This implies that byx = k
1
We know that, r2 = byx × bxy = k × 4. Now if r < 1, then r2 <1 and hence 4k < 1 or k <
4
If k = 1/16, then y = (1/16)x + 4. Thus the regression coefficient of y on x, becomes, byx = 1/16.
Then the correlation coefficient is

r= byx × bxy = (1/16) × 4 = 1/4 = ± 0.5.

Since both bxy and byx are positive, r = +0.5.
Solve two repression equations for x and y to get x = 28 and y = 5.76.

Example 10.16: In a partially destroyed laboratory record of an analysis of regression data, the
following results only are legible:
Variance of x = 9
Regression equations : 8x – 10y + 66 = 0 and 40x – 18y = 214
Find on the basis of the above information:
(a) The mean values of x and y,
(b) Coefficient of correlation between x and y, and
(c) Standard deviation of y.
Solution: (a) Since two regression lines always intersect at a point ( x , y ) representing mean values of the
variables involved, solving given regression equations to get the mean values x and y as shown below:
8x – 10y = – 66
40x – 18y = 214
Multiplying the first equation by 5 and subtracting from the second, we have
32y = 544 or y = 17, i.e. y = 17
Substituting the value of y in the first equation, we get
8x – 10(17) = – 66 or x = 13, that is, x = 13
(b) To find correlation coefficient r between x and y, we need to determine the regression coefficients
bxy and byx.
Rewriting the given regression equations in such a way that the coefficient of dependent variable
is less than one at least in one equation.
 REGRESSION ANALYSIS 353

66 8
8x – 10 y = – 66 or 10 y = 66 + 8x or y= + x
10 10
That is, byx = 8/10 = 0.80
214 18
40x – 18y = 214 or 40x = 214 + 18y or x = + y
40 40
That is, bxy = 18/40 = 0.45
Hence coefficient of correlation r between x and y is given by

r= bxy × byx = 0.45 × 0.80 = 0.60

(c) To determine the standard deviation of y, consider the formula:
σy byx σ x 0.80 × 3
byx = r or σy = = =4
σx r 0.6

Example 10.17: There are two series of index numbers, P for price index and S for stock of a
commodity. The mean and standard deviation of P are 100 and 8 and of S are 103 and 4 respectively.
The correlation coefficient between the two series is 0.4. With these data, work out a linear equation
to read off values of P for various values of S. Can the same equation be used to read off values of S for
various values of P?
Solution: The regression equation to read off values of P for various values S is given by
σp
P = a + bS or (P – P ) = r (S − S)
σs
Given P = 100, S = 103, σp = 8, σs = 4, r = 0.4. Substituting these values in the above equation,
we have
8
P – 100 = 0.4 (S − 103) or P = 17.6 + 0.8 S
4
This equation cannot be used to read off values of S for various values of P. Thus to read off values of
S for various values of P we use another regression equation of the form:
σs
S = c + dP or S−S = (P − P)
σp
Substituting given values in this equation, we have
4
S – 103 = 0.4 (P – 100) or S = 83 + 0.2P
8
Example 10.18: A panel of Judges A and B graded seven debators and independently awarded the
following marks:
Debator : 1 2 3 4 5 6 7
Marks of A : 40 34 28 30 44 38 31
Marks of B : 32 39 26 30 38 34 28
An eighth debator was awarded 36 marks by Judge A while Judge B was not present. If Judge B
was also present, how many marks would you expect him to award to eighth debator assuming same
degree of relationship exists in judgement? [Delhi Univ., BCom (Hons) 1993]
Solution: Let marks of A be denoted by x and that of B by y. A = 30 and B = 30 be assumed as mean
value of x-series and y-series, respectively. The calculation required for regression equations are
shown below:
354 BUSINESS S T A T I S T I C S

Marks of dx = x – 30 d x2 Marks of dY = y – 30 dy 2 dxdy

A(x) B(y)

40 10 100 32 2 4 20
34 4 16 39 9 81 36
28 –2 4 26 –4 16 8
30 ← A 0 0 30 ← B 0 0 0
44 14 196 38 8 64 112
38 8 64 34 4 16 32
31 1 1 28 –2 4 2
245 35 381 227 17 185 206

Σx 245 Σy 227
x = = = 35 and y= = = 32.43
n 7 n 7
Regression coefficient of y on x:
n Σ dx d y − Σ dx Σ d y 7(206) − (35)(17) 1442 − 595 847
byx = = = = = 0.587
2
n Σ dx − (Σ dx )
2
7(381) − (35)2 2667 − 1225 1442

Regression equation of y on x is:

y − y = byx ( x − x )
y – 32.43 = 0.587(x – 35)
y = 0.587x – 20.545 + 32.43 = 0.587x + 11.885
Marks awarded to eighth debator by Judge B when marks awarded by Judge A is x = 36 is:
y = 0.587(36) + 11.885 = 33 marks
Example 10.19: The lines of regression of a bivariate distribution are as follows: 5x – 145 = –10 y
and 14y – 208 = – 8 x, variance of x = 4. Find out mean values of x and y and standard deviation of
y. Also find correlation coefficient between x and y. [Delhi Univ., BCom (Hons), 2001]
Solution: Let the regression equation of x on y be:
5x – 145 = –10y
−10 y 145
x =
+ = –2y + 29
5 5
Hence, regression coefficient of x on y becomes: bxy = –2
Let the regression equation of y on x be:
14y – 208 = –8x
208 8 x
y = −
14 14
−8
Hence regression coefficient of y on x becomes: byx =
14
Now the coefficient of correlation r between x and y is given by

−8
r = byx × bxy = × (−2) =1.06
14
 REGRESSION ANALYSIS 355

Since value of r cannot exceed one, consider another set of regression equations:
Regression equation of x on y is
208 14
14y – 208 = –8x or x= −
8 8
−14
Thus bxy =
8
Regression equation of y on x is:
145 5
5x – 145 = –10y 10y = 145 – 5x or y = − x
10 10
−5 1
Thus byx = =−
10 2

1  −14 
Now r = byx × bxy = − ×   = 0.875 = ± 0.93
2  8 
Since bxy and byx both are negative, r should also be negative. Hence r = – 0.93.
rσ y 1 − 0.93 σ y
We know that, byx = or − = , where σx = 2
σx 2 2

2
or σy = = 1.075
0.93 × 2

Solve the two regression equations simultaneously to get x and y .

5x – 145 = –10y or 5x + 10y = 145 (i)
14y – 208 = –8x or 8x + 14y = 208 (ii)
Multiply (i) by 8 and (ii) by 5 and subtracting, we get 10y = 120 or y = 12
Substitute y = 12 is (i), we get 5x + 10(2) = 145 or x = 5 Hence x = 5 and y = 12.
Example 10.20: The two regression lines obtained in a correlation analysis of 60 observations are:
5x = 6x + 24 and 1000y = 768 x – 3708
What is the correlation coefficient and what is its probable error? Show that the ratio of the
coefficient of variability of x to that of y is 5/24. What is the ratio of variances of x and y?
Solution: Rewriting the regression equations
6 24
5x = 6y + 24 or x = y+
5 5
That is, bxy = 6/5
768 3708
1000y = 768x – 3708 or y = x −
1000 1000
That is, byx = 768/1000

σx 6 σy 768
We know that bx y = r = and byx = r = , therefore
σy 5 σx 1000
356 BUSINESS S T A T I S T I C S

6 768
bx y byx = r2 = × = 0.9216
5 1000
Hence r = 0.9216 = 0.96.
Since both b xy and b yx are positive, the correlation coefficient is positive and hence
r = 0.96.
1 − r2 1 − (0.96)2
Probable error of r = 0.6745 = 0.6745
n 60
0.0528
= = 0.0068
7.7459
Solving the given regression equations for x and y, we get x = 6 and y = 1 because regression
lines passed through the point ( x , y ).
σx 6 σ 6 σ 6 5
Since r = or 0.96 x = or x = =
σy 5 σy 5 σy 5 × 0.96 4

σx / x y σx 1 5 5
Also the ratio of the coefficient of variability = = ⋅ = × = .
σy / y x σy 6 4 24

S e l f-P r a c t i c e P r o b l e m s 10A

10.1 The following calculations have been made for 10.3 The following data give the experience of
prices of twelve stocks (x) at the Calcutta Stock machine operators and their performance
Exchange on a certain day along with the ratings given by the number of good parts
volume of sales in thousands of shares (y). From turned out per 100 pieces:
these calculations find the regression equation Operator : 11 12 13 4 5 16 7 18
of price of stocks on the volume of sales of experience (x) : 16 12 18 4 3 10 5 12
shares.
Performance
Σ x = 580, Σ y = 370, Σ xy = 11494,
ratings (y) : 87 88 89 68 78 80 75 83
Σ x2 = 41658, Σ y2 = 17206.
Calculate the regression lines of performance
10.2 A survey was conducted to study the ratings on experience and estimate the probable
relationship between expenditure (in Rs.) on performance if an operator has 7 years
accommodation (x) and expenditure on food experience.
and entertainment (y) and the following results 10.4 A study of prices of a certain commodity at Delhi
were obtained: and Mumbai yield the following data:
Mean Standard
Delhi Mumbai
Deviation
• Expenditure on 173..1 63.15 • Average price per kilo (Rs.) 2.463 2.797
• accommodation • Standard deviation 0.326 0.207
• Expenditure on food 47.8 22.98 • Correlation coefficient
• and entertainment between prices at Delhi
• Coefficient of correlation r = 0.57 and Mumbai r = 0.774
Write down the regression equation and
Estimate from the above data the most likely
estimate the expenditure on food and
price (a) at Delhi corresponding to the price of Rs.
entertainment if the expenditure on
2.334 per kilo at Mumbai (b) at Mumbai
accommodation is Rs. 200.
corresponding to the price of 3.052 per kilo at
Delhi
 REGRESSION ANALYSIS 357

10.5 The following table gives the aptitude test do this he selects a random sample of 10
scores and productivity indices of 10 workers applicants. They are given the test and later
selected at random: assigned a production rating. The results are
Aptitude as follows:
scores (x) : 60 62 65 70 72 48 53 73 65 82
Worker : A B C D E F G H I J
Productivity
index (y) : 68 60 62 80 85 40 52 62 60 81 Test score : 53 36 88 84 86 64 45 48 39 69
Production
Calculate the two regression equations and rating : 45 43 89 79 84 66 49 48 43 76
estimate (a) the productivity index of a worker
whose test score is 92, (b) the test score of a Fit a linear least squares regression equation of
worker whose productivity index is 75. production rating on test score.
10.6 A company wants to assess the impact of R&D 10.9 Find the regression equation showing the
expenditure (Rs. in 1000s) on its annual profit capacity utilization on production from the
(Rs. in 1000’s). The following table presents following data:
the information for the last eight years:
Average Standard
Year R & D expenditure Annual profit Deviation
1991 9 45 • Production 35.6 10.5
1992 7 42 (in lakh units) :
1993 5 41 • Capacity utilization
1994 10 60 (in percentage) : 84.8 8.5
1995 4 30 • Correlation coefficient r = 0.62
1996 5 34 Estimate the production when the capacity
1997 3 25 utilization is 70 per cent.
1998 2 20 10.10 Suppose that you are interested in using past
expenditure on R&D by a firm to predict current
Estimate the regression equation and predict expenditures on R&D. You got the following
the annual profit for the year 2002 for an allo- data by taking a random sample of firms, where
cated sum of Rs. 1,00,000 as R&D expendi- x is the amount spent on R&D (in lakh of rupees)
ture. 5 years ago and y is the amount spent on R&D
10.7 Obtain the two regression equations from the (in lakh of rupees) in the current year:
following bivariate frequency distribution: x : 30 50 20 180 10 20 20 40
y : 50 80 30 110 20 20 40 50
Sales Revenue Advertising Expenditure (Rs. in thousand) (a) Find the regression equation of y on x.
(Rs. in lakh) 5–15 15–25 25–35 35–45 (b) If a firm is chosen randomly and x = 10,
75–125 3 4 4 8 can you use the regression to predict the
value of y? Discuss.
125–175 8 6 5 7
175–225 2 2 3 4 10.11 The following data relates to the scores
obtained by a salesmen of a company in an
225–275 3 3 2 2
intelligence test and their weekly sales (in Rs.
1000’s ):
Estimate (a) the sales corresponding to
advertising expenditure of Rs. 50,000, (b) the Salesman
advertising expenditure for a sales revenue of intelligence : A B C D E F G H I
Rs. 300 lakh, (c) the coefficient of correlation.
Test score : 50 60 50 60 80 50 80 40 70
10.8 The personnel manager of an electronic Weekly sales: 30 60 40 50 60 30 70 50 60
manufacturing company devises a manual test
for job applicants to predict their production (a) Obtain the regression equation of sales on
rating in the assembly department. In order to intelligence test scores of the salesmen.
358 BUSINESS S T A T I S T I C S

(b) If the intelligence test score of a salesman is 10.14 In trying to evaluate the effectiveness in its ad-
65, what would be his expected weekly sales? vertising compaign, a firm compiled the
[HP Univ., M.Com., 1996] following information:
Calculate the regression equation of sales on
10.12 Two random variables have the regression
advertising expenditure. Estimate the probable
equations:
sales when advertisement expenditure is Rs.
3x + 2y – 26 = 0 and 6x + y – 31 = 0 60 thousand.
(a) Find the mean values of x and y and coeffi-
cient of correlation between x and y. Year Adv. expenditure Sales
(b) If the varaince of x is 25, then find the stand- (Rs. 1000’s) (in lakhs Rs.)
ard deviation of y from the data. 1996 12 5.0
10.13 For a given set of bivariate data, the following 1997 15 5.6
results were obtained 1998 17 5.8
x = 53.2, y = 27.9, 1999 23 7.0
Regression coefficient of y on x = – 1.5, and 2000 24 7.2
Regression coefficient of x and y = – 0.2. 2001 38 8.8
2002 42 9.2
Find the most probable value of y when x = 60.
2003 48 9.5

Hints and Answers

10.1 x = Σx/n = 580/12 = 48.33; For x = 200, we have y = 11.917 + 0.207(200)

= 53.317
y = Σy/n = 370/12 = 30.83
10.3 Let the experience and performance rating be
Σ xy − n x y represented by x and y respectively.
bxy =
Σ y2 − n ( y )2 x = Σ x/n = 80/8 = 10;
11494 − 12 × 48.33 × 30.83 y = Σ y/n = 648/8 = 81
= = – 1.102
17206 − 12(30.83)2
n Σ dx dy − Σ dx Σ dy 247
Regression equation of x on y: byx = = = 1.133;
n Σ dx2 − (Σ dx ) 2 218
x – x = bxy (y – y )
x – 48.33 = – 1.102 (y – 30.83) where dx = x – x , dy = y – y
or x = 82.304 – 1.102y Regression equation of y on x
10.2 Given x = 172, y = 47.8, σx = 63.15, y – y = byx (x – x )
σy = 22.98, and r = 0.57
or y – 81 = 1.133 (x – 10)
Regression equation of food and entertainment
or y = 69.67 + 1.133x
(y) on accomodation (x) is given by
When x = 7, y = 69.67 + 1.133 (7)
σy = 77.60 ≅ 78
y– y =r (x − x )
σx 10.4 Let price at Mumbai and Delhi be represented
by x and y, respectively
22.98
y – 47.8 = 0.57 (x – 173) (a) Regression equation of y on x
63.15
σy
or y = 11.917 + 0.207x y– y =r (x – x )
σx
 REGRESSION ANALYSIS 359

0.326 8 × 238 − (− 3) (1)

y – 2.463 = 0.774 (x – 2.797) = = 4.266 ;
0.207 8 × 57 − (− 3)2
For x = Rs. 2.334, the price at Delhi would where dx = x – 6, dy = y – 37
be y = Rs. 1.899. Regression equation of annual profit on R&D
(b) Regression on equation of x on y expenditure
y – y = byx (x – x )
σx y – 37.125 = 4.26 (x – 5.625)
x– x = r (y − y)
σy or y = 13.163 + 4.266x
For x = Rs. 1,00,000 as R&D expenditure, we
0.207
or x – 2.791 = 0.774 (y – 2.463) have from above equation y = Rs. 439.763 as
0.326 annual profit.
For y = Rs. 3.052, the price at Mumbai would 10.7 Let sales revenue and advertising expenditure
be x = Rs. 3.086. be denoted by x and y respectively
10.5 Let aptitude score and productivity index be
represented by x and y respectively. Σ fdx 12
x =A+ × h = 150 + × 50 = 159.09
n 66
x = Σx/n = 650/10 = 65;
y = Σy/n = 650/10 = 65 Σ fd y 26
y =B+ × k = 30 – × 10 = 26.06
n 66
n Σ dx dy − (Σ dx ) (Σ dy ) 1044
bxy = = = 0.596; n Σ fdx d y − (Σ fdx ) (Σ fd y ) h
n Σ dy2 2
− Σ (dy ) 1752 bxy = ×
nΣ fd 2y − (Σ fdy ) 2
k
where dx = x – x ; dy = y – y
66 (−14) − 12(− 26) 50
(a) Regression equation of x on y = 2
× = – 0.516
66(100) − ( − 26) 10
x – x = bxy (y – y )
(a) Regression equation of x on y
or x – 65 = 0.596 (y – 65)
x – x = bxy (y – y )
or x = 26.26 + 0.596y
x – 159.09 = – 0.516 (y – 26.06)
When y = 75, x = 26.26 + 0.596 (75) = 70.96
or x = 172.536 – 0.516y
≅ 71
For y = 50, x = 147.036
n Σ dx dy − (Σdx ) (Σdy )
(b) byx = (b) Regression equation of y on x
n Σdx2 − (Σdx )2
n Σ fdx d y − (Σ fdx ) (Σ fd y ) k
1044 byx = ×
= = 1.168 nΣ fdx2 − (Σ fdx ) 2
h
894
y – y = byx (x – x ) 66 (− 14) − 12( − 26) 10
= 2
× = – 0.027.
66 (70) − (12) 50
or y – 65 = 1.168 (x – 65)
or y = – 10.92 + 1.168x y – y = byx (x – x )
When x = 92, y = – 10.92 + 1.168 (92) = y – 26.06 = – 0.027 (x – 159.09)
96.536 ≅ 97 y = 30.355 – 0.027x
10.6 Let R&D expenditure and annual profit be For x = 300, y = 22.255
denoted by x and y respectively
x = Σ x/n = 40/8 = 5.625; y = Σy/n = 297/8 (c) r = bxy × byx = – 0.516 × 0.027
= 37.125
= – 0.1180
n Σ dx d y − (Σ dx )(Σ d y ) 10.8 Let test score and production rating be denoted
byx =
n Σ dx2 − (Σ dx )2 by x and y respectively.
360 BUSINESS S T A T I S T I C S

x = Σ x/n = 612/10 = 61.2; Regression equation of y on x

y = Σ y/n = 622/10 = 62.2 y – y = byx (x – x )
n Σ dx d y − (Σ dx ) (Σd y ) y – 50 = 1.338 (x – 33.75)
byx =
n Σ dx2 − (Σ dx ) 2 y = 4.84 + 1.338x

10 × 3213 − 2 × 2 For x = 10, y = 18.22

= = 0.904
10 × 3554 − (2)2 10.11 Let intelligence test score be denoted by x and
weekly sales by y
Regression equation of production rating (y)
on test score (x) is given by x = 540/9 = 60; y = 450/9 = 50,

y – y = byx (x – x ) n Σ dx dy − (Σ dx)(Σ dy)

byx =
y – 62.2 = 0.904 (x – 61.2) n Σ dx2 − (Σ dx )2

y = 6.876 + 0.904x 9 × 1200

= = 0.75
10.9 Let production and capacity utilization be 9 × 1600
denoted by x and y, respectively. Regression equation of y on x :
(a) Regression equation of capacity utilization
y − y = byx ( x − x )
(y) on production (x)
σy y – 50 = 0.75 (x – 60)
y– y = r (x − x )
σx y = 5 + 0.75x
For x = 65, y = 5 + 0.75 (65) = 53.75
8.5
y – 84.8 = 0.62 (x – 35.6) 10.12 (a) Solving two regression lines:
10.5
3x + 2y = 6 and 6x + y = 31
y = 66.9324 + 0.5019x
we get mean values as x = 4 and y = 7
(b) Regression equation of production (x) on
capacity utilization (y) (b) Rewritting regression lines as follows:

σx 3x + 2y = 26 or y = 13 – (3/2)x,
x – x =r (y − y)
σy So byx = – 3/2
6x + y = 31 or x = 31/6 – (1/6)y,
10.5
x – 35.6 = 0.62 (y – 84.8) So bxy = – 1/6
8.5
x = – 29.3483 + 0.7659y Correlation coefficient,

When y = 70, x = – 29.3483+0.7659(70) = r= bxy × byx = − (3 / 2)(1 / 6) = – 0.5

24.2647
Given, Var(x) = 25, so σx = 5. Calculate σy using
Hence the estimated production is 2,42,647 the formula:
units when the capacity utilization is 70 per cent.
σy
10.10 x = Σ x/n = 270/8 = 33.75; y = Σ y/n = byx = r
σy
400/8 = 50
n Σ dx dy − (Σ dx )(Σ dy ) 3 σy
byx = or = 0.5
− or σy = 15
n Σ dx2 − (Σ dx )2 2 5
10.13 The regression equation of y on x is stated as:
8 × 4800 − 6 × 0
= = 1.338; σy
8 × 3592 − (6)2 y − y = bxy ( x − x ) = r ⋅ (x − x )
σx
where dx = x – 33 and dy = y – 50
Given, x = 53.20; y = 27.90, byx = – 1.5;
 REGRESSION ANALYSIS 361

bxy = – 0.2 n∑ dx dy − (∑ dx)(∑ dy)

byx =
Thus y – 27.90 = – 1.5(x – 53.20) n∑ dx2 − (∑ dx)2
or y = 107.70 – 1.5x
8(172.2) − (25)(2.1) 1325.1
For x = 60, we have y = 2
= = 0.125
8(1403) − (25) 10599
= 107.70 – 1.5(60) = 17.7
Thus regression equation of y on x is:
Also r = byx × bxy = – 1.5 × 0.2 = – 0.5477 y − y = byx ( x − x )
10.14 Let advertising expenditure and sales be or y – 7.26 = 0.125(x – 27.125)
denoted by x and y respectively.
y = 3.86 + 0.125x
x = Σ x/n = 217/8 = 27.125; y = Σ y/n = 58.2/8
When x = 60, the estimated value of y = 3.869
= 7.26 + 0.125(60) = 11.369

Conceptual Questions
1. (a) Explain the concept of regression and point 10. (a) Distinguish between correlation and
out its usefulness in dealing with business regression analysis.
problems. (b) The coefficient of correlation and coefficient
(b) Distinguish between correlation and of determination are available as measures of
regression. Also point out the properties of association in correlation analysis. Describe
regression coefficients. the different uses of these two measures of
2. Explain the concept of regression and point out association.
its importance in business forecasting. 11. What are regression coefficients? State some of
3. Under what conditions can there be one the important properties of regression
regression line? Explain. coefficients.
4. Why should a residual analysis always be done as 12. What is regression? How is this concept useful to
part of the development of a regression model? business forecasting?
13. What is the difference between a prediction
5. What are the assumptions of simple linear
interval and a confidence interval in regression
regression analysis and how can they be
analysis?
evaluated?
14. Explain what is required to establish evidence of
6. What is the meaning of the standard error of
a cause-and-effect relationship between y and x
estimate?
with regression analysis.
7. What is the interpretation of y-intercept and the 15. What technique is used initially to identify the kind
slope in a regression model? of regression model that may be appropriate.
8. What are regression lines? With the help of an 16. (a) What are regression lines? Why is it necessary
example illustrate how they help in business to consider two lines of regression?
decision-making.
(b) In case the two regression lines are identical,
9. Point out the role of regression analysis in business prove that the correlation coefficient is either + 1
decision-making. What are the important or – 1. If two variables are independent, show
properties of regression coefficients? that the two regression lines cut at right angles.

Formulae Used
1. Simple linear regression model 3. Regression coefficient in sample regression
y = β0 + β1x + e equation b = ŷ
2. Simple linear regression equation based on
sample data y = a + bx a = y − bx
362 BUSINESS S T A T I S T I C S

Chapter Concepts Quiz

True or False
1. A statistical relationship between two variables 8. Correlation coefficient is the geometric mean of
does not indicate a perfect relationship. regression coefficients.
2. A dependent variable in a regression equation is a 9. If the sign of two regression coefficients is
continuous random variable. negative, then sign of the correlation coefficient is
positive.
3. The residual value is required to estimate the
10. Correlation coefficient and regression coefficient
amount of variation in the dependent variable
are independent.
with respect to the fitted regression line.
11. The point of intersection of two regression lines
4. Standard error of estimate is the conditional represents average value of two variables.
standard deviation of the dependent variable. 12. The two regression lines are at right angle when
5. Standard error of estimate is a measure of scatter the correlation coefficient is zero.
of the observations about the regression line. 13. When value of correlation coefficient is one, the
6. If one of the regression coefficients is greater than two regression lines coincide.
one the other must also be greater than one. 14. The product of regression coefficients is always
more than one.
7. The signs of the regression coefficients are always
15. The regression coefficients are independent of
same.
the change of origin but not of scale.

Concepts Quiz Answers

1. T 2. T 3. T 4. T 5. T 6. F 7. T 8. T 9. F
10. F 11. T 12. T 13. T 14. F 15. T

R e v i e w S e l f-P r a c t i c e P r o b l e m s
10.15 Given the following bivariate data: 10.17 The coefficient of correlation between the ages
x: – 1 5 3 2 1 1 7 3 of husbands and wives in a community was
found to be + 0.8, the average of husbands
y: – 6 1 0 0 1 2 1 5 age was 25 years and that of wives age 22 years.
(a) Fit a regression line of y on x and predict y Their standard deviations were 4 and 5 years
if x = 10. respectively. Find with the help of regression
equations:
(b) Fit a regression line of x on y and predict x
if y = 2.5. (a) the expected age of husband when wife’s
age is 16 years, and
10.16 Find the most likely production corresponding (b) the expected age of wife when husband’s
to a rainfall of 40 inches from the following data: age is 33 years.
10.18 You are given below the following information
Rainfall Production about advertisement expenditure and sales:
(in inches) (in quintals)
Adv. Exp. (x) Sales (y)
Average 30 50
(Rs. in crore) (Rs. in crore)
Standard deviation 5 10
Mean 20 120
Coefficient of correlation r = 0.8. Standard deviation 5 25
 REGRESSION ANALYSIS 363

Correlation coefficient 0.8 Age of cars : 2 4 6 8

(a) Calculate the two regression equations. (in years)
(b) Find the likely sales when advertisement Maintenance costs : 10 20 25 30
expenditure is Rs. 25 crore. (Rs. in 100’s)
(c) What should be the advertisement budget
10.22 An analyst in a certain company was studying
if the company wants to attain sales target
the relationship between travel expenses in
of Rs. 150 crore?
rupees (y) for 102 sales trips and the duration
10.19 For 50 students of a class the regression in days (x) of these trips. He has found that the
equation of marks in Statistics (x) on the marks relationship between y and x is linear. A
in Accountancy (y) is 3y – 5x + 180 = 0. The summary of the data is given below:
mean marks in Accountancy is 44 and the
Σx = 510; Σy = 7140; Σx2 = 4150; Σxy =
variance of marks in Statistics is 9/16th of the
54,900, and Σy2 = 7,40,200
variance of marks in Accountancy. Find the
mean marks in Statistics and the coefficient of (a) Estimate the two regression equations
correlation between marks in the two subjects. from the above data.
(b) A given trip takes seven days. How much
10.20 The HRD manager of a company wants to find money should a salesman be allowed so
a measure which he can use to fix the monthly that he will not run short of money?
income of persons applying for a job in the
production department. As an experimental 10.23 The quantity of a raw material purchased by
project, he collected data on 7 persons from ABC Ltd. at specified prices during the post 12
that department referring to years of service months is given below:
and their monthly income.
Years of service : 11 7 9 5 8 6 10 Month Price per Quantity Month Price per Quantity
Income (Rs. in 1000’s): 10 8 6 5 9 7 11 kg (in Rs.) (in kg) kg (in Rs.) (in kg)
(a) Find the regression equation of income on Jan 96 250 July 112 220
years of service. Feb 110 200 Aug 112 220
(b) What initial start would you recommend March 100 250 Sept 108 200
for a person applying for the job after April 90 280 Oct 116 210
having served in a similar capacity in
May 86 300 Nov 86 300
another company for 13 years?
June 92 300 Dec 92 250
(c) Do you think other factors are to be
considered (in addition to the years of
service) in fixing the income with reference (a) Find the regression equations based on the
to the above problems? Explain. above data.
(b) Can you estimate the approximate
10.21 The following table gives the age of cars of a
quantity likely to be purchased if the price
certain make and their annual maintenance
shoots up to Rs. 124 per kg?
costs. Obtain the regression equation for costs
related to age. (c) Hence or otherwise obtain the coefficient
of correlation between the price prevailing
and the quantity demanded.

Hints and Answers

10.15 x = Σ x = 21/8 = 2.625; 8 × 30 − ( − 3) ( − 12)

= = 0.568;
y = Σ y/n = 4/8 = 0.50 8 × 45 − (− 1)2
n Σ dx dy − (Σ dx ) (Σ dy ) dx = x – 3; dy = y – 3.
byx = Regression equation:
n Σ dx2 2
− (Σ dx )
y – y = byx (x – x )
364 BUSINESS S T A T I S T I C S

or y – 0.5 = 0.568 (x – 2.625) y = 40 + 4x

y = – 0.991 + 0.568x (b) When advertisement expenditure is of Rs.
n Σ dx dy − (Σ dx )(Σ dy ) 25 crore, likely sales is
(b) bxy = y = 40 + 4x = 40 + 4 (25) = 140 crore.
n Σ d2y − (Σ dy )2
(c) For y = 150, x = 0.8 + 0.16y
8 × 30 − (− 3) ( −12)
= = 0.386 = 0.8 + 0.16(150)
8 × 84 − (− 12)2
= 24.8
Regression equation: 10.19 Let x = marks in Statistics and y = marks in
x – x = bxy (y – y ) Accountancy,
or x – 2.625 = 0.386 (y – 5) Given: 3y – 5x + 180 = 0
x = 0.695 + 0.386y or x =(3/5) y + (180/5)
10.16 Let x = rainfall y = production by y. The For y = 44, x = (3/5) × 44 + (180/5) = 62.4
expected yield corresponding to a rainfall of 40 Regression coefficient of x on y, bxy = 3/5
inches is given by regression equation of y on x. Coefficient of regression
Given y = 50, σy = 10, x = 30, σx = 5, r = 0.8 σx 9
bxy = r = (given)
σy σy 16
y – y =r (x − x) ;
σx
3 9 3 3r
10 or = r or =
y – 50 = 0.8 (x – 30) 5 16 5 4
5 Hence 3r = 2.4 or r = 0.8
y = 2 + 1.6x
10.20 Let x = years of service and y = income.
For x = 40,y = 2 + 1.6 (40) = 66 quintals.
(a) Regression equation of y on x
10.17 Let x = age of wife y = age of husband. n Σ xy − (Σ x)(Σ y)
byx =
Given x = 25, y = 22, σx = 4, σy = 5, r = 0.8 n Σ x 2 − (Σ x)2
(a) Regression equation of x on y 7 × 469 − 56 × 56
= = 0.75
σ
x − x = r x ( y − y) 7 × 476 − (56)2
σy y – y = byx (x – x )
4 y – 8 = 0.75 (x – 8)
x – 25 = 0.8 (y – 22)
5 y = 2 + 0.75x
x = 10.92 + 0.64 y (b) When x = 13 years, the average income
When age of wife is y = 16; x = 10.92 + 0.64 would be
(16) = 22 approx.(husband’s age) y = 2 + 0.75x = 2 + 0.75(13) = Rs. 11,750
(b) Left as an exercise 10.21 Let x = age of cars and y = maintainance costs.
10.18 (a) Regression equation of x on y The regression equation of y on x
σx x = Σ x/n = 20/4 = 5;
x – x = bxy (y – y ) = r ( y − y)
σy y = Σ y/n = 85/4 = 21.25
5 n Σ xy − Σ x Σ y 7 × 490 − 20 × 85
x – 20 = 0.8 (y – 120) and byx = 2 2
=
25 n Σ x − (Σ x) 7 × 120 − (20)2
x = 0.8 + 0.16y = 3.25
Regression equation of y on x y − y = byx ( x – x )
σy y – 21.25 = 3.25 (x – 5)
y – y =r (x − x ) y = 5 + 3.25x
σx
10.22 x = Σx/n = 510/102 = 5;
25
y – 120 = 0.8 (x – 20) y = Σy/n = 7140/102 = 70
5
 REGRESSION ANALYSIS 365

Regression coefficients: y = Σy/n = 2980/12 = 248.33

n (Σ xy) − (Σ x)(Σ y) (a) Regression coefficients:
bxy =
n Σ y2 − (Σ y)2
n Σ dx dy − (Σ dx)(Σ dy)
102 × 54900 − 510 × 7140 bxy = = – 0.26
= = 0.08 n Σ d2y − (Σ d y )2
102 × 740200 − (7140)2
n Σ dx dy − (Σ dx )(Σ dy )
n (Σ xy) − (Σ x) (Σ y) bxy = = – 3.244
byx = 2
n Σ x − (Σ x) 2 n Σ dx2 − (Σ dx )2

102 × 54900 − 510 × 7140 Regression lines:

= = 12
102 × 4150 − (510)2 x – x = bxy (y – y )
Regression lines: x – 100 = – 0.26 (y – 248.33)
x − x = bxy ( y − y ) or x = – 0.26y + 164.56
x – 5 = 0.08 (y – 70) or x = 0.08y – 0.6 and y – y = byx (x – x )
and y − y = byx (x – x ) y – 248.33 = – 3.244 (x – 100)
y – 70 = 12 (x – 5) or y = 12x + 10 y = – 3.244x + 572.73
When x = 7, y = 12×7 + 10 = 94 (b) For x = 124,
y = – 3.244 × 124 + 572.73
10.23 Let price be denoted by x and quantity by y
= 170.474
x = Σx/n = 1200/12 = 100 ;
 LEARNING OBJECTIVES

After studying this chapter, you should be able to

z understand the pattern of the historical data and then extrapolate the pattern into the
future.
z understand the different approaches to forecasting that can be applied in business.
z gain a general understanding of time-series forecasting techniques.
z learn how to decompose time-series data into their various components and to forecast
by using decomposition techniques.

11.1 INTRODUCTION
The increasing complexity of the business environment together with changing demands and
expectations, implies that every organization needs to know the future values of their key decision
variables. Forecasting takes the historical data and project them into the future to predict the
occurrence of uncertain events. This may help organizations to assess the future consequences of
existing decisions and to evaluate the consequences of decisions (actions or strategies). For example,
inventory is ordered without certainty of future sales; new equipment is purchased despite uncer-
tainty about the demand for products; investments are made without knowing profits in future;
alternative staff mix is made without knowing the increase in the level of service that can be
provided, and so on.
Forecasting is essential to make reliable and accurate estimates of what will happen in the
future in the face of uncertainty. A flow chart of forecasts and the decision-making process is
shown in Fig. 11.1. In general, the decisions are influenced by the chosen strategy with regard
to an organization’s future priorities and activities. Once decisions are taken, the consequences are
measured in terms of expectation to achieve the desired products/services levels.
 FORECASTING AND TIME SERIES ANALYSIS 367

Figure 11.1 Decision-Making Process and Forecasts

Decisions also get influenced by the additional information obtained from the forecasting
method used. Such information and the perceived accuracy of the forecasts may also affect the
strategy formulation of an organization. Thus an organization needs to establish a monitoring
system to compare planned performance with the actual. Divergence, if any, and no matter what
the cause of such divergence between the planned and actual performance, should be fed back into
the forecasting process, to generate new forecasts. A few objectives of forecasting are as follows:
(i) The creation of plans of action, because it is not possible to evolve a system of business
control without an acceptable system of forecasting.
(ii) Monitoring of the continuing progress of action plans based on forecasts.
(iii) The forecast provides a warning system of the critical factors to be monitored regularly
because they might drastically affect the performance of the plan.

11.2 FORECASTING METHODS

Forecasting methods may be classified as either quantitative or qualitative (opinion or judgmental).
Figure 11.2 provides an overview of the types of forecasting methods.

11.2.1 Quantitative Forecasting Methods

These methods can be used when
(i) past information about the variable being forecast is available,
(ii) information can be quantified, and
(iii) a reasonable assumption is that the pattern of the past will continue into the future.
The quantitative methods of forecasting are further classified into two categories:
Time Series Forecasting Methods A time series is a set of measurements of a variable that are ordered
through time. The time variable does not fluctuate arbitrarily. It moves uniformly always in the same
direction, from past to future, Thus we can exercise some freedom of choice as to the times at which
observations can be made. The time-series data are gathered on a given variable characteristic over a
period of time at regular intervals.
368 BUSINESS S T A T I S T I C S

Figure 11.2 Forcasting Methods

The time series forecasting methods attempt to account for changes over a period of time at
regular intervals by examining patterns, cycles or trends to predict the outcome for a future time
period.
Causal Forecasting Methods These methods are based on the assumptions that the variable value
which we intend to forecast has a cause-effect relationship with one or more other variables. A
linear regression analysis which depends upon the causal relationship or interaction of two or
more variables is called causal forcasting method (Fig. 11.2).

11.2.2 Qualitative Forecasting Methods

These methods consist of collecting the opinions and judgments of individuals who are expected
to have the best knowledge of current activities or future plans of the organization. For example,
knowledge of demand trend and customer plans are often known to marketing executives or
product managers. Through regular contact with customers, the marketing and sales personnel
are presumably familiar with individual customers or retail market segment. Management usually
maintains broader market information on trends by product line, geographic area, customer
groups, and so on.
Qualitative forecasting methods have the advantage that they can incorporate subjective
experience as inputs along with objectives data. It is the human brain that permits assimilation
of all types of information and the ultimate issuance of a prediction.
Since each human being has different knowledge, experience, and perspective of reality,
intuitive forecasts are likely to differ from one individual to another. Furthermore, the less they
are based upon fact and quantified data, the less they lend themselves to analysis and resolution
of differences of opinion. The quantification of data gives them a more precise meaning than
words which are inexact and are capable of being misunderstood. Also, if the forecasts prove to
be inaccurate there is an objective basis for improvement the next time around.
 FORECASTING AND TIME SERIES ANALYSIS 369

11.3 TIME SERIES ANALYSIS

A time series is a set of numerical values of some variable obtained at regular period over time.
The series is usually tabulated or graphed in a manner that readily conveys the behaviour of the
variable under study. Figure 11.3 presents the export of cement (in tonnes) by a cement company
between 1994 and 2004. The graph suggests that the series is time dependent. The management
of the company is interested in determining how the series is dependent on time and in developing
a means of predicting future levels with some degree of reliability. The nature of the time
dependence is often analysed by decomposing the time series into its components.

Year Export (tonnes)

1994 2
1995 3
1996 6
1997 10
1998 8
1999 7
2000 12
2001 14
2002 14
2003 18
2004 19

Figure 11.3 Export of Cement

11.3.1 Objectives of Time Series Analysis
1. The assumption underlying time series analysis is that the future will look like the past, that
is, the various factors which have already influenced the patterns of change in the value of
the variable under study will continue to do so in more or less the same manner in the future.
In other words, some underlying pattern exists in historical data. Thus one of the objective
of time-series analysis is to identify the pattern and isolate the influencing factors (or effects)
for prediction purposes as well as for future planning and control.
2. The review and evaluation of progress made on the basis of a plan are done on the basis of
time-series data. For example the progress of our Five-Year Plans is judged by the annual
growth rates in the Gross National Product (GNP). Similarly the evaluation of our policy of
controlling inflation and price rise is done by the study of various price indices which are
based on the analysis of time-series.

11.3.2 Time Series Patterns

We assume that time series data consist of an underlying pattern accompanied by random
fluctuations. This may be expressed in the following form:
Actual value of the = Mean value of the + Random deviation from mean value
variable at time t variable at time t of the variable at time t
y = Pattern + e
370 BUSINESS S T A T I S T I C S

where y is the forecast variable at period t; pattern is the mean value of the forecast variable at
period t and represents the underlying pattern, and e is the random fluctuation from the pattern
that occurs of the forecast variable at period t.

11.3.3 Components of a Time Series

The time-series data contain four components: trend, cyclicality, seasonality and irregularity. Not all
time-series have all these components. Figure 11.4 shows the effects of these time-series components
over a period of time.
Trend Sometimes a time-series displays a steady tendency of either upward or downward movement
in the average (or mean) value of the forecast variable y over time. Such a tendency is called a
trend. When observations are plotted against time, a straight line describes the increase or
decrease in the time series over a period of time.
Cycles An upward and downward movement in the variable value about the trend time over a
time period are called cycles. A business cycle may vary in length, usually more than a year but
less than 5 to 7 years. The movement is through four phases: from peak (prosperity) to contradiction
(recession) to trough (depression) to expansion (recovery or growth) as shown in Fig. 11.4 .

Figure 11.4 Time-series Effects

Seasonal It is a special case of a cycle component of time series in which fluctuations are repeated
usually within a year (e.g. daily, weekly, monthly, quarterly) with a high degree of regularity. For
example, average sales for a retail store may increase greatly during festival seasons.
Irregular Irregular variations are rapid charges or bleeps in the data caused by short-term
unanticipated and non-recurring factors. Irregular fluctuations can happen as often as day to day.

11.4 TIME SERIES DECOMPOSITION MODELS

The analysis of time series consists of two major steps:
1. Identifying the various factors or influences which produce the variations in the time series,
and
2. Isolating, analysing and measuring the effect of these factors independently, by holding other
things constant.
The purpose of decomposition models is to break a time series into its components: Trend (T),
Cyclical (C), Seasonality (S), and Irregularity (I ). Decomposition of time series aims to isolate influence
of each of the four components on the actual series so as to provide a basis for forecasting. There are
 FORECASTING AND TIME SERIES ANALYSIS 371

many models by which a time series can be analysed; two models commonly used for decomposition
of a time series are discussed below.

11.4.1 Multiplicative Model

The actual values of a time series, represented by Y can be found by multiplying four components
at a particular time period. The effect of four components on the time series is interdependent.
The multiplicative time series model is defined as:
Y = T × C × S × I ← Multiplicative model
The multiplicative model is appropriate in situations where the effect of C, S, and I is
measured in relative sense and is not in absolute sense. The geometric mean of C, S, and I is
assumed to be less than one. For example, let the actual sales for period of 20 months be Y20 =
423.36. Further let this value be broken down into its components as: trend component (mean
sales) 400; effect of current cycle (0.90) which decreases sales by 10 per cent; seasonality of the
series (1.20) that increases sales by 20 per cent. Thus besides the random fluctuation, the expected
value of sales for this period is: 400 × 0.90 × 1.20 = 432. If the random factor decreases sales
by 2 per cent in this period, then the actual sales value will be 432 × 0.98 = 423.36.

11.4.2 Additive Model

In this model, it is assumed that the effect of various components can be estimated by adding the
various components of a time-series. It is stated as:
Y = T + C + S + I ← Additive model
Here C, S, and I are absolute quantities and can have positive or negative values. It is assumed
that these four components are independent of each other. However, in real-life time series data
this assumption does not hold good.

C o n c e p t u a l Q u e s t i o n s 11A
1. Briefly describe the steps that are used to 8. Explain what you understand by time series.
develop a forecasting system. Why is time-series considered to be an effective
2. What is forecasting? Discuss in brief the various tool of forecasting?
theories and methods of business forecasting. 9. Explain briefly the additive and multiplicative
models of time series. Which of these models is
3. For what purpose do we apply time series
more popular in practice and why?
analysis to data collected over a period of
time? 10. Identify the four principal components of a time-
series and explain the kind of change, over time,
4. How can one benefit from determining past
to which each applies.
patterns?
11. What is the advantage of reducing a time series
5. What is the difference between a causal into its four components?
model and a time series model?
12. Despite great limitations of statistical forecasting,
6. What is a judgmental forecasting model, and forecasting techniques are invaluable to the
when is it appropriate? economist, the businessman, and the
7. Explain clearly the different components into government. Explain.
which a time series may be analysed. Explain 13. (a) Why are forecasts important to organiza-
any method for isolating trend values in a time tions?
series.
372 BUSINESS S T A T I S T I C S

(b) Explain the difference between the terms: examples where you believe the seasonality
seasonal variation and cyclical variation. may change.
(c) Give reasons why the seasonal component 14. Identify the classical components of a time
in the time-series is not constant? Give series and indicate how each is accounted for
in forecasting.

11.5 QUANTITATIVE FORECASTING METHODS

The quantitative forecasting methods fall into two general categories:
• Time series methods
• Causal methods
The time series methods are concerned with taking some observed historical pattern for some
variable and projecting this pattern into the future using a mathematical formula. These methods do
not attempt to suggest why the variable under study will take some future value. This limitation of the
time-series approach is taken care by the application of a causal method. The causal method tries to
identify factors which influence the variable is some way or cause it to vary in some predictable
manner. The two causal methods, regression analysis and correlation analysis, have already been
discussed previously.
A few time series methods such as freehand curves and moving averages simply describe the given
data values, while other methods such as semi-average and least squares help to identify a trend
equation to describe the given data values.

11.5.1 Freehand (or Graphical) Method

A freehand curve drawn smoothly through the data values is often an easy and, perhaps, adequate
representation of the data. From Fig. 11.3, it appears that a straight line connecting the 1994 and
2004 exports volumes is a fairly good representation of the given data.
The forecast can be obtained simply by extending the trend line. A trend line fitted by the
freehand method should confirm to the following conditions:
(i) The trend line should be smooth—a straight line or mix of long gradual curves.
(ii) The sum of the vertical deviations of the observations above the trend line should equal
the sum of the vertical deviations of the observations below the trend line.
(iii) The sum of squares of the vertical deviations of the observations from the trend line
should be as small as possible.
(iv) The trend line should bisect the cycles so that area above the trend line should be equal
to the area below the trend line, not only for the entire series but as much as possible
for each full cycle.

Example 11.1: Fit a trend line to the following data by using the freehand method.
Year : 1997 1998 1999 2000 2001 2002 2003 2004
Sales turnover : 80 90 92 83 94 99 92 104.
(Rs. in lakh)
Solution: Figure 11.5 presents the freehand graph of sales turnover (Rs. in lakh) from 1997 to
2004. Forecast can be obtained simply by extending the trend line
 FORECASTING AND TIME SERIES ANALYSIS 373

Figure 11.5 Graph of Sales Turnover

Limitations of freehand method

(i) This method is highly subjective because the trend line depends on personal judgment
and therefore what happens to be a good-fit for one individual may not be so for another.
(ii) The trend line drawn cannot have much value if it is used as a basis for predictions.
(iii) It is very time-consuming to construct a freehand trend if a careful and conscientious job
is to be done.

11.5.2 Smoothing Methods

The objective of smoothing methods is to smoothen out the random variations due to irregular
components of the time series and thereby provide us with an overall impression of the pattern
of movement in the data over time. In this section, we shall discuss three smoothing methods:
(i) Moving averages
(ii) Weighted moving averages
(iii) Semi-averages
The data requirements for the techniques to be discussed in this section are minimal and these
techniques are easy to use and understand.

Moving Averages
If we attempt to observe the movement of some variable values over a period of time and try to
project this movement into the future, then it is essential to smooth out first the irregular pattern
in the historical values of the variable, and later use this as the basis for a future projection. This
can be done by using the technique of moving averages.
This method is a subjective method and depends on the length of the period chosen for
calculating moving averages. To remove the effect of cyclical variations, the period chosen should
be an integer value that corresponds to or is a multiple of the estimated average length of a cycle
in the series.
The moving averages which serve as an estimate of the next period’s value of a variable given
a period of length n is expressed as:
374 BUSINESS S T A T I S T I C S

Σ{Dt + Dt − 1 + Dt − 2 + ... + Dt − n + 1}
Moving average, MA t + 1 =
n
where t = current time period
D = actual data which is exchanged each period
n = length of time period
In this method, the term ‘moving’ is used because it is obtained by summing and averaging the
values from a given number of periods, each time deleting the oldest value and adding a new
value.
The major advantage of a moving average is the opportunity it provides to focus on the long-
term trend (and cyclical) movements in a time series without the obscuring effect of short-term
‘noise’ influences.
The limitation of this method is that it is highly subjective and dependent on the length of
period chosen for constructing the averages. Moving averages have the following three limitations:
(i) As the size of n (the number of periods averaged) increases, it smoothens the variations
better, but it also makes the method less sensitive to real changes in the data.
(ii) It is difficult to choose the optimal length of time for which to compute the moving
average. Moving averages can not be found for the first and last k/2 periods in a k-period
moving average.
(iii) Moving averages cannot pick-up trends very well. Since these are averages, it will always
stay within past levels and will not predict a change to either a higher or lower level.
(iv) It causes a loss of information (data values) at either end of the original time series.
(v) Moving averages do not usually adjust for such time-series effects as trend, cycle or
seasonality.
Example 11.2: Shown is production volume (in ’000 tonnes) for a product. Use these data to
compute a 3-year moving average for all available years. Also determine the trend and short-term
error.

Year Production Year Production

(in ’000 tonnes) (in ’000 tonnes)
1995 21 2000 22
1996 22 2001 25
1997 23 2002 26
1998 25 2003 27
1999 24 2004 26

Solution: The first average is computed for the first 3 years as follows:

21 + 22 + 23
Moving average (year 1–3) = = 22
3
The first 3-year moving average can be used to forecast the production volume in fourth year,
1998. Because 25,000 tonnes production was made in 1998, the error of the forecast is Error1998
= 25,000 – 22,000 = 3000 tonnes.
Similarly, the moving average calculation for the next 3 years is:
22 + 23 + 25
Moving average (year 2–4) = = 23.33
3
 FORECASTING AND TIME SERIES ANALYSIS 375

A complete summary of 3-year moving average calculations is given in Table 11.1.

Table 11.1 Calculation of Trend and Short-term Fluctuations

Year Production 3-Year Moving 3-Yearly Moving Forecast Error

y Total Average ( y – y )
(Trend values)
y
1995 21 — — —
1996 22 → (21 + 22 + 23) = 66 66/3 = 22.00 0
1997 23 → (22 + 23 + 25) = 70 70/3 = 23.33 – 0.33
1998 25 → (23 + 25 + 24) = 72 72/3 = 24.00 1.00
1999 24 71 23.67 0.33
2000 22 71 23.67 – 1.67
2001 25 73 24.33 0.67
2002 26 78 26.00 0
2003 27 → (26 + 27 + 26) = 79 79/3 = 26.33 0.67
2004 26 — — —

Odd and Even Number of Years When the chosen period of length n is an odd number, the moving
average period is centred on i (middle period in the consecutive sequence of n periods). For
instance with n = 5, MA3(5) is centred on the third year, MA4(5) is centred on the fourth year...,
and MA9(5) is centred on the ninth year.
No moving average can be obtained for the first (n – 1)/2 years or the last (n – 1)/2 year of
the series. Thus for a 5-year moving average, we cannot make computations for the just two years
or the last two years of the series.
When the chosen period of length n is an even numbers, equal parts can easily be formed
and an average of each part is obtained. For example, if n = 4, then the first moving average M3
(placed at period 3) is an average of the first four data values, and the second moving average M 4
(placed at period 4) is the average of data values 2 through 5. The average of M 3 and M4 is placed
at period 3 because it is an average of data values for period 1 through 5.
Example 11.3: Assume a four-year cycle and calculate the trend by the method of moving average
from the following data relating to the production of tea in India:

Year Production Year Production

(million lbs) (million lbs)
1987 464 1992 540
1988 515 1993 557
1989 518 1994 571
1990 467 1995 586
1991 502 1996 612
376 BUSINESS S T A T I S T I C S

Solution: The first 4-year moving average is:

464 + 515 + 518 + 467 1964

MA3(4) = = = 491.00
4 4
This moving average is centred on the middle value, that is, the third year of the series. Similarly,

515 + 518 + 467 + 502 2002

MA4(4) = = = 500.50
4 4
This moving average is centred on the fourth year of the series.
Table 11.2 presents the data along with the computations of 4-year moving averages.

Table 11.2 Calculation of Trend and Short-term Fluctuations

Year Production 4-Yearly 4-Yearly Moving 4-Yearly Moving

(mn lbs) Moving Totals Average Average Centred
1987 464 — — —
1988 515 — — —
→1964 491.00
1989 518 → 495.75
→ 2002 500.50
1990 467 → 503.62
→ 2027 506.75
1991 502 → 511.62
2066 512.50
1992 540 529.50
2170 542.50
1993 557 553.00
2254 563.50
1994 571 572.50
2326 581.50
1995 586 — — —
1996 612 — — —

Weighted Moving Averages

In moving averages, each observation is given equal importance (weight). However, it may be desired
to place more weight (importance) on certain periods of time than on others. So a moving average in
which some time periods are weighted differently than others is called a weighted moving average. In such a case
different values may be assigned to compute a weighted average of the most recent n values. Choice of
weights is somewhat arbitrary because there is no set formula to determine them. In most cases, the
most recent observation receives the most weightage, and the weight decreases for older data values.
A weighted moving average is computed as:

Σ (Weight for period n) (Data value in period n)

Weighted moving average =
Σ Weights
 FORECASTING AND TIME SERIES ANALYSIS 377

Example 11.4: Vacuum cleaner sales for 12 months is given below. The owner of the supermarket
decides to forecast sales by weighting the past three months as follows:

Weight Applied Month

3 Last month
2 Two months ago
1 Three months ago
6

Months : 1 2 3 4 5 6 7 8 9 10 11 12
Actual sales : 10 12 13 16 19 23 26 30 28 18 16 14
(in units)

Solution: The results of 3-month weighted average are shown in Table 11.3
xweighted = 3M
t – 1 + 2Mt – 2 + 1Mt – 3

1
= [3 × Sales last month + 2 × Sales two months ago +1 × Sales three months ago]
6

Table 11.3 Weighted Moving Average

Month Actual Sales Three-month Weighted Moving Average

1 10 —
2 12 —
3 13 —
1 121
4 16 [(3 × 13) + (2 × 12) + (1 × 10)] =
6 6
1 141
5 19 [(3 × 16) + (2 × 13) + (1 × 12)] =
6 3
1
6 23 [(3 × 19) + (2 × 16) + (1 × 13)] = 17
6
1 201
7 26 [(3 × 23) + (2 × 19) + (1 × 16)] =
6 2
1 235
8 30 [(3 × 26) + (2 × 23) + (1 × 19)] =
6 6
1 271
9 28 [(3 × 30) + (2 × 26) + (1 × 23)] =
6 2
1 289
10 18 [(3 × 28) + (2 × 30) + (1 × 26)] =
6 3
1 231
11 16 [(3 × 18) + (2 × 28) + (1 × 30)] =
6 3
1 182
12 14 [(3 × 16) + (2 × 18) + (1 × 28)] =
6 3
378 BUSINESS S T A T I S T I C S

Example 11.5: A food processor uses a moving average to forecast next month’s demand. Past
actual demand (in units) is shown below:
Month : 43 44 45 46 47 48 49 50 51
Actual
demand : 105 106 110 110 114 121 130 128 137
(a) Compute a simple five-month moving average to forecast demand for month 52.
(b) Compute a weighted three-month moving average where the weights are highest for the latest
months and descend in order of 3, 2, 1.
Solution: Calculations for five-month moving average are shown in Table 11.4.
Table 11.4 Five-month Moving Average

Month Actual Demand 5-month 5-month

Moving Total Moving Average
43 105 — —
44 106 — —
45 110 545 109.50
46 110 561 112.2
47 114 585 117.0
48 121 603 120.6
49 130 630 126.0
50 128 — —
51 137 — —

(a) Five-month average demand for month 52 is

∑x 114 + 121 + 130 + 128 + 137
= = 126 units
Number of periods 5
(b) Weighted three-month average as per weights is as follows:
Σ Weight × Data value
xweighted =
Σ weight

where Month Weight × Value = Total

51 3 × 137 = 411
50 2 × 128 = 256
49 1 × 130 = 130
6 797
797
xweighted = = 133 units.
6

Semi-Average Method
The semi-average method permits us to estimate the slope and intercept of the trend line quite easily
if a linear function will adequately describe the data. The procedure is simply to divide the data into
two parts and compute their respective arithmetic means. These two points are plotted corresponding
to their midpoint of the class interval covered by the respective part and then these points are joined
 FORECASTING AND TIME SERIES ANALYSIS 379

by a straight line, which is the required trend line. The arithmetic mean of the first part is the
intercept value, and the slope is determined by the ratio of the difference in the arithmetic mean of
the number of years between them, that is, the change per unit time. The resultant is a time series
of the form : ŷ = a + bx. The ŷ is the calculated trend value and a and b are the intercept and slope
values respectively. The equation should always be stated completely with reference to the year where
x = 0 and a description of the units of x and y.
The semi-average method of developing a trend equation is relatively easy to commute and may
be satisfactory if the trend is linear. If the data deviate much from linearity, the forecast will be
biased and less reliable.

Example 11.6: Fit a trend line to the following data by the method of semi-average and forecast
the sales for the year 2002.

Year Sales of Firm Year Sales of Firm

(thousand units) (thousand units)
1993 102 1997 108
1994 105 1998 116
1995 114 1999 112
1996 110

Solution: Since number of years are odd in number, therefore divide the data into equal parts
(A and B) of 3 years ignoring the middle year (1996). The average of part A and B is

102 + 105 + 114 321

yA = = = 107 units
3 3
108 + 116 + 112 336
yB = = = 112 units
3 3
Part A is centred upon 1994 and part B on 1998. Plot points 107 and 112 against their middle
years, 1994 and 1998. By joining these points, we obtain the required trend line as shown
Fig. 11.6. The line can be extended and be used for prediction.

Figure 11.6 Trend Line by the Method of Semi-Average

380 BUSINESS S T A T I S T I C S

To calculate the time-series ŷ = a + bx, we need

∆y change in sales
Slope = b = =
∆x change in year

112 − 107 5
= = = 1.25
1998 − 1994 4
Intercept = a = 107 units at 1994
Thus, the trend line is: ŷ = 107 + 1.25x
Since 2002 is 8-year distant from the origin (1994), therefore we have
ŷ = 107 + 1.25(8) = 117
Example 11.7: In the study of sales, a company obtained the following trend equation: yc = 16 + 2x
(Origin 1995, x unit = 1 year, y = total number of units sold).
The company has the physical facilities to provide only 30 units in a year and it believes that at
least for the next decade trend will continue as before. Find:
(a) What is the average annual increase in the number of units sold?
(b) By which year the company’s expected sales have equalled to its present capacity?
(c) Estimate the sales for the year 1998. [Delhi Univ., BCom(Hons), 2003]
Solution: (a) Trend equation is yc = 16 + 2x. Since slope of this line is b = 2, therefore average annual
increase is 2 units.
(b) Since the company’s present capacity is 30 units, substituting y = 30 in the trend equation, we
get 30 = 16 + 2x or x = 7. Thus, in seven years, the company’s expected sales have equalled
the present capacity. Since 1995 is taken as origin, therefore required year would be 1995 +
7 = year 2002.
(c) Since 1995 is origin, therefore, for estimating sales of 1998, putting x = 1998–1995 = 3 in the
trend equation we get y = 16 + 2(3) = 22 units.

Example 11.8: Trend equation for yearly sales (in ‘000 Rs.) for a commodity is : y = 81.6 + 28.8 x
(unit of x = 1 year, origin is July 16, 1991). Adjust the trend equation to find the monthly trend
values with Jan. 1992 as origin and find the trend values for March 1992.[Delhi Univ., B Com (Hons),
2003]

Solution: Annual trend equation is y = 81.6 + 28.8x. Therefore, monthly trend equation is
81.6 28.5
y= + x
12 12 × 12
Here x unit = one month and origin = July 16, 1991
Since the required origin is Jan. 1992 i.e., Jan. 16, 1992, the trend equation will be obtained by
increasing x by 6 months because, Jan. 16, 1992– July 16, 1991 = 6 months
y = 6.8 + 0.2 (x + 6) = 0.2x + 8
To find trend value for March 1992, we will put x = 2 in trend equation as March 16, 1992– Jan.
16, 1992 = 2 months
y = 0.2 (2) + 8 = 8.4 (Rs., in thousand)
 FORECASTING AND TIME SERIES ANALYSIS 381

Example 11.9: Give below is the quarterly trend equation for sales (Rs. in thousand) of a commodity:
yC = 130 + 1.8x
[Origin: first quarter of 2002; x unit = 1 quarter, y unit = average quarterly sales (Rs. in thousand)]
Convert the above equation to annual trend equation and estimate the sales for the year 2006.
[Delhi Univ., B Com (Hons), 2005]
Solution: Quarterly trend equation is: yC = 130 + 1.8 x; origin as first quarter of 2002, i.e. February
15, 2002. To convert it into annual trend, shift origin to June 30, 2002 (middle of year 2002). That is
shift x by June 30, 2002–February 15, 2002 = 4.5 months or 1.5 quarters
Thus, the trend equation with June 30, 2002 as origin becomes:
yC = 130 + 1.8 (x + 1.5) = 132.7 + 1.8 x
The annual trend equation then is
yC = 132.7 × 4 + (1.8) (16 x) = 530.8 + 28.8 x
Putting x = 4 to get expected sales for 2006 : yC = 530.8 + 28.8(4) = 530.8 + 115.2 = Rs.. 646.

S e l f-P r a c t i c e P r o b l e m s 11A

11.1 The owner of a small company manufactures

Month Accidents
a product. Since he started the company, the
number of units of the product he has sold is 1 280
represented by the following time series: 2 300
3 280
Year : 1995 1996 1997 1998 1999 2000 2001 4 280
Units : 100 120 95 105 108 102 112 5 270
sold
6 240
Find the trend line that describes the trend by 7 230
using the method of semi-averages. 8 230
11.2 Fit a trend line to the following data by the 9 220
freehand method: 10 200
11 210
Year Production of Steel Year Production of Steel
(million tonnes) (million tonnes) 12 200

1995 20 2000 25 Find the trend line that describes the trend by
1996 22 2001 23 using the method of semi-averages.
1997 24 2002 26 11.4 Calculate the three-month moving averages
1998 21 2003 25 from the following data:
1999 23 Jan. Feb. March April May June
57 65 63 72 69 78
11.3 A State Govt. is studying the number of traffic
July Aug. Sept. Oct. Nov. Dec.
fatalities in the state resulting from drunken
driving for each of the last 12 months: 82 81 90 92 95 97
[Osmania Univ., B.Com, 1996]
382 BUSINESS S T A T I S T I C S

11.5 Gross revenue data (Rs. in million) for a Travel Calculate a 5-year moving average for the
Agency for a 11-year period is as follows: unit cost of the product.

Ye a r Revenue 11.7 The following data provide a time series of

the number of Commercial and Industrial
1995 3 units failures during the period 1989–2004.
1996 6
1997 10
Year No. of Failures Year No. of Failures
1998 8
1999 7 1989 23 1997 9
2000 12 1990 26 1998 13
2001 14 1991 28 1999 11
2002 14 1992 32 2000 14
2003 18 1993 20 2001 12
2004 19 1994 12 2002 9
1995 12 2003 3
Calculate a 3-year moving average for the 1996 10 2004 1
revenue earned.
11.6 The owner of small manufacturing company Calculate a 5-year and 7-year moving average
has been concerned about the increase in for the number of units failure.
manufacturing costs over the past 10 years. 11.8 Estimate the trend values using the data given
The following data provide a time series of by taking a four-year moving average :
the cost per unit for the company’s leading
product over the past 10 years.
Year Value Year Value
Year Cost per Unit Year Cost per Unit 1990 12 1997 100
1995 332 2000 405 1991 25 1998 82
1996 317 2001 410 1992 39 1999 65
1997 357 2002 427 1993 54 2000 49
1998 392 2003 405 1994 70 2001 34
1999 402 2004 438 1995 87 2002 20
1996 105 2003 7

Hints and Answers

11.2
11.1

Year ( y) Units Sold (x)

1995 100 
1996 120  315/3 = 105.00 = a
1997 95 
1998 105
1999 108

2000 102  322/3 = 107.33 = b
2001 112 
Trend line y = 105 + 107.33x.
 FORECASTING AND TIME SERIES ANALYSIS 383

11.3 11.6
Month Accidents Year Per Unit Cost 5-year 5-year
1 280 Moving Total Moving Average
2 300 1995 332 — —
3 280
4 280 1996 317 — —
5 270
—→
1997 357 1800 1800/5 = 360.0
6 240 —→
7 230 1998 392 1873 1873/5 = 374.6
8 230 1999 402 1966 1966/5 = 393.2
9 220
2000 405 2036 407.2
10 200
11 210 2001 410 2049 409.8
12 200 2002 427 2085 417.0
Average of first 6 months, 2003 405 — —
a = 1650/6 = 275 2004 438 — —
Average of last 6 months, b = 1290/6 = 215
Trend line y = 275 + 215 x. 11.7
11.4 Year Number 5-year 5-year 7-year 7-year
of Failures Moving Moving Moving Moving
Month Values 3-month 3-month Moving
Total Average Total Average
Total Average
1989 23 — — — —
Jan. 57 — —
Feb. 65 185 185/3 = 61.67 1990 26 — — — —
March 63 200 200/3 = 66.67 1991 28
—→ 129 25.8 — —
April 72 204 204/3 = 68.00 —→
1992 32 118 23.6 153 21.9
May 69 219 73.00
June 78 229 76.33 1993 20 104 20.8 140 20.0
July 82 241 80.33 1994 12 86 17.2 123 17.6
Aug. 81 253 84.33
1995 12 63 12.6 108 15.4
Sept. 90 263 87.67
Oct. 92 277 92.38 1996 10 56 11.2 87 12.4
Nov. 95 284 94.67 1997 9 55 11.0 81 11.6
Dec. 97 — —
1998 13 57 11.4 81 11.6
11.5 1999 11 59 11.8 78 11.1

Year Revenue 3-year 3-year 2000 14 59 11.8 71 10.1

Moving Total Moving Average 2001 12 69 9.8 63 5.0
1995 3 — 2002 29 39 7.9 — —
1996 6 —→ 19 19/3 = 6.33 2003 23 — — — —
1997 10 —→ 24 24/3 = 8.00 2004 21 — — — —
1998 8 —→ 21 21/3 = 7.00
1999 7 25 8.33
2000 12 32 10.66
2001 14 34 11.33
2002 14 46 15.33
2003 18 51 17.00
2004 19 — —
384 BUSINESS S T A T I S T I C S

11.8 1996 105 92.00

374 93.5
Year Value 4-year 4-year 4-year Moving 1997 100 90.75
Moving Moving 352 88.0
Average Centred 1998 82 81.00
Total Average 296 74.0
1990 12 — — — — 1999 65 65.75
1991 25 — — — — 230 57.5
2000 49 49.75
—→ 130 130/4=32.5
1992 39 (32.5 + 47)/2 168 42.0
= 39.75 2001 34 34.75
188 188/4=47.0 110 27.5
—→ 2002 20 — — —
1993 54 (47 + 62.5)/2
= 54.75 2003 7 — — —
250 250/4=62.5
→
1994 70 70.75
316 79.0
1995 87 84.75
362 90.5

11.6 TREND PROJECTION METHODS

A trend is the long-run general direction (upward, downward or constant) of a business climate over
a period of several year. It is best represented by a straight line.
The trend projection method fits a trend line to a time series data and then projects medium-to-
long-range forecasts. Several possible trend fits can be explored (such as exponential and quadratic),
depending upon movement of time-series data. In this section, we will discuss linear, quadratic and
exponential trend models. Since seasonal effects can compound trend analysis, it is assumed that no
seasonal effects occur in the data or are removed before establishing the trend.

Reasons to study trend: A few reasons to study trends are as follows:

1. The study of trend helps to describe the long-run general direction (upward, downward, constant)
of a business climate over a period of several years.
2. The study allows us to use trends as an aid in making intermediate and long-range forecasting
projections in the future.
3. The study of trends help to esolate and then eliminate its influencing effects on the time-series
model.

11.6.1 Linear Trend Model

The method of least squares from regression analysis is used to find the trend line of best fit to a time series
data. The regression trend line (y) is defined by the following equation:
ŷ = a + bx
where ŷ = predicted value of the dependent variable
a = y-axis intercept
b = slope of the regression line (or the rate of change in y for a given change in x)
x = independent variable (which is time in this case)
 FORECASTING AND TIME SERIES ANALYSIS 385

The trend line of best fit has the properties that (i) the summation of all vertical deviations about
it is zero, that is, Σ (y – ŷ ) = 0, (ii) the summation of all vertical deviations squared is a minimum, that
is, Σ (y – ŷ ) is least, and (iii) the line goes through the mean values of variables x and y. For linear
equations, it is found by the simultaneous solution for a and b of the two normal equations:
Σ y = na + bΣ x and Σ x y = aΣx + bΣx2
where the data can be coded so that Σ x = 0, two terms in these equations drop out and we have
Σ y = n a and Σ x y = bΣx2
Coding is easily done with time-series data. For coding the data, we choose the centre of the time
period as x = 0 and have an equal number of plus and minus periods on each side of the trend line
which sum to zero.
Alternately, we can also find the values of constants a and b for any regression line as:
Σxy− n x y
b = and a = y − bx
Σ x 2 − n ( x )2
Example 11.10: Below are given the figures of production (in thousand quintals) of a sugar factory:
Year : 1995 1996 1997 1998 1999 2000 2001
Production : 80 90 92 83 94 99 92
(a) Fit a straight line trend to these figures.
(b) Plot these figures on a graph and show the trend line.
(c) Estimate the production in 2004. [Bangalore Univ., B.Com, 1998]
Solution: (a) Using normal equations and the sugar production data we can compute constants a and
b as shown in Table 11.5:

Table 11.5 Calculation for Least Squares Equation

Year Time Period Production x2 xy Trend Values

(x) (x) y
1995 1 80 1 80 84
1996 2 90 4 180 86
1997 3 92 9 276 88
1998 4 83 16 332 90
1999 5 94 25 470 92
2000 6 99 36 594 94
2001 7 92 49 644 96
28 630 140 2576

Σx 28 Σy 630
x = = = 4, y = = = 90
n 7 n 7
Σx y − nx y 2576 − 7(4) (90) 56
b= = 2
= =2
2
Σ x − n (x ) 2
140 − 7(4) 28
a = y – b x = 90 – 2(4) = 82
Therefore, linear trend component for the production of sugar is:
ŷ = a + bx = 82 + 2x
386 BUSINESS S T A T I S T I C S

The slope b = 2 indicates that over the past 7 years, the production of sugar had an average growth
of about 2 thousand quintals per year.

Figure 11.7 Linear Trend for Production of Sugar

(b) Plotting points on the graph paper, we get an actual graph representing production of sugar
over the past 7 years. Join the point a = 82 and b = 2 (corresponds to 1996) on the graph we get
a trend line as shown in Fig. 11.7.
(c) The production of sugar for year 2004 will be
ŷ = 82 + 2 (10) = 102 thousand quintals

Example 11.11: The following table relates to the tourist arrivals (in millions) during 1994 to 2000
in India:
Year : 1994 1995 1996 1997 1998 1999 2000
Tourists arrivals : 18 20 23 25 24 28 30
Fit a straight line trend by the method of least squares and estimate the number of tourists that
would arrive in the year 2004.
Solution: Using normal equations and the tourists arrival data we can compute constants a and b as
shown in Table 11.6:
Table 11.6 Calculations for Least Squares Equation

Year Time Tourist

Scale Arrivals
(x) y xy x2
1994 –3 18 – 54 9
1995 –2 20 – 40 4
1996 –1 23 – 23 1
1997 0 25 0 0
1998 1 24 24 1
1999 2 28 56 4
2000 3 30 90 9
168 53 28
 FORECASTING AND TIME SERIES ANALYSIS 387

Σx Σy 168
x = = 0, y = = = 24
n n 7
Σx y − nx y 53
b= = = 1.893;
2
Σ x − n (x ) 2 28

a = y – b x = 24 – 1.893(0) = 24
Therefore, the linear trend component for arrival of tourists is
y = a + bx = 24 + 1.893x
The estimated number of tourists that would arrive in the year 2004 are:
y = 24 + 1.893 (7) = 37.251 million (measured from 1997 = origin)
Example 11.12: From the following data, calculate trend by method of least squares:
Year : 1970 1971 1972 1973 1974 1975 1976
Profit (’000 Rs.) : 300 700 600 800 900 700 1000
[Delhi Univ., BCom (P) 1985]
Solution: Using normal equations the calculations required to determine trend are shown below:

Year (t) Profit (Rs. 000) y x = t – 1973 x2 xy

1970 300 –3 9 –900
1971 700 –2 4 –1400
1972 600 –1 1 –600
1973 800 0 0 0
1974 900 1 1 900
1975 700 2 4 1400
1976 1000 3 9 3000
5000 0 28 2400

Equation of straight line trend, yc = a + bx

Σ y 500 Σ xy 2400
where a = = = 714.28 and b = = = 85.71
n 7 Σ x2 28
Hence the trend line becomes:
y = a + bx = 714.28 + 85.71x
Y1970 = 714.28 + 85.71 (–3) = 457.15; Y1971 = 714.28 + 85.71 (–2) = 542.86
Y1972 = 714.28 + 85.71 (–1) = 628.57; Y1973 = 714.28 + 85.71(0) = 714.28
Y1974 = 714.28 + 85.71(1) = 799.99; Y1975 = 714.28 + 85.71(2) = 885.70
Y1976 = 714.28 + 85.71(3) = 971.41
Example 11.13: Calculate trend values by method of least squares from the data given below and also
estimate the sales for year 1991:
Year : 1986 1987 1988 1989 1990
Sales (in crores) : 12 18 20 23 27 [Delhi Univ., BCom (P), 1992]
Solution: Using normal equations, calculations required to estimate the sales for the year 1991 are
shown below:
388 BUSINESS S T A T I S T I C S

Year (t) x = t – 1988 y (sales) x2 xy

1986 –2 12 4 –24
1987 –1 18 1 –18
1988 0 20 0 0
1989 1 23 1 23
1990 2 27 4 54

Σ y 100 Σ xy 35
Let the straight line trend be: y + a + bx, where a = = = 20, and b = 2
= = 3.5
n 5 Σx 10
Hence, y = 20 + 3.5x. Putting x = 3, in the trend line to estimate sales for year 1991 as follows:
Y1991 = 20 + 3.5(3) = Rs. 30.5 crore.
Example 11.14: Fit a straight line trend to the following data by least squares method after summing
the given quarterly data to yearly data. Also tabulate short term fluctuations.
Export of Cotton Textile (Million Rs.)

Year Quarter I QuarterII Quarter III Quarter IV

1998 10 13 14 12
1999 12 14 15 13
2000 13 15 18 14
2001 15 18 21 18
2002 15 22 23 20

Plot the trend values and actual values and draw the trend line. [Delhi Univ., BCom (Hons), 1989]
Solution: Let the year 2000 be origin. Also x represents time in years and y represents exports in
millions of rupees.
Convert the quarterly data into yearly data as follows:

Year Quarter I QuarterII Quarter III Quarter IV Yearly total

1998 10 13 14 12 49
1999 12 14 15 13 54
2000 13 15 18 14 60
2001 15 18 21 18 72
2002 15 22 23 20 80

Calculations required to fit a straight line trend are shown below:

Year x x2 y xy xc = 63 + 8x Short Term Fluctuations
(t – 2000) y – yc

1998 –2 4 49 –98 63 + 8 (–2) = 47 49 – 47 = 2

1999 –1 1 54 –54 63 + 8 (–1) = 55 54 – 55 = –1
2000 0 0 60 0 63 + 8(0) = 63 60 – 63 = –3
2001 1 1 72 72 63 + 8(1) = 71 72 – 71 = 1
2002 2 4 80 160 63 + 8(2) = 79 80 – 79 = 1
0 10 315 80
 FORECASTING AND TIME SERIES ANALYSIS 389

Σy 315 Σxy 80
Let the stright line trend be, y = a + bx, where a = = = 63 and b = = =8
n 5 Σx2 10
Hence, y = a + bx = 63 + 8x
Plotting trend values on a graph, the trend line so obtained is shown below:

11.6.2 Exponential Trend Model

When the given values of dependent variable y form approximately a geometric progression while the
corresponding independent variable x values form an arithmetic progression, the relationship between
variables x and y is given by an exponential function, and the best fitting curve is said to describe the
exponential trend. Data from the fields of biology, banking, and economics frequently exhibit such a
trend. For example, growth of bacteria, money accumulating at compound interest, sales or earnings
over a short period, and so on, follow exponential growth.
The characteristic property of this law is that the rate of growth, that is, the rate of change of y
with respect to x is proportional to the values of the function. The following function has this property.
y = a bc x, a > 0
The letter b is a fixed constant, usually either 10 or e, where a is a constant to be determined from the data.
To assume that the law of growth will continue is usually unwarranted, so only short range
predictions can be made with any considerable degree of reliability.
If we take logarithms (with base 10) of both sides of the above equation, we obtain
log y = log a + (c log b) x
For b = 10, log b = 1, but for b = e, log b = 0.4343 (approx.). In either case, this equation is of the form
y′ = c + dx (11-1)
where y′ = log y, c = log a, and d = c log b.
Equation (11-1) represents a straight line. A method of fitting an exponential trend line to a set
of observed values of y is to fit a straight trend line to the logarithms of the y-values.
390 BUSINESS S T A T I S T I C S

In order to find out the values of constants a and b in the exponential function, the two normal
equations to be solved are
Σ log y = n log a + log b Σ x
Σ x log y = log a Σ x + log b Σ x2
When the data is coded so that Σx = 0, the two normal equations become
1
Σ log y = n log a or log a = Σ log y
n
Σ x log y
and Σ x log y = log b Σ x2 or log b =
Σ x2
Coding is easily done with time-series data by simply designating the center of the time period as
x = 0, and have equal number of plus and minus period on each side which sum to zero.
Example 11.15: The sales (Rs. in million) of a company for the years 1995 to 1999 are:
Year : 1997 1998 1999 2000 2001
Sales : 1.6 4.5 13.8 40.2 125.0
Find the exponential trend for the given data and estimate the sales for 2004.
Solution: The computational time can be reduced by coding the data. For this consider
u = x – 3. The necessary computations are shown in Table 11.7.
Table 11.7 Calculation for Least Squares Equation

Year Time u=x–3 u2 Sales Log y u Log y

Period y
x
1997 1 –2 4 1.60 0.2041 – 0.4082
1998 2 –1 1 4.50 0.6532 – 0.6532
1999 3 0 0 13.80 1.1390 0
2000 4 1 1 40.20 1.6042 1.6042
2001 5 2 4 125.00 2.0969 4.1938
10 5.6983 4.7366

1 1
log a = Σ log y = (5.6983) = 1.1397
n 5
Σ u log y 4.7366
log b = 2
= = 0.4737
Σu 10
Therefore log y = log a + (x + 3) log b = 1.1397 + 0.4737x
For sales during 2004, x = 3, and we obtain
log y = 1.1397 + 0.4737 (3) = 2.5608
or y = antilog (2.5608) = 363.80
Example 11.16: Fit an exponential trend to the following data:
Year : 2001 2002 2003 2004 2005 2006 2007
Sales (in lakhs of Rs.) : 32 47 65 92 132 190 275
 FORECASTING AND TIME SERIES ANALYSIS 391

Solution: Calculations to fit an exponential trend to the given data are shown below:
Year x Sales (y) log y x2 x log y

2001 –3 32 1.5051 9 – 4.5153

2002 –2 47 1.6721 4 – 3.3442
2003 –1 65 1.8129 1 1.8129
2004 0 92 1.9638 0 0
2005 1 132 2.1206 1 2.1206
2006 2 190 2.2788 4 4.5576
2007 3 275 2.4393 9 7.3179
0 833 13.7926 28 4.3237

Let year 2004 be the origin and the exponential trend equation be y = abx. Then normal equations
are
Σ log y = n log a + log bΣx or 13.7926 = 7log a
Σx log y = log a Σx + log b Σx2 or log a = 1.9704
Also 4.3237 = 0 + 28 log b or log b = 0.154
Then log y = 1.9704 + 0.154x
For sales in 2008, x = 4. Thus log y = 1.9704 + 0.154(4) = 2.5864. Hence
y = Antilog (2.5864) = 385.9

11.6.3 Changing the Origin and Scale of Equations

When a moving average or trend value is calculated it is assumed to be centred in the middle of the
month (fifteenth day) or the year (July 1). Similarly, the forecast value is assumed to be centred in the
middle of the future period. However, the reference point (origin) can be shifted, or the units of
variables x and y are changed to monthly or quarterly values if desired. The procedure is as follows:
(i) Shift the origin, simply by adding or subtracting the desired number of periods from
independent variable x in the original forecasting equation.
(ii) Change the time units from annual values to monthly values by dividing independent variable
x by 12.
(iii) Change the y units from annual to monthly values, the entire right-hand side of the equation
must be divided by 12.
Example 11.17: The following forecasting equation has been derived by a least-squares method:
ŷ = 10.27 + 1.65x (Base year: 1997; x = years; y = tonnes/year)
Rewrite the equation by
(a) shifting the origin to 2002.
(b) expressing x units in months, retaining y in tonnes/year.
(c) expressing x units in months and y in tonnes/month.
Solution: (a) Shifting of origin can be done by adding the desired number of period
5 (1997 to 2002) to x in the given equation. That is
ŷ = 10.27 + 1.65 (x + 5) = 18.52 + 1.65x
where 2002 = 0, x = years, y = tonnes/year.
392 BUSINESS S T A T I S T I C S

(b) Expressing x units in months

1.65 x
ŷ = 10.27 +
= 10.27 + 0.14x
12
where July 1, 1997 = 0, x = months, y = tonnes/year.
(c) Expressing y in tonnes/month, retaining x in months
1
(10.27 + 0.14 x) = 0.86 + 0.01x
ŷ =
12
where July 1, 1997 = 0, x = months, y = tonnes/month.
Remarks
1. If both x and y are to be expressed in months together, then divide constant ‘a’ by 12 and constant
‘b’ by 24. It is because data are sums of 12 months. Thus monthly trend equation becomes
a b
Linear trend : ŷ = + x
12 24
a b c
Parabolic trend : + x+
ŷ = x2
12 144 1728
But if data are given as monthly averages per year, then value of ‘a’ remains unchanged, ‘b’ is
divided by 12 and ‘c’ by 144.
2. The annual trend equation can be reduced to quarterly trend equation as:
a b a b
ŷ = + x = + x
4 4 × 12 4 48

S e l f-P r a c t i c e P r o b l e m s 11B

11.9 The general manager of a building materials (b) Determine a point estimate for
production plant feels that the demand for plasterboard shipments when the number
plasterboard shipments may be related to the of construction permits is 30.
number of construction permits issued in the 11.10 A company that manufactures steel observed
country during the previous quarter. The the production of steel (in metric tonnes)
manager has collected the data shown in the represented by the time-series:
table. Year : 1996 1997 1998 1999 2000 2001 2002
Production
Construction Plasterboard of steel : 60 72 75 65 80 85 95
Permits Shipments (a) Find the linear equation that describes the
trend in the production of steel by the
15 6
company.
9 4 (b) Estimate the production of steel in 2003.
40 16 11.11 Fit a straight line trend by the method of least
20 6 squares to the following data. Assuming that
25 13 the same rate of change continues, what would
25 9
be the predicted earning (Rs. in lakh) for the
15 10
year 2004?
35 16
Year : 19951996199719981999200020012002
(a) Use the normal equations to derive a Earnings: 38 40 65 72 69 60 87 95
regression forecasting equation. [Agra Univ., BCom 1996; MD Univ., BCom, 1998]
 FORECASTING AND TIME SERIES ANALYSIS 393

11.12 The sales (Rs. in lakh) of a company for the the percentage of private industry jobs that
years 1990 to 1996 are given below: are managerial. The following data show the
percentage of females who are managers
Year : 1998 1999 2000 2001 2002 2003 2004
from 1996 to 2003.
Sales : 32 47 65 88 132 190 275
Find trend values by using the equation yc Years : 1996 97 98 99 00 01 02 03
= a bx and estimate the value for 2005. Percentage : 6.7 5.3 4.3 6.1 5.6 7.9 5.8 6.1
[Delhi Univ., B.Com, 1996] (a) Develop a linear trend line for this
11.13 A company that specializes in the production of time series through 2001 only.
petrol filters has recorded the following (b) Use this trend to estimate the
production (in ’000 units) over the last 7 years. percentage of females who are
managers in 2004.
Years : 1995 96 97 98 99 00 01 11.15 A company develops, markets, manufactures,
Production : 42 49 62 75 92 122 158 and sells integrated wide-area network access
(a) Develop a second-degree estimating products. The following are annual sales (Rs.
equation that best describes these in million) data from 1998 to 2004.
data. Year : 1998 1999 2000 2001 2002 2003 2004
(b) Estimate the production in 2005. Sales : 16 17 25 28 32 43 50
11.14 In 1996 a firm began downsizing in order to (a) Develop the second-degree estimating
reduce its costs. One of the results of these equation that best describes these data.
cost cutting measures has been a decline in (b) Use the trend equation to forecast sales
for 2005.

Hints and Answers

11.9 (a) 11.10 a = Σ y/n =532/7 = 76; b = Σ xy/Σ x2 = 136/28

= 4.857
x y xy x2 y2
(a) Trend line ŷ = a + bx = 76 + 4.857x
15 6 90 225 36
9 4 36 81 16 (b) For 2003, x = 4, ŷ = 76 + 4.857(4)
40 16 640 1,600 256 = 95.428 metric tonnes.
20 6 120 400 36 11.11 a = Σ y/n = 526/8 = 65.75;
25 13 325 625 169
b = Σ x y/ Σ x2 = 616/168 = 3.667
25 9 225 625 81
15 10 150 225 100 Trend line : ŷ = a + bx = 65.75 + 3.667x
35 16 560 1,225 256 For 2004, x = 11; ŷ = 65.75 + 3.667 (11)
184 80 2,146 5,006 950 = Rs. 106.087 lakh.
n= 8 pairs of observations; 1 1
11.12 log a = Σ log y = (13.7926) = 1.9704
x = 184/8 = 23; y = 80/8 = 10 n 7
Σ y = na + bΣ x Σ x log y 4.3237
log b = = = 0.154
or 80 = 8a + 184b Σ x2 28
Σ x y = Σ x + bΣ x2
Thus log y = log a + x log b
or 2,146 = 184 a + 5,006 b
After solving equations we get a = 0.91 and = 1.9704 + 0.154x
b = 0.395. Therefore the equation is: For 2005, x = 4;
y = 0.91 + 0.395x log y = 1.9704 + 0.154 (4) = 2.5864
(b) For x = 30, we have y = 0.91 + 0.395 (30) y = Antilog (2.5864)
= 13 shipments (approx.) = Rs. 385.9 lakh.
394 BUSINESS S T A T I S T I C S

11.13 (a) Solving the equations

Year Period Deviation x2 x4 y xy x2y

Σ y = na + bΣx
from or 47.8 = 8 a – 12b
1998 (x)
Σxy = a Σ x + b Σ x2
1995 1 –3 9 81 42 – 126 378
or – 67.4 = – 12a + 60b
1996 2 –2 4 16 49 – 98 196
1997 3 –1 1 1 62 – 62 62 We get a = 6.28 and b = 0.102
1998 4 0 0 0 75 0 0 Hence ŷ = a + bx = 6.128 + 0.102x
1999 5 1 1 1 92 + 92 92
2000 6 2 4 16 122 + 244 488 (b) For 2004, x = 3; ŷ = 6.128 + 0.102 (3)
2001 7 3 9 81 158 + 474 1422 = 6.434 per cent.
0 28 196 600 524 2638 11.15
(a) Solving the equations Year Time Deviation Sales xy x2 x4 x2 y
Σ y = n a + c Σ x2 Period from y
or 600 = 7a + 28c 2001 (x)
Σ x 2 y = aΣ x2 + cΣ x4 1998 1 –3 16 – 48 9 81 144
or 2638 = 28 a + 196 c
1999 2 –2 17 – 34 4 16 68
Σ x y = b Σ x2
2000 3 –1 25 – 25 1 1 25
or 524 = 28 b
2001 4 0 28 0 0 0 0
We get a = 80.05, b = 18.71 and
2002 5 1 32 32 1 1 32
c = – 1.417
2003 6 2 43 86 4 16 172
Hence ŷ = a + bx + cx2
2004 7 3 50 150 9 81 450
= 80.05 + 18.71x – 1.417x2
0 211 161 28 196 891
(b) For 2005, x = 8;
ŷ = 80.05 + 18.71(8) – 1.417(8)2 (a) Solving the equations
= Rs. 139.042 thousand. Σ y = n a + cΣ x2
11.14 or 211 = 7a + 28 c
Σ x 2 y = a Σ x2 + cΣ x4
Year Time Deviation Percentage xy x2
Period from of or 891 = 28 a + 196 c
2001 Females Σ x y = b Σ x2
x y
or 161 = 28 b
1996 1 –5 6.7 – 33.5 25 We get a = 27.904, b = 5.75 and
1997 2 –4 5.3 – 21.2 16
c = 0.559
1998 3 –3 4.3 – 12.9 9
1999 4 –2 6.1 – 12.2 4 ŷ = a + bx + cx2
2000 5 –1 5.6 – 6.6 1 = 27.904 + 5.75x + 0.559 x2
2001 6 0 7.9 0 0 For 2005, x = 4;
2002 7 1 5.8 5.8 1 y = 27.904 + 5.75(4) + 0.559 (4)2
2003 8 2 6.1 12.2 4 = 59.848
– 12 47.8 –68.4 60
 FORECASTING AND TIME SERIES ANALYSIS 395

11.7 MEASUREMENT OF SEASONAL EFFECTS

As mentioned earlier that time-series data consists of four components: trend, cyclical effects, seasonal
effects and irregular fluctuations. In this section, we will discuss techniques for identifying seasonal
effects in a time-series data. Seasonal effect is defined as the repetitive and predictable pattern of data
behaviour in a time-series around the trend line during particular time intervals of the year. In order to
measure (or detect) the seasonal effect, time period must be less than one year such as days, weeks,
months, or quarters.
Seasonal effects arises as the result of natural changes in the seasons during the year or may result
due to habits, customs, or festivals that occur at the same time year after year.
We have three main reasons to study seasonal effects:
(i) The description of the seasonal effect provides a better understanding of the impact this
component has upon a particular time-series.
(ii) Once the seasonal pattern that exists is established, seasonal effect can be eliminated from the
time-series in order to observe the effect of the other components, such as cyclical and irregular
components. Elimination of seasonal effect from the series is referred to as deseasonalizing or
seasonal adjusting of the data.
(iii) Trend analysis may be adequate for long-range forecast, but for short-run predictions,
knowledge of seasonal effects on time-series data is essential for projection of past pattern
into the future.
Remarks:
1. In an additive time-series model, we can estimate the seasonal component as:
S = Y – (T + C + I)
In the absence of C and I, we have S = Y – T. That is, the seasonal component is the difference
between actual data values in series and the trend values.
2. One of the technique for isolating the effects of seasonality is decomposition. The process of
decomposition begins by determining T.C for each and dividing the time-series data (T.C.S.I) by
T.C. The resulting expression contains seasonal effects along with irregular fluctuations
T.C.S.I
= S.I.
T.C
A method for eliminating irregular fluctuations can be applied, leaving only the seasonal
effects as shown below:
T.S.C.I Y
Seasonal effect = = × 100%
T.C.I T.C.I
3. The process of eliminating the effects of seasonality from a time-series data is referred to as de-
seasonalization or seasonal adjustment. The data can be deseasonalized by dividing the actual values
Y by final adjusted seasonal effects, and is expressed as:
Y T.S.C.I
= = T. C. I × 100% ← Multiplicative
S S
Y – S = (T + S + C + I) – S = T + C + I ← Additive Model
Each adjusted seasonal index measures the average magnitude of seasonal influence on the
actual values of the time series for a given period within a year. By subtracting the base index of
100 (which represents the T and C components) from each seasonal index, the extent of the
influence of seasonal force can be measured.
396 BUSINESS S T A T I S T I C S

11.7.1 Seasonal Index

Seasonal effects are measured in terms of an index, called seasonal index, attached to each period of the
time series within a year. Hence, if monthly data are considered, there are 12 separate seasonal
indexes, one for each month. Similarly for quarterly data, there are 4 separate indexes. A seasonal
index is an average that indicates the percentage deviation of actual values of the time series from a base value
which excludes the short-term seasonal influences. The base time series value represents the trend/cyclical
influences only.
The following four methods are used to construct seasonal indexes to measure seasonal effects in
the time-series data:
(i) Method of simple averages
(ii) Ratio-to-trend method
(iii) Ratio-to-moving average method
(iv) Link relatives method.

11.7.2 Method of Simple Averages

This method is also called average percentage method because this method expresses the data of each month
or quarter as a percentage of the average of the year. The steps of the method are summarized below:
(i) Average the unadjusted data by years and months (or quarters if quarterly data are given).
(ii) Add the figures of each month and obtain the averages by dividing the monthly totals by the
number of years. Let the averages for 12 months be denoted by x1 , x2 , ..., x12 .
(iii) Obtain an average of monthly averages by dividing the total of monthly averages by 11. That is
x1 + x2 + . . . + x12
x =
12
(iv) Compute seasonal indexes for different months by expressing monthly averages as percentages
of the grand average x as follows:
Monthly average for month i
Seasonal index for month i = × 100
Average of monthly averages

xi
= × 100 (i = 1, 2, ...,12)
x
It is important to note that the average of the indexes will always be 100, that is, sum of the
indexes should be 1200 for 12 months, and sum should be 400 for 4 quarterly data. If the sum of
these 12 months percentages is not 1200, then the monthly percentage so obtained are adjusted by
multiplying these by a suitable factor [1200 ÷ (sum of the 12 values)].
Example 11.18: The seasonal indexes of the sale of readymade garments in a store are given below:

Quarter Seasonal Index

January to March 98
April to June 90
July to September 82
October to December 130

If the total sales of garments in the first quarter is worth Rs. 1,00,000, determine how much
worth of garments of this type should be kept in stock to meet the demand in each of the remaining
quarters. [Delhi Univ., B.Com, 1996]
 FORECASTING AND TIME SERIES ANALYSIS 397

Solution: Calculations of seasonal index for each quarter and estimated stock (in Rs.) is shown in
Table 11.8.
Table 11.8 Calculation of Estimated Stock

Quarter Seasonal Index (SI) Estimated Stock (Rs.)

Jan. —March 98 1,00,000.00
April —June 90 91,836.73*
July —Sept. 82 83,673.45
Oct. —Dec. 130 1,32,653.06

* These figures are calculated as follows:

Figure for first quarter × SI for second quarter
Seasonal index for second quarter =
SI for first quarter

Figure for first quarter × SI for third quarter

Seasonal index for third quarter =
SI for first quarter

Example 11.19: Use the method of monthly averages to determine the monthly indexes for the data
of production of a commodity for the years 2002 to 2004.

Month 2002 2003 2004

January 15 23 25
February 16 22 25
March 18 28 35
April 18 27 36
May 23 31 36
June 23 28 30
July 20 22 30
August 28 28 34
September 29 32 38
October 33 37 47
November 33 34 41
December 38 44 53

Solution: Computation of seasonal index by average percentage method based on the data is shown
in Table 11.9.
Monthly Average : 1080/20 = 90; 360/12 = 30; 1200/2 = 100
Monthly Average : 1080/20 = 90; 360/12 = 30; 1200/2 = 100
398 BUSINESS S T A T I S T I C S

Table 11.9 Calculation of Seasonal Indexes

Month 2002 2003 2004 Monthly Monthly Percentage

Total for Averages Average of
3 Years for 3 Years Monthly
Averages
Jan. 15 23 25 63 21 70
Feb. 16 22 25 63 21 70
March 18 28 35 81 27 90
April 18 27 36 81 27 90
May 23 31 36 90 30 100
June 23 28 30 81 27 90
July 20 22 30 72 24 80
Aug. 28 28 34 90 30 100
Sept. 29 32 38 99 33 110
Oct. 33 37 47 117 39 130
Nov. 33 34 41 108 36 120
Dec. 38 44 53 135 45 150
1080 360 1200

The average of monthly averages is obtained by dividing the total of monthly averages by 12. In
column 7 each monthly average for 3 years have been expressed as a percentage of the averages. For
example, the percentage for January is:
Monthly index for January = 21/30 = 70;
February = (21/30) × 100 = 70
March = (27/30) × 100 = 90, and so on
Example 11.20: The data on prices (Rs. in per kg) of a certain commodity during 2000 to 2004 are
shown below:
Quarter Years
2000 2001 2002 2003 2004
I 45 48 49 52 60
II 54 56 63 65 70
III 72 63 70 75 84
IV 60 56 65 72 66

Compute the seasonal indexes by the average percentage method and obtain the deseasonalized
values.
Solution: Calculations for quarterly averages are shown in Table 11.10.

50.8 + 61.6 + 72.8 + 63.8 249

Average of quarterly averages = = = 62.25
4 4
 FORECASTING AND TIME SERIES ANALYSIS 399

Table 11.10 Calculation Seasonal Indexes

Year Quarters
I II III IV
2000 45 54 72 60
2001 48 56 63 56
2002 49 63 70 65
2003 52 65 75 72
2004 60 70 84 66
Quarterly total 254 308 364 319
Quarterly average 50.8 61.6 72.8 63.8
Seasonal index 81.60 98.95 116.94 102.48

50.8
Thus, Seasonal index for quarter I = × 100 = 81.60
62.25
61.6
Seasonal index for quarter II = × 100 = 98.95
62.25
72.8
Seasonal index for quarter III = × 100 = 116.94
62.25
63.8
Seasonal index for quarter IV = × 100 = 102.48
62.25
Deseasonalized Values Seasonal influences are removed from a time-series data by dividing the actual
y value for each quarter by its corresponding seasonal index:
Actual quarterly value
Deseasonalized value = × 100
Seasonal index of corresponding quarter
The deseasonalized y values which are measured in the same unit as the actual values, reflect the
collective influence of trend, cyclical and irregular forces. The deseasonalized values are given in
Table 11.11.
Limitations of the method of simple averages This method is the simplest of all the methods for
measuring seasonal variation. However, the limitation of this method is that it assumes that
there is no trend component in the series, that is, C ⋅ S ⋅ I = 0 or trend is assumed to have
little impact on the time-series. This assumption is not always justified.

Table 11.11 Calculation for Least Squares Equation

Year Quarters
I II III IV
2000 55.14 54.57 61.57 58.54
2001 58.82 56.59 53.87 54.64
2002 60.00 63.66 59.85 63.42
2003 63.72 65.68 64.13 70.25
2004 73.52 70.74 71.83 64.40
400 BUSINESS S T A T I S T I C S

11.7.3 Ratio-to-Trend Method

This method is also known as the percentage trend method. This method is an improvement
over the method of simple averages. Because here it is assumed that seasonal variation for a
given month is a constant fraction of trend. The ratio-to-trend method isolates the seasonal
factor when the following ratios are computed:
T ⋅S ⋅ C ⋅I
= S⋅C⋅I
T
The steps of the method are summarized as follows:
(i) Compute the trend values by applying the least-squares method.
(ii) Eliminate the trend value. In a multiplicative model the trend is eliminated by dividing
the original data values by the corresponding trend values and multiplying these ratios
by 100. The values so obtained are free from trend.
(iii) Arrange the percentage data values obtained in Step (ii) according to months or
quarters as the case may be for the various years.
(iv) Find the monthly (or quarterly) averages of figures arranged in Step (iii) with any one
of the usual measures of central tendency—arithmetic mean, median.
(v) Find the grand average of monthly averages found in Step (iv). If the grand average
is 100, then the monthly averages represent seasonal indexes. Otherwise, an adjustment
is made by multiplying each index by a suitable factor [1200/(sum of the 12 values)] to
get the final seasonal indexes.
Example 11.21: Quarterly sales data (Rs. in million) in a super bazar are presented in the following
table for a four-year period

Year Quarters
I II III IV
2000 60 80 72 68
2001 68 104 100 88
2002 80 116 108 96
2003 108 152 136 124
2004 160 184 172 164

Calculate the seasonal index for each of the four quarters using the ratio-to-trend method.
Solution: Calculations to obtain annual trend values from the given quarterly data using the method
of least-squares are shown in Table 11.12.
Solving the following normal equations, we get
Σ y = na + bΣ x or 560 = 5a or a = 112
Σ xy = aΣ x + bΣ x2 or 240 = 10b or b = 24
Thus the yearly fitted trend line is: y = 112 + 24x. The value of b = 24 indicates yearly increase in
sales. Thus the quarterly increment will be 24/4 = 6.
To calculate quarterly trend values, consider first the year 2000. The trend value for this year is
64. This is the value for the middle of the year 2000, that is, half of the 2nd quarter and half of the
3rd quarter. Since quarterly increment is 6, the trend value for the 2nd quarter of 2000 would be
 FORECASTING AND TIME SERIES ANALYSIS 401

Table 11.12 Calculation of Trend Values

Year Yearly Yearly Deviation x2 xy Trend

Total Average From Mid-Year Values
(1) y = (2)/4 x y
2000 280 70 –2 4 – 140 64
2001 360 90 –1 1 – 90 88
2002 400 100 0 0 0 0
2003 520 130 1 1 130 112
2004 680 170 2 4 340 160
560 10 240

64 – (6/2) = 61 and for the 3rd quarter it would be 64 + (6/2) = 67. The value for the 1st quarter of
2000 would be 61 – 6 = 55 and for the 4th quarter it would be 67 + 6 = 73. Similarly, trend values
of the various quarters of other years can be calculated as shown in Table 11.13.

Table 11.13 Quarterly Trend Values

Year Quarters
I II III IV
2000 55 61 67 73
2001 79 85 91 97
2002 103 109 115 121
2003 127 133 139 145
2004 151 157 163 169

After getting the trend values, the given data values in the time-series are expressed as percent-
ages of the corresponding trend values in Table 11.13. Thus for the 1st quarter of 2000, this percent-
age would be (60/55) × 100 = 109.09; for the 2nd quarter it would be (80/61) × 100 = 131.15, and
so on. Other values can be calculated in the same manner as shown in Table 11.14.
The total of average of seasonal indexes is 403.12 (>400). Thus we apply the correction factor
(400/403.12) = 0.992. Now each quarterly average is multiplied by 0.992 to get the adjusted seasonal
index as shown in Table 11.14.
The seasonal index 92.02 in the first quarter means that on average sales trend to be depressed by
the presence of seasonal forces to the extent of approx. (100 – 92.02) = 7.98%. Alternatively, values of
time series would be approx. (7.98/92.02)×100 = 8.67% higher had seasonal influences not been
present.
402 BUSINESS S T A T I S T I C S

Table 11.14 Ratio-to-Trend Values

Year Quarters
I II III IV
2000 109.09 131.15 107.46 93.15
2001 86.08 122.35 109.89 90.72
2002 77.67 106.42 93.91 79.34
2003 85.04 114.29 97.84 85.52
2004 105.96 117.20 105.52 97.04
Total 463.84 591.41 514.62 445.77
Average 92.77 118.28 102.92 89.15
Adjusted = 403.12
seasonal index 92.02 117.33 102.09 88.43

Example 11.22: The production of a commodity during 1993-98 is given below. Fit the second
degree parabola to these data and estimate the production for the year 2000:
Year : 1993 1994 1995 1996 1997 1998
Production : 10 12 13 15 18 20
(’000 tonnes) [Delhi Univ., BCom), 2002]
Solution: Second degree parabolic trend equation is given by
yC = a + bx + cx2
To find the values of constants a, b and c, the normal equations are:
Σy = na + bΣx + cΣx2
Σxy = aΣx + bΣx2 + cΣx3
Σx 2 y = aΣx2 + bΣx3 + cΣx4
Calculations required to calculate values of constants considering 1995 as origin are shown
below:
Year y x x2 x3 x4 xy x2y

1993 10 –2 4 –8 16 –20 40
1994 12 –1 1 –1 1 –12 12
1995 13 0 0 0 0 0 0
1996 15 1 1 1 1 15 15
1997 18 2 4 8 16 36 72
1998 20 3 9 27 81 60 180
88 3 19 27 115 79 319

Putting values to the normal equations, we get

88 = 6a + 3b + 19c (i)
79 = 3a + 19b + 27c (ii)
319 = 19a + 27b + 115c (iii)
Multiply equation (i) by 9 and subtracting from (iii), we get 35a + 56c = 473 (iv)
 FORECASTING AND TIME SERIES ANALYSIS 403

Multiply eqn. (i) by 19 and (ii) by 3 and subtract, we get 105a + 280c = 1435 (v)
Multiply eqn. (iv) by 3 and subtract (v) from it, we get 112c = 16 or c = 0.143
Putting c = 0.143 in (iv) we get 35a + 56(0.143) = 473 or a = 13.285
Multiply eqn. (ii) by 19 and (iii) by 3 and add we get 280b – 168c = 5752
or 35b – 21c = 719 (vi)
Putting c = 0.143 in (vi), we get
35b – 21(0.143) = 719 or b = 20.40
Again putting values of b and c in Eqn. (i), we get
6a + 3(20.40) + 19(0.143) = 88 or a = 4.46
Hence the parabolic equation becomes:
y= 4.46 + 20.40x + 0.143x2
Also, for x = 7
y2000 = 4.46 + 20.40 (7) + 0.143(7)2 = 4.46 + 142.80 + 7.00 = 154.26

Example 11.23: The prices of a commodity during 2001-2006 are given below. Fit a parabola
Y = a + bx + cx2 to these data. Estimate the price for the year 2007.
Year : 2001 2002 2003 2004 2005 2006
Price (Rs.) : 100 107 128 144 181 192

[Delhi Univ., B. Com (Hons), 2006]

Solution: Normal equations are given by:
Σ y = na + bΣx + c Σ x2
Σ xy = aΣ x + bΣx2 + c Σ x3
Σ x 2 y = aΣ x2 + bΣx3 + c Σ x4
Let the year 2004 be taken as origin. Then x unit = years, y unit = price in Rs. Calculations
required to calculated values of constants a, b and c in normal equation is shown below.

Year y x x2 x3 x4 xy x2 y

2001 100 –3 9 –27 81 –300 900

2002 107 –2 4 –8 16 –214 428
2003 128 –1 1 –1 1 –128 128
2004 144 0 0 0 0 0 0
2005 181 1 1 1 1 181 181
2006 192 2 4 8 16 384 768
848 –3 19 –27 115 –77 2405

Putting values to normal equations, we get

848 = 6a – 3b + 19c (i)
–77 = –3a + 19b – 27c (ii)
2405 = 19a – 27b + 115c (iii)
Multiply equation (ii) by 2 and add to (i)
–154 = – 6a + 38b – 54c or 35b – 35a = 694 (iv)
404 BUSINESS S T A T I S T I C S

Put a = 13.285 and c = 0.143 in (i), we get

6(13.285) + 3b + 19(0.143) = 88
79.7 + 3b + 2.717 = 88 or b = 1.857
Parabolic trend equation is given by
x = 13.285 + 1.857 x + 0.143x2
Put x = 5 for the year 2000, we get
y = 13.285 + 1.857(5) + 0.143(5)2 = 26.145 or y2000 = 26145 tonnes.

Example 11.24: (a) The trend equation for the yearly sales of a commodity with 1st July, 1991 as
origin is yC = 96 + 28.8x + 4x2, where x unit = 1 year. Determine the monthly trend equation with
Jan 1992 as origin.
(b) Compute trend values for August 1991. [Delhi Univ., B.Com(Hons), 2002]

Solution: Given trend equation yC = 96 + 28.8x + 4x2 (origin: 1st July, 1991, x unit = one year)
(a) To obtain monthly trend equation divide 96(i.e. a) by 12, 41(i.e. b) by (12×12) and 4(i.e. c) by
(12×12×12):
96 28.8 4
yC = + x+ x2 = 8 + 0.2x + 0.0023x2 (i)
12 12 × 12 12 × 12 × 12
(origin: Ist July, 1991, x unit 1 month)
To change the origin from 1st July 1991 to January 1992, x shall be increased by 6.5. That is
yC = 8 + 0.2 (x + 6.5) + 0.0023 (x + 6.5)2
= 8 + 0.2 (x + 6.5) + 0.0023 (x2 + 13x + 42.25)
= 8 + 0.2x + 1.3 + 0.0023 x2 + 0.03 x + 0.097
= 9.397 + 0.23x + 0.023 x2 (ii)
(b) To get trend value for August 1991, replace x by 1.5 in (i)
yC = 8 + 0.2(1.5) + 0.0023×(1.5)2 = 8.305.

11.7.4 Ratio-to-Moving Average Method

This method is also called the percentage moving average method. In this method, the original values
in the time-series data are expressed as percentages of moving averages instead of percentages of
trend values in the ratio-to-trend method. The steps of the method are summarized as follows:
(i) Find the centred 12 monthly (or 4 quarterly) moving averages of the original data values in the
time-series.
(ii) Express each original data value of the time-series as a percentage of the corresponding cen-
tred moving average values obtained in Step (i). In other words, in a multiplicative time-series
model, we get
Original data values T ⋅ C ⋅S ⋅I
× 100 = × 100 = (S ⋅ I) × 100%
Trend values T ⋅C
This implies that the ratio-to-moving average represents the seasonal and irregular compo-
nents.
(iii) Arrange these percentages according to months or quarter of given years. Find the averages
over all months or quarters of the given years.
(iv) If the sum of these indexes is not 1200 (or 400 for quarterly figures), multiply them by a
correction factor = 1200/(sum of monthly indexes). Otherwise, the 12 monthly averages will be
considered as seasonal indexes.
 FORECASTING AND TIME SERIES ANALYSIS 405

Example 11.25: Calculate the seasonal index by the ratio-to-moving method from the following data:

Year Quarters
I II III IV
2001 75 60 53 59
2002 86 65 63 80
2003 90 72 66 85
2004 100 78 72 93

Solution: Calculations for 4 quarterly moving averages and ratio-to-moving averages are shown in
Tables 11.15 and 11.16.

Table 11.15 Calculation of Ratio-to-Moving Averages

Year Quarter Original 4-Quarter 4-Quarter 2× 4-Quarter Ratio-to-Moving

Values Moving Moving Moving Average (Percent)
Y = T.C.S.I Total Average Average Y
=(S.I)100%
T.C T .C
2001 1 75 — — — —
2 60 — —
→ 248
3 54
→ → 507 63.375 54/63.375 = 85.20
4 59 259 → 523 65.375 59/65.375 = 90.25
→ 264
2002 1 86 → 537 67.125 128.12
273
2 65 567 70.875 91.71
3 62 294 592 74.000 85.13
298
4 80 603 75.375 106.14
305
2003 1 90 613 76.625 117.43
308
2 72 521 77.625 92.75
313
3 66 636 79.500 83.02
323
4 85 652 81.500 104.29
2004 1 100 329 664 84.750 92.03
335
2 78 678 84.750 92.03
343
3 72 — — — —
4 93 — — — —

Table 11.16 Calculation of Seasonal Index

Year Quarters
I II III IV
2001 — — 85.21 90.25
2002 128.12 91.71 85.13 106.14
2003 117.45 92.75 85.13 104.29
2004 120.48 92.03 — —
Total 366.05 276.49 255.47 300.68
Seasonal average 91.51 69.13 63.87 75.17 = 299.66
Adjusted
seasonal index 122.07 92.22 85.20 100.30 ≅ 400
406 BUSINESS S T A T I S T I C S

The total of seasonal averages is 299.66. Therefore the corresponding correction factor would be
400/299.68 = 1.334. Each seasonal average is multiplied by the correction factor 1.334 to get the
adjusted seasonal indexes shown in Table 11.17.
Example 11.26: Calculate the seasonal indexes by the ratio-to-moving average method from the
following data:

Year Quarter Actual 4-quarterly Year Quarter Given 4-quarterly

Values Moving Values Moving
(Y = T.C.S.I) Average (Y) Average
2000 1 75 — 2002 1 90 76.625
2 60 — 2 72 77.625
3 54 63.375 3 66 79.500
4 59 65.375 4 85 81.500
2001 1 86 67.125 2003 1 100 83.000
2 65 70.875 2 78 84.750
3 63 74.000 3 72 —
4 80 75.375 4 93 —

Solution: Calculations of ratio-to-moving averages are shown in Table 11.16.

Table 11.17 Calculation of Seasonal Indexes

Year Quarter Actual 4-quarterly Ratio to Moving

Values Moving Average (Percentage)
Y
(Y = T.C.S.I) (T.C) × 100
T .C
2000 1 75 — —
2 60 — —
3 54 63.375 85.21
4 59 65.375 90.25
2001 1 86 67.125 128.12
2 65 70.875 91.71
3 63 74.000 85.14
4 80 75.375 106.14
2002 1 90 76.625 117.46
2 72 77.625 92.75
3 66 79.500 83.02
4 85 81.500 104.29
2003 1 100 83.000 120.84
2 78 84.750 92.04
3 72 — —
4 93 — —

Rearranging the percentages to moving averages, the seasonal indexes are calculated as shown in
Table 11.18.
 FORECASTING AND TIME SERIES ANALYSIS 407

Table 11.18 Seasonal Indexes

Year Quarter (Percentages to Moving Averages)

1 2 3 4
2000 — — 85.21 90.25
2001 128.12 91.71 85.14 106.14
2002 117.46 92.75 83.02 104.30
2003 120.48 92.04 — —
Total 366.06 276.50 253.37 300.69
Average 122.02 92.17 84.46 100.23 = 398.88
Adjusted 122.02 92.17 84.46 100.23
seasonal index ×100 ×100 ×100 ×100
99.72 99.72 99.72 99.72
= 122.36 = 92.43 = 84.70 = 100.51 = 400

Since the total of average indexes is less than 400, the adjustment of the seasonal index has been
done by calculating the grand mean value as follows:

122.02 + 92.17 + 84.46 + 100.23

x = = 99.72
4
The seasonal average values are now converted into adjusted seasonal indexes using
x = 99.72 as shown in Table 11.18.

Advantages and Disadvantages of Ratio-to-Moving Average Method This is the most widely used method for
measuring seasonal variations because it eliminates both trend and cyclical variations from the time-
series. However, if cyclical variations are not regular, then this method is not capable of eliminating
them completely. Seasonal indexes calculated by this method will contain some effect of cyclical
variations.
The only disadvantage of this method is that six data values at the beginning and the six data
values at the end are not taken into consideration for calculation of seasonal indexes.

11.7.5 Link Relative Method

This method is also known as Pearson’s method. The percentages obtained by this method are called
link relatives as these link each month to the preceding one. The steps involved in this method are
summarized below:
(i) Convert the monthly (or quarterly) data into link relatives by using the following formula:

Data value of current month

Link relative for a particular month = × 100
Data value of preceding month
(ii) Calculate the average of link relatives of each month using either median or arithmetic
mean.
(iii) Convert the link relatives (L.R.) into chain relatives (C.R.) by using the formula:
408 BUSINESS S T A T I S T I C S

[L.R. of current month (or quarter)

× C.R. of preceding month (or quarter)]
C.R. for a particular month =
100
The C.R. for the first month (or quarter) is assumed to be 100.
(iv) Compute the new chain relative for January (first month) on the basis of December (last month)
using the formula:
C.R. of January × C.R. of December
New C.R. for January =
100
The new C.R. is usually not equal to 100 and therefore needs to be multiplied with the monthly
correction factor
1
d= (New C.R. for January − 100)
12
If the figures are given quarterly, then the correction factor would be
1
d= (New C.R. of first quarter – 100)
4
The corrected C.R. for other months can be calculated by using the formula:
Corrected C.R. for kth month = Original C.R. of kth month – (k – 1) d
where k = 1, 2, 3, ..., 12
(v) Find the mean of the corrected chain index. If it is 100, then the corrected chain indexes
represent the seasonal variation indexes. Otherwise divide the corrected C.R. of each month
(or quarter) by the mean value of corrected C.R. and then multiply by 100 to get the seasonal
variation indexes.
Example 11.27: Apply the method of link relatives to the following data and calculate seasonal
indexes.

Year Quarters
I II III IV
1999 68 62 61 63
2000 65 58 56 61
2001 68 63 63 67
2002 70 59 56 62
2003 60 55 51 58

Solution: Computations of link relatives (L.R.) are shown in Table 11.19 by using the following
formula:
Data value of current quarter
Link relative of any quarter = × 100
Data value of preceeding quarter
 FORECASTING AND TIME SERIES ANALYSIS 409

Table 11.19 Computation of Link Relatives

Year Quarters
I II III IV
1999 — 91.18 98.39 103.28
2000 103.18 89.23 96.55 108.93
2001 111.48 92.65 100.00 106.35
2002 104.48 84.29 94.91 110.71
2003 96.78 91.67 92.73 113.73
Total of L.R. 415.92 449.02 482.58 543.00
Arithmetic
mean of L.R. 103.98 89.80 96.52 108.60
89.80 × 100 96.52 × 89.80 108.60 × 86.67
Chain relatives (C.R.) 100
100 100 100
= 89.80 = 86.67 = 94.12

The new chain relatives for the first quarter on the basis of last quarter is calculated as follows:
L.R. of first quarter × C.R. of previous quarter 103.98 × 94.12
New C.R = = = 97.9
100 100
Since new C.R. is not equal to 100, therefore we need to apply quarterly correction factor as:

1 1
d= (New C.R. of first quarter – 100) = (97.9 – 100) = – 0.53
4 4
Thus the corrected (or adjusted) C.R. for other quarters is shown in Table 11.20. For this we use the
formula:
Corrected C.R. for kth quarter = Original C.R. of kth quarter – (k – 1) d
where k = 1, 2, 3, 4.

Table 11.20 Calculation of Link Relatives

Quarter I II III IV

Corrected C.R. 100 89.80 – (– 0.53) 86.67 – 2(– 0.53) 94.13 – 3(– 0.53)
= 90.33 = 87.73 = 95.71

100 90.33 87.73 95.71

Seasonal indexes × 100 × 100 × 100 × 100
93.44 93.44 93.44 93.44
= 107.02 = 96.67 = 93.89 = 102.42

100 + 90.33+ 87.73+ 95.71

Mean of corrected C.R. = = 93.44
4
Corrected C.R.
Seasonal variation index = × 100
Mean of corrected C.R.
410 BUSINESS S T A T I S T I C S

Example 11.28: Apply the method of link relatives to the following data and calculate the seasonal
index:

Year Quarters

I II III IV

2000 45 54 72 60
2001 48 56 63 56
2002 49 63 70 65
2003 52 65 75 72
2004 60 70 84 86

Solution : Computations of link relatives (L.R.) using the following formula are shown in
Table 11.21.
Data value of current quarter
L.R. of any quarter = × 100
Data value of preceding quarter

Table 11.21 Computation of Link Relatives

Year Quarters
I II III IV
2000 — 120 133.33 83.33
2001 80.00 116.67 112.50 88.89
2002 87.50 128.57 111.11 92.86
2003 80.00 125.00 115.38 96.00
2004 85.71 116.67 120.00 78.57
Total of L.R. 333.21 606.91 592.32 439.65
Arithmetic
mean of L.R. 83.30 121.38 118.46 87.93
121.38 × 100 118.46 × 121.38 87.93 × 143.78
Chain relatives 100
100 100 100
(C.R.) = 121.38 = 143.78 = 126.42

The new chain relatives for the first quarter on the basis of the preceding quarter is calculated as
follows:
L.R. of first quarter × C.R. of previous quarter
New C.R. =
100
83.30 × 126.42
= = 105.30
100
Since the new C.R. is more than 100, therefore we need to apply a quarterly correction factor as :
1
d= (New C.R. of first quarter – 100)
4
1
= (105.30 – 100) = 1.325
4
 FORECASTING AND TIME SERIES ANALYSIS 411

Thus the corrected (or adjusted) C.R. for other quarters is shown in Table 11.22. For this we use the
formula
Corrected C.R. for kth quarter = Original C.R. of kth quarter – (k – 1) d
where k = 1, 2, 3, 4.

Table 11.22 Corrected C.R.

Quarters I II III IV
Corrected C.R. 100 121.38 – 1.32 143.78 – 2(1.32) 126.42 – 3(1.32)
= 120.06 = 141.14 = 122.46

100 120.06 141.14 122.46

Seasonal × 100 × 100 × 100 × 100
indexes 120.92 120.92 120.92 120.92
= 82.70 = 99.30 = 116.72 = 101.27

100 + 120.06 + 141.14 + 122.46

Mean of corrected C.R. = = 120.92
4
Corrected C.R.
Seasonal variation index = × 100
Mean of corrected C.R.

Advantages and Disadvantages of Link Relative Method This method is much simpler than the ratio-to-
trend or the ratio-to-moving average methods. In this method the L.R. of the first quarter (or month)
is not taken into consideration as compared to ratio-to-trend method, where 6 values each at the
beginning and at the end periods (month) are lost.
This method eliminates the trend but it is possible only if there is a straight line (linear) trend in the
time-series—which is generally not formed in business and economic series.

11.8 MEASUREMENT OF CYCLICAL VARIATIONS—RESIDUAL METHOD

As mentioned earlier that a typical time-series has four components: secular trend (T), seasonal
variation (S), cyclical variation (C), and irregular variation (I). In a multiplicative time-series model,
these components are written as:
y = T. C. S. I
The deseasonalization data can be adjusted for trend analysis by dividing these by the corresponding
trend and seasonal variation values. Thus we are left with only cyclical (C) and irregular (I) variations
in the data set as shown below:
y T.C.S.I
= = C. I
T.S T.S
The moving averages of an appropriate period may be used to eliminate or reduce the effect of
irregular variations and thus left behind only the cyclical variations.
The procedure of identifying cyclical variation is known as the residual method. Recall that cyclical
variations in time-series tend to oscillate above and below the secular trend line for periods longer
than one year. The steps of residual method are summarized as follows:
412 BUSINESS S T A T I S T I C S

(i) Obtain seasonal indexes and deseasonalized data.

(ii) Obtain trend values and expressed seasonalized data as percentages of the trend values.
(iii) Divide the original data (y) by the corresponding trend values (T) in the time-series to get S. C.
I. Further divide S. C. I by S to get C. I.
(iv) Smooth out irregular variations by using moving averages of an appropriate period but of
short duration, leaving only the cyclical variation.

11.9 MEASUREMENT OF IRREGULAR VARIATIONS

Since irregular variations are random in nature, no particular procedure can be followed to isolate
and identify these variations. However, the residual method can be extended one step further by
dividing C. I by the cyclical component (C) to identify the irregular component (I).
Alternately, trend (T), seasonal (S), and cyclical (C) components of the given time-series are esti-
mated and then the residual is taken as the irregular variation. Thus, in the case of multiplicative
time-series model, we have
Y T.C.S.I
= =I
T.C.S T.C.S
where S and C are in fractional form and not in percentages.

C o n c e p t u a l Q u e s t i o n s 11B
15. (a) Under what circumstances can a trend 18. Briefly describe the moving average and least
equation be used to forecast a value in a series squares methods of measuring trend in time-
in the future? Explain. series.
(b) What are the advantages and disadvantages 19. Explain the simple average method of
of trend analysis? When would you use this calculating indexes in the context of time-series
method of forecasting? analysis.
16. What effect does seasonal variability have on a
20. Distinguish between ratio-to-trend and ratio-
time-series? What is the basis for this variability
for an economic time-series? to-moving average as methods of measuring
seasonal variations. Which is better and why?
17. What is measured by a moving average? Why
are 4-quarter and 12-month moving averages 21. Distinguish between trend, seasonal variations,
used to develop a seasonal index? and cyclical variations in a time-series. How
can trend be isolated from variations?
 FORECASTING AND TIME SERIES ANALYSIS 413

S e l f-P r a c t i c e P r o b l e m s 11C
11.16 Apply the method of link relatives to the 11.19 Calculate seasonal index numbers from the
following data and calculate seasonal indexes. following data:

Quarter 1999 2000 2001 2002 2003

Quarter
Year
I 6.0 5.4 6.8 7.2 6.6
I II III IV
II 6.5 7.9 6.5 5.8 7.3
III 7.8 8.4 9.3 7.5 8.0 1998 108 130 107 93
IV 8.7 7.3 6.4 8.5 7.1 1999 86 120 110 91
2000 92 118 104 88
11.17 A company estimates its sales for a particular
year to be Rs. 24,00,000. The seasonal indexes 2001 78 100 94 78
for sales are as follows: 2002 82 110 98 86
2003 106 118 105 98
Month Seasonal Index Month Seasonal Index
11.20 Calculate seasonal index for the following
January 75 July 102 data by using the average method:
February 80 August 104
Quarter
March 98 September 100 Year
April 128 October 102 I II III IV
May 137 November 82
2000 72 68 80 70
June 119 December 73
2001 76 70 82 74
Using this information, calculate estimates of 2002 74 66 84 80
monthly sales of the company. (Assume that 2003 76 74 84 78
there is no trend). 2004 78 74 86 82
11.18 Calculate the seasonal index from the following
data using the average method: 11.21 On the basis of quarterly sales (Rs. in lakh) of a
certain commodity for the years 2003—2004,
Quarter the following calculations were made:
Year
Trend : y = 20 + 0.5 t with origin at first
I II III IV
quarter of 2003
2000 72 68 80 70 where t = time unit (one quarter),
2001 76 70 82 74 y = quarterly sales (Rs. in lakh)
2002 74 66 84 80 Seasonal variations:
2003 76 74 84 78
Quarter : 1 2 3 4
2004 78 74 86 82
Seasonal index : 80 90 120 110
[Kerala Univ., B.Com, 1996] Estimate the quarterly sale for the year 2003
using multiplicative model.
414 BUSINESS S T A T I S T I C S

Hints and Answers

11.16

Year Quarters
I II III IV
1999 — 108.3 120.0 111.5
2000 62.1 146.3 106.3 89.9
2001 93.2 95.6 143.1 68.8
2002 112.5 80.6 129.3 113.3
2003 77.6 110.6 109.6 88.8

345.4 541.4 608.3 469.3

Arithmetic average: = 86.35 = 108.28 = 121.66 = 93.86
4 5 5 5
100 × 108.28 121.66 × 108.28 93.86 × 131.73
Chain relatives: 100
100 100 100
0= 108.28 = 131.73 = 123.65
Corrected chain 100 108 – 1.675 131.73 – 3.35 123.64 – 5.025
relatives: 00= 106.325 = 128.38 = 118.615
100 × 100 106.605 128.38 118.615
Seasonal indexes: ×100 ×100 ×100
113.4 113.4 113.4 113.4
= 88.18 = 94.01 = 113.21 = 104.60

11.17 Seasonal indexes are usually expressed as per-

Month Seasonal Seasonal Effect Estimated Sales
centages. The total of all the seasonal indexes Index (3) = (2) ÷ 100 (4) = (3) × 2,00,000
is 1200. (1) (2)
Seasonal effect = Seasonal index + 100 January 75 0.75 1,50,000
The yearly sales being Rs. 24,00,000, the esti- February 80 0.80 1,60,000
mated monthly sales for a specified month: March 98 0.98 1,96,000
April 128 1.28 2,56,000
Estimated sales May 137 1.37 2,74,000
Annual sales June 119 1.19 2,38,000
= × Seasonal effect July 102 1.02 2,04,000
12
August 104 1.04 2,08,000
24,00,000 September 100 1.00 2,00,000
= × Seasonal effect
12 October 102 1.02 2,04,000
= 2,00,000 × Seasonal effect November 82 0.82 1,64,000
December 73 0.73 1,46,000
1200 12.00 24,00,000
 FORECASTING AND TIME SERIES ANALYSIS 415

11.18 Sales in different quarters:

I: Rs. 20,000; II: 20,000×1.16 = Rs. 23,200;
Year Quarterss III: 20,000 × 1.03 = Rs. 20,600;
I II III IV IV: 20,000 × 0.89 = Rs. 17,800
11.20
2000 72 68 80 70
2001 76 70 82 74 Year Quarters
2002 74 66 84 80
2003 76 74 84 78 I II III IV
2004 78 74 86 82
2000 72 68 80 70
Total 376 352 416 384 2001 76 70 82 74
Average 75.2 70.4 83.2 76.8
2002 74 66 84 80
Seasonal index 98.43 92.15 108.9 100.52
2003 76 74 84 78
75.2 + 70.4 + 83.2 + 76.8 2004 78 74 86 82
Grand average =
4 Total 376 352 416 384
305.6 Average 75.2 70.4 83.2 76.8
= = 76.4
4 75.2 70.4 83.2 76.8
Seasonal index for quarter Seasonal ×100 ×100 ×100 × 100
76.4 76.4 76.4 76.4
Average of quarter k
k= × 100 Index = 98.43 = 92.15 = 108.90 = 100.52
Grand average
11.19 11.21
Year Quarters
Quarter Time Trend (T) Seasonal Estimated
I II III IV of 2003 Unit Values Effect or Sales
y = 20 + 0.5t Seasonal (Rs. in
1998 108 130 107 93
1999 86 120 110 91 Index lakh)
2000 92 118 104 88 (S ) T⋅S
2001 78 100 94 78
2002 82 110 98 86 1 4 20 + 0.5 × 4 = 22.0 0.80 17.60
2003 106 118 105 98 2 5 20 + 0.5 × 5 = 22.5 0.90 20.25
3 6 20 + 0.5 × 6 = 23.0 1.20 27.60
Total 552 696 618 534
Average 92 116 103 89 4 7 20 + 0.5 × 7 = 23.5 1.10 25.85
92 116 103 89
Seasonal ×100 ×100 ×100 ×100
100 100 100 100
Index = 92 0 = 116 11 = 103000 = 890

Formulae Used
y = ab x;
1. Secular trend line 1 Σ x log y
log a = Σ log y; log b =
• Linear trend model n Σ x2
y = a + bx
• Parabolic trend model
Σx y − nx y y = a + bx + cx2
• where a = y – b x ; b =
Σ x 2 − n ( x )2 Σ y − c Σ x2 Σx y
• where a = ; b=
• Exponential trend model n Σ x2
416 BUSINESS S T A T I S T I C S

n Σ x2 y − Σ x2 Σ y where t = current time period

c= 4 2 2 D = actual data value
n Σ x − (Σ x ) n = length of time period
2. Moving average
Σ {D t + D t − 1 + ... + D t − n + 1}
MAt + 1 =
n

R e v i e w S e l f-P r a c t i c e P r o b l e m s

11.22 A sugar mill is committed to accepting beets (c) Change the sales (y) scale to monthly and
from local producers and has experienced the forecast the monthly sales rate at July 1,
following supply pattern (in thousands of tons/ 2003, and also at one year later.
year and rounded).
11.25 Data collected on the monthly demand for an
item were as shown below:
Year Tonnes Year Tonnes
1990 100 1995 400 January 100
1991 100 1996 400 February 90
1992 200 1997 600 March 80
1993 600 1998 800 April 150
1994 500 1999 800 May 240
June 320
The operations manager would like to project
July 300
a trend to determine what facility additions will
August 280
be required by 2004
September 220
(a) Sketch a freehand curve and extend it to
(a) What conclusion can you draw with respect
2004. What would be your 2004 forecast
to the length of moving average versus
based upon the curve? smoothing effect?
(b) Compute a three-year moving average and (b) Assume that the 12-month moving aver-
plot it as a dotted line on your graph. age centred on July was 231. What is the
value of the ratio-to-moving average that
11.23 Use the data of Problem 11.22 and the normal would be used in computing a seasonal
equations to develop a least squares line of best index?
fit. Omit the year 1990. 11.26 Consider the following time-series data:
(a) State the equation when the origin is 1995. Week : 1 2 3 4 5 6
(b) Use your equation to estimate the trend Value : 8 13 15 17 16 9
value for 2004. (a) Develop a 3-week moving average for this
time-series. What is the forecast for week
11.24 A forecasting equation is of the form: 7?
y c = 720 + 144x (b) Use α = 0.2 to compute the exponential
[2003 = 0, x unit = 1 year, y = annual sales] smoothing values for the time-series. What
is the forecast for week 7?
(a) Forecast the annual sales rate for 2003 and
11.27 Below are given the figures of production (in
also for one year later.
million tonnes) of a cement factory:
(b) Change the time (x) scale to months and
Year : 1990 1992 1993 1994 1995 1996 1999
forecast the annual sales rate at July 1,
Production : 77 88 94 85 91 98 90
2003, and also at one year later.
 FORECASTING AND TIME SERIES ANALYSIS 417

(a) Fit a straight line trend by the ‘least squares 11.29 Fit a parabolic curve of the second degree to
method’ and tabulate the trend values. the data given below and estimate the value for
(b) Eliminate the trend. What components of 2002 and comment on it.
the time series are thus left over? Year : 1996 1997 1998 1999 2000
(c) What is the monthly increase in the Sales
production of cement?
(Rs. in ’000): 10 12 13 10 8
11.28 The sale of commodity in tonnes varied from
January 2000 to December, 2000 in the 11.30 Given below are the figures of production of a
following manner: sugar (in 1000 quintals) factory:
Year : 1991 1992 1993 1994 1995 1996 1997
280 300 280 280 270 240
Production : 40 45 46 42 47 49 46
230 230 220 200 210 200
Fit a straight line trend by the method of least
Fit a trend line by the method of semi-averages. squares and estimate the value for 2001.

Hints and Answers

11.22 (a) Forecasts is around 1200 (thousand) (b)
tonnes
(b) Averages are: 133, 300, 433, 500, 433, Week Values Forecast Forecast Squared
466, 600 and 733. (t) yt Ft Error Error
y t – Ft (yt – Ft )2
11.23 (a) ŷ = 489 + 75x [1995 = 0, x = years,
y = tonnes in thousand] 1 18 — — —
(b) 11,64,000 tonnes 2 13 18.00 5.00 25.00
3 15 19.00 6.00 36.00
11.24 (a) 720 units when x = 0, 864 units when
4 17 10.20 6.80 46.24
x = 1.
5 16 11.56 4.44 19.71
(b) ŷ = 720 + 12x [ July 1, 2003 = 0; x 6 19 12.45 – 3.45 11.90
unit = 1 month;
y = annual sales rates in units] 138.85
720 units per year; 864 units per year. Forecast for week 7 is: 0.2 (9)+(1 – 0.2) (12.45)
= 11.76.
(c) ŷ = 60 + x [ July 1, 2003 = 0, x unit = 1
month; 11.27 (a)
y = monthly sales rates in units]
60 units per month; 72 units per month. Year Time Production Deviation Trend
Period (in m. From 1994 Values
11.25 (a) Longer average yield more smoothing;
tonnes)
(b) 1.3
y x xy x y
11.26 (a)
1990 – 4 77 –4 – 308 16 83.299
Week Values Forecast Forecast Squared 1992 – 2 88 –2 – 176 4 86.051
(1) (2) (3) Error Forecast 1993 – 1 94 –1 – 94 1 87.427
(4) = (2) – (3) Error
1994 – 0 85 0 0 0 88.803
1 18 — — —
1995 – 1 91 1 91 1 90.179
2 13 — — —
3 15 — — — 1996 – 2 98 2 196 4 91.555
4 17 12 –5 25 1999 – 5 90 5 450 25 95.683
5 16 15 –1 21
6 19 16 –7 49 623 1 159 51

Forecast for week 7 is: (17 + 16 + 9)/3 = 14.

418 BUSINESS S T A T I S T I C S

Solving the normal equations Parabolic trend line : y = a + bx + bx2

Σ y = na + bΣ x 623 = 7a + b
Σy − cΣx 2 53 − 0.857 × 10
2 a= = = 8.886
Σ x y = aΣ x + b Σ x 159 = a + 5b n 5
we get a = 88.803 and b = 1.376 x. Thus
Σ xy −6
ŷ = a + bx = 88.803 + 1.376 x b= = = – 0.6 ;
Σ x2 10
Substituting x = – 4, – 2, – 1, 0, 1, 2, 5 to get
trend values as shown above in the table.
n Σ x 2 y − Σ x 2 Σy 5(94) − 10 (53)
(b) After eliminating the trend, we are left with c= 4 2 2
=
n Σ x − (Σ x ) 5(34) − (10)2
S, C, and I components of time-series.
(c) Monthly increase in the production of ce- = – 0.857
ment in given by b/12 = 1.376/12 = 0.115. ∴ y = 8.886 – 0.6x – 0.857x2
11.28 For 2002, x = 4;
Month Sales y = 8.886 – 0.6 (4) – 0.857 (4)2
(in tonnes)
= – 7.226
January 280 Total = 1650 of first six months;
February 300 11.30
March 280
1650 Year Production Deviations
April 280 Average = = 275
6 (’000 qtls) from 1994
May 270
y x xy x2
June 240
July 230 Total = 1290 of last six months; 1991 40 –3 – 120 9
August 230 1992 45 –2 – 90 4
September 220 1993 46 –1 – 46 1
1290
October 200 Average = = 215 1994 42 0 0 0
6
1995 47 1 47 1
Plot 275 and 215 in the middle of March-April 1996 49 2 98 4
2000 and that of September-October 2000. By
1997 46 3 138 9
joining these two points we get a trend line
which describes the given data. 315 0 27 28
11.29
ŷ = a + bx; a = Σ y/n = 315/7 = 45;
Year Sales Period Σ xy 27
b= 2
= = 0.964
Σx 28
y x xy x2 x2 y x4
ŷ = 45 + 0.964x
1996 10 –2 – 20 4 40 16
y2001 = 45 + 0.964 (7)
1997 12 –1 – 12 1 12 1
1998 13 0 0 0 0 0 = 45 + 6.748 = 51.748
1999 10 1 10 1 10 1
2000 8 2 16 4 32 16
53 0 –6 10 94 34
 FORECASTING AND TIME SERIES ANALYSIS 419

G l o s s a r y o f Te r m s

Causal forecasting methods: Forecasting methods that relate a time-series to other variables which are used to explain
cause and effect relationship.
Delphi method: A quantitative forecasting method that obtains forecasts through group consensus.
Time-series: A set of observations measured at successive points in time or over successive periods of time.
Trend: A type of variation in time-series that reflects a long-term movement in time-series over a long period of time.
Cyclical variation: A type of variation in time-series, in which the value of the variable fluctuates above and below a
trend line and lasting more than one year.
Seasonal variation: A type of variation in time-series that shows a periodic pattern of change in time-series within a
year; patterns tend to be repeated from year to year.
Irregular variation: A type of variation in time-series that reflects the random variation of the time-series values which
is completely unpredictable.
Moving averages: A quantitative method of forecasting or smoothing a time-series by averaging each successive group
of data values.
Weighted moving average: A quantitative method of forecasting or smoothing a time-series by computing a weighted
average of past data values; sum of weights must equal one.
Deseasonalization: A statistical process used to remove the effect of seasonality from a time-series by dividing each
original series observation by the corresponding seasonal index.
 LEARNING OBJECTIVES

After studying this chapter, you should be able to

z explain the purpose of index numbers.
z compute indexes to measure price changes and quantity changes over time.
z revise the base period of a series of index numbers
z explain and derive link relatives
z discuss the limitations of index number construction

12.1 INTRODUCTION
We know that most values change and therefore may want to knowhow much change has taken place
over a period of time. For example, we may want to know how much the prices of different items
essential to a household have increased or decreased so that necessary adjustments can be made in
the monthly budget. An organization may be concerned with the way in which prices paid for raw
materials, annual income and profit, commodity prices, share prices, production volume, advertising
budget, wage bills, and so on, have changed over a period of time. However, while prices of a few
items may have increased, others may have decreased over a given period of time. Consequently in all
such situations, an average measure needs to be defined to compare such differences from one time
period to another. Index numbers are yardsticks for describing such difference.
An index number can be defined as a relative measure describing the average changes in any
quantity over time. In other words, an index number measures the changing value of prices, quantities,
or values over a period of time in relation to its value at some fixed point in time, called the base
period. This resulting ratio of the current value to a base value is multiplied by 100 to express the
index as a percentage. Since an index number is constructed as a ratio of a measure taken during one
time period to that same measure taken during another time period (called base period), it has no
unit and is always expressed as a percentage term as follows:
 INDEX NUMBERS 421

Current period value

Index number = × 100
Base period value

Indexes may be based at any convenient period, which is occasionally adjusted, and these
are published at any convenient frequency. Examples of some indexes are:
Daily Stock market prices
Monthly Unemployment figures
Yearly Gross National Product (GNP)
Index numbers were originally developed by economists for monitoring and comparing
different groups of goods. For decision-making in business, it is sometimes essential to understand
and manipulate different published index series and to construct one’s own index series.

12.2 INDEX NUMBER DEFINED

Definition of index numbers can be classified into the following three broad categories:
1. A measure of change
• It is a numerical value characterizing the change in complex economic phenomena over
a period of time or space.` —Maslow
• An index number is a quantity which, by reference to a base period, shows by its
variations the changes in the magnitude over a period of time. In general, index
numbers are used to measure changes over time in magnitudes which are not capable
of direct measurement. —John I. Raffin
• An index number is a statistical measure designed to show changes in variables or a
group of related variables with respect to time, geographic location or other characteristics.
—Speigel
• Index number is a single ratio (usually in percentages) which measures the combined
(i.e., averaged change of several variables between two different times, places or situations.
—A. M. Tuttle
2. A device to measure change
• Index numbers are devices measuring differences in the magnitude of a group of
related variables. —Corxton and Cowden
• An index number is a device which shows by its variation the changes in a magnitude
which is not capable of accurate measurement in itself or of direct valuation in practice.
—Wheldom
3. A series representing the process of change
• Index numbers are series of numbers by which changes in the magnitude of a phenomenon
are measured from time to time or place to place. —Horace Secris
• A series of index numbers reflects in its trend and fluctuations the movements of some
quantity of which it is related. —B. L. Bowley
• An index number is a statistical measure of fluctuations in a variable arranged in the
form of a series, and using a base period for making comparisons. —L. J. Kaplan
422 BUSINESS S T A T I S T I C S

12.3 TYPES OF INDEX NUMBERS

Index numbers are broadly classified into three categories: (i) price indexes,
(ii) quantity indexes, and (iii) value indexes. A brief description of each of these is as follows:
Price Indexes These indexes are of two categories:
• Single price index
• Composite prices index
The single price index measures the precentage change in the current price per unit of a product to
its base period price. To facilitate comparisons with other years, the actual per unit price is
converted into a price relative, which expresses the unit price in each period as a percentage
of unit price in a base period. Price relatives are very helpful to understand and interpret
changing economic and business conditions over time. Table 12.1 illustrates the calculations
of price relatives,

Table 12.1 Calculation of Price Index (Base year = 1996)

Year Total Wage Bill Ratio Price Index or

(Rs. millions) Percentage Relative
(1) (2) (3) = (2)/ 11.76 (4) = (3) × 100
2000 11.76 11.76/11.76 = 1.0 100.0
2001 12.23 12.23/11.76 = 1.039 103.9
2002 12.84 12.84/11.76 = 1.091 109.1
2003 13.35 13.35/11.76 = 1.135 113.5
2004 13.82 13.82/11.76 = 1.175 113.5

From Table 12.1, it is observed that the price relative of 113.5 in 2003 shows an increase of
13.5% in wage bill compared to the base year 2000.
A composite price index measures the average price change for a basket of related items from a base
period to the current period. For example, the wholesale price index reflects the general price level
for a group of items (or a basket of items) taken as a whole.
The retail price index reflects the general changes in the retail prices of various items including
food, housing, clothing, and so on. In India, the Bureau of Labour statistics, publishes retail price
index. The consumer price index, a special type of retail price index, is the primary measure of the
cost of living in a country. The consumer price index is a weighted average price index with fixed
weights. The weightage applied to each item in the basket of items is derived from the urban and
rural families.
Quantity Index A quantity index measures the relative changes in quantity levels of a group (or
basket) of items consumed or produced, such as agricultural and industrial production, imports and
exports, between two time periods. The method of constructing quantity indexes is the same as that of
price index except that the quantities are vary from period to period.
The two most common quantity indexes are the weighted relative of aggregates and the weighted
average of quantity relative index.
Value Index A value index measures the relative changes in total monetary worth of an item, such as
inventories, sales, or foreign trade, between the current and base periods. The value of an item is
 INDEX NUMBERS 423

determined by multiplying its unit price by the quantity under consideration. The value index can
also be used to measure differences in a given variable in different locations. For example, the
comparative cost of living shows that in terms of cost of goods and services, it is cheaper to live in a
small city than in metro cities.
Special Purpose Indexes A few index numbers such as industrial production, agricultural production,
productivity, etc. can also be constructed separately depending on the nature and degree of relationship
between groups and items.
• Index number, almost alone in the domain of social sciences, may truly be called an exact
science, if it be permissible to designate as science the theoretical foundations of a useful art.
—Irving Fisher.

12.4 CHARACTERISTICS AND USES OF INDEX NUMBERS

Based on the definitions and types of index numbers discussed earlier in this chapter, the following
characteristics and uses of index number emerge.

12.4.1 Characteristics of Index Numbers

1. Index numbers are specialized averages : According to R. L. Corner, ‘An index number represents a
special case of an average, generally weighted average, compiled from a sample of items judged to be
representative of the whole’.
‘Average’ is a single figure representing the characteristic of a data set. This figure can be used
as a basis for comparing two or more data sets provided the unit of measurement of observations
in all sets is the same. However, index numbers which are considered as a special case of average
can be used for comparison of two or more data sets expressed in different units of measurement.
The consumer price index, for example, which represents a price comparison for a group of
items—food, clothing, fuel, house rent, and so on, are expressed in different units. An average of
prices of all these items expressed in different units is obtained by using the technique of price
index number calculation.
2. Index numbers measure the change in the level of phenomena in percentages : Since index numbers
are considered as a special case of an average, these are used to represent, in one single figure,
the increase or decrease (expressed in terms of percentage) in the value of a variable. For example,
a quantity index number of 110 for cars sold in a given year when compared with that of a base
year would mean that cars sales in the given year were 10 per cent higher than in the base year
(value of index number in base period is always equal to 100). Similarly, a quantity index
number of 90 in a given year would indicate that the number of cars sold in the given were 10 per
cent less than in the base year.
3. Index numbers measure changes in a variety of phenomena which cannot be measured
directly : According to Bowley, ‘Index numbers are used to measure the changes in some quantity which we
cannot observe directly. . .’
It is not possible, for example, to directly measure the changes in the import-export activities
of a country. However, it is possible to study relative changes in import and export activities by
studying the variations in factors such as raw materials available, technology, competitors, quality,
and other parameters which affect import and export, and are capable of direct measurement.
Similarly, cost of living cannot be measured in quantitative terms directly, we can only study
relative changes in it by studying the variations in certain other factors connected to it.
424 BUSINESS S T A T I S T I C S

4. Index numbers measure the effect of changes in relation to time or place : Index numbers are used to
compare changes which take place over periods of time, between locations, and in categories. For
example, cost of living may be different at two different places at the same or cost of living in one
city can be compared across two periods of time.

12.4.2 Uses of Index Numbers

According to G. Simpson and F. Kafka ‘Today Index numbers are one of the most widely used statistical tools.
They are used to feel the pulse of the economy and they have come to be used as indicators of inflationary or
deflationary tendencies’. Other important uses of index number can be summarized as follows:
1. Index numbers act as economic barometers : A barometer is an instrument that is used to measure
atmospheric pressure. Index numbers are used to feel the pressure of the economic and business
behaviour, as well as to measure ups and downs in the general economic condition of a country.
For example, the composite index number of indexes of prices, industrial output, foreign
exchange reserves, and bank deposits, could act as an economic barometer.
2. Index numbers help in policy formulation : Many aspects of economic activity are related to price
movements. The price indexes can be used as indicators of change in various segments of the
economy. For example, by examining the price indexes of different segments of a firm’s operations,
the management can assess the impact of price changes and accordingly take some remedial and/
or preventive actions.
In the same way, by examining the population index, the government can assess the need to
formulate a policy for health, education, and other utilities.
3. Index numbers reveal trends and tendencies : An index number is defined as a relative measure
describing the average change in the level of a phenomenon between the current period and a
base period. This property of the index number can be used to reflect typical patterns of change
in the level of a phenomenon. For example, by examining the index number of industrial
production, agricultural production, imports, exports, and wholesale and retail prices for the
last 8–10 years, we can draw the trend of the phenomenon under study and also draw conclusions
as to how much change has taken place due to the various factors.
4. Index numbers help to measure purchasing power : In general, the purchasing power is not associated
with a particular individual; rather it is related to an entire class or group. Furthermore, it is not
associated with the cost of a single item, because individuals purchase many different items in
order to live. Consequently, earnings of a group of people or class must be adjusted with a price
index that provides an overall view of the purchasing power for the group.
For example, suppose a person earns Rs. 1000 per month in 1990. If an item costs Rs. 100 in that
year, the person could purchase 1000 ÷ 100 = 10 units of the item with one month’s earnings, But
if in year 2000, the same person earns Rs. 2000 per month but the item cost is Rs. 250, then he could
purchase 2000 ÷ 250 = 8 units of the item. Hence, the effect of monthly earning relative to the
particular item is less in year 2000 than in 1990 as a lesser number of units of the items can be
purchased with current earnings. By dividing the item price in both the years, we can eliminate the
effect of price and determine the real purchasing power for that item. For instance, in 1990, the
purchasing power was 10 ÷ 1000 = 0.10 or 10 paise which it was Rs. 0.125 or 12.5 paise in 2000.
5. Index numbers help in deflating various values : When real rupee value is computed, the base period
is earlier than the given years for which this value is being determined Thus the adjustment of
current rupee value to real terms is referred to as deflating a value series because prices typically
increase over time.
 INDEX NUMBERS 425

The price index number is helpful in deflating the national income to remove the effect of
inflation over a long term, so that we may understand whether there is any change in the real
income of the people or not. The retail price index is often used to compute real changes in
earnings and expenditure as it compares the purchasing power of money at different points in
time. It is generally accepted as a standard measure of inflation even though calculated from a
restricted basket of goods.

C o n c e p t u a l Q u e s t i o n s 12A
1. Explain the significance of index numbers. 6. Index numbers are economic barometers. Explain
2. Explain the differences among the three principal this statement and mention the limitations of index
types of indexes: price, quantity, and value. numbers (if any).
3. How are index numbers constructed? What is 7. What are the basic characteristics of an index
their purpose? number?
4. What is an index number? Describe briefly its 8. Since value of the base year is always 100, it does
applications in business and industry. not make any difference which period is selected
as the base on which to construct an index.
5. What does an index number measure? Explain
Comment.
the nature and uses of index numbers.
9. What are the main uses of an index number?
10. What is meant by the term deflating a value series?

12.5 METHODS FOR CONSTRUCTION OF PRICE INDEXES

Various types of price indexes and their methods of construction can be classified into broad categories
as shown in the chart below:

12.6 UNWEIGHTED PRICE INDEXES

The unweighted price indexes are further classified into three groups as shown above in the chart.
The method of calculating each of these is discussed below:
426 BUSINESS S T A T I S T I C S

12.6.1 Single Price Index

A single unweighted price index number measures the percentage change in price for a single item or a basket of
items between any two time periods. Unweighted implies that all the values considered in calculating the
index are of equal importance.
An unweighted single price index is calculated by dividing the price of an item in the given
period by the price of the same item in the base period. To facilitate comparison with other years, the
actual price of the item can be converted into a price relative, which expresses the unit price in each
year (period) as a percentage of the unit price in a base year.
The general formula for calculating the single price index or price relative index is
pn
Single price index in period n = × 100
p0
where pn = price per unit of an item in the nth year
p0 = price per unit of an item in the base year
Example 12.1: The retail price of a typical commodity over a period of four years is given below:

Year : 2000 2001 2002 2003

Price (Rs.) : 24.60 25.35 26.00 26.50

(a) Find the price index based on 2000 prices

(b) Find the percentage change in price between consecutive years (base year = 2000) and the
percentage increase between consecutive years
Solution: (a) For the prices of the commodity with base year 2000, the price relatives for one unit of
the commodity in the years 2000 to 2003 are given in Table 12.2.

Table 12.2 Price Relatives

Year Price (Rs.) Price Relatives Percentage Change

2000 24.60 100 —
25.35
2001 25.35 × 100 = 103.04 3.04
24.60

26
2002 26.00 × 100 = 105.69 2.65
24.60

26.50
2003 26.50 × 100 = 107.72 2.03
24.60

(b) The percentage change in price relative is divided by the index it has come from and multiplied
by 100 for finding percentage increase.
103.04 − 100
For year 2001: × 100 = 3.04 per cent
100
105.69 − 103.04
For year 2002: × 100 = 2.57 per cent
103.04
107.72 − 105.69
For year 2003: × 100 = 1.92 per cent
105.69
 INDEX NUMBERS 427

12.6.2 Aggregate Price Index

An aggregate index price or composite price index measures the average price change for a basket of related
items from the base period to the current period. For example, to measure the change in the cost of living
over a period of time, we need the index that measures the change based on the price changes for a
variety of commodities including food, housing, clothing, transportation, health care, and so on.
Since the number of commodities is large, therefore a sample of commodities should be selected for
calculating the aggregate price index.
Irrespective of the units of measurement in which prices of several commodities are quoted, the
steps of the method to calculate an aggregate price index are summarized as follows:
(i) Add the unit prices of a group of commodities in the year of interest.
(ii) Add the unit prices of a group of commodities in the base year.
(iii) Divide the sum obtained in step (i) by the sum obtained in step (ii), and multiply the quotient
by 100.
From the sample of commodities or items included in the calculation of index, we cannot expect
a true reflection of price changes for all commodities. This calculation provides us with only a rough
estimate of price change.
A formula of calculating an unweighted aggregate price index is defined as:
Σ p1
Aggregate price index P01 = × 100 (12-1)
Σ p0
where p1 = unit price of a commodity in the current period of interest
p0 = unit price for a commodity in the base period
Example 12.2: The following are two sets of retail prices of a typical family’s shopping basket. The
data pertain to retail prices during 2001 and 2002.

Commodity Unit Price (Rs.)

2001 2002
Milk (1 litre) 18 20
Eggs (1 dozen) 15 18
Butter (1 kg) 120 150
Bread (500 gm) 9 11

Calculate the simple aggregate price index for 2002 using 2000 as the base year.
Solution: Calculations for aggregate price index are shown in Table 12.3.

Table 12.3 Calculation of Aggregate Price Index

Commodity Unit Price (Rs.)

2000 ( p0) 2002 ( p1)
Milk (1 litre) 18 20
Eggs (1 dozen) 15 18
Butter (1 kg) 120 150
Bread (500 gm) 9 11
Total 162 199
428 BUSINESS S T A T I S T I C S

The unweighted aggregate price index for expenses on a few food items in 2002 is given by
Σ p1 199
P01 = × 100 = × 100 = 122.83
Σ p0 162
The value P01 = 122.83 implies that the price of food items included in the price index has
increased by 22.83% over the period 2000 to 2002.

Limitations of an Unweighted Aggregate Price Index

1. The unweighted aggregate approach of calculating a composite price index is heavily influenced
by the items with large per unit price. Consequently items with relatively low unit price are
dominated by the high unit price items.
2. Equal weights are assigned to every commodity included in the index irrespective of the relative
importance of the commodity in terms of the amount purchased by a typical consumer. In other
words, it did not attach more weight or importance to the price change of a high-use commodity
than it did to a low-use commodity. For example, a family may purchase 30 packets of 500 gm
bread in a month while it is unusual to buy 30 kg butter every month. A substantial price change
for slow-moving items like butter, ghee can distort an index.
Due to these limitations, the unweighted index is not widely used in statistical analyses. These
limitations suggest the use of weighted index. There are two methods to calculate weighted
index, and these will be discussed later in the chapter.

12.6.3 Average Price Relative Index

This index is an improvement over the aggregate price index because it is not affected by the unit in
which prices are quoted. However, it also suffers from the problem of equal importance (weight)
given to all the items or commodities included in the index.
Steps of the method to calculate average price relative index are summarized as follows:
(i) Select a base year, and then divide the price of each commodity in the current year by the
price in the base year, to obtain price relatives.
(ii) Divide the sum of the price relatives of all commodities by the number of commodities used
in the calculation of the index.
(iii) Multiply the average value obtained in step (ii) by 100 to express it in percentage.
The formula for computing the index is as follows:

1  p1 
Average price relative index P01 = Σ   100 (12-2)
n  p0 

where n = number of commodities included in the calculation of the index.

The average used in computing the index of price relatives could be arithmetic mean or geometric
mean. When geometric mean is used for averaging the price relatives, the formula (12-1) becomes

1  p   1 p 
log P01 = Σ log  1  100  = Σ log P ; P =  1  100
n p
 0   n  p0 

Then P01 = antilog { 1

n }
Σ log p
 INDEX NUMBERS 429

Advantages: This index has the following advantages over the aggregate price index:
(i) The value of this index is not affected by the units in which prices of commodities are
quoted. The price relatives are pure numbers and therefore are independent of the original
units in which they are quoted.
(ii) Equal importance is given to each commodity and extreme commodities do not influence the
index number.

Limitations: Despite the few advantages mentioned above, this index is not popular on account of the
following limitations.
(i) Since it is an unweighted index, therefore each price relative is given equal importance.
However in actual practice a few price relatives are more important than others.
(ii) Although arithmetic mean is often used to calculate the average of price relatives, it also has
a few biases. The use of geometric mean is computationally difficult. Other measures of
central tendency such as median, mode and and harmonic mean, are almost never used for
calculating this index.
(iii) Index of price relatives does not satisfy all criteria such as identity, time reversal, and circular
properties, laid down for an ideal index. These criteria will be discussed later in the chapter.

Example 12.3: From the data given below, construct the index of price relatives for the year 2002
taking 2001 as base year using (a) arithmetic mean and (b) geometric mean.

Expenses on : Food Rent Clothing Education Misc.

Price (Rs.), 2001 : 1800 1000 700 400 700
Price (Rs.), 2002 : 2000 1200 900 500 1000

Solution: Calculations of Index number using arithmetic mean (A.M.) is shown in Table 12.4.

Table 12.4 Calculation of Index Using A.M.

Expenses on Price in Price in Price Relatives

2001( p0) 2000( p1) p1
× 100
p0
Food 1800 2000 111.11
Rent 1000 1200 120.00
Clothing 700 900 128.57
Education 400 500 125.00
Miscellaneous 700 1000 142.86
627.54

1  p1  1
Average of price relative index P01 = Σ   100 = (627.54) = 125.508
n  p0  5
Hence, we conclude that prices of items included in the calculation of index have increased by
25.508% in 2002 as compared to the base year 2001.
430 BUSINESS S T A T I S T I C S

(b) Index number using geometric mean (G.M.) is shown in Table 12.5
Table 12.5: Calculations of Index Using G.M.

Expenses on Price in Price in Price Relatives Log P

p1
2001( p 0) 2002( p2) P= × 100
p0
Food 1800 2000 111.11 2.0457
Rent 1000 1200 120.00 2.0792
Clothing 700 900 128.57 2.1090
Education 400 500 125.00 2.0969
Miscellaneous 700 1000 142.86 2.1548
10.4856

Average price relative index P01 = antilog {1

n } 1
{
Σ log p = antilog (10.4856)
5 }
= antilog (2.0971) = 125.00

S e l f-P r a c t i c e P r o b l e m s 12A
12.1 The following data concern monthly salaries (b) Calculate the percentage points change
for the different classes of employees within a between consecutive years.
small factory over a 3-year period.
12.3 A State Govt. had compiled the information
Employee Salary per Month shown below regarding the price of the three
essential commodities: wheat, rice, and sugar.
Class 1998 1999 2000 From the commodities listed, the corresponding
price indicates the average price for that year.
A 2300 2500 2600 Using 1998 as the base year, express the price
B 1900 2000 2300 for the years 2000 to 2002, in terms of
C 1700 1700 1800 unweighted aggregate index.
D 1000 1100 1300
Commodity 1998 1999 2000 2001 2002
Using 1998 as the base year, calculate the simple
aggregate price index for the years 1999 and Wheat 4 6 8 10 12
2000. Rice 16 20 24 30 36
12.2 The following data describe the average salaries Sugar 8 10 16 20 24
(Rs. in ’000) for the employees in a company
over ten consecutive years. 12.4 Following are the prices of commodities in 2003
and 2004. Calculate a price index based on price
Year : 1 2 3 4 5 relatives, using the geometric mean.
Average salary : 10.9 11.4 12.0 12.7 13.6
Year Commodity
Year : 6 7 8 9 10
Average salary : 14.4 15.0 15.5 16.3 13.6 A B C D E F
2003 45 60 20 50 85 120
(a) Calculate an index for these average 2004 55 70 30 75 90 130
salaries using year 5 as the base year.
 INDEX NUMBERS 431

12.5 A textile worker in the city of Mumbai earns

Commodity Quantity (in units) Price (in Rs.) per unit
Rs. 3500 per month. The cost of living index in 1995
for a particular month is given as 136. Using 1995 1996
the following information, find out the amount A 110 8.00 12.00
of money he spent on house rent and clothing.
B 25 6.00 7.50
Group Expenditure (Rs.) Group Index C 10 5.00 5.25
D 20 48.00 60.00
Food 1400 180 E 25 15.00 16.50
Clothing x 150 F 30 9.00 27.00
House rent y 100
Food and lighting 1560 110 [Karnataka Univ., B.Com, 1997]
Misc. 1630 180 12.8 The following table gives the annual income of
a teacher and the general index of price during
[Delhi Univ., B.Com, 1997] 1990–97. Prepare the index number to show
12.6 In 1996, for working class people, wheat the change in the real income of the teacher
was selling at an average price of Rs. 160 and comment on price increase:
per 10 kg, cloth at Rs. 40 per metre, house
rent Rs. 10,000 per house, and other items Year Income Index
at Rs. 100 per unit. By 1997 the cost of 1990 4000 100
wheat rose by Rs. 40 per 10 kg, house rent 1991 4400 130
by Rs. 1500 per house, and other items
1992 4800 160
doubled in price. The working class cost of
living index for the year 1997 (with 1996 as 1993 5200 220
base) was 160. By how much did the cloth 1994 5600 270
price rise during the period 1996–97? 1995 6000 330
12.7 From the following data calculate an index 1996 6400 400
number using family budget method for the 1997 6800 490
year 1996 with 1995 as the base year.
[HP Univ., B.Com, 1997]

Hints and Answers

12.1 Simple aggregate price index
(b) Ye a r Index Percentage
7300 Number Point Change
P0, 89 = × 100 = 105.8 for the year 1999
6900 1 80.1 —
8000 2 83.8 3.7
P0, 90 = × 100 = 115.9 for the year 2000
6900 3 88.2 4.4
12.2 (a) 4 93.4 5.2
5 100.0 6.6
Year : 1 2 3 4 5 6 105.9 5.9
Index number : 80.1 83.8 88.2 93.4 100 7 110.3 4.4
Year : 6 7 8 9 10 8 114.0 3.7
Index number : 105.9110.3114.0119.9129.4 9 119.9 5.9
10 129.4 9.5
For example, index for year 1: 12.3 Aggregate price
(10.9 ÷ 13.6)100 = 80.1;
1998 1999 2000 2001 2002
year 2: (11.4 ÷ 13.6)100 = 83.8
100 133.33 137.78 125 120
432 BUSINESS S T A T I S T I C S

12.4 The index for 1997 as given is 160. Therefore,

the sum of the index numbers of the four
P1 commodities would be 160 × 4 = 640. Thus
Commodity P= ×100 Log P
P0 440 + 2.5x = 640 or x = 80. Hence the rise in
A 122.22 2.0872 the price of cloth was Rs. 40 (80 – 40) per
B 116.67 2.0669 metre.
C 150.00 2.1761 12.7
D 150.00 2.1761
E 105.88 2.0248 p1
Commodity Quantity p0 p1 P = × 100 PQ
F 108.33 2.0348 p0
Q

P01 = antilog
1
n { }
log P = antilog
1
6 {
(12.5659) } A
B
100
25
8
6
12.00
7.50
150
125
15,000
3,125
= antilog (2.0948) = 124.4
C 10 5 5.25 105 1,050
12.5 Let expenditure on clothing be x and on house
rent be y. Then as per conditions given, we have D 20 48 60.00 125 2,500
3500 = 1400 + x + y + 560 + 630 E 25 15 16.50 110 2,750
or x + y = 910 (i) F 30 9 27.00 300 9,000
Multiplying expenditure with group index and Total 210 33,425
equating it to 136, we get
(1400 × 180) + ( x × 150) + ( y × 100) ΣPQ 33, 425
+(500 × 110) + (630 × 80) Index number = = = 159.17
136 = ΣQ 210
3500
2,52,000 + 150 x + 100 y + 61,600 + 50, 400
12.8
136 =
3500 Year Income Index Real Income Real Income
4,76,000 = 2,52,000 + 150x + 100y + 61,600
(Rs.) (Rs.) Index
+ 50,400
150x + 100y = 1,12,000 (ii) 4000
1990 4000 100 ×100 = 4000.00 100.00
Multiplying Eqn. (i) by 150 and subtracting it 100
from (ii), we get
4400
50y = 24,500 or y = Rs. 490 (house rent) 1991 4400 130 ×100 = 3384.62 84.62
130
Substituting the value of y in Eqn. (i):
x + 490 = 910 or x = Rs. 420 (clothing) 4800
1992 4800 160 ×100 = 3000.00 75.00
12.6 Let the rise in price of cloth be x. 160

Commodity Price Index Price 1997 Index 5200

1993 5200 220 ×100 = 2363.64 59.09
220
200
Wheat 160 100 200 × 100
160 5600
1994 5600 270 ×100 = 2074.07 51.85
= 125 270
x
Cloth 40 100 x × 100
40 Σp1 q1
1995 6000 330 ×100 = 1818.18 45.45
= 2.5x Σp0 q1
11,500
House 10,000 100 11,500 × 100
10,000 6400
rent 1996 6400 400 ×100 = 1600.00 40.00
= 115 400
200
Miscellaneous 100 100 200 × 100
100 6800
1997 6800 490 ×100 = 1387.76 34.69
= 200 490
Total 440 + 2.5x
 INDEX NUMBERS 433

12.7 WEIGHTED PRICE INDEXES

While constructing weighted price indexes, rational weights are assigned to all items or commodities in
an explicit manner. Such weights indicate the relative importance of items or commodities included in
the calculation of an index. The weights used are of two types, quantity weights and value weights. There
are two price indexes that are commonly in use
1. Weighted aggregate price index
2. Weighted average of price relative index

12.7.1 Weighted Aggregate Price Index

In a weighted aggregate price index, each item in the basket of items chosen for calculation of the
index is assigned a weight according to its importance. In most cases, the quantity of usage is the best
measure of importance. Hence, we should obtain a measure of the quantity of usage for the various
items in the group. This explicit weighting allows us to gather more information than just the change
in price over a period of time as well as improve the accuracy of the general price level estimate.
Weight is assigned to each item in the basket in various ways and the weighted aggregates are also
used in different ways to calculate an index. A few methods (or approaches) to determine weights (value)
to be assigned to each item in the basket are as follows:
• Laspeyre’s method • Marshall-Edgeworth’s method
• Paasche’s method • Walsch’s method
• Dorbish and Bowley’s method • Kelly’s method
• Fisher’s ideal method

Laspeyre’s Weighting Method

This method suggests to treat quantities as constant at base period level and are used for weighting
price of each item or commodities both in base period and current period. Since this index number
depends upon the same base price and quantity, therefore one can directly compare the index of one
period with another. The formula for calculating Laspeyre’s price index, named after the statistician
Laspeyre’s is given by
Σ p1 q0
Laspeyre’s price index = × 100
Σ p0 q0
where p1 = prices in the current period
p0 = prices in the base period
q0 = quantities consumed in the base period

Advantages: The main advantage of this method is that it uses only one quantity measure based on the
base period and therefore we need not keep record of quantity consumed in each period. Moreover,
having used the same base period quantity, we can compare the index of one period with another
directly.
Disadvantages: We know that the consumption of commodities decreases with relatively large increases
in price and vice versa. Since in this index the fixed quantity weights are determined from the base
period usage, it does not adjust such changes in consumption and therefore tends to result in a bias
in the value of the composite price index.
Example 12.4: Compute the cost of living index number using Laspeyre’s method, from the following
information:
434 BUSINESS S T A T I S T I C S

Commodity Unit Consumption Price in Price in

in Base Period Base Period Current Period
Wheat 200 1.0 1.2
Rice 50 3.0 3.5
Pulses 50 4.0 5.0
Ghee 20 20.0 30.0
Sugar 40 2.5 5.0
Oil 50 10.0 15.0
Fuel 60 2.0 2.5
Clothing 40 15.0 18.0

Solution: Calculation of cost of living index by Laspeyre’s method is shown in Table 12.6.

Table 12.6 Laspeyre’s Method

Commodity Base Period Quantity Base Period Price Current Price

(q0) ( p0 ) ( p1) p1q0 p0q0
Wheat 200 1.0 1.2 240 200
Rice 50 3.0 3.5 175 150
Pulses 50 4.0 5.0 250 200
Ghee 20 20.0 30.0 600 400
Sugar 40 2.5 5.0 200 100
Oil 50 10.0 15.0 750 500
Fuel 60 2.0 2.5 150 120
Clothing 40 15.0 18.0 720 600
Total 510 3085 2270

Σ p1 q0 3085
Cost of living index = × 100 = × 100 = 135.9
Σ p0 q0 2270

Paasche’s Weighting Method

In the Paasche’s method, the price of each item or commodity is weighted by the quantity in the
current period instead of the base year as used in Laspeyre’s method. Paasche’s formula for calculating
the index is given by
Σ p1q1
Paasche price index =
Σ p0 q1
where p1 = prices in current year
p0 = prices in base year
q1 = quantities in current year
Advantages: The Paasche’s method combines the effects of changes in price and quantity consumption
patterns during the current year. It provides a better estimate of changes in the economy than
Laspeyre’s method. If the prices or quantities of all commodities or items change in the same ratio,
then the values of the Laspeyre’s and Paasche’s indexes will be same.
Disadvantages: This method requires knowledge of the quantities consumed of all commodities in
each period. Getting the data on the quantities for each period is either expensive or time-consuming.
 INDEX NUMBERS 435

Moreover, each year the index number for the previous year requires recomputation to reflect the
effect of the new quantity weights. Thus, it is difficult to compare indexes of different periods when
calculated by the Paasche’s method.
Example 12.5: For the following data, calculate the price index number of 1999 with 1998 as the
base year, using: (a) Laspeyre’s method, and (b) Paasche’s method.

Commodity 1998 1999

Price Quantity Price Quantity
A 20 18 40 16
B 50 10 60 15
C 40 15 50 15
D 20 20 20 25

Solution: Table 12.7 presents the information necessary for both Laspeyre’s and Paasche’s methods.

Table 12.7 Calculation of Laspeyre’s and Paasche’s and Paasche’s Indexes

Commodity Base Period, 1998 Current year, 1999

Price Quantity Price Quantity
( p0) (q0) ( p 1) (q1) p1 q 0 p0 q 0 p1 q 1 p0 q 1
A 20 18 40 26 320 160 240 120
B 50 10 60 25 600 500 300 250
C 40 15 50 15 750 600 750 600
D 20 20 20 25 400 400 500 500
2070 1660 1790 1470
Σp1q 0 2070
Laspeyre’s price index = × 100 = × 100 = 124.7
Σp 0 q 0 1660
Σp1q1 1790
Paasche’s price index = × 100 = × 100 = 121.77
Σp 0 q1 1470
The Paasche’s price index shows a price level increase of 21.77 per cent, while Laspeyre’s index
shows a price level increase of 24.7 per cent. Hence, we may conclude that Paasche’s index shows a
trend towards less expensive commodities.
Dorbish and Bowley’s Method
This method (or approach) is the simple arithmetic mean of the Laspeyre’s and Paasche’s indexes. This
index takes into account the influence of quantity weights of both base period and current period.
The formula for calculating the index using Dorbish and Bowley method is given by
1  Σ p1 q0 Σ p1 q1 
Dorbish and Bowley’s price index =  +  100
2  Σ p0 q0 Σ p0 q1 

Fisher’s Ideal Method

This method (or approach) is the geometric mean of the Laspeyre’s and Paasche’s indexes and the
formula is given by
Σ p1q0 Σ p1q1
Fisher’s ideal price index = × × 100
Σ p0 q0 Σ p0 q1
436 BUSINESS S T A T I S T I C S

Advantages: Fisher’s method is also called ideal method due to following reasons:
(i) The formula is based on geometric mean which is considered to be the best average for
constructing index numbers.
(ii) The formula takes into account both base year and current year quantities as weights. Thus
it avoids the bias associated with the Laspeyre’s and Paasche’s indexes.
(iii) This method satisfies essential tests required for a index, that is, time reversal test and factor
reversal test.
Disadvantages: The calculation of index using this method requires more computation time. Although
the index number is theoretically better than others discussed previously, it is not fit for common use
because it requires current quantity weights every time an index is calculated.

Example 12.6: Compute index number from the following data using Fisher’s ideal index formula.

Commodity 1999 2000

Price Quantity Price Quantity
A 12 10 15 12
B 15 27 20 25
C 24 25 20 29
D 25 16 25 14

Solution: Table 12.8 presents the information necessary for Fisher’s method to calculate the index.

Table 12.8 Calculations of Fisher Ideal Index

Commodity Base Year, 1999 Current Year, 2000

( q 0) ( p0 ) (q1) ( p1) p 1 q0 p 0 q0 p 1 q1 p 0 q1
A 12 10 15 12 144 120 180 150
B 15 7 20 5 75 105 100 140
C 24 5 20 9 216 120 180 100
D 5 16 5 14 70 80 70 80
505 425 530 470

Σ p1q0 Σ p1q1 505 530

Fisher’s ideal price index = × × 100 = × × 100
Σ p0 q0 Σ p0 q1 425 470

= 1.3399 ×100 = 1.1576×100 = 115.76

Hence, we conclude that the price level has increased by 15.76% in the year 2000.
Example 12.7: Calculate from the following data, the Fisher’s ideal index number for the year 2000:

1999 2000
Commodity Price Expenditure on Quantity Price Expenditure on Quantity
(Rs.) Consumed (Rs.) (Rs.) Consumed (Rs.)
A 8 200 65 1950
B 20 1400 30 1650
C 5 80 20 900
D 10 360 15 300
E 27 2160 10 600
 INDEX NUMBERS 437

Solution: Table 12.9 presents the information necessary for Fisher’s method to calculate the index.

Table 12.9 Calculations of Fisher’s Ideal Index

Commodity Base Year, 1999 Current Year, 2000

( p0) (q0) ( p 1) (q1) p1 q0 p0 q 0 p1 q 1 p0 q 1

A 8 200/8 = 25 65 1950/65 = 30 200 1950 240 240

B 20 1400/20 = 70 30 1650/30 = 55 2100 1400 1650 1100
C 5 80/5 = 16 20 900/20 = 45 320 80 900 225
D 10 360/10 = 36 15 300/15 = 20 540 360 300 200
E 27 2160/27 = 80 10 600/10 = 60 800 2160 600 1620
5385 4200 5400 3385

Σ p1q0 Σ p1q1 5385 5400

Fisher’s ideal price index = × × 100 = × × 100
Σ p0 q0 Σ p0 q1 4200 3385

= 1.430 × 100 = 143.

Hence we conclude that the price level has increased by 43% in the year 2000.
Example 12.8: Compute index numbers from the following data using (i) Laspeyre’s, (ii) Paasche’s
(iii) Fisher’s Ideal formulae:
Base Year Current Year
Commodity Quantity Price Quantity Price
A 8 4 10 9
B 7 3 8 4
C 6 4 7 8
D 5 2 5 4

[Delhi Univ., B.Com (P), 1990]

Solution: The following table presents information necessary for calculating required index numbers
Base Year Current Year
Commodity Price p0 Qty. q0 Price p1 Qty. q1 p1 q 0 p0 q 0 p1 q 1 p0 q 1

A 4 8 9 10 72 32 90 40
B 3 7 4 8 28 21 32 24
C 4 6 8 7 48 24 56 28
D 2 5 4 5 20 10 20 10
168 87 198 102

Σp1 q0 168
Laspeyre’s index = × 100 = × 100 =193.10.
Σp0 q0 87
Σp1 q1 198
Paasche’s index = × 100 = × 100 =194.12.
Σp0 q1 102

Σp1 p0 Σp1 q1 168 198

Fisher’s index = × × 100 = × × 100 =193.61.
Σp0 q0 Σp0 q1 87 102
438 BUSINESS S T A T I S T I C S

Example 12.9: Give that Σ p1 q1 = 250, Σ p0 q1 = 150; Paasche’s index number = 150 and Dorbish
and Bowley’s index number = 145. Find out Fisher’s ideal index number and Marshal Edgeworth
index number.[Delhi Univ., B.Com (Hons), 1992, 2005]

Solution: Given, Σ p1 q1 = 250, Σ p0 q0 = 150

Σ p1 q1 250 × 100
Paasche’s index number = × 100 or 150 =
Σ p0 q1 Σ p0 q1

25000 500
or 150 Σ p0 q1 = 25000, i.e. Σ p0 q1 = = =167 (approx)
150 3

 Σ p1 q1 + Σ p1 q1  × 100
Σ p q 
 0 0 Σ p0 q1 
Dorbish-Bowley’s index number =
2

Σ p q 250  145 Σ p1 q0
or 145 =  1 0 +  50 or = + 1.497
 150 167  50 150

Σp1 q0 Σp1 q0
or 2.90 – 1.50 = or 1.40 = [Taking 1.497 as 1.50]
150 150
or Σ p1 q0 = 1.40×150 = 210

Σp1 q0 Σp1 q1 210 250

Fisher’s index number = × × 100 = × × 100
Σp0 q0 Σp0 q1 150 167

= 100 × 1.40 × 1.50 = 100 2.10 = 100×1.449 = 144.9

Example 12.10: Give that Σp1 q1 = 250, Σ p0 q0 = 150, Paasche’s index number = 150 and Dorbish
and Bowley’s index number = 145. Find Fisher’s Ideal index number and Σ p1 q0.
[Delhi Univ., B.Com (Hons), 2005]
Σ p1 q1 250
Solution: Paashe’s Index number = × 100 or 150 = × 100
Σ p0 q1 Σ p0 q1

250 500
or Σ p0 q1 = × 100 = =166.6 ~ 167
150 3
L+P L + 150
Dorbish and Bowley’s index number = or 145 = , i.e. L = 140
2 2

Σp1 q0 Σp1 q0
But L = × 100 , or 140 = × 100
Σp0 q0 150

140 × 150
or Σ p1 q0 = = 210
100

Fisher’s ideal index = L × P = 140 × 150 =144.91

 INDEX NUMBERS 439

Marshall-Edgeworth Method
In this method the sum of base year and current year quantities are considered as the weight to
calculate the index. The formula for constructing the index is:
Σ ( q0 + q1 ) p1 Σ q0 p1 + Σ q1 p1
Marshall-Edgeworth price index = × 100 = × 100
Σ ( q0 + q1 ) p0 Σ q0 p0 + Σ q1 p0
where notations have their usual meaning.
The disadvantage with this formula is the same as that of Paasche index and Fisher’s ideal index
in the sense that it also needs current quantity weights every time an index is constructed.

Walsch’s Method
In this method the quantity weight used is the geometric mean of the base and current year quantities.
The formula for constructing the index is
Σ p1 q0 q1
Walsch’s price index = × 100
Σ p0 q0 q1
Although this index satisfies the time reversal test, it needs current quantity weight every time an
index is constructed.

Kelly’s Method
The method suggested by T.L. Kelly for the construction of index number is
Σ p1q
Kelly’s price index = × 100
Σ p0 q
where q = fixed weight.
This method is also called the fixed weight aggregate method because instead of using base period or
current period quantities as weights, it uses weights from a representative period. The representative
weights are referred to as fixed weight. The fixed weights and the base period prices do not have to
come from the same period.
Advantages and Disadvantages of Kelly’s Method
Advantages: An important advantage of this index is that it does not need yearly changes in the
weights. One can select a different period for fixed weight other than base period. This can improve
the accuracy of the index. Moreover, the base period can also be changed without changing the fixed
weight. The weights should be appropriate and should indicate the relative importance of various
commodities. This weight may be kept fixed until new data are available to revise the index.
Disadvantages: One disadvantage with this index is that it does not take into account the weight either
of the base year or of the current year.
Example 12.11: It is stated that the Marshall-Edgeworth index number is a good approximation of
the ideal index number. Verify this statement using the following data:

Commodity 2002 2003

Price Quantity Price Quantity
A 2 174 3 182
B 5 125 4 140
C 7 140 6 133
440 BUSINESS S T A T I S T I C S

Solution: Table 12.10 presents the information necessary to calculate Fisher and Marshall-Edgeworth
indexes.

Table 12.10 Calculations of Fisher’s Ideal and Marshall-Edgeworth’s Index

Commodity Base Year, 2002 Current Year, 2003

( p0 ) (q0) ( p1 ) (q1) p1 q0 p0 q 0 p1 q 1 p0 q 1
A 2 174 3 182 222 148 246 164
B 5 125 4 140 500 625 560 700
C 7 140 6 133 240 280 198 231
962 1053 1004 1095

Σ p1q0 Σ p1q1 962 1004

Fisher ideal price index = × ×100 = × ×100
Σ p0 q0 Σ p0 q1 1053 1095
0.836 ×100 = 0.9144×100 = 91.44
=
Σ p1 ( q0 + q1 ) Σ p1 q0 + Σ p1 q1
Marshall-Edgeworth price index = ×100 = ×100
Σ p0 ( q0 + q1 ) Σ p0 q0 + Σp0 q1
962 + 1004
= ×100 = 0.9152×100 = 91.52
1053 + 1095
Hence, we conclude that Fisher’s method and Marshall-Edgeworth method provide almost the same
value of the index.
Example 12.12: From the following data, construct quantity index number by (i) Fisher’s method;
and (ii) Marshall-Edgeworth’s method:

Base Year Current Year

Commodities Price Quantity Expenditure Quantity
(Rs.) (Rs.) (Rs.)

A 25 40 2,000 50
B 22 18 1,200 30
C 54 16 1,320 44
D 20 40 1,350 45
E 18 30 630 15

[Delhi Univ., BCom(H), 1990]

Solution: The following table presents necessary information to calculate required index numbers.
Base Year Current Year
Commodity p0 q0 p 1* q1 q 1 p0 q 0 p0 q 1 p1 q 0 p1

A 25 40 40 50 1250 1000 2000 1600

B 22 18 40 30 660 396 1200 720
C 54 16 30 44 2376 864 1320 480
D 20 40 30 45 900 800 1350 1200
E 18 30 42 15 270 540 630 1260
5456 3600 6500 5260
*
Current year prices have been calculated by dividing the expenditure by quantity.
 INDEX NUMBERS 441

Σ q1 p0 Σ q1 p1
Fisher’s quantity index number, Q01 = × × 100
Σ q0 p0 Σ q0 p1
5456 6500 35,464
= × × 100 = × 100 =136.85
3600 5260 18,936
Σ q1( p1 + p0 )
Marshall Edgeworth quantity index, Q01 = × 100
Σ q0 ( p1 + p0 )
Σ q1 p1 + Σ q1 p0 6500 + 5456
= × 100 = × 100 =134.94.
Σ q0 p1 + Σ q0 p0 5260 + 3600
Example 12.13: Compute Laspeyre’s, Paasche’s, Fisher’s, and Marshall-Edgeworth’s index num-
bers from the following data:

Item 1998 1999

Price Quantity Price Quantity
A 5 25 6 30
B 3 8 4 10
C 2 10 3 8
D 10 4 3 5

[Bangalore Univ., BCom, 2000]

Solution: Table 12.11 presents the information necessary to calculate several indexes.

Table 12.11 Calculations of Indexes

Item Base Year, 1998 Current Year, 1999

( p0) (q0) ( p 1) (q1) p1 q0 p0 q0 p1 q1 p0 q1
A 5 25 6 30 150 125 180 150
B 3 8 4 10 32 24 40 30
C 2 10 3 8 30 20 24 16
D 10 4 3 5 12 40 15 50
224 209 259 246

Σ p1 q0 224
Laspeyre’s price index = ×100 = ×100 = 107.17
Σ p0 q0 209

Σ p1 q1 259
Paasche’s price index = ×100 = ×100 = 105.28
Σ p0 q1 246

Fisher’s ideal price index = L×P = 107.17 × 105.28 = 106.22

Σ p1 ( q0 + q1 ) Σ p1 q0 + Σ p1 q1
Marshall-Edgeworth’s price index = ×100 = ×100
Σ p0 ( q0 + q1 ) Σ p0 q0 + Σp0 q1

244 + 259
= ×100 = 110.55
209 + 246
442 BUSINESS S T A T I S T I C S

12.7.2 Weighted Average of Price Relative Index

Unlike the unweighted average of price relative, the weighted average of price relative is determined
by using the quantity consumed in the base period for weighting the items or commodities. The value
(in rupees) of each item or commodity included in the calculation of composite index is determined
by multiplying the price of each item by its quantity consumed.
The formula for constructing the weight average of price relatives index using base values is:

Σ {( p1 / p0 ) × 100} ( p0 q0 ) Σ PV
Weighted average of price relative index, P01 = =
Σ p0 q0 ΣV

Σ p1 q0
= ×100
Σ p0 q0
where V(= p0q0) = base period value
P(= (p0/q0) × 100 = price relative
This formula is equivalent to Laspeyre’s method for any given problem.
If we wish to compute a weighted average of price relative using V = p0q1, then the above formula
becomes
Σ {( p1 / p0 ) × 100}( p0 q1 ) Σ p1q1
P01 = = ×100
Σ p0 q1 Σ p0 q1

This formula is equivalent to Paasche’s method for any given problem.

If instead of using weighted arithmetic average, we wish to use weighted geometric mean, then
the above formula becomes
Σ V × log P p1
P01 = ; P= ×100 and V = p0q0
ΣV p0

Advantages of Weighted Average Price Relatives

(i) Different index numbers constructed using average price relative with same base can be
combined to form a new index.
(ii) Weighted average of price relative method is suitable to construct an index by selecting one
item from each of the many subgroups of items. In such a case, the values of each subgroup
may be used as weights.
Example 12.14: Compute price index by using weighted average of price relative method based on
the following data.

Quantity (kg) Price per Unit (Rs.)

Items (q0) p0 p1

A 3 20 4.0
B 1.5 40 1.6
C 1.0 10 1.5

Solution: The following table presents the necessary information to calculate the weighted average
price relative index.
 INDEX NUMBERS 443

Price Relatives Weighted Average

Items q0 p0 p1 V = p0 q0 P = (p1/p0) × 100 Relatives PV

4
A 20 3 4 60 × 100 =133.33 8000
3

1.6
B 40 1.5 1.6 60 × 100 =106.67 6400
1.5

1.5
C 10 1 1.5 10 × 100 =150 1500
1

130 15,900

ΣPV 15,900
Weighted average of price relative index, P01 = = =122.31.
ΣV 130
Example 12.15: A large manufacturer purchases an identical component from three different suppliers
that differ in unit price and quantity supplied. The relevant data for 2000 and 2001 are given below:

Supplier Quantity Index Unit Price (Rs.)

in (2000) 2000 2001
A 20 18 20
B 40 12 14
C 10 15 16

Construct a weighted average price relative index using (a) arithmetic mean and (b) geometric mean.
Solution: Table 12.12 presents the information necessary to calculate the weight average price relative
index.
(a) Weighted average of price relative index
[Σ( p1 / p0 ) 100] p0 q0
1,12,001.70
P01 = == 113.13
Σ p0 q0 990
The value of P01 implies that there has been 13.13% increase in price from year 2000 to 2001.

Table 12.12 Calculations of Weighted Average of Price Relatives

Supplier Prices in Quantity in Percentage Base Value Weighted Percentage

2000 2001 2000 Price Relative V = p0 q 0 Relative PV
p
p0 p1 q0 P = 1 × 100
p0

A 18 20 20 (20/18) × 100 = 111.11 360 39,999.60

B 12 14 40 (14/12) × 100 = 116.67 480 56,001.60
C 15 16 10 (16/15) × 100 = 106.67 150 16,000.50
990 1,12,001.70
444 BUSINESS S T A T I S T I C S

(b)
Table 12.13 Calculations of Weighted Geometric Mean of Price Relatives

Supplier Prices in Quantity in Base Percentage Log P V log P

2000 2001 2000 Value Price Relative
p1
p0 p1 q0 V = p0 q 0 P = p × 100
0

A 18 20 20 360 111.11 2.046 736.56

B 12 14 40 480 116.67 2.067 992.16
C 15 16 10 150 106.67 2.028 304.20
990 2032.92

Weighted geometric mean of price relatives (Table 12.13)

 ΣV × logP   2032.92 
P01 = antilog   = antilog  
 ΣV   990 
= antilog (2.0535) = 113.11

12.8 QUANTITY OR VOLUME INDEXES

A quantity index measures the percentage change in consumption, production or distribution level
of either an individual item or a basket of items from one time period to another. When constructing
quantity indexes, it is necessary to hold price levels constant over time to isolate the effect of quantity
(consumption level) changes only. For example, agricultural production is measured using a quantity
index because it eliminates effects of fluctuating prices. Any of the methods, such as the relative
method (both simple and weighted) or the aggregative method, which take into account weights to
construct price indexes can also be used to calculate quantity indexes. The weights in quantity index
numbers are prices. Therefore quantity indexes can be easily derived from a price indexes by
interchanging the p’s and q’s.
Seven quantity indexes analogous to the seven price indexes already discussed in the previous
section can be constructed as given below:

Σ V0 ( q1 / q0 ) Σ q1 p0
Laspeyre’s quantity index QL = × 100 = × 100
Σ V0 Σ q0 p0

where V0 = p0 q0, values of base year consumption at base year prices

Σ V1 ( q1 / q0 ) Σ q1 p1
Similarly, Paasche’s quantity index QP = × 100 = × 100
Σ V1 Σ q0 p1

where V1 = p0 q0, values of base year consumption at current year prices

Σ q1 p0 Σ q1 p1
Fisher’s quantity index QF = QL × QP = × × 100
Σ q0 p1 Σ q0 p1
 INDEX NUMBERS 445

The formula for computing a weighted average of quantity relative index is also the same as used
to compute a price index. The formula for this type of quantity index is
q 
Σ  1 × 100  ( q0 p0 )
 q0 
Weighted average of quantity relative index =
Σ q0 p0
where q 1 = quantities for the current period
q0 = quantities for the base period

Example 12.16: Obtain Laspeyre’s price index number and Paasche’s quantity index number from
the following data:

Price (Rs. per Unit) Quantity (Units)

Item Base Year Current Year Base Year Current Year
1 2 5 20 15
2 4 8 4 5
3 1 2 10 12
4 5 10 5 6

[Mangalore Univ., BCom, 1997]

Solution: Table 12.14 presents the information necessary to calculate Laspeyre’s price index and
Paasche’s quantity indexes.

Table 12.14 Calculations on Laspeyre’s Price Index and Paasche’s Quantity Index

Price Quantity
Item p0 p1 q0 q1 p1 q 0 p0 q 0 q1 p 1 q0 p 1
1 2 5 20 15 100 40 75 100
2 4 8 4 5 32 16 40 32
3 1 2 10 12 20 10 24 20
4 5 10 5 6 50 25 60 50
202 91 199 202

Σ p1 q0 202
Laspeyre’s price index = × 100 = × 100 = 221.98
Σ p0 q0 91
Σ q1 p1 199
Paasche’s quantity index = × 100 = × 100 = 98.51
Σ q0 p1 202
Example 12.17: Compute the quantity index by using Fisher’s formula from the data given below:

Commodity 2002 2003

Price (Rs./Unit) Total Value Price (Rs./Unit) Total Value
A 5 50 4 48
B 8 48 7 49
C 6 18 5 20
446 BUSINESS S T A T I S T I C S

Solution: The base year quantity q0 and current year quantity q1 for individual commodity can be
calculated as follows (Table 12.15):
Total value 50 48 18
q0 (for 2002) = = = 10; = 6; =3
Price 5 4 6
Total value 48 49 20
q1 (for 2003) = = = 12; = 7; =4
Price 4 7 5

Table 12.15 Calculations for Fisher’s Quantity Index

Price, 2002 Quantity, 2003

Commodity p0 p1 q0 q1 p1 q 0 p0 q 0 q 1 p1 q 0 p1
A 5 10 4 12 60 50 48 40
B 8 6 7 7 56 48 49 42
C 6 3 5 4 24 18 20 15
140 116 117 97

Substituting values in the formula, we get

Σ q1 p0 Σ q1 p1
Fisher’s quantity index= × × 100
Σ q0 p0 Σ q0 p1

140 117
= × × 100 = 120.65
116 97
Example 12.18: Calculate the weighted average of quantity relative index from the following data:
Quantity (Units) Price (Rs./Unit)
Commodity
2000 2002 2000
A 10 12 100
B 15 20 75
C 8 10 80
D 20 25 60
E 50 60 500

Solution: Table 12.16 presents the information necessary to calculate the weighted average of quantity
relative index.
Table 12.16 Calculations of a Weighted Average of Quantity Relatives Index

Commodity Quantity (units) Price (Rs./unit) Percentage Base Weighted

2000 2002 2000 Relatives Value Relatives
q0 q1 p0 (q1/q2)×100 q0 p0 {(q1/q0)×100}×q0p0

A 10 12 100 (12/10) × 100 = 120 1000 1,20,000.00

B 15 20 75 (20/15) × 100 = 133.33 1125 1,49,996.25
C 8 10 80 (10/8) × 100 = 125 640 80,000.00
D 20 25 60 (25/20) × 100 = 125 1200 1,50,000.00
E 50 60 500 (60/50) × 100 = 120 25,000 30,00,000.00
28,965 34,99,996.25
 INDEX NUMBERS 447

Σ {( q1 / q0 ) × 100}( q0 p0 )
Weighted average of quantity relatives index =
Σ q0 p0
34,99,996.25
= = 120.835
28,965

12.9 VALUE INDEXES

A value index number measures the percentage change in the total value of either an individual item
or a basket of items from one time period to another. The value of an item or commodity is obtained
by multiplying its price and quantity. Since value is determined both by price and quantity, a value
index measures the combined effects of price and quantity changes. A simple value ratio is equal to
the value of the current year divided by the value of the base year. If this ratio is multiplied by 100, we
get the value index as:
Σ p1 q1
Value index, V = × 100
Σ p0 q0
If the values are given directly, then the value index number is given by
Σ V1
Value index, V =
Σ V0
where V0 = values at the base year or period
V1 = values at the current year or period
Such indexes are not weighted because they take into account both the price and quantity. These
indexes are, however, not very popular because the situation revealed by price and quantities are not
fully revealed by the values. A value index does not distinguish between the effects of its components,
namely price and quantity.

S e l f-P r a c t i c e P r o b l e m s 12B
12.9 The following table contains information from employees of an industrial centre for a particular
the raw material purchase records of a small year (with base 1990 = 100) were:
factory for the year 2002 and 2003:
Food 200
Commodity 2002 2003 Clothing 130
Price Total Price Total Fuel and Lighting 120
(Rs./Unit) Value (Rs./Unit) Value Rent 150
A 15 50 6 72 Miscellaneous 140
B 17 84 10 80
The weights are 60, 8, 7, 10, and 15 respectively.
C 10 80 12 96
It is proposed to fix dearness allowance in such
D 14 20 5 30
a way as to compensate fully the rise in the
E 18 56 8 64
prices of food and house rent. What should
be the dearness allowance, expressed as a
Calculate Fisher’s ideal index number.
percentage of wage?
12.10 The subgroup indexes of the consumer price
index number for urban non-manual
448 BUSINESS S T A T I S T I C S

12.11 The owner of a small shop selling food items

Commodity 1996 2000
collected the following information regarding
the price and quantity sold of a particular item. Price Quantity Price Quantity
A 2 174 3 182
Average Price (Rs./Unit) Quantity Sold (Units) B 5 125 4 140
Item
2002 2003 2002 2003 C 7 140 6 133
A 1 2 10 5 12.16 In preparation for an appropriations hearing,
B 1 x 15 2 the DCP of a city zone has collected the following
information:
If the ratio between Laspeyre’s (L) and
Paasche’s (P) Index number is: L:P = 28:27, Type of Crime 2000 2001 Weight
then find the value of x.
12.12 An increase of 50 per cent in the cost of a certain Robberies 13 8 6
consumable product raises the cost of living of Car thefts 15 22 5
a certain family by 5 per cent. What percentage Cycle thefts 249 185 4
of its cost of living was due to buying that Pocket picking 328 259 1
product before the change in the price? Theft by servants 497 448 2
12.13 Calculate Fisher’s ideal index from the data
given below: Calculate the index of crime for 2001, using
2000 as the base period.
Base Year, 2000 Current Year, 2001
Commodity 12.17 Using Paasche’s formula compute the quantity
Price Value Price Value index for the year 1993 with 1985 as base
A 10 30 12 48 year.
B 15 60 15 75 Quantity (in Units) Value (in Rs.)
C 15 50 18 96 Commodity
D 12 10 13 25 1985 1993 1985 1992

[HP Univ., MCom, 1995] A 100 150 500 900

B 080 100 320 500
12.14 Using the data given below, calculate the price
index number for the year 1998 by (i) C 060 072 150 360
Laspeyre’s formula, (ii) Paasche’s formula, and D 030 033 360 297
(iii) Fisher’s formula considering 1989 as the
base year. 12.18 Calculate a weighted average of relative
quantity index using 1995 as base period.
Price (Rs./Unit) Quantity (in 1000 kg)
Commodity Quantity (in 1000 kg) Price (Rs./kg)
1989 1998 1989 1998 Commodity
1995 1999 1995
Rice 9.3 4.5 100 90
Wheat 6.4 3.7 111 10 Wheat 29 24 3.80
Pulses 5.1 2.7 115 13 Corn 3 2.5 2.91
Soyabeans 12 14 6.50
12.15 It is stated that the Marshall-Edgeworth’s index
is a good approximation of the ideal index
number. Verify using the following data:
 INDEX NUMBERS 449

Hints and Answers

12.9 Divide the values by price and obtain quantity 12.12 Let the cost of the article before the increase be
figures and then calculate Fisher’s ideal price x. After increase it will be 150x ÷ 100 = 1.5x.
index. The rise 1.5x – x = 0.5x is equivalent to an
increase of 5 per cent in the cost of living. The
Commodity p0 q0 p1 q1 p1 q 0 p0 q 0 q 1 p1 q 0 p1 increases in the cost of living was 1.05y – y =
A 35 10 36 12 360 350 372 360 0.05y.
B 37 12 10 38 120 384 380 356 Hence 0.5x = 0.5y or x = 0.5 y/0.5 = 0.1y
= 10 per cent of y. Thus the expenditure on
C 10 38 12 38 396 380 396 380 that item was 10 per cent of the cost of living.
D 34 35 35 36 325 320 330 324 12.13
E 38 37 38 38 356 356 364 364
357 290 342 284 Commodity p0 q0 p1 q1 p1 q 0 p0 q 0 p1 q 1 p0 q 1
A 10 53 12 54 536 530 548 540
Fisher’s ideal price index: B 15 54 15 55 560 560 575 575
C 55 10 58 12 550 550 596 560
Σ p1 q0 Σ p1 q1 D 52 55 53 58 510 510 515 510
\ × × 100
Σ p0 q0 Σ p0 q1 191 150 234 185

357 342
= × × 100 = 121.96 Σ p1 q0 Σ p1 q1
290 284 P01 = × × 100
Σ p0 q0 Σ p0 q1
12.10 Let the income of the consumer be Rs. 100.
He spent Rs. 60 on food and Rs. 10 on house 191 234
= × × 100 = 126.9
rent in 1990. The index of food is 200 and the 150 185
house rent Rs. 150 for the particular year for 12.14
which the data are given. In order to maintain
the same consumption standards regarding two Comm- Price (Rs.) Quantity
items, he will have to spend Rs. 120 on food odity (in 1000 kg)
and Rs. 15 on house rent. Further the weights 1989 1998 1989 1998
of other items are constant; in order to maintain p0 p1 q0 q1 p0 q0 p1 q0 p0 q1 p1q1
the same standard he will have to spend Rice 9.3 4.5 100 90 4930.0 450.0 837.0 405.0
120+8+7+15+5 = Rs. 155. Hence the Wheat 6.4 3.7 111 10 0470.4 440.7 464.0 437.0
dearness allowance should be 55 per cent. Pulses 5.1 2.7 115 13 4425.5 413.5 415.3 48.1
12.11 1025.9 504.2 916.3 450.1
Item p0 q0 p1 q1 p1 q0 p0 q0 p1 q1 p0 q1
Σ p1 q0 504.2
A 1 10 2 5 20 10 10 5 Laspeyre’s index = × 100 = × 100
Σ p0 q0 1025.9
B 1 35 x 2 5x 35 2x 2
= 49.15
20 + 5x 15 10 + 2x 7
Σ p1 q1
Σ p1q1 20 + 5 x Paasche’s index = × 100
Laspeyre’s index = = ; Σ p0 q1
Σ p0 q0 15
450.1
Σ p1 p0 10 + 2 x = × 100 = 49.12
Paasche’s index = = 916.3
Σ p0 q1 7
Σ p1 q0 Σ p1 q1
(20 + 5x) /15 28 Fisher’s index = × × 100
Given = Σ p0 q0 Σ p0 q1
(10 + 2x) / 7 27
20 + 5 x 7 28 = 49.15 × 49.12 = 49.134
or × = or x = 4
15 10 + 2x 27
450 BUSINESS S T A T I S T I C S

12.15 12.17
Comm- 1996 2000 Commodity Quantity Price
odity p0 q0 p1 q1 p0q0 p0q1 p1q0 p1q1 1985 1993 1993
A 2 174 3 182 1148 1165 222 1246 q0 q1 p1 q1p1 q0 p1
B 5 125 4 140 1625 1700 500 1560 A 100 150 900/150 = 6 900 600
C 7 140 6 133 1280 1231 240 1198 B 80 100 500/100 = 5 500 400
1053 1095 962 1004 C 60 72 360/72 = 5 360 300
D 30 33 297/33 = 9 297 270
Marshall-Edgeworth index 2057 1570
Σ p1q0 + Σ p1q1
= × 100
Σ p0 q0 + Σ p0 q1 Σ q1 p1
Paasche’s quantity index = × 100
Σ q0 p1
962 + 1004
= × 100 = 91.53
1053 + 1095 2057
= × 100
Fisher’s Ideal index 1570
= 131.02
Σ p1q0 Σ p1q1 12.18
= × × 100 = 91.523
Σ p0 q0 Σ p0 q1
Commodity Quantity Price Percent Base Weighted
12.16 1995 1999 1995 Relatives Value Relatives
q1
Type of 2000 2001 Weight, Crime Relative, RW q0 q1 p0 × 100 q0 p0 (6)=
Crime W R
q0 (4)×(5)
(1) (2) (3) (4) (5)
Robberies 13 8 6 (8/13)×100 369.24
Wheat 29 24 3.80 83 110.20 9,146.60
= 61.54
Corn 13 2.5 2.91 83 8.73 724.59
Car 15 22 5 (22/15)×100 733.50
thefts = 146.70 Soyabeans 12 14 6.50 117 78.00 9,126.00
Cycle 249 185 4 (185/249)×100 297.16 196.93 18,997.19
thefts = 74.290
Pocket 328 259 1 (259/328)×100 78.96 Weighted average of relative quantity index
picking = 78.960
Thefts 497 448 2 (448/497)×100 180.28
Σ {( q1 / q0 ) × 100}( q0 p0 )
=
by servants = 90.15 Σ q0 p0
18 1659.14 18, 997.19
= = 96.
196.93
Σ RW 1659.14
Crime index = = = 92.17
ΣW 18

12.10 TESTS OF ADEQUACY OF INDEXES

So for we have discussed several methods to construct unweighted and weighted index numbers.
However, the problem still remains of selecting an appropriate method for the construction of an
index number in a given situation. The following tests have been suggested to select the adequacy of
an index number:
• Time reversal test
• Factor reversal test
• Circular test
 INDEX NUMBERS 451

12.10.1 Time Reversal Test

The time reversal test is used to test whether a given method will work both backwards and forwards
with respect to time. The test is that the formula for calculating the index number should be such that
it will give the same ratio between one point of comparison and another no matter which of the two
is taken as base. In other words, a price or quantity index for a given period with respect to the
preceding period is equal to the reciprocal of the price or quantity index when periods are
interchanged. For example, if P01 is a price index in the current year ‘1’ with base of preceding year
‘0’ and P10 is a price index for the base year ‘0’ based on the current year ‘1’, then the following
relation should be satisfied:
1
P01 = or P01 × P10 = 1 and Q01 × Q10 = 1
P01
This test is not satisfied by the Laspeyre’s Index and the Paasche’s Index. The methods which
satisfy the time reversal test are:
• Fisher’s ideal index method
• Simple geometric mean of price relatives
• Aggregates with fixed weights (Kelly’s formula)
• Marshall-Edgeworth’s method
• Weighted geometric mean of price relatives when fixed weights are used
• Walsch’s formula
For example, let us see how Fisher’s ideal index formula satisfies the time reversal test.
Σ p1q0 Σ p1q1
P01 = ×
Σ p0 q0 Σ p0 q1
For calculating P10 the time is interchanged so that p0 becomes p1 and p1 becomes p0. Similarly q0
becomes q1 and q1 becomes q0, we get
Σ p0 q1 Σ p0 q0
P10 = ×
Σ p1q1 Σ p1q0

Σ p1q0 Σ p1q1 Σ p0 q1 Σ p0 q0
P01 × P10 = × × × =1
Σ p0 q0 Σ p0 q1 Σ p1 q1 Σ p1q0
Since P01 × P10 = 1, Fisher’s ideal index satisfies the test.

12.10.2 Factor Reversal Test

According to Fisher, ‘Just as each formula should permit the interchange of two items without giving
inconsistent result so it ought to permit interchanging the prices and quantities without giving inconsistent
result, i.e. the two results multiplied together should give the true value ratio.’ In other words, the change
in the price when multiplied by the change in quantity should represent the total change in value.
Thus, if the price of a commodity has doubled during a certain period and in this period the
quantity has trebled, then the total change in the value should be six times the former level. That
is, if p1 and p0 represent the prices and q1 and q0 the quantities in the current and the base periods
respectively, then the price index for period ‘1’ with base year ‘0’ and the quantity index for
period ‘1’ with base year ‘0’ is given by
452 BUSINESS S T A T I S T I C S

Σ p1 q1
P01 × Q01 =
Σ p0 q0
The factor reversal test is satisfied only by the Fisher’s ideal price index as shown below:
Σ p1q0 Σ p1q1
P01 = ×
Σ p0 q0 Σ p0 q1
Changing p to q and q to p, we get the quantity index:
Σ q1 p0 Σ q1 p1
Q01 = ×
Σ q0 p0 Σ q0 p1
Multiplying P10 and Q01, we get
Σ p1q0 Σ p1q1 Σ q1 p0 Σ q1 p1 Σ p1q1 Σ q1 p1 Σ p1 q1
× × × = × =
Σ p0 q0 Σ p0 q1 Σ q0 p0 Σ q0 p1 Σ p0 q0 Σ q0 p0 Σ p0 q0
This result implies that the Fisher’s index formula could be used for constructing both price and
quantity indexes.

12.10.3 Circular Test

This test is concerned with the measurement of price change over a period of years when the shifting
of base is desirable in a circular fashion. It may therefore be considered as an extension of time
reversal test. For example, if an index is constructed for the year 2000 with the base of 1999 and
another index for 1999 with the base of 1998, then it should be possible for us to directly get an index
for the year 2000 with the base of 1998. If the index calculated directly does not give an inconsistent
value, the circular test is said to be satisfied. If Pab is price index for period ‘b’ with base period ‘a’; Pbc
is the price index for period ‘c’ with base period ‘b’ and Pca is the price index for period ‘a’ with base
period ‘c’, then an index is said to satisfy the circular test provided
Pab × Pbc × Pca = 1
This test is not satisfied by most of the common formulae used in the construction of indexes.
Even Fisher’s ideal formula does not satisfy this test. This test is satisfied by simple aggregative index,
simple geometric mean of price relatives and weighted aggregate (with weight) index.
Example 12.19: Calculate Fisher’s ideal index from the following data and prove that it satisfy both
Time Reversal and Factor Reversal test.
Base Year Current Year
Items Price Quantity Price Quantity
A 6 50 10 60
B 2 100 2 120
C 4 60 6 60

[Delhi Univ., BCom (Pass), 2005]

Solution: The following table presents the necessary information for constructing the Fisher’s ideal
index number:
 INDEX NUMBERS 453

Base Year Current Year

Items p0 q0 p1 q1 p0 q 0 p0 q1 p1 q0 p1 q 1

A 6 50 10 60 300 360 500 600

B 2 100 2 120 200 240 200 240
C 4 60 6 60 240 240 360 360
740 840 1060 1200

Σ p1 q0 × Σ p1 q1 1060 × 1200
Fisher’s index, P01 = × 100 = × 100 = 143
Σ p0 q0 × Σ p0 q1 740 × 840
Time Reversal Test, i.e. P01 × P10 = 1

Σ p1 q0 × Σ p1 q1 1060 × 1200
P01 = =
Σ p0 q0 × Σ p0 q1 740 × 840

Σ p0 q1 Σ p0 q0 840 × 740
and P10 = × =
Σ p1q1 Σ p1q0 1200 × 1060

1060 1200 840 740

P01 × P10 = × × × =1
740 840 1200 1060
Hence, the time reversal test is satisfied.
Σ p1 q1
Factor Reversal Test, i.e. P01 × Q01 =
Σ p0 q0

Σ p1q0 Σ p1q1 Σ q1 p0 Σ q1 p1
P01 × Q01 = × × ×
Σ p0 q0 Σ p0 q1 Σ q0 p0 Σ q0 p1

1060 1200 840 1200 (1200)2 1200 Σp1 q1

= × × × = 2
= =
740 840 740 1060 (740) 740 Σp0 q0
Hence, factor reversal test is satisfied.
Example 12.20: Construct Fisher’s price index using following data and show how it satisfies the
time and factor reversal tests.

Commodity 2002 2003

Quantity Price Quantity Price
A 20 12 30 14
B 13 14 15 20
C 12 10 20 15
D 18 16 10 14
E 15 18 15 16

Solution: Table 12.17 presents all the necessary information for constructing the Fisher’s ideal index
number.
454 BUSINESS S T A T I S T I C S

Table 12.17 Calculation of Fisher’s Ideal Index

2002 2003
Commodity q0 p0 q1 p1 p1 q 0 p0 q 0 q 1 p1 p0 q 1
A 20 12 30 14 280 240 420 360
B 13 14 15 20 260 182 300 210
C 12 10 20 15 180 120 300 200
D 8 6 10 4 32 48 40 60
E 5 8 5 6 30 40 30 40
782 630 1090 870

Fisher’s ideal price index

Σ p1 q 0 Σ p1q1 782 1090
P01 = × = × = 1.2471
Σ p 0 q 0 Σ p 0 q1 630 870
Time Reversal Test: This test is satisfied when P01 × P10 = 1

Σ p0 q1 Σ p0 q0 870 630
P10 = × = × = 0.8019
Σ p1q1 Σ p1q0 1090 782

782 1090 870 630

P01 × P10 = × × × = 1.2471 × 0.8019 = 1
630 870 1090 782
Hence, time reversal test is satisfied.
Σ p1 q1
Factor Reversal Test: This test is satisfied when P01 × Q01 =
Σ p0 q0
Σ p1q0 Σ p1q1 Σ q1 p0 Σ q1 p1
P01 = × and Q01 = ×
Σ p0 q0 Σ p0 q1 Σ q0 p0 Σ q0 p1

782 1090 870 1090 1090 Σ p1 q1

Thus, P01 × Q01 = × × × = which is the value of
630 870 630 782 630 Σ p0 q0
Hence, factor reversal test is also satisfied.
Example 12.21: Calculate Fisher’s ideal index from the data given below and show that it satisfies the
time reversal test.

Commodity 2000 2001

Price Quantity Price Quantity
A 10 49 12 50
B 12 25 15 20
C 18 10 20 12
D 20 5 40 2

Solution: Table 12.18 presents information necessary for Fisher’s method to calculate the index.

Σ p1q0 Σ p1q1 1363 1220

Fisher’s index P01 = × × 100 = × × 100 = 124.9
Σ p0 q0 Σ p0 q1 1070 996
 INDEX NUMBERS 455

Table 12.18 Calculation of Fisher’s Ideal Index

2000 2001
Commodity p0 q0 p1 q1 p1 q 0 p0 q 0 p1 q 1 p0 q 1
A 10 49 12 50 588 490 600 500
B 12 25 15 20 375 300 300 240
C 18 10 20 12 200 180 240 216
D 20 5 40 2 200 100 80 40
1363 1070 1220 996

Time Reversal Test: This test is satisfied when P01 × P10 = 1

Σ p0 q1 Σ p0 q0
P10 = ×
Σ p1q1 Σ p1q0

Σ p1q0 Σ p1q1 Σ p0 q1 Σ p0 q0
Thus P01 × P10 = × × ×
Σ p0 q0 Σ p0 q1 Σ p1 q1 Σ p1q0

1363 1220 996 1070

= × × × = 1 =1
1070 996 1220 1363
Hence, the time reversal test is satisfied.

Example 12.22: With the help of the following data, show that the index number calculated on the
basis of arithmetic mean is not reversible while the index number calculated on the basis of geometric
mean is reversible. Make comparison between arithmetic mean and geometric mean:

Commodity Price Price

in 1998 in 1999

A 40 60
B 50 80
C 20 40
D 20 10
[Delhi Univ., B Com (Hons), 2003]
Solution: Calculations for price relatives using 1998 and 1999 as base year are shown in the table
below:

Price in Price in Price Relative Price Relative

Commodity 1998 1999 (p1/p0) × 100 (p0/p1) × 100
(p0) (p1) Using 1998 as Base Using 1999 as Base

A 40 60 150 66.67
B 50 80 160 62.50
C 20 40 200 50.00
D 20 10 50 200.00
560 379.17

560
A.M. of price relatives [with 1998 as Base] = =140 (= P01 )
4
456 BUSINESS S T A T I S T I C S

379.17
A.M. of price relatives [with 1999 as base] = = 94.79 (= P10 )
4
Time Reversal Test is said to be satisfied if P01 × P10 = 1 [Ignoring the factor 100] In the given
situation the A.M. of price relatives: P01×P10 = 1.4 × 0.9479 ≠ 1
G.M. of price relatives (ignoring the factor 100) with 1998 as base is
= (1.5 × 1.6 × 2 × 0.5)1/4 = 1.244666 (= P01)
G.M. of price relatives (ignoring the factor 100) with 1999 as base is
= (0.6667 × 0.6250 × 0.5 × 2)1/4 = 0.8034384 (= P10)
In case of G.M. the relation: P01×P10 = 1. Further it may be noted that the index number using
A.M. is higher than the index number using G.M.

Example 12.23: If the ratio between Laspeyers and Paasche’s index number is 28 : 27. Find missing
figure in the following table.
Base Year Current Year
Commodities Price Quantity Price Quantity

X 1 10 2 5
Y 1 5 – 2

Solution: Calculations required to find the missing figure are shown below:

Base Year Current Year

Items p0 q0 p1 q1 p0 q0 p0 q1 p1 q0 p1 q1

X 1 10 2 5 10 5 20 10
Y 1 5 x 2 5 2 5x 2x

Σ p1 q0 20 + 5x
Laspeyers index, L = × 100 = × 100
Σ p0 q0 15
Σ p1q1 10 + 2x
Paasche’s index, P = × 100 = × 100
Σ p0 q1 7
L 28
Since = , therefore
P 27
20 + 5x
× 100
15 28
=
10 + 2x 27
× 100
7
20 + 5x 7 28
× =
15 10 + 2x 27
4+x 7 28
× =
3 2(5 + x) 27
4+x 8
=
5+ x 9
9(4 + x) = 8(5 + x)
36 + 9x = 40 + 8x or x = 40 – 36 = 4
Hence, the missing figure is: 4.
 INDEX NUMBERS 457

Example 12.24: Calculate Fisher’s price index for the following data and prove that it satisfies both
Time Reversal and Factor Reversal test:

Quantity (kg) Price (Rs.)

Items 2001 2002 2001 2002

Wheat 8 10 20 30
Sugar 6 9 14 18
Tea 2 5 15 20

[Delhi Univ., BCom (Pass), 2004]

Solution: Calculations required to calculate Fisher’s price index are shown below:

Quantity (kg) Price (Rs. per kg)

Commodity 2001 2002 2001 2002 p1 q0 p0 q0 p1 q1 p0 q1
(q0) (q1) (p0)

Wheat 8 10 20 30 240 160 300 200

Sugar 6 9 14 18 108 84 162 126
Tea 2 5 15 20 40 30 100 75
388 274 562 401

Σp1q0 Σp1q1 388 562

P01 = × × 100 = × × 100 = 1.985 × 100 =140.88 (App.)
Σp0 q0 Σp0 q1 274 401

Time Reversal Test: P01 × P10 = 1

Σp1q0 Σp1q1 Σp0 q1 Σp0 q0 388 562 401 274

P01 × P10 = × × × = × × × = 1 =1
Σp0 q0 Σp0 q1 Σp1 q1 Σp1q0 274 401 562 388
Hence, Fisher’s index satisfies time reversal test.

Σp1q1
Factor Reversal Test: P01 × Q01 = = V01
Σp0 q0

Σp1 q0 Σp1 q1 Σq1 p0 Σq1 p1 388 562 401 562

P01 × Q01 = × × × = × × ×
Σp0 q0 Σp0 q1 Σq0 p0 Σq0 p1 274 401 274 388

562 562 562 Σp1q1

= × = = = V01
274 274 274 Σp0 q0

12.11 CHAIN INDEXES

The various formulae discussed so far assumed that the base period is any fixed period. The base period
is the immediately preceding year of the current year. Moreover, the index of a given period on a given
fixed base was not affected by changes in the relevant values of any other year. But in the chain base
458 BUSINESS S T A T I S T I C S

method, the data of each period is related with that of the immediately preceding period and not with
any fixed period. This means that for the index of 2000 the base would be 1999 and for the index of 1999
the base would be 1998 and similarly of the index of 1998 the base would be 1997. Such index numbers
are very useful in comparing current period data with the preceding period’s data. Fixed base index in
such a case does not give an appropriate comparison, because all prices are based on the fixed base
period which may be far away for the current period and the preceding period.
For constructing an index by the chain base method, a series of indexes are computed for each
period with preceding period as the base. These indexes are known as link index or link relatives. The
steps of calculating link relatives are summarized below:
(i) Express the data of a particular period as a percentage of the preceding period’s data. This
is called the link relative.
(ii) These link relatives can be chained together. This is done by multiplying the link relative of
the current year by the chain index of the previous year and dividing the product by 100.
Thus
Chain index Link relative of current year × Chain index of previous year
for current year = 100
The chain index is useful for long-term comparison whereas link relatives are used for a comparison
with the immediately preceding period. The fixed base indexes compiled from the original data and
the chain indexes compiled from link relatives give the same value of index provided there is only
one commodity whose indexes are being constructed.
Remarks Chain relatives differ from fixed base relatives in computation. Chain relatives are computed
from link relatives whereas fixed base relatives are computed directly from original data.

Price relative for the current period

Link relative = ×100
Price relative for the previous period

Current period's link relative × Previous period's price relatives

Price relative =
100
Multiplying the link relatives P01, P12, P23, . . ., P(n – 1) n successively is known as the chaining
process that gives link relatives with a common base:
P01 = First link
P02 = P01 × P12
P03 = (P01 × P12) × P23 = P02 × P23
.
.
.
P0n = P0(n – 1) × P(n – 1)n

Advantages: The following are a few advantages of chain base index

(i) The chain base indexes enable us to make comparisons with the previous and not any distant
past period. Thus these index are very useful in the analysis of business data.
(ii) The chain base method permits us to introduce new commodities and delete the existing
ones which are obsolete without any recalculation of the entire series.
(iii) The index numbers calculated by the chain base method are relatively free from cyclical and
seasonal variations.
 INDEX NUMBERS 459

Disadvantage: The main disadvantage of the chain base index is that it is not useful for long term
comparisons of chained percentages in a time series. The process of chaining link relatives is
computationally difficult.

Conversion of chain base index (CBI) to fixed base index (FBI)

Current period CBI × Previous period FBI
Current period FBI =
100
Example 12.25: Construct an index by the chain base method based on the following data of the
wholesale prices of a certain commodity.

Year Price Year Price

1994 37 2000 48
1995 39 2001 49
1996 43 2002 54
1997 48 2003 56
1998 48 2004 87
1999 52

Solution: Computation of the chain base index number is shown in Table 12.19.

Table 12.19 Chain Base Indexes

Year Price Link Relative Chain Base Index Numbers

(Base Year 1985 = 100)
1994 37 100.00 100
1995 39 (39/37)×100 = 105.41 (105.41/100)×100 = 105.41
1996 43 (43/39)×100 = 110.26 (110.76/100)×105.41 = 116.23
1997 48 (48/43)×100 = 111.63 (111.63/100)×116.23 = 129.75
1998 48 (48/48)×100 = 100.00 (100/100)×129.75 = 129.75
1999 52 (52/48)×100 = 108.33 (108.33/100)×129.75 = 140.56
2000 48 (48/48)×100 = 100.00 (100/100)×140.56 = 140.56
2001 49 (49/48)×100 = 102.08 (102.08/100)×140.56 = 143.48
2002 54 (54/49)×100 = 110.20 (110.20/100)×158.11 = 158.11
2003 56 (56/54)×100 = 103.70 (103.70/100)×158.11 = 163.96
2004 57 (57/56)×100 = 101.79 (110.79/100)×163.96 = 166.90

Example 12.26: Prepare fixed base index numbers from the chain base index numbers given below:

Year : 1996 1997 1998 1999 2000 2001

Chain Index : 94 104 104 93 103 102

Solution: Computation of fixed base indexes is shown in Table 12.20 using the following formula:
Current period CBI × Previous period FBI
Fixed base index (FBI) =
100
460 BUSINESS S T A T I S T I C S

Table 12.20 Fixed Base Index Numbers

Year Chain Index Fixed Base Index

1996 94 94
104 × 94
1997 104 = 97.76
100
104 × 97.76
1998 104 = 101.67
100
93 × 101.67
1999 93 = 94.55
100
103 × 94.55
2000 103 = 97.39
100
102 × 97.39
2001 102 = 99.34
100

Example 12.27: Calculate the chain base index number and fixed base index number from the
following data:

Commodity 1998 1999 2000 2001 2002

A 14 16 18 10 12
B 16 20 24 30 36
C 18 10 16 20 24

Solution: Computation of the chain index number and fixed base index number is shown in Tables
12.21 and 12.22.

Table 12.21 Chain Base Index Number

Commodity Link Relatives Based on Preceding Year

1998 1999 2000 2001 2002
6 8 10 12
A 100 × 100 =150 × 100 =133.33 ×100 =125 × 100 =120
4 6 8 10
20 24 30 36
B 100 × 100 =125 × 100 =120 × 100 =125 × 100 =120
16 20 24 30
10 16 20 24
C 100 × 100 =125 × 100 =160 × 100 =125 × 100 =120
8 10 16 20
Total link 300 400 413.33 375 360
relatives
Average 100 133.33 137.78 125 120
link relatives
133.33 × 100 137.78 × 133.33 125 × 183.70 120 × 229.63
Chain 100
100 100 100 100
base index
= 133.3300. = 183.70 = 229.63 = 275.5600
 INDEX NUMBERS 461

Table 12.22 Fixed Base Index Number

Commodity Price Relatives (Base 1998 = 100)

1998 1999 2000 2001 2002
6 8 10 12
A 100 × 100 =150 × 100 =200 × 100 =250 × 100 =300
4 4 4 4
20 24 30 36
B 100 × 100 =125 × 100 =150 × 100 =187.5 × 100 =225
16 16 16 16
10 16 20 24
C 100 × 100 =125 × 100 =200 × 100 =250 × 100 =300
8 8 8 8

Total 300 400.33 550.33 687.51 825

Average 100 133.33 183.33 229.17 275
(Fixed base
index number)

Example 12.28: Shift the base from 2004 to 2006 in the data given below:
Year Index (2004 = 100)

2001 87.27
2002 90.91
2003 95.40
2004 100.00
2005 104.00
2006 106.00
2007 112.00

Solution: Calculations required to shift the base from 2004 to 2006 are shown below:

Year Index (2004 = 100) Index (2006 = 100)

87.27
2001 87.27 × 100 = 82.33
106
90.91
2002 90.91 × 100 = 85.76
106
95.4
2003 95.40 × 100 = 90
106
100
2004 100.00 × 100 = 94.34
106
104
2005 104.00 × 100 = 98.11
106
2006 106.00 100
112
2007 112.00 × 100 = 105.66
106
462 BUSINESS S T A T I S T I C S

Example 12.29: Given below are two sets of indices. For the purpose of continuity of records, you
are required to construct a combined series with the year 1993 as the base:

Year First Set Second set

Price Relative Link Relatives

1990 100 –
1991 120 –
1992 125 –
1993 150 –
1994 – 110
1995 – 120
1996 – 95
1997 – 105

Link relative × Previous year price relative

Solution: Price relative =
100
110 × 500
For 1994: Price Relative = =165
100
120 × 165
For 1995: Price Relative = =198 and so on.
100
Year Price Relative With Price Relative With
1990 = 100 1993 as Base

100
1990 100 100 × = 66.66
150
100
1991 120 120 × = 80
150
100
1992 125 125 × = 83.33
150
100
1993 150 150 × = 100
150
150 × 110 100
1994 =165 165 × = 110
100 150
165 × 120 100
1995 =198 198 × = 132
100 150
198 × 95 100
1996 =188.1 188.1 × = 125.4
100 150
105 100
1997 188.1 × =197.5 197.5 × = 131.66
100 150
Example 12.30: Prepare a spliced series of index numbers with 2003 as base from the following
series:
Year 1998 1999 2000 2001 2002 2003 2004
Index A 100 120 135
Index B 100 115 125 145
Index C 100 110

[Delhi Univ., BCom (Hons), 2005]

 INDEX NUMBERS 463

Solution: The spliced series of index numbers with 2003 as base is shown below:

Splicing Splicing All Index

Year Index Index Index A to B Index C Number with
A B with Base 2000 Base 2003

100 100
1998 100 – × 100 = 74.07 × 74.07 = 51.082
135 145
100 100
1999 120 – × 120 = 88.89 × 88.89 = 61.3
135 145
100
2000 135 100 – 100 × 100 = 68.97
145
100
2001 – 115 – 115 × 115 = 79.3
145
100
2002 – 125 – 125 × 125 = 86.2
145
2003 – 145 – 145 100 100
2004 – – – 110 110

Example 12.31: A price index number series was started in 1997 as base. By 2001, it rose by 15%.
The link relative for year 2002 was 95. In this year, a new series was started. This new series rose by
25% in next year. During year 2004, the price level was 5% higher than 2003. However in 2005, they
were 6% higher than 2004. Splice the two series and calculate the index number for various years by
shifting base to 2003.
Solution: Calculations required for splicing of indices are shown below:

Year Index A Index B Splicing A to B

(1997 = 100) (2002 = 100) (2002 = 100)

100
1997 100 × 100 = 91.53
109.25
 15  100
2001 100 +  × 100  = 115 × 115 = 105.26
 100  109.25
95 × 115
2002 = 109.25* 100 100
100
25
2003 100 + × 100 = 125 125
100
5
2004 125 + × 125 = 131.25 131.25
100
6
2005 131.25 + × 131.25 = 139.125 139.125
100

Link relative for 2002 × Previous year price

* Price for year 2002 =
100
464 BUSINESS S T A T I S T I C S

Shifting Base to 2003

Year Index (2002 = 100) Index (2003 = 100)

91.53
1997 91.53 × 100 = 73.224
125
105.26
2001 105.26 × 100 = 84.21
125
100
2002 100 × 100 = 80
125
2003 125 100
131.25
2004 131.25 × 100 = 105
125
139.125
2005 139.125 × 100 = 111.3
125

Example 12.32: The following table gives the annual income of a person and the general price index
number for the period 1988 to 1996. Prepare index number to show the changes in the real income
of the person.

Year Annual Income Price Index

(Rs.) Number

1988 36,000 100

1989 42,000 120
1990 50,000 145
1991 55,000 160
1992 60,000 250
1993 64,000 320
1994 68,000 450
1995 72,000 530
1996 75,000 600

[Delhi Univ., BCom (H) 2000]

Solution: Calculations required to calculate real income index number are shown below:

Annual income
Year Annual Income Price Real Wage = × 100 Real Wage Index
Price index
(Rs.) Index Number 1988 = 100

36,000
1988 36,000 100 × 100 = 36,000 1000
100
42,000 35,000
1989 42,000 120 × 100 = 35,000 × 100 = 97.22
120 36,000
50,000 34,482.7
1990 50,000 145 × 100 = 34482.7 × 100 = 95.78
145 36,000
55,000 34,375
1991 55,000 160 × 100 = 34,375 × 100 = 95.49
160 36,000

Contd...
 INDEX NUMBERS 465

Annual income
Year Annual Income Price Real Wage = × 100 Real Wage Index
Price index
(Rs.) Index Number 1988 = 100

60,000 24,000
1992 60,000 250 × 100 = 24,000 × 100 = 66.67
250 36,000
64,000 20,000
1993 64,000 320 × 100 = 20,000 × 100 = 55.56
320 36,000
68,000 15,111.1
1994 68,000 450 × 100 = 15,111.1 × 100 = 41.97
450 36,000
72,000 13,584.9
1995 72,000 530 × 100 = 13,584.9 × 100 = 37.74
530 36,000
75,000 12,500
1996 75,000 600 × 100 = 12,500 × 100 = 34.72
600 36,000

S e l f-P r a c t i c e P r o b l e m s 12C

12.19 Calculate Fisher’s Ideal index from the data 12.22 The following table gives the annual income of
given below and show that it satisfies the time a clerk and the general index number of price
reversal and factor reversal tests. during 1994–98. Prepare the index number to
show the changes in the real income of the
Commodity Base Year Current Year teacher.
Quantity Price Quantity Price
Year Income Price Year Income Price
A 12 10 15 12
B 15 27 20 25 (Rs.) Index No. (Rs.) Index No.
C 24 25 20 29 1994 36,000 100 1999 64,000 290
D 25 16 25 14
1995 42,000 104 2000 68,000 300
12.20 Splice the following two index number series, 1996 50,000 115 2001 72,000 320
continuing series A forward and series B
backward. 1997 55,000 160 2002 75,000 330
1998 60,000 280 — — —
Year : 1998 1999 2000 2001 2002 2003
Series A : 100 120 150 — — — 12.23 From the following average price of groups
Series B : — — 100 110 120 150 of commodities given in rupees per unit, find
the chain base index number with 1994 as the
12.21 Calculate the chain base index number chained base year:
to 1994 from the average price of following
three commodities: Group 1994 1995 1996 1997 1998

Commodity 1999 2000 2001 2002 2003 I 2 13 14 15 16

Wheat 14 16 18 10 12 II 8 10 12 15 18
Rice 16 20 24 30 36 III 4 15 18 10 12
Sugar 18 10 16 20 24
[Agra Univ., BCom, 1998]
466 BUSINESS S T A T I S T I C S

12.24 Given the following data: C 5 7 8 10

D 15 20 12 15
Year Weekly Take-home Consumer Price
E 20 25 15 10
Pay (Wages) Index
1998 109.50 112.8
12.26 From the data given below, calculate Fisher’s
1999 112.20 118.2
ideal index and show that it satisfies the time
2000 116.40 127.4 reversal and factor reversal test:
2001 125.08 138.2
2002 135.40 143.5 Commodity 1998 1999
2003 138.10 149.8 Price Quantity Price Quantity

(a) What was the real average weekly wage A 12 20 14 30

for each year? B 14 13 20 15
(b) In which year did the employees have the C 10 12 15 20
greatest buying power? D 6 8 4 10
(c) What percentage increase in the weekly E 8 5 6 5
wages for the year 2003 is required (if any)
to provide the same buying power that the 12.27 Calculate the index number by using Paasche’s
employees enjoyed in the year in which method, and Fisher’s method.
they had the highest real wages?
12.25 Using the following data construct Fisher’s ideal Commodity p1 q1 p0 q0
index and show that it satisfies the factor A 5 14 3 8
reversal and time reversal tests: B 8 18 6 25
C 3 25 1 40
Price (in Rs./Unit) Number of Units D 15 36 12 48
Base Current Base Current E 9 14 7 18
Commodity
Year Year Year Year F 7 13 5 19
A 6 8 10 12
B 10 10 5 8

Hints and Answers

12.19
Σ p0 q1 Σ p0 q0 470 425
P10 = × = × = 0.8638
Commodity q0 p0 q1 p1 p1q0 p0q0 p1q1 p0 q1 Σ p1q1 Σ p1q0 530 505
A 12 10 15 12 144 120 180 150
Time Reversal Test
B 15 7 20 5 75 105 100 140
C 24 5 20 9 216 120 180 100 505 530 470 425
D 5 16 5 14 70 80 70 80 P01 × P10 = × × × = 1 =1
425 670 530 505
505 425 530 470
Factor Reversal Test

Σ p1q0 Σ p1q1 505 530 Σ p1q1

P01 = × = × = 1.1576 P01 × Q01 = ;
Σ p0 q0 Σ p0 q1 425 470 Σ p0 q0
 INDEX NUMBERS 467

12.21
Σ q1 p0 Σ q1 p1
Q01 = ×
Σ q0 p0 Σ q0 p1 Comm- Relatives Based on the Preceding Year
odity
1999 2000 2001 2002 2003
470 530
= × = 0.9693
525 505 Wheat 100 150 133.33 125 120
Rice 100 125 120.00 125 120
505 530 470 530 Sugar 100 125 160.00 125 120
P01 × Q01 = × × ×
425 470 425 505 Total 300 400 413.33 375 360
Average 100 133.33 137.78 125 120
of link
530 530
= × relatives
425 425 133.33×100 137.78 ×133.33 125×183.70 120 × 229.63
Chain 100 100 100 100 100
index
503 Σ p1q1
= which is equal to (1999 = 133.33 =183.70 = 229.63= 275.55
425 Σ p0 q0 = 100)
12.20
12.22
Year Series Series Series B Series A
A B Spliced to A Spliced to B Year Income Price Real Real Income
(Rs.) Index Income Index
 100 
1998 100 — —   ×100
 150  1994 360 100 (360/100) ×100 100.00
= 66.66 =360.00
1995 420 104 (420/104) × 100 112.18
 100  = 403.85
1999 120 — —   ×120
 150  1996 500 115 (500/115) × 100 120.77
= 180.00 = 434.78
1997 550 160 (550/160) ×100 95.49
 150   100  = 343.75
2000 150 100   ×100   ×150
 100   150  1998 600 280 (600/280) × 100 59.52
= 150 = 100.00 = 214.29
 150  1999 640 290 (640/290) × 100 61.30
2001 — 110   ×110 = 165 = 220.69
 100 
2000 680 300 (680/300) × 100 62.96
 150  = 226.67
2002 — 120   ×120 = 180 2001 720 320 (720/320) × 100 62.52
 100 
= 225.00
 150  2002 750 330 (750/330) × 100 63.13
2003 — 150   ×150 = 225
 100  = 227.27

12.23
Group 1994 1995 1996 1997 1998
Price Link Price Link Price Link Price Link Price Link
Relative Relative Relative Relative Relative
I 2 100 13 150 14 133.3 15 125 16 120
II 8 100 10 125 12 120.0 15 125 18 120
III 8 100 15 125 18 160.0 10 125 12 120
Total 300 400 413.3 375 360
Average of link 100 133.33 137.77 125 120
relatives
133.33 137.77 125 120
Chain index 100 × 100 × 133.33 × 183.69 × 229.61
(1994 = 100) 100 100 100 100
= 133.33 = 183.69 = 229.61 = 275.53
468 BUSINESS S T A T I S T I C S

12.24 (a) Average weekly wage can be obtained by Time Reversal Test: P01 × P10 = 1
using the following formula: 801 796 627 630
Money wage P01 × P10 = × × × = 1 =1
Real wage = × 100 630 627 796 801
Price index
Σ p1q1 796
Factor Reversal Test: P01 × Q01 = =
Year Weekly Take-home Consumer Real wages Σ p0 q0 630
Pay (Rs.) Price Index
Σ q1 p0 Σ q0 p1 627 796
Q01 = × = × P × Q01
109.5 Σ q0 p0 Σ q0 p1 630 801 01
1998 109.50 112.8 × 100
112.8
= 97.07 801 796 627 796 796
= × × × =
112.2 630 627 630 801 630
1999 112.20 118.2 × 100
118.2 Σ p1q1
which is equal to
= 94.92 Σ p0 q0
116.4
2000 116.40 127.4 × 100 12.26
127.4
= 91.37 1998 1999
125.08 Comm p0 q0 p1 q1 p1 q0 p0 q0 p1 q1 p0 q1
2001 125.08 138.2 × 100
138.2
= 90.51 A 12 20 14 30 280 240 420 360
B 14 13 20 15 260 182 300 210
135.4
2002 135.40 143.5 × 100 C 10 12 15 20 180 120 300 200
143.5 D 6 8 4 10 32 48 40 60
= 94.36 E 8 5 6 5 30 40 30 40
138.10
2003 138.10 149.8 × 100 782 630 1090 870
149.8
= 92.19 Fisher’s Ideal Index:
(b) Since real wage was maximum in the year 1998,
Σ p1 q0 Σ p1 q1
the employees had the greatest buying power in P01 = × × 100
that year. Σ p0 q0 Σ p0 q1
(c) The percentage increase in the weekly wages for
the year 2003 required to provide the same 782 1090
= × × 100 = 124.7
buying power that the employees had in 1998: 630 870
Absolute difference = 97.07 – 92.19 = 6.88. Time Reversal Test:
12.25
Σ p1 q0 Σ p1 q1 Σ p0 q1 Σ p0 q0
Comm p0 p1 q0 q1 p1 q0 p0 q0 p1 q1 p0 q1 P01 × P10 = × × ×
Σ p0 q0 Σ p0 p1 Σ p1 q1 Σ p1 q0
A 6 8 10 12 80 60 96 72
782 1090 870 630
B 10 10 5 8 50 50 80 80 = × × × =1
630 870 1040 782
C 5 7 8 10 56 40 70 60
12.27
D 15 20 12 15 240 180 300 225
E 20 25 15 10 375 300 250 200 Comm 1p1 q1 1p0 1q011p1q1 1p0q1 11p1q0 p0q0
801 630 796 627 A 15 14 13 18 1170 142 1140 1124
B 18 18 16 25 1144 108 1200 1150
Σ p1q0 Σ p1q1 C 13 25 11 40 1175 125 1120 1140
P01 = × ;
Σ p0 q0 Σ p0 q1 D 15 36 12 48 1540 432 1720 1576
E 09 14 17 18 1126 198 1162 1126
Σ p0 q1 Σ p0 q0 F 17 13 15 19 1191 165 1133 1195
P10 = ×
Σ p1 q1 Σ p1 q0 1046 770 1375 1011
 INDEX NUMBERS 469

(i) Paasche’s Index, Σ p1 q0 Σ p1 q1

P01 = × × 100
Σ p1 q1 1046 Σ p0 q0 Σ p0 q1
P01 = × 100 = × 100 = 135.84
Σ p0 q1 770
1375 1046
(ii) Fisher’s Index, = × × 100 = 135.92
1011 770

12.13 CONSUMER PRICE INDEXES

The consumer price index, also known as the cost of living index or retail price index, is constructed to measure
the amount of money which consumers of a particular class have to pay to get a basket of goods and services
at a particular point of time in comparison to what they the paid for the same in the base year.
The need for constructing consumer price indexes arises because the general indexes do
not highlight the effects of rise or fall in prices of various commodities consumed by different
classes of people on their cost of living. Moreover, different classes of people consume different
types of commodities and even the same type of commodities are not consumed in the same
proportion by different classes of people. To study the effect of rise or fall in prices of different
types of commodities, the Cost of Living Index (CLI) are constructed separately for different
classes of people.
The problem in constructing consumer price indexes arise because variations in prices of
commodities have to be studied from the point of view of consumers living in different regions or
places. Since retail prices in different places differ and the pattern of consumption is also not identical
at different places, therefore people living in different regions, pay different prices to purchase
various commodities. Moreover, the relative importance of various commodities to all people is not
identical. Therefore we cannot construct one CLI for the whole country.

12.13.1 Uses of Consumer Price Index (CPI) Number

The importance of the CPI can be seen from the following:
(i) The CPI are used to formulate economic policy, escalate income payments, and measure real
earnings.
(ii) The CPI are used to measure purchasing power of the consumer in rupees. The purchasing
power of the rupee is the value of a rupee in a given year as compared to a base year. The
formula for calculating the purchasing power of the rupee is:
1
Purchasing power = × 100
Consumer price index
(iii) When a time series is concerned with such rupee values as retail sales amounts or wage rates,
the price index is most frequently used to achieve deflation of such time-series. The process
of deflating can be expressed in the form of a formula as:
Money value
Real wage = × 100
Consumer price index
(iv) The CPI is used in wage negotiations and wage contracts. Automatic adjustment of wages or the
dearness allowance component of the wages is done on the basis of the consumer price index.

12.13.2 Construction of a Consumer Price Index

The CPI is a weighted aggregate price index with fixed weights. The need for weighting arises
because the relative importance of various commodities or items for different classes of people is not
470 BUSINESS S T A T I S T I C S

the same. The percentage of expenditure on different commodities by an average family constitutes
the individual weights assigned to the corresponding price relatives, and the percentage expenditure
on five well-accepted groups of commodities namely: (i) food, (ii) clothing, (iii) fuel and lighting, (iv)
house rent, (v) miscellaneous.
The weight applied to each commodity in the market basket is derived from a usage survey of
families throughout the country. The consumer price index or cost of living index numbers are
constructed by the following two methods:
Aggregate expenditure method or weighted aggregate method
This method is similar to the Laspeyre’s method of constructing a weighted index. To apply this
method, the quantities of various commodities consumed by a particular class of people are assigned
weights on the basis of quantities consumed in the base year. Mathematically it is stated as:
Total expenditure in current period Σ p1 q0
Consumer price index = × 100 = × 100
Total expenditure in base period Σ p0 q0
where p1 and p0 = prices in the current period and base period, respectively
q0 = quantities consumed in the base period
Family budget method or method of weighted average of price relatives
To apply this method the family budget of a large number of people, for whom the index is meant, are
carefully studied. Then the aggregate expenditure of an average family on various commodities is
estimated. These values constitute the weights. Mathematically, consumer price index is stated as:
Σ PV
Consumer price index = × 100
ΣV
when P = price relatives, p1/p0×100
V = Value weight, p0q0
Example 12.33: Owing to change in prices the consumer price index of the working class in a
certain area rose in a month by one quarter of what it was prior to 225. The index of food became 252
from 198, that of clothing from 185 to 205, of fuel and lighting form 175 to 195, and that of
miscellaneous from 138 to 212. The index of rent, however, remained unchanged at 150. It was
known that the weight of clothing, rent and fuel, and lighting were the same. Find out the exact
weight of all the groups.
Solution: Suppose the weights of items inclnded in the group are as follows:
• Food x • Fuel and Lighting z • Rent z
• Miscellaneous y • Clothing z
Therefore, the weighted index in the beginning of the month would be:

Index Weight IW
I W
Food 198 x 198x
Clothing 185 z 185z
Fuel and Lighting 175 z 175z
Rent 150 z 150z
Miscellaneous 138 y 138y
x + y + 3z 198x+138y+510z
 INDEX NUMBERS 471

198 x + 138 y + 510 z

Index number =
x + y + 3z
Similarly the weighted index at the end of the month would be:
I W IW
Food 252 x 252x
Clothing 205 z 205z
Fuel and Lighting 195 z 195z
Rent 150 z 150z
Miscellaneous 212 y 212y

x + y + 3z 252x+212y+550z

252x + 212 y + 550 z

Index number =
x + y + 3z
The weighted index at the end of the month was 225 (given). This index is a rise from the first
index by one quarter. Therefore, the index at the beginning was (4/5)th of 225 = 180.
Hence the weighted index at the beginning of the month was
198x + 138y + 510z
180 =
x + y + 3z
180 + 180y + 540z = 198x + 138y + 510z
18x – 42y – 30z = 0 (i)
Similarly the weighted index at the end of month was
252x + 212y + 550z
225 =
x + y + 3z
225x + 225y + 675z = 252x + 212y + 550z
27z – 13y – 125z = 0 (ii)
Let the total weight be equal to 100. Hence
x + y + 3z = 100 (iii)
Multiplying Eqn. (iii) by 18 and subtracting from (i), we get
– 60y – 84z = – 1800 or 60y + 84z = 1800 (iv)
Multiplying (iii) by 27, and subtracting from Eqn. (ii), we get
– 40y – 206z = – 2700 or 40y + 206z = 2700
Multiplying Eqn. (iv) by 20, and Eqn. (v) by 30 and subtracting, we get
– 4500z = – 45000 or z = 10
Substituting the value of z in Eqn. (iv), we have
60y + (84 × 10) = 180 or y = 10
Substituting the value of y and z in Eqn. (iii), we have
x + 16 + (3 × 10) = 100 or x = 54
Thus, the exact weights are:
• Food 54 • Clothing 10
• Fuel and Lighting 10 • Rent 10
• Miscellaneous 16
472 BUSINESS S T A T I S T I C S

Example 12.34: Incomplete information obtained from a partially destroyed records on cost of living
analysis is given below:
Group Group Index Percent of Total
Expenditure
Food 268 60
Clothing 280 Not available
Housing 210 20
Fuel and Electricity 240 5
Miscellaneous 260 Not available

The cost of living index with percent of total expenditure as weight was found to be 255.8.
Estimate the missing weights. [Delhi Univ., B.Com (Hons) 2005]
Solution: Let the weights for clothing be x1 and for miscellaneous be x2. Then
60 + x1 + 20 + 5 + x2 = 100
x1 + x2 = 15 ...(i)
268 × 60 + 280 x1 + 210 × 20 + 240 × 5 + 260 x2
255.8 =
100
25580 = 16080 + 280 x1 + 4200 + 1200 + 260 x2
= 21480 + 280 x1 + 260 x2
4100 = 260 (x1 + x2) + 20 x1
= 260 × 15 + 20 x1 = 3900 + 20 x1 [Since, x1 + x2 = 15]
200 = 20 x1 or x1 = 10
x 2 = 15 – 10 = 5 and, then x1 = 10, x2 = 5.
Example 12.35: Calculate the index number using (a) Aggregate expenditure method, and (b) Family
budget method for the year 2000 with 1995 as the base year from the following data:
Commodity Quantity (in Units) Price (in Rs./Unit) Price (in Rs./Unit)
1990 1990 2000
A 100 8.00 12.00
B 25 6.00 7.50
C 10 5.00 5.25
D 20 48.00 52.00
E 25 15.00 16.50
F 30 9.00 27.00
Solution: Calculations of cost of living index are shown in Tables 12.23 and 12.24.
Table 12.23 Index Number by Aggregative Expenditure Method

Commodity Price (Rs. per unit) in Quantity (in units)

1900 2000 in 1990
p0 p1 q0 p0 q 0 p1 q 0
A 8.00 12.00 100 800.00 1200.00
B 6.00 7.50 25 150.00 187.50
C 5.00 5.25 10 50.00 52.50
D 48.00 52.00 20 960.00 1040.00
E 15.00 16.50 25 375.00 412.50
F 9.00 27.00 30 270.00 810.00
2605.00 3702.50
 INDEX NUMBERS 473

Table 12.24 Index Number by Family Budget Method

Price (Rs. per unit) in Quantity (in units) Price Relatives Weights
Commodity 1990 2000 in 1990 P (P1/p0)×100 W = p0 q0 PW
p0 p1 q0
A 8.00 12.00 100 150.00 800 1,20,000
B 6.00 7.50 25 125.00 150 18,750
C 5.00 5.25 10 105.00 50 5,250
D 48.00 52.00 20 108.33 960 1,03,996.80
E 15.00 16.50 25 110.00 375 41,250
F 9.00 27.00 30 300.00 270 81.00
2605 3,70,246.8

Σ p1 q0 3702.5
Cost of living index = × 100 = × 100 = 142.13
Σ p0 q0 2605

Σ PW 3,70,246.8
Cost of living index = = = 142.123
ΣW 2605
The small difference observed between the index by the Aggregative Method (142.13) and the
index by the Family Budget Method (142.123) is due to the approximation in the value of price
relatives (= 108.33) in commodity D.

Example 12.36: The monthly income of a person is Rs. 10,500. It is given that cost of living index for
a particular month is 136. Find out the money spent by that person on food and on clothing.
Item Expenditure (Rs.) Index
Food — 180
Rent 1470 100
Clothing — 150
Fuel and Power 1680 110
Misc 1890 80

[Delhi Univ., B.Com(Hons), 2001]

Solution: Let amount spent on food be x and on clothing be y

Item Expenditure Index

E I I×E

Food x 180 180X

Rent 1470 100 147000
Clothing y 150 150Y
Fuel and Power 1680 110 184800
Miscellaneous 1890 80 151200
10500 180x + 150y + 483000

ΣI × E
Consumer Price Index (CPI) =
ΣE
474 BUSINESS S T A T I S T I C S

180 x + 150 y + 483000

136 =
10500
136 (10,500) = 180x + 150y + 4,83,000
180x + 150y = 14,28,000 – 4,83,000 = 9,45,000 (i)
Given 10,500 (Expenditure) = x + 1,470 + y + 1,680 + 1,890
or x + y = 10,500 – 5,040 = 5,460 (ii)
Multiplying (ii) by 180 and subtracting from (i), we get
180x + 150y = 9,45,000
180x + 180y = 9,82,800
–30y = – 37,800 or y = 1,260
Putting y = 1,260 in (ii), we get x = 4,200.

Example 12.37: The consumer price index in a particular town and the weights according to differ-
ent groups of items were as follows:
Food 55, Fuel 15, Clothing 10, Rent 12 and Miscellaneous 8. In October 1999, the dearness
allowance was fixed by a mill of that town at 182 per cent of worker’s wages which fully compensated
for the rise in the prices of food and rent but did not compensate for anything else. Another mill of
the same town paid dearness allowance of 46.5 per cent which compensated for the rise in fuel and
miscellaneous groups. It is known that the rise in food is double the rise in fuel and the rise in
miscellaneous group is double the rise in rent. Find the rise in food, fuel, rent and miscellaneous
groups. [Delhi Univ., B.Com (Hons), 2002]

Solution: Let rise in fuel be x and rise in rent be y. Then, rise in food will be 2x and rise in miscella-
neous group will be 2y.

First mill compensated fully for rise in food and rent but did not compensate for anything else.
Dearness allowance was fixed at 182%, i.e. Rs. 282 paid against Rs. 100.
Index after rise for first mill is 282.

Index (I) Weight (W) W×I

Food 2X 55 110x
Fuel 100 15 1500
Clothing 100 10 1000
Rent Y 12 12y
Miscellaneous 100 8 800
100 3300 + 110x + 12y

ΣW × I
Index =
ΣW

3300 + 110 x + 12 y
282 =
100
28200 – 3300 = 110x + 12y
110x + 12y = 24900
Second mill paid dearness allowance at the rate of 46.5%.
 INDEX NUMBERS 475

Index (I) Weight (W) WI

Food 100 55 5500

Fuel x 15 15x
Clothing 100 10 1000
Rent y 12 1200
Miscellaneous 2y 8 16y
Total 100 7700 + 15x + 16y

ΣW × I
Index =
ΣW

7700 + 15x + 16 y
146.5 =
100
14650 – 7700 = 15x + 16y
15x + 16y = 6950 (ii)
By multiplying (i) by 4 and (ii) by 3 and subtracting (ii) from (i), we get
440x + 48y = 99600
45x + 48y = 20850
395x = 78750, i.e. x = 199.367
Substituting x = 199.367 in (ii), we get 16y = 6950 – 2990 = 360 or y = 247.5
(a) Hence rise in fuel shall be 199.37 and rise in food shall be 398.74
(b) Rise in rent shall be 247.5 and rise in Miscellaneous groups shall be 495.

C o n c e p t u a l Q u e s t i o n s 12B
18. What is the chain base method of construction of
11. (a) Discuss the various problems faced in the
index numbers and how does it differ from the
construction of index numbers.
fixed base method?
(b) Explain the problem faced in the construction
19. It is said that index numbers are a specialized
of cost of living index.
type of averages. How far do you agree with this
12. Discuss the importance and use of weights in the statement? Explain briefly the Time Reversal and
construction of general price index numbers. Factor Reversal Tests.
13. What is Fisher’s Ideal index? Why is it called ideal? 20. What are the Factor Reversal and Circular tests of
Show that it satisfies both the time reversal test as consistency in the selection of an appropriate index
well as the factor reversal test. formula? Verify whether Fisher’s Ideal Index
14. Laspeyre’s price index generally shows an upward satisfies such tests.
trend in the price changes while Paasche’s method 21. What is the major difference between a weighted
shows a downward trend on them. Elucidate the aggregate index and a weighted average of
statement. relatives index?
15. Explain the Time Reversal Test and Factor 22. What are the tests to be satisfied by a good index
Reversal Test with the help of suitable examples. number? Examine how far they are met by
16. Distinguish between deflating and splicing of Fisher’s Ideal index number.
index numbers. 23. What are the tests prescribed for a good index
17. What is the cost of living index number? Is it the number? Describe the index number which
same as the consumer price index number? satisfies these tests.
476 BUSINESS S T A T I S T I C S

Formulae Used
pn Weighted average of price relatives
1. Price relatives in period n, P0n = × 100
p0
p 
qn Σ  1 × 100  ( p0 q0 )
Quantity relative in period n, Q0n = × 100 p
 0 
q0 P0n =
Σ p0 q0
Σ pn qn
Value relative in period n, V0n = × 100
Σ p0 q0 (base year value as weights)
2. Unweighted aggregate price index in period n Weighted average of price relatives

Σ pn  p 
P0n = × 100 Σ  1 × 100  ( p1q1 )
Σ p0 p
P0n =  0 
Simple average of price relative Σ p1q1
1 p  (current year value as weights)
P0n =
2
∑  p0n  × 100
 
4. Quantity indexes
Simple G.M. of price relative
(a) Unweighted quantity index in period n
1  p  
P0n = antilog  Σ  n  × 100  Σ qn
 n  p0   Q0n =
Σ q0
× 100
Simple aggregate quantity index
Σ qn Simple average of quantity relative
Q0n = × 100
Σ q0
1  qn 
3. Weighted aggregate price indexes Q0n = Σ  × 100 
n  q0 
(a) Weighted aggregate method in period n
Σ pn q (b) Weighted quantity index in period n
P0n = × 100
Σ p0 q
Σqn W
Σ pn q0 Q0n = × 100
Laspeyre’s index, Ip (L) = × 100 Σq0W
Σ p0 q0
5. Tests for adequacy or consistency
Σ pn qn
Paasche’s index, Ip (P) = × 100 Time reversal test: P0n × Pn0 = 1
Σ p0 qn
Marshall-Edgeworth’s index Σ pn qn
Factor reversal test: P0n × Q0n =
Σ p0 q0
Σ pn ( q0 + qn )
Ip (M-E) = × 100
Σ p0 ( q0 + qn ) Circular test: P01 × P12 × P23 × ... × P(n – 1)n
Dorbish and Bowley’s index × Pn0 = 1
1 6. Link relative
Ip (D-B) = (L + P) × 100
2 Current period price
= × 100
Fisher’s ideal index, = L × P × 100 Price of the preceding period
(b) Weighted average of price relatives in period n
p  Current period's link relative
Σ  n × 100  W × Preceding period's chain index
p Chain index =
P0n =  0  100
ΣW
 INDEX NUMBERS 477

Chapter Concepts Quiz

True or False
1. Like all statistical tools, index numbers must be 7. Index numbers measure change in magnitude of
used with great caution. a group of distinct but related variables.
2. For constructing index numbers, the best method 8. Prices should be for the same unit of quantity in
on theoretical grounds is not the best method index numbers.
from practical point of view. 9. Quantity relatives are used to measure changes
3. The Fisher Ideal Index number is a compromise in the volume of consumption.
between two well known indexes—not a right 10. Weighting of index number makes them more
compromise, economically, for the statistician. representative.
11. Cost of living index numbers are based on retail
4. The real problem while constructing a index
prices of items of consumption.
number is whether he shall leave weighting to
12. Splicing means constructing one continuous series
chance or seek to rationalize it.
from two index series on the basis of a common
5. Like relatives are based on the idea that one series base.
can be converted into another because time 13. Chain indexes give the same result as do fixed
reversibility holds. base index numbers.
6. Index numbers are the signs and guide-posts along 14. Weighted average of relatives and weighted
the business highway that indicate to the aggregative methods render the same result.
businessman how he should drive or manage. 15. Paasche’s formula is a weighted aggregate index
with quantity weights in the base year.
Concepts Quiz Answers

1. T 2. F 3. F 4. F 5. T 6. T 7. T 8. F 9. T
10. T 11. T 12. T 13. T 14. T 15. F

R e v i e w S e l f-P r a c t i c e P r o b l e m s
12.28 Construct an index number for each year from
Comm Unit Weight Price (in Rs. per unit)
the following average annual price of cotton
with 1989 as the base year. (Rs. 1000) 1996 1997
A Kg 5 2.00 4.50
Year Price (Rs.) Year Price (Rs.) B Quintal 7 2.50 3.20
1989 75 1994 70 C Dozen 6 3.00 4.50
D Kg 2 1.00 1.80
1990 50 1995 69
1991 65 1996 75 12.30 From the chain base index numbers given
1992 60 1997 84 below, find the fixed base index numbers.
1993 72 1998 80 Year : 1996 1997 1998 1999 2000
Chain base index : 80 110 120 90 140
12.29 The price quotations of four different
commodities for 1996 and 1997 are given 12.31 The following are the group index numbers
below. Calculate the index number for 1997 and the group weights of an average working
with 1996 as base by using (i) the simple average class family’s budget. Construct the cost of living
of price relatives and (ii) the weighted average number.
of price relatives.
478 BUSINESS S T A T I S T I C S

12.35 In calculating a certain cost of living index

Group Index Number Weight
number, the following weights were used: Food
Food 330 50 15, Clothing 3, Rent 4, Fuel and Light 2,
Clothing 208 10 Miscellaneous 1. Calculate the index for the
Fuel and lighting 200 12 period when the average percentage increases
House rent 162 12 in prices of items in the various groups over the
base period were 32, 54, 47, 78, and 58
Miscellaneous 180 16 respectively.
12.32 In 1988, for working class people, wheat was Suppose a business executive was earning Rs.
selling at an average price of Rs. 120 per 20 kg. 2050 in the base period, what should be his
Cloth Rs. 20 per metre, house rent Rs. 300 per salary in the current period if his standard of
house and other items Rs. 100 per unit. By living is to remain the same?
1998 cost of wheat rose by Rs. 160 per 20 kg,
12.36 Construct the cost of living index number from
rent by Rs. 450 house and other items doubled
the following data:
in price. The working class cost of living index
for the year 1998 with 1988 as base was 160.
Group Weights Group Index
By how much did the price of cloth rise during
the period? Food 47 247
12.33 Calculate the cost of living index from the Fuel and Lighting 47 293
following data: Clothing 48 289

Items Quantity Consumed Price (in Rs. per Unit) House rent 13 100
per Year in the Base Year Given Year Miscellaneous 14 236
Given Year
[Vikram Univ., MBA, 1996]
Rice (qtl) 2.50 × 12 12 25
Pulses (kg) 3 × 12 4 0.6 12.37 During a certain period the cost of living index
Oil (litre) 2 × 12 1.5 2.2 goes up from 110 to 200 and the salary of a
Clothing worker is also raised from Rs. 3250 to Rs. 5000.
(metres) 6 × 12 0.75 1.0 Does the worker really gain, and if so, by how
Housing much in real terms?
(per month) — 20 30 12.38 An enquiry into the budgets of middle class
Miscellaneous families in a certain city gave the following
(per month) — 10 15 information.
12.34 Compute the Consumer Price Index number Expenses Food Fuel ClothingRent Miscellaneous
from the following: 35% 10% 20% 15% 20%
Group Base Year Current Year Weight Prices (Rs.)
Price (Rs.) Price (Rs.) (Per cent) 1990 : 150 25 75 30 40
Food 400 550 35 Prices (Rs.)
Rent 250 300 25 1991 : 145 23 65 30 45
Clothing 500 600 15
Fuel 200 350 20 What is the Cost of Living Index number of
Entertainment 150 225 25 1991 as compared with that of 1990?

[Mangalore Univ., BCom, 1997]

 INDEX NUMBERS 479

Hints and Answers

12.30 The formula for converting a Chain Base Index
12.28 Year Price Index Number (Base 1989)
(CBI) number to a Fixed Base Index (FBI)
1989 75 100 Number is
50 Current years FBI
1990 50 × 100 = 166.67
75 Current years CBI × Previous years FBI
=
65 100
1991 65 × 100 = 186.57
75 Conversion of CBI to FBI
60
1992 60 × 100 = 180.00 Year Chain Base Index Fixed Base Index
75
72 1996 280 80
1993 72 × 100 = 196.00
75 110 × 80
1997 110 = 88
70 100
1994 70 × 100 = 193.33
75 120 × 88
69 1998 120 = 105.60
1995 69 × 100 = 192.00 100
75 90 × 105.6
75 1999 290 = 95.04
1996 75 × 100 = 100.00 100
75 140 × 95.04
84 2000 140 = 133.06
1997 84 × 100 = 112.00 100
75
80 12.31
1998 80 × 100 = 106.67
75 Groups Index No. 2Weights PW
12.29 P W
Food 330 250 16,500
p Clothing 208 210 2080
Comm Unit Weight Price P = 1 × 100 PW
p0 Fuel and Lighting 200 212 2400
(Rs., 1000) 1996 1997 House Rent 162 212 1944
W p0 p1 Miscellaneous 180 216 2880
100 25,804
A Kg 25 2.00 4.50 225 1,125
Σ PW 25,804
B Quint 27 2.50 3.20 128 896 Cost of living index = = = 258.04
ΣW 100
C Dozen 26 3.00 4.50 150 900 12.32
D Kg 22 1.00 1.80 180 360
Commodity Price Index No. Price Index No.
20 283 3,281
1998 1998
(i) Simple average of price relatives 160
Wheat 120 100 160 × 100 =150
120
1  p1  683 x
P01 = Σ × 100  = = 170.75 Cloth 320 100 x × 100 = 5x
n  p0 4 20
 450
House rent 300 100 450 × 100 = 150
(ii) Weighted average of price relatives 300
method: 200
Miscell 100 100 200 × 100 =200
Σ PW 3281 100
P01 = = = 164.05
ΣW 20 500 + 5x
480 BUSINESS S T A T I S T I C S

The index for 1998 is 160. Thus the sum of the ΣP W 3544
index numbers of the four commodities would Cost of living index = = =
ΣW 25
be 160 × 4 = 640. Hence, 500 + 5x = 640 or x 141.76
= 28. The rise in the price of cloth was Rs. 8 per For maintaining the same standard, the
metre.
2050 × 141.76
12.33 business executive should get
100
= Rs. 2906.08.
Items Quantity
12.36
Consumed q1 p0 p1 p1 q 1 p0 q 1
Rice (qtl) 2.50× 12 12.00 25.0 750.0 360.0 Group Weights (W) Group Index (P) PW
Pulses (kg) 3×12 0.40 0.6 21.6 14.0 Food 47 247 11609
Oil (litres) 2×12 1.50 2.2 52.8 36.0 Fuel and 7 293 2051
Clothing (mt) 6×12 0.75 1.0 72.0 54.0 lighting
Housing Clothing 8 289 2312
(per month) — 20 30 360.0 240.0 House rent 13 100 1300
Miscellaneous Miscellaneous 14 236 3304
(per month) — 10 15 180.0 120.0
Total 89 20,576
1436.4 824.4
ΣP W 20,576
Σ p1 q1 1436.4 Cost of living = = = 231.19
Cost of living index = × 100 = × 100 ΣW 89
Σ p0 q1 824.4
12.37 Real wage of Rs. 3250
= 174.24
Actual wage
12.34 = × 100
Cost of living index
p1 3250
Group p0 q1 P = × 100 Weight PW = × 100 = Rs. 2954.54
q0 110
W
5000
Food 400 550 137.5 35 4812.5 Real wage of Rs. 5000 = × 100
200
Rent 250 300 120.0 25 3000.0 = Rs. 2500 which is less than Rs. 2954.54
Clothing 500 600 120.0 15 1800.0
Since the real wage of Rs. 5000 is less than
Fuel 200 350 175.0 20 3500.0
that of Rs. 3250, the worker does not really
Entertainment 150 225 150.0 5 750.0
gain, real wage decrease by Rs. (2954.54 –
Total 100 13,862.5 2500) = Rs. 45.45.
12.38
Σ PW 13, 862.5
Consumer price index = =
ΣW 100 p
Expenses 1990 1991 P= 1 × 100
= 138.63 q
on 0
12.35 p0 p1 W PW
Group Average Group Index Price PW Food 150 145 96.67 35 3383.45
Per cent Increase in Weight
Fuel 125 123 92.00 10 920.00
P W
Clothing 175 165 89.67 20 1733.40
Food 32 132 15 1980 Rent 130 130 100.00 15 1500.00
Clothing 54 154 3 462 Miscell 140 145 112.50 20 2250.00
Rent 47 147 4 588
Total 100 9786.85
Fuel and light 78 178 2 356
Miscellaneous 58 158 1 158 Σ PW 9786.85
Cost of living index for 1991 = Σ W =
Total 25 2544 100
= 97.86.