UNIT 5 CORRELATION AND REGRESSION*

Structure
5.0 Objectives
5.1 Introduction
5.2 Scatter Diagram
5.3 Covariance
5.4 Correlation Coefficient
5.5 Interpretation of Correlation Coefficient
5.6 Rank Correlation Coefficient
5.7 The Concept of Regression
5.8 Linear Relationship: Two-Variables Case
5.9 Minimisation of Errors
5.10 Method Least Squares
5.11 Prediction
5.12 Relationship between Regression and Correlation
5.13 Multiple Regressions
5.14 Non-Linear Regression
5.15 Let Us Sum Up
5.16 Answers/Hints to Check Your Progress Exercises

5.0 OBJECTIVES
After going through this unit you will be in a position to
 plot scatter diagram;
 compute correlation coefficient and state its properties;
 compute rank correlation;
 explain the concept of regression;
 explain the method of least squares;
 identify the limitations of linear regression;
 apply linear regression models to given data; and
 use the regression equation for prediction.

5.1 INTRODUCTION
The word ‘bivariate’ is used to describe situations in which two character are
measured on each individual or item, the character being represented by two
variables. For example, the measurement of height ( X i ) and weight ( Yi ) of
students in a school. The subscript i in this case represents the student concerned.
*
Prof. Kaustuva Barik, School of Social Sciences, Indira Gandhi National Open University.
Summarisation of Thus, for example, X 5 , Y5 represent the height and weight of the fifth student.
Bivariate and Multi-
variate Data Statistical data relating to simultaneous measurement of two variables are called
bivariate data. The observation on each individual are paired, one for each
variable (X1, Y1), (X2, Y2), ......, (Xn, Yn).
In statistical studies with several variables, there are generally two types of
problems. In some problems it is of interest to study how the variables are
interrelated; such problems are tackled by using correlation technique. For
instance, an economist may be interested in studying the relationship between the
stock prices of various companies; for this he may use correlation techniques. In
other problems there is a variable y of basic interest and the problem is to find out
what information the other variable provides on Y, such problems are tackled
using regression techniques. For instance, an economist may be interested in
studying what factors determine the pay of an employed person and in particular,
he may be interested in exploring what role the factors such as education,
experience, market demand, etc. play in determining the pay. In the above
situation he may use regression techniques to set up a prediction formula for pay
based on education, experience, etc.

5.2 SCATTER DIAGRAM

We first illustrate how the relationship between two variables is studied.
A teacher is interested in studying the relationship between the performance in
Statistics and Economics of a class of 20 students. For this he compiles the scores
on these subjects of the students the last semester examination. Some data of this
type are presented in Table 5.1.
Table 5.1: Scores of 20 Students in Statistics and Economics
Serial Score in Serial Score in
Number Statistics Economics Number Statistics Economics
1 82 64 11 76 58
2 70 40 12 76 66
3 34 35 13 92 72
4 80 48 14 72 46
5 66 54 15 64 44
6 84 56 16 86 76
7 74 62 17 84 52
8 84 66 18 60 40
9 60 58 19 82 60
10 86 82 20 90 60

A representation of data of this type on a graph is a useful device which will help
us to understand the nature and form of the relationship between the two
variables, whether there is a discernible relationship or not and if so whether it is
linear of not. For this let us denote score in Economics by X and the score in
Statistics by Y and plot the data of Table 5.1 on the x-y plane. It does not matter
which is called X and which Y for this purpose. Such a plot is called Scatter Plot
or Scatter Diagram. For data of Table 5.1 the scatter diagram is given in Fig. 5.1.
98
 Correlation and
Regression

Fig. 5.1: Scatter Diagram of Scores in Statistics and Economics.

An inspection of Table 5.1 and Fig. 5.1 shows that there is a positive relationship
between x and y. This means that larger values of x associated with larger values
of y and smaller values of y. Further, the points seem to lie scattered around both
sides of a straight line. Thus, it appears that a linear relationship exists between x
and y. This relationship, however, in not perfect in the sense that there are
deviations from such a relationship in the case of certain observations. It would
indeed be useful to get a measure of the strength of this linear relationship.

5.3 COVARIANCE
In the case of a single variable we have learnt the concept of variance, which is
defined as
𝜎 = ∑ (𝑋 − 𝑋) … (5.1)

In the above we use a subscript x to specify that  22 represents the variance in x. In

a similar manner we can represent  y2 as the variance in y and  x and  y as the
standard deviation in x and y respectively.
As you know, variance measures the dispersion from mean. In the case of
bivariate data we have to reach a single figure which will present the deviation in
both the variables from their respective means. For this purpose we use a concept
termed covariance, which is defined as follows:
𝜎 = ∑ (𝑋 − 𝑋)(𝑌 − 𝑌 ) … (5.2)

You may recall that standard deviation is always positive since it is defined as the
positive square root of variance. In the case of covariance there are two terms
( X i  X ) and (Yi  Y ) which represent the deviations in x from X and Y from Y .
99
Summarisation of Moreover, ( X i  X ) can be positive or negative depending on whether xi is less
Bivariate and Multi-
variate Data than or greater than X . Similarly (Yi  Y ) can be positive or negative. It is not
necessary that whenever ( X i  X ) is positive (Yi  Y ) will also be positive.
Therefore, the product ( X i  X ) (Yi  Y ) can be either positive or negative. A
positive value for ( X i  X ) (Yi  Y ) implies the whenever X i  X , we have
Yi  Y . Thus a higher value of xi is associated with a relatively higher value in yi .
On the other hand, ( X i  X )(Yi  Y )  0 implies that a lower value in X i is
associated with a relatively higher value in yi . when we sum it over all the
observations and ivied by the number of observations, we may obtain a negative
or positive value. Therefore, covariance can assume both positive and negative
values.
When covariance between x and y is negative ( xy  0) we can say that the
relationship could be inverse. Similarly, ( xy  0 ) implies a positive relationship
between x and y. A major limitation of covariance is that it is not independent of
unit of measurement. It means that if we change the unit of measurement of the
variables we will get a difference value for  xy .

The computation of  xy as given in (5.2) often involves large numbers. Therefore,

it is derived further as

1 n 1 n
 xy  
n i 1
( X i  X )(Yi  Y )   ( X iYi  XYi  XY )
n i1

By further simplification we find that

1 n 1 n 1 n 1 n
 xy   i i n
n i 1
X Y 
i 1
X Yi   i n
n i1
X Y 
i 1
XY

1 n 1
Since 
n i 1
XYi  X iY  XY we have
n

1 n
 xy   X iYi  XY
n i 1
… (5.3)

5.4 CORRELATION COEFFICIENT

The task before us is to measure the linear relationship between x and y. It is
desirable to have this measure of strength of linear relationship independent of the
scale chosen for measuring the variables. For instance, if we are measuring the
relationship between height and weight, we should get the same measure whether
height is measured in inches or centimetres and weight in pounds or kilograms.
Similarly, if a variable is temperature, it should not matter whether it is recorded
in Celsius or Fahrenheit.

100
This can be achieved by standardizing each variable, that is by considering Correlation and
Regression
XX Y Y
and where X and Y are the means of X and Y respectively and  x
x y
and  y are standard deviations.

Let us denote these standardised variables by u and y respectively. Let us also use
the notation ( X i , Yi ) to denote the score ith student in Economics and Statistics
respectively, i ranging from 1 to n, the number of students, n being 20 in our
example. Similarly, let (ui , vi ) denote the standardised scores of ith student. Then
recall the following formulae for mean and standard deviation:
1 n 1 n
X 
n i 1
X i ;  x2   ( X i  X ) 2 ;
n i 1
1 n 1 n
Y  i Y n
n i 1
X ;  2

i 1
(Yi  Y ) 2

Fig. 5.2: Scatter Diagram of Standardised Scores in Statistics and Economics

Fig. 5.2 is the scatter diagram in terms of standardised variables u and v. Let us
observe that in this example there is a positive association between the two
scores. The larger one score is, the larger the other score also is; the smaller one
score is the smaller the other score is, on the whole. In view of this, most of the
points are either in the first quadrant or in the third quadrant. The first quadrant
represents the cases where both scores are above their respective means and third
quadrant represents the cases where both scores are below their respective means.
There are only a very few points in second and fourth quadrants, which represent
the cases where one score is above its mean and the other is below its mean. Thus
the product of the u, v values is a suitable indicator of the strength of the
relationship; this product is positive in the first and third quadrants and negative
in the second and fourth. Thus the product of u, v averaged over all the points
may be considered to be suitable measure of the strength of linear relationship
between X and Y.

101
Summarisation of This measure is called the correlation coefficient between X and Y and is usually
Bivariate and Multi-
variate Data denoted by rxy or simply by r, when it is clear what x and y in the context are.
This is also called the Pearson’s Product-Moment Correlation Coefficient to
distinguish it from other types of correlation coefficients.

Thus the formula for r is

1 n
r  ui vi
n i 1
… (5.4)

If we substitute the variables x and y in (5.4) above

1 n
1 n  Xi  X  Yi  Y   X i  X Yi  Y 
n i1
r 
n i1   x




  y   x y

In the above expression, the term

1 n
 X i  X Yi  Y 
n i 1

is the covariance between x and y ( xy ).

Thus, the formula for correlation coefficient is

 xy
r … (5.5)
 x  y

Incorporating the formulae for x , y ,  x ,  y it becomes

1 n 1 n
 i
n i1
( X  x )(Yi  Y ) Xi  X (Yi  Y )
n i1
r  ... (5.6)
1 n 2 1
n
1 n 2 1
n

X i  X  n 
n i1 i1
(Yi  Y ) 2
( X i  X ) n 
n i1 i1
(Yi  Y ) 2

Or, alternatively
n n n
n X iYi   X i  Yi
r i 1 i 1 i 1
… (5.7)
 2
 n 2

 n X i2    X i   n Yi 2    Yi  
n n n

 i 1  i1    i 1  i 1  


Let us go back to the data given in Table 5.1 and work out the value of r. You can
use any of the formulae (5.4), (5.5) or (5.7) to get the value of r. Since all the
formulae are derived from the same concept we obtain the same value for r
whichever formulae we use. For the data set in Table 5.1 we have calculated it by
using (5.4) and (5.7). We construct Table 5.2 for this purpose.

102
 Table 5.2: Calculation of Correlation Coefficient Correlation and
Regression
Observation X Y X2 Y2 XY
No.
1 82 64 6724 4096 5248
2 70 40 4900 1600 2800
3 34 35 1156 1225 1190
4 80 48 6400 2304 3840
5 66 54 4356 2916 3564
6 84 56 7056 3136 4704
7 74 62 5476 3844 4588
8 84 66 7056 4356 5544
9 60 52 3600 2704 3120
10 86 82 7396 6724 7052
11 76 58 5776 3364 4408
12 76 66 5776 4356 5016
13 92 72 8464 5184 6624
14 72 46 5184 2116 3312
15 64 44 4096 1936 2816
16 86 76 7396 5776 6536
17 84 52 7056 2704 4368
18 60 40 3600 1600 2400
19 82 60 6724 3600 4920
20 90 60 8100 3600 5400
Total 1502 1133 116292 67141 87450

From Table 5.2 we note that

20

X
i 1
i  150; X  75.1;

20

Y
i 1
i  1133; Y  56.65;

20
1  15022 
 X i2  116292; x2 
i 1 20 
116292 
20 
 174.59; x  13.21;

20
1  11332 
Y
i 1
i
2
 67141; x2  
20 
67141 
20 
 147.83;  y  12.16;

1  1502  1133 
XY i i  87450;  xy  
20 
87450 
20   118.09

Thus, using formula given at (5.4), we have

118.09
r  0.735
13.21 12.16
Now let us use the formula 5.7. We have
20  87450
r  0.735
(20  116292  1502 2 )(20  67141  11332 )

103
Summarisation of Thus we see that both the formulae provide the same value of the correlation
Bivariate and Multi-
variate Data coefficient r. You can check yourself that the same value of r is obtained by using
the formula (5.5). For this purpose you will need values on

(X i  X ) 2 ,  (Yi  Y ) 2 and  (X i  X )(Yi  Y )

Hence you can have five columns on

( X i  X ), (Yi  Y ), ( X i  X ) 2 , (Yi  Y ) 2 and ( X i  X )(Yi  Y ) in a table and find the
totals.

5.5 INTERPRETATION OF CORRELATION

COEFFICIENT
It is a mathematical fact that the value of r as defined above lies between –1and +
1. The extreme values of –1and + 1are obtained only in situations where there is a
perfect linear relationship between X and Y. The –1is obtained when this
relationship is perfectly negative (i.e., inverse) and +1 when this is perfect
positive (i.e., direct). The value of 0 is obtained when there is no linear
relationship between x and y.
We can make some guess work about the sign and degree of the correlation
coefficient from the scatter diagram. Fig. 5.3 gives example of scatter diagrams
for various values of r. Fig. 5.3 (a) is a scatter diagram for the case r = 0; here
there is no linear relationship between x and y. Fig. 5.3(b) is also and example of
scatter diagram for the case r = 0; here there is discernible relationship between X
and Y but it is not of the linear type. Here, initially, Y increases with X but later Y
decreases as X increases resulting in a definitive quadratic relationship. But the
correlation coefficient in the case is zero. Thus the correlation coefficient is only a
measure of linear relationship. This sort of scatter diagram is obtained, if we plot,
for instance, body weight (Y) of individuals against there are (X). Fig. 5.3.(c) is an
example of a scatter diagram where there is a perfect positive linear relationship
between X and Y. We get this sort of scatter diagram if we plot, for instance,
height of individuals in inches (X) against their heights in centimetres (Y); in that
case Y = 2.54X, which is a deterministic and perfect linear relationship. Figures
5.3(d) to 5.3(k) are scatter diagrams for other values of r. Form these scatter
diagrams we get an idea of the nature of relationship and associated values of r.
From these is would seem that a value of 0.81 indicates a fair degree of linear
relationship between scores in Statistics and Economics of these candidates. Such
a quantification of relationship or association between variables is helpful for
natural and social scientists to understand the phenomena they are investigating
and explore these phenomena further. In an example of this sort, an educational
psychologist may compute correlation coefficients between scores in various
subjects and by further statistical analysis of the correlation coefficients and using
psychological techniques may be able to form a theory as to what mental and
other faculties are involved in making students good in various disciplines.
104
 Correlation and
Regression

Fig. 5.3: Scatter Plots for Various Values of Correlation Coefficient

You should remember that

 Correlation coefficient shows the linear relationship between X and Y. Thus,
even if there is a strong non-linear relationship between X and Y, correlation
coefficient may be low.
 Correlation coefficient is independent of scale and origin. If we subtract some
constant from one (or both) of the variables, correlation coefficient will
remain unchanged from one (or both) of the variables by some constant,
correlation coefficient will not change.
 Correlation coefficient varies between –1 and +1. It means that r cannot be
smaller than –1 and cannot be greater than +1.
The existence of a linear relationship between two variables is not to be
interpreted to mean a cause-effect relationship between the two.
105
Summarisation of For instance, if you work out the correlation between family expenditure on petrol
Bivariate and Multi-
variate Data
and chocolates, you may find it to be fairly high indicating a fair degree of linear
relationship. However, both of these are luxury items and richer families can
afford them while poorer ones cannot. Thus the high correlation here is caused by
the high correlation of each of the variables with family income. To consider
another example, suppose for each of the last twenty years, you work out the
average height of an Indian and the average time per week an Indian watches
television; you are likely to find a positive correlation. This does not, however,
imply that watching television increases one’s height or that taller people tend to
watch television longer. Both these variables have an increasing trend over time
and this is reflected in the high correlation. This kind or correlation between two
variables is caused by the effect of a third variable on each of them rather than a
direct linear cause-effect of a third variable on each of them rather than a direct
linear cause-effect relationship between them is called spurious correlation.
Another aspect of the computation of correlation coefficient that we should be
aware of is that the correlation coefficient like any other quantity computed from
sample, varies from sample to sample and these sample fluctuations should be
taken into account in making use of the computed coefficient. We do not discuss
these techniques here.
Whether the presence of a linear relationship between two variables and hence a
high correlation between them is genuine or spurious, such a situation is helpful
to predict one variable from the other.
Check Your Progress 1
1) Calculate r from the following given results:
n  10;  X  125;  X 2  1585;  Y  80;  Y 2  650;  XY  1007.
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
2) Calculate the coefficient of correlation for the ages of husband and wife:
Age of husband : 23 27 28 29 30 31 33 35 36 39
Age of wife : 18 22 23 24 25 26 28 29 30 32
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
106
3) Specimens of similarly treated alloy steel containing various percentages of Correlation and
Regression
nickel are tested for toughness with the following results:

4)

Find the correlation coefficient between toughness and nickel content and
comment on the result.
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
4) Determine the correlation coefficient between x and y.

x : 5 7 9 11 13 15

y : 1.7 2.4 2.8 3.4 3.7 4.4

……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
5) The following table gives the saving bank deposits in billions of dollars and
strikes and lock-outs, in thousands, over a number of years. Compute the
correlation coefficient and comment on the result.
Saving deposits : 5.1 5.4 5.5 5.9 6.4 6.0 7.2
Strikes and lock-outs : 3.8 4.4 3.3 3.6 3.3 2.3 1.0
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
……………………………………………………………………………..
107
Summarisation of
Bivariate and Multi- 5.6 RANK CORRELATION COEFFICIENT
variate Data
The Pearson’s product moment correlation coefficient (or simply, the correlation
coefficient) described above is suitable if both the variables involved are
measureable (numerical) and the relationship between the variables is linear.
However, there are situations where variables are not numerical but various items
can be ranked according to the characteristics (i.e., ordinal). Sometimes even
when the original variables are measurable, they are converted into ranks and a
measure of association is computed. Consider for instance the situation when two
examiners are asked to judge ten candidates on the basis of an oral examination.
In this case, it may be difficult to assign scores to candidates, but the examiners
find it reasonably easy to rank the candidates in order of merit. Before using the
resulted it may be advisable to find out if rankings are in reasonable concordance.
For this, a measure of association between the ranks assigned by the two
examiners may by computed. The Karl Pearson’s correlation coefficient is not
suitable in this situation. One may use the following called Spearman’s Rank
Correlation Coefficient for this purpose.
Table 5.3: Ranks of 10 Candidates by two Examiners
S. No Rank given by Difference
Examiner I Examiner II Di Di2
1 6.0 6.5 –0.5 0.25
2 2.0 3.0 –1.0 1.00
3 8.5 6.5 2.0 4.00
4 1.0 1.0 0.0 0.00
5 10.0 2.0 8.0 64.00
6 3.0 4.0 –1.0 1.00
7 8.5 9.5 –1.0 1.00
8 4.0 5.0 –1.0 1.00
9 5.0 8.0 –3.0 9.00
10 7.0 9.5 –2.5 6.25

D i 0 D1
2
 87.50

Let us consider the data of Table 5.3. Here there are some ties; the tied cases are
given the same rank in such a way their total is the same as when there is no tie.
For example, when there are two cases with rank 6, each is given a rank of 6.5
and there is no case with rank either 6 or 7. Similarly, if there are three cases with
rank 5, then each is given a rank of 6 and there is no case with rank 5 or 7.
Spearman’s rank correlation coefficient, called Spearman’s Rho, denoted by  , is
based on the difference Di (i for ith observation) between the two rankings. If the
two rankings completely coincide, then Di is zero for every case. The larger the
value of Di, the greater is the difference between the two rankings and smaller is
the association. Thus, the association can be measured by considering the
magnitudes of Di. Since the sum of Di is always zero, to find a single index on the
basis of Di values, we should remove the sign of Di and consider only the
magnitude. In Spearman’s  , this is done by taking Di2 .
108
 n Correlation and
However, the largeness or smallness of D
i 1
i
2
, where n is the number of cases, Regression

will depend on n. thus, in order to be able to interpret this value, we could create a
ratio by dividing this sum by the largest possible value, which depends only on n,
n
6   Di2
n( n  1 )2
i 1
which is . However, is zero for perfect association and 2 for
6 n( n 2  1 )
lack of association, i.e., perfect negative association, while we would like it to be
other way around. So we subtract this ratio from 1. Thus
n
6   Di2
  1 i 1
… (5.8)
n( n 2  1 )
is defined as Spearman’s rank correlation.
Let us calculated the value of  from the data given in Table 5.3.
6  87.5 525
   1  1  0.53  0.47.
10( 10  1 )
2
990
Like Karl Pearson’s coefficient of correlation the Spearman’s rank correlation has
a value +1 for perfect matching of ranks, –1for perfect mismatching of ranks and
0 for the lack of relation between the ranks.
There are other measures of association suitable for use when the variables are of
nominal, ordinal and other types. We do not discuss them here.
Check Your Progress 2
1) In a contest, two judges ranked eight candidates A, B, C, D, E, F, G and H in
order of their preference, as shown in the following table. Find the rank
correlation coefficient.
A B C D E F G H
First Judge 5 2 8 1 4 6 3 7
Second Judge 4 5 7 3 2 8 1 6
…………………………………………………………………………….
…………………………………………………………………………….
…………………………………………………………………………..…
…………………………………………………………………………..…
…………………………………………………………………………..…
…………………………………………………………………………..…
……………………………………………………………………………..

2) Compute the correlation coefficient of the following ranks of a group of

students in two examinations. What conclusion do you draw from the results?

109
Summarisation of
Bivariate and Multi- Roll Nos. 1 2 3 4 5 6 7 8 9 10
variate Data
Rank in B. Com. Exam. 1 5 8 6 7 4 2 3 9 10

Rank in M. Com Exam. 2 1 5 7 6 3 4 8 10 9

…………………………………………………………………………….........
………………………………………………………………………........…….
……………………………………………………………………….......…..…
……………………………………………………………………….......…..…
………………………………………………………………………….........…
……………………………………………………………………….......…..…

3) Ten competitors in a musical contest were ranked by 3 judges A, B and C in

the following order:

Ranks by A : 1 6 5 10 3 2 4 9 7 8

Ranks by B : 3 5 8 4 7 10 2 1 6 9

Ranks by C : 6 4 9 8 1 2 3 10 5 7

Using the rank correlation method, discuss which pair of judges has the
nearest approach to common liking in music.
…………………………………………………………………....………….....
………………………………………………………………….......…………..
………………………………………………………………….......……..……
………………………………………………………………….......………..…
………………………………………………………………….......……..……

4) Ten students obtained the following marks in Mathematics and Statistics.

Calculate the rank correlation coefficient.
Student 1 2 3 4 5 6 7 8 9 10
(Roll No.)
Marks in 78 36 98 25 75 82 90 62 65 39
Mathematics
Marks in 84 51 91 60 68 62 86 58 53 47
Statistics
…………………………………………………………………………….........
…………………………………………………………………………........….
………………………………………………………………………….........…
………………………………………………………………………….........…
………………………………………………………………………….........…

110
 Correlation and
5.7 THE CONCEPT OF REGRESSION Regression
In the previous section we noted that correlation coefficient does not reflect cause
and effect relationship between two variables. Thus we cannot predict the value
of one variable for a given value of the other variable. This limitation is removed
by regression analysis. In regression analysis, the relationship between variables
are expressed in the form of a mathematical equation. It is assumed that one
variable is the cause and the other is the effect. You should remember that
regression is a statistical tool which helps understand the relationship between
variables and predicts the unknown values of the dependent variable from known
values of the independent variable.
In regression analysis we have two types of variables: i) dependent (or explained)
variable, and ii) independent (or explanatory) variable. As the name (explained
and explanatory) suggests the dependent variable is explained by the independent
variable.
In the simplest case of regression analysis there is one dependent variable and one
independent variable. Let us assume that consumption expenditure of a household
is related to the household income. For example, it can be postulated that as
household income increases, expenditure also increases. Here consumption
expenditure is the dependent variable and household income is the independent
variable.
Usually we denote the dependent variable as Y and the independent variable as X.
Suppose we took up a household survey and collected n pairs of observations in X
and Y. The next step is to find out the nature of relationship between X and Y.
The relationship between X and Y can take many forms. The general practice is
to express the relationship in terms of some mathematical equation. The simplest
of these equations is the linear equation. This means that the relationship between
X and Y is in the form of a straight line and is termed linear regression. When the
equation represents curves (not a straight line) the regression is called non-linear
or curvilinear.
Now the question arises, ‘How do we identify the equation form?’ There is no
hard and fast rule as such. The form of the equation depends upon the reasoning
and assumptions made by us. However, we may plot the X and Y variables on a
graph paper to prepare a scatter diagram. From the scatter diagram, the location of
the points on the graph paper helps in identifying the type of equation to be fitted.
If the points are more or less in a straight line, then linear equation is assumed. On
the other hand, if the points are not in a straight line and are in the form of a
curve, a suitable non-linear equation (which resembles the scatter) is assumed.
We have to take another decision, that is, the identification of dependent and
independent variables. This again depends on the logic put forth and purpose of
analysis: whether ‘Y depends on X’ or ‘X depends on Y’. Thus there can be two
regression equations from the same set of data. These are i) Y is assumed to be

111
Summarisation of dependent on X (this is termed ‘Y on X’ line), and ii) X is assumed to be
Bivariate and Multi-
variate Data dependent on Y (this is termed ‘X on Y’ line).
Regression analysis can be extended to cases where one dependent variable is
explained by a number of independent variables. Such a case is termed multiple
regression. In advanced regression models there can be a number of both
dependent as well as independent variables.
You may by now be wondering why the term ‘regression’, which means ‘reduce’.
This name is associated with a phenomenon that was observed in a study on the
relationship between the stature of father (x) and son (y). It was observed that the
average stature of sons of the tallest fathers has a tendency to be less than the
average stature of these fathers. On the other hand, the average stature of sons of
the shortest fathers has a tendency to be more than the average stature of these
fathers. This phenomenon was called regression towards the mean. Although this
appeared somewhat strange at that time, it was found later that this is due to
natural variation within subgroups of a group and the same phenomenon occurred
in most problems and data sets. The explanation is that many tall men come from
families with average stature due to vagaries of natural variation and they produce
sons who are shorter than them on the whole. A similar phenomenon takes place
at the lower end of the scale.

5.8 LINEAR RELATIONSHIP: TWO-VARIABLES

CASE
The simplest relationship between X and Y could perhaps be a linear
deterministic function given by
Yi  a  bX i …(5.9)

In the above equation X is the independent variable or explanatory variable and Y

is the dependent variable or explained variable. You may recall that the subscript i
represents the observation number, i ranges from 1 to n. Thus Y1 is the first
observation of the dependent variable, X 5 is the fifth observation of the
independent variable, and so on.
Equation (5.9) implies that Y is completely determined by X and the parameters a
and b. Suppose we have parameter values a = 3 and b = 0.75, then our linear
equation is Y = 3 + 0.75 X. From this equation we can find out the value of Y for
given values of X. For example, when X = 8, we find that Y = 9. Thus if we have
different values of X then we obtain corresponding Y values on the basis of (5.9).
Again, if X i is the same for two observations, then the value of Yi will also be
identical for both the observations. A plot of Y on X will show no deviation from
the straight line with intercept ‘a’ and slope ‘b’.
If we look into the deterministic model given by (5.9) we find that it may not be
appropriate for describing economic interrelationship between variables. For
example, let Y = consumption and X = income of households. Suppose you
record your income and consumption for successive months.

112
For the months when your income is the same, do your consumption remain the Correlation and
Regression
same? The point we are trying to make is that economic relationship involves
certain randomness.
Therefore, we assume the relationship between Y and X to be stochastic and add
one error term in (5.9). Thus our stochastic model is
Yi  a  bX i  ei …(5.10)
where ei is the error term. In real life situations ei represents randomness in
human behaviour and excluded variables, if any, in the model. Remember that
the right hand side of (5.10) has two parts, viz., i) deterministic part (that is,
a  bX i ), and ii) stochastic or randomness part (that is, ei ). Equation (5.10)
implies that even if X i remains the same for two observations, Yi need not be the
same because of different ei . Thus, if we plot (5.10) on a graph paper the
observations will not remain on a straight line.
Example 5.1
The amount of rainfall and agricultural production for ten years are given in Table
5.4.
Table 5.4: Rainfall and Agricultural Production
Rainfall Agricultural
(in mm.) production (in
tonne)
60 33
62 37
65 38
71 42
73 42
75 45
81 49
85 52
88 55
90 57

Fig. 5.4: Scatter Diagram

113
Summarisation of We plot the data on a graph paper. The scatter diagram looks something like Fig.
Bivariate and Multi-
variate Data 5.4. We observe from Fig. 5.4 that the points do not lie strictly on a straight line.
But they show an upward rising tendency where a straight line can be fitted. Let
us draw the regression line along with the scatter plot.

Fig. 5.5: Regression Line

The vertical difference between the regression line and the observations is the
error ei . The value corresponding to the regression line is called the predicted
value or the expected value. On the other hand, the actual value of the dependent
variable corresponding to a particular value of the independent variable is called
the observed value. Thus ‘error’ is the difference between predicted value and
observed value.
A question that arises is, ‘How do we obtain the regression line? The procedure of
fitting a straight line to the data is explained below.

5.9 MINIMISATION OF ERRORS

As mentioned earlier, a straight line can be represented by
Yi  a  bX i

where b is the slope and a is the intercept on y-axis. The location of a straight line
depends on the value of a and b, called parameters. Therefore, the task before us
is to estimate these parameters from the collected data. (You will learn more
about the concept of estimation in Block 4). In order to obtain the line of best fit
to the data we should find estimates of a and b in such a way that the error ei is
minimum.
In Fig. 5.4 these differences between observed and predicted values of Y are
marked with straight lines from the observed points, parallel to y-axis, meeting
the regression line. The lengths of these segments are the errors at the observed
points.
Let us denote the n observations as before by ( X i , Yi ), i = 1, 2, ....., n. In Example
5.1 on agricultural production and rainfall, n=10.

114
Let us denote the predicted value of Yi at X i by Ŷi (the notation Ŷi is pronounced Correlation and
Regression
as ‘ Yi -cap’ or ‘ Yi -hat’). Thus

Yˆi  a  bX i , i = 1, 2, ....., n.

The error at the ith point will then be

ei  Yi  Yˆi ……(5.11)

It would be nice if we can determine a and b in such a way that each of the ei , i =
1, 2, ....., n is zero. But this is impossible unless it so happens that all the n points
lie on a straight line, which is very unlikely. Thus we have to be content with
minimising a combination of ei , i = 1, 2, ....., n. What are the options before us?
n
 It is tempting to think that the total of all the ei , i = 1, 2, ….., n, that is,  ei
i1

is a suitable choice. But it is not. Because, ei for points above the line are
positive and below the line are negative. Thus by having a combination of
n
large positive and large negative errors, it is possible for  ei to be very
i 1

small.
 A second possibility is that if we take a = y (the arithmetic mean of the Yi ’s)
n
and b = 0,  ei could be made zero. In this case, however, we do not need the
i 1

value of X at all for prediction! The predicted value is the same irrespective of
the observed value of X. This evidently is wrong.
n
 What then is wrong with the criterion  ei ? It takes into account the sign of
i 1
ei . What matters is the magnitude of the error and whether the error is on the
n
positive side or negative side is really immaterial. Thus, the criterion  ei is
i 1

a suitable criterion to minimise. Remember that ei means the absolute value

of ei . Thus, if ei = 5 then ei = 5 and also if ei = –5 then ei =5. However,
this option poses some computational problems.
 For theoretical and computational reasons, the criterion of least squares is
preferred to the absolute value criterion. While in the absolute value criterion
the sign of ei is removed by taking its absolute value, in the least squares
criterion it is done by squaring it. Remember that the squares of both 5 and –5
are 25. This device has been found to be mathematically and computationally
more attractive.

We explain in detail the least squares method in the following section.

115
Summarisation of
Bivariate and Multi- 5.10 METHOD OF LEAST SQUARES
variate Data
In the least squares method we minimise the sum of squares of the error terms,
n
that is,  ei2 .
i 1

From (5.9) we find that ei  Yi  Yˆi

which implies ei  Yi  (a  bX i )  Yi  a  bX i .
n n
Hence,  ei2   (Yi  a  bX i ) 2 …(5.12)
i 1 i 1

The next question is: How do we obtain the values of a and b to minimise (5.12)?
 Those of you who are familiar with the concept of differentiation will
remember that the value of a function is minimum when the first derivative of
the function is zero and second derivative is positive. Here we have to choose
n
the value of a and b. Hence,  ei2 will be minimum when its partial derivatives
i 1
n
with respect to a and b are zero. The partial derivatives of  ei2 are obtained as
i 1

follows:
  ei2   (Yi  a  bX i ) 2
i
 i
 2  (1)   (Yi  a  bX i ) …(5.13)
a a i

  ei2   (Yi  a  bX i ) 2
i
 i
 2  ( X i )   (Yi  a  bX i ) …(5.14)
b b i

By equating (5.13) and (5.14) to zero and re-arranging the terms we get the
following two equations:
n n
 Yi  na  b  X i …(5.15)
i 1 i 1

n n n
2
 X i Yi  a  X i  b  X i …(5.16)
i 1 i 1 i 1

These two equations, (5.15) and (5.16), are called the normal equations of
least squares. These are two simultaneous linear equations in two unknowns.
These can be solved to obtain the values of a and b.
 Those of you who are not familiar with the concept of differentiation can use
a rule of thumb (We suggest that you should learn the concept of
differentiation, which is so much useful in Economics). We can say that the
normal equations given at (5.15) and (5.16) are derived by multiplying the
coefficients of a and b to the linear equation and summing over all
observations. Here the linear equation is Yi  a  bX i . The first normal
equation is simply the linear equation Yi  a  bX i summed over all
observations (since the coefficient of a is 1).
 Yi   a   bX i or  Yi  na  b  X i
116
The second normal equation is the linear equation multiplied by X i (since the Correlation and
Regression
coefficient of b is X i )
2 2
 X iYi   aX i   bX i or  X i Yi  a X i  b X i

After obtaining the normal equations we calculate the values of a and b from the
set of data we have.
Example 5.2: Assume that quantity of agricultural production depends on the
amount of rainfall and fit a linear regression to the data given in Example 5.1.
In this case dependent variable (Y) is quantity of agricultural production and
independent variable (X) is amount of rainfall. The regression equation to be
fitted is
Yi  a  bX i  ei

For the above equation we find out the normal equations by the method of least
squares. These equations are given at (5.15) and (5.16). Next we construct a table
as follows:
Table 5.5: Computation of Regression Line

Xi Yi X i2 X i Yi Ŷi ei

60 33 3600 1980 33.85 -0.85

62 37 3844 2294 35.34 1.66

65 38 4225 2470 37.57 0.43

71 42 5041 2982 42.03 -0.03

73 42 5329 3066 43.51 -1.51

75 45 5625 3375 45.00 0.00

81 49 6561 3969 49.46 -0.46

85 52 7225 4420 52.43 -0.43

88 55 7744 4840 54.66 0.34

90 57 8100 5130 56.15 0.85

 Yi =450  X i Yi =34526  Yˆi =450  ei =0

2
Total  Xi
i i i i i
 X i = 750 =57294
i

By substituting values from Table 5.5 in the normal equations (5.15) and (5.16)
we get the following:

117
Summarisation of 450 = 10a + 750b
Bivariate and Multi-
variate Data
34526 = 750a + 57294b

By solving these two equations we obtain a = –10.73 and b = 0.743.

So the regression line is Yˆi  10.73  0.743X i .

Notice that the sum of errors  ei for the estimated regression equation in zero
i

(see the last column of Table 5.5).

The computation given in Table 5.5 often involves large numbers and poses
difficulty. Hence we have a short-cut method for calculating the values of a and b
from the normal equations.

Let us take
x  X  X and y  Y  Y where X and Y are the arithmetic means of X and Y
respectively.
Hence xy  ( X  X )(Y  Y )

By re-arranging terms in the normal equations we find that

n
 xy
i 1
b n
…(5.17)
2
x
i 1

a  Y  bX …(5.18)

1n
You may recall that covariance is given by  xy  (Xi  X)(Yi Y ) = 1 n
 xi y i .
n i1 n i 1

1 n

n
Moreover, variance of X is given by  x2  ( Xi  X )2 = 1  xi2
n i1 n i 1

n
 xy xy
Since b 
i1 b
n
x
2
we can say that
x2 …(5.19)
i1

Since these formulae are derived from the normal equations we get the same
values for a and b in this method also. For the data given in Table 5.4 we compute
the values of a and b by this method. For this purpose we construct Table 5.6.

118
 Table 5.6: Computation of Regression Line (short-cut method) Correlation and
Regression
Xi Yi xi yi xi2 xi y i
60 33 -15 -12 225 180
62 37 -13 -8 169 104
65 38 -10 -7 100 70
71 42 -4 -3 16 12
73 42 -2 -3 4 6
75 45 0 0 0 0
81 49 6 4 36 24
85 52 10 7 100 70
88 55 13 10 169 130
90 57 15 12 225 180
Total = 750 450 0 0 1044 776

On the basis of Table 5.6 we find that

750 450
X   75 and Y   45
10 10
n
 xy
i 1 776
b n
  0.743
2 1044
x
i 1

a  Y  b X  45  0.743  10   10 .73

Thus the regression line in this method also Yˆi  10.73  0.743X i …(5.20)

Coefficient b in (5.20) is called the regression coefficient. This coefficient reflects

the amount of increase in Y when there is a unit increase in X. In regression
equation (5.20) the coefficient b = 0.743 implies that if rainfall increases by 1
mm., agricultural production will increase 0.743 thousand tonne.
Regression coefficient is widely used. It is also an important tool of analysis. For
example, if Y is aggregate consumption and X is aggregate income, b represents
marginal propensity to consume (MPC).

5.11 PREDICTION
A major interest in studying regression lies in its ability to forecast. In Example
5.1 we assumed that the quantity of agricultural production is dependent on the
amount of rainfall. We fitted a linear equation to the observed data and got the
relationship
Yˆi  10.73  0.743X i

From this equation we can predict the quantity of agricultural output given the
amount of rainfall. Thus when rainfall is 60 mm. agricultural production is
(  10 .73  0.74  60 )  33 .85 thousand tonnes. This figure is the predicted value on
the basis of regression equation. In a similar manner we can find the predicted
values of Y for different values of X.

119
Summarisation of Let us compare the predicted value with the observed value. From Table 5.4,
Bivariate and Multi-
variate Data where observed values are given, we find that when rainfall is 60 mm.
agricultural production is 33 thousand tonnes. In fact, the predicted values Ŷi for
observed values of X are given in the fifth column of Table 5.5. Thus when
rainfall is 60 mm. Predicted value is 33.85 thousand tonnes. Thus the error value
ei is –0.85 thousand tonne.

Now a question arises, ‘Which one, between observed and predicted values,
should we believe?’ In other words, what will be the quantity of agricultural
production if there is a rainfall of 60 mm. in future? On the basis of our regression
line it is given to be 33.85 tonnes. And we accept this value because it is based on
the overall data. The error of –0.85 is considered as a random fluctuation which
may not be repeated.
The second question that comes to our mind is, ‘Is the prediction valid for any
value of X?’ For example, we find from the regression equation that when rainfall
is zero, agricultural production is –10.73 thousand tonne. But common sense tells
us that agricultural production cannot be negative! Is there anything wrong with
our regression equation? In fact, the regression equation here is estimated on the
basis of rainfall data in the range of 60-90 mm. Thus prediction is be valid in this
range of X. Our prediction should not be for far off values of X.
A third, question that arises here is, ‘Will the predicted value come true?’ This
depends upon the coefficient of determination. If the coefficient of determination
is closer to one, there is greater likelihood that the prediction will be realised.
However, the predicted value is constrained by elements of randomness involved
with human behaviour and other unforeseen factors.

5.12 RELATIONSHIP BETWEEN REGRESSION AND

CORRELATION
In regression analysis the status of the two variables (X, Y) are different such that
Y is the variable to be predicted and X is the variable, information on which is to
be used. In the rainfall-agricultural production problem, it makes sense to predict
agricultural production on the basis of rainfall and it would not make sense to try
and predict rainfall on the basis of agricultural production. However, in the case
of scores in Economics and Statistics (see Table 5.1), either one could be X and
the other Y. Hence we consider the two prediction problems: (i) predicting
Economics score (Y) from Statistics score (X); and (ii) predicting Statistics score
(X) from Economics score (Y).
Thus, we can have two regression coefficients from a given set of data depending
upon the choice of dependent and independent variables. These are:

a) Y on X line, Yi  a  bX i

b) X on Y line, X i    Yi

120
You may ask, ‘What is the need for having two different lines? By rearrangement Correlation and
Regression
a 1
of terms of the Y on X line we obtain X i    Yi . Thus we should have
b b
a 1
 and   . However, the observations are not on a straight line and the
b b
relation between X and Y is not a mathematical one. You may recall that
estimates of the parameters are obtained by the method of least squares. Thus the
regression line Yˆi  a  bX i is obtained by minimising  (Yi  a  bX i ) 2 whereas
i

the regression line Xˆ i    Yi is obtained by minimising  ( X i    Yi ) 2 .

i

However, there is a relationship between the two regression coefficients b and  .


We have noted earlier that b  xy2 . By a similar formula by interchanging the
x
 xy
roles of X and Y we find   2 . But by definition we notice that  xy   yx .
y

 xy2
Thus b    , which is the same as r2.
 x2   2y

This r2 is called the coefficient of determination. Thus the product of the two
regression coefficients of Y on X and X on Y is the square of the correlation
coefficient. This gives a relationship between correlation and regression. Notice,
however, that the coefficient of determination of either regression is the same,
i.e., r2; this means that although the two regression lines are different, their
predictive powers are the same. Note that the coefficient of determination r2
ranges between 0 and 1, i.e., the maximum value it can assume is unity and the
minimum value is zero; it cannot be negative.
From the previous discussions, two points emerge clearly:
1) If the points in the scatter lie close to a straight line, then there is a strong
relationship between X and Y and the correlation coefficient is high.
2) If the points in the scatter diagram lie close to a straight line, then the observed
values and predicted values of Y by least squares are very close and the
prediction errors (Yi  Yˆi ) are small.

Thus, the prediction errors by least squares seem to be related to the correlation
coefficient. We explain this relationship here. The sum of squares of errors at the
various points upon using the least squares linear regression is  Yi  Yˆi
n

i 1
 
2
.

On the other hand, if we had not used the value of observed X to predict Y, then
the prediction would be a constant, say, a. The best value of a by least squares
n
criterion is such an a that minimises  Yi  a  ; the solution to this a is seen to be
2

i 1

Y . Thus the sum of squares of errors of prediction at various points without using
X is  Yi  Y  .
n 2

i 1

121
Summarisation of n n
Bivariate and Multi- The ratio,  (Yi  Yˆi ) 2  (Yi  Y ) 2 can then be used as an index of how much has
i 1 i 1
variate Data
been gained by the use of X. In fact, this ratio is the coefficient of determination
and same as r 2 mentioned above. Since both the numerator and denominator of
this ratio are non-negative, the ratio is greater than or equal to zero.

Check Your Progress 3

1) From the following data find the coefficient of linear correlation between X
and Y. Determine also the regression line of Y on X, and then make an
estimate of the value of Y when X = 12.
X 1 3 4 6 8 9 11 14
Y 1 2 4 4 5 7 8 9
.............................................................................................................................
…………………………………………………………………....………….....
………………………………………………………………….......…………..
………………………………………………………………….......……..……
………………………………………………………………….......………..…
………………………………………………………………….......……..……

2) Obtain the lines of regression for the following data:

(X) 1 2 3 4 5 6 7 8 9
(Y) 9 8 10 12 11 13 14 16 15
…………………………………………………………………....………….....
………………………………………………………………….......…………..
………………………………………………………………….......……..……
………………………………………………………………….......………..…
………………………………………………………………….......……..……
3) Find the two lines of regression from the following data:
Age of Husband (X) 2 2 2 2 3 2 2 4 2 1
5 2 8 6 5 0 2 0 0 8
Age of Wife (Y) 1 1 2 1 2 1 1 2 1 1
8 5 0 7 2 4 6 1 5 4
Hence estimate (i) age of husband when the age of wife is 19, (ii) the age of
wife when the age of husband is 30.
…………………………………………………………………....………….....
………………………………………………………………….......…………..
………………………………………………………………….......……..……
………………………………………………………………….......………..…
………………………………………………………………….......……..……
122
4) From the following data, obtain the two regression equations : Correlation and
Regression
Sales : 91 97 108 121 67 124 51 73 111 57
Purchases : 71 75 69 97 70 91 39 61 80 47
…………………………………………………………………....………….....
………………………………………………………………….......…………..
………………………………………………………………….......……..……
………………………………………………………………….......………..…
………………………………………………………………….......……..……

5) Obtain the equation of the line of regression of yield of rice (y) on water (x)
from the data given in the following table :
Water in inches (x) 12 18 24 30 36 42 48
Yield in tons (y) 5.27 5.68 6.25 7.21 8.02 8.71 8.42
Estimate the most probable yield of rice for 40 inches of water.
…………………………………………………………………....………….....
………………………………………………………………….......…………..
………………………………………………………………….......……..……
………………………………………………………………….......………..…
………………………………………………………………….......……..……

5.13 MULTIPLE REGRESSION

So far we have considered the case of the dependent variable being explained by
one independent variable. However, there are many cases where the dependent
variable is explained by two or more independent variables. For example, yield
of crops (Y) being explained by application of fertilizer (X 1 ) and irrigation
water(X 2 ). This sort of models is termed multiple regression. Here, the equation
that we consider is
Y     X1   X 2  e …(5.21)

Where Y is the explained variable, X 1 and X 2 are explanatory variables, and e is

the error term. In order to make the presentation simple we have dropped the
subscripts. A regression equation can be fitted to (5.21) by applying the method
of least squares. Here also we minimise  e 2 and obtain the normal equations as
follows:
Y  n  X 1  X 2
X 1Y  X 1  X 12  X 1 X 2 … (5.22)
X 2Y  X 2  X 1 X 2  X 22

123
Summarisation of By solving the above equations we obtain estimates for α, β and γ. The regression
Bivariate and Multi-
variate Data equation that we obtain is
Yˆ    X 1  X 2 …(5.23)

Remember that we obtain predicted or forecast values of Y (that is Yˆ ) through

(5.23) by applying various values for X1 and X 2 .

In the bivariate case (Y,X) we could plot the regression line on a graph paper.
However, it is quite complex to plot the three variable case (Y, X 1 , X 2 ) on graph
paper because it will require three dimensions. However, the intuitive idea
remains the same and we have to minimise the sum of errors. In fact when we
add all the error terms ( e1 , e2 ,........en ) it sum up to zero.

In many cases the number of explanatory variables may be more than two. In
such cases we have to follow the basic principle of least squares: minimize e2 .
Thus if Y  a0  a1 X1  a2 X 2  ...............  an X n  e then we have to minimize

e 2   (Y  a0  a1 X 1  a2 X 2 ........  an X n ) 2

and find out the normal equations.

Now a question arises, ‘How many variables should be added in a regression
equation?’ It depends on our logic and what variables are considered to be
important. Whether a variable is important or not can be identified on the basis of
statistical tests also. These tests will be discussed later in Block 4.
We present a numerical example of multiple regression below.
Example 5.3
A student tries to explain the rent charged for housing near the University. She
collects data on monthly rent, area of the house and distance of the house from the
university campus and fits a linear regression model.

Rent (in Rs.’000) Area (in sq.mt.) distance(in Km.)

Y X1 X2
20 65 5.7
25 66 3.2
26 70 7.5
28 70 6.5
30 75 5.0
31 76 4.0
32 72 6.0
33 75 6.2
35 78 3.5
40 103 2.4
In the above example rent charged (Y) is the dependent variable while area of the
house ( X 1 ) and distance of the house from the university campus ( X 2 ) are
independent variables.

124
The steps involved in estimation of regression line are: Correlation and
Regression
i) Find out the regression equation to be estimated. In this case it is given by
Y    X 1  X 2  e .

ii) Find out the normal equations for the regression equation to be estimated.
In this case the normal equations are
Y  n  X 1  X 2
X 1Y  X 1  X 12  X 1 X 2
X 2Y  X 2  X 1 X 2  X 22

iii) Construct a table as given in Table 9.4.

iv) Put the values from the table in the normal equations.
v) Solve for the estimates of  ,  and  .

Table 5.7: Computation of Multiple Regression

Y X1 X2 X 1Y X 2Y X 12 X 22 X1X 2 Yˆ ei

20 65 5.7 1300 114 4225 32.49 370.5 25.49 -5.49

25 66 3.2 1650 80 4356 10.24 211.2 25.71 -0.71

26 70 7.5 1820 195 4900 56.25 525 27.94 -1.94

28 70 6.5 1960 182 4900 42.25 455 27.85 0.15

30 75 5 2250 150 5625 25 375 30.00 0.00

31 76 4 2356 124 5776 16 304 30.37 0.63

32 72 6 2304 192 5184 36 432 28.72 3.28

33 75 6.2 2475 204.6 5625 38.44 465 30.11 2.89

35 78 3.5 2730 122.5 6084 12.25 273 31.24 3.76

40 103 2.4 4120 96 10609 5.76 247.2 42.58 -2.58

300 750 50 225000 15000 562500 2500 37500 300 0

By applying the above mentioned steps we obtain the estimated regression line as
Yˆ  4.80  0.45 X 1  0.09 X 2 .

125
Summarisation of
Bivariate and Multi- 5.14 NON-LINEAR REGRESSION
variate Data
The equation fitted in regression can be non-linear or curvilinear also. In fact, it
can take numerous forms. A simpler form involving two variables is the
quadratic form. The equation is
Y = a + bX + cX 2
There are three parameters here viz., a, b and c and the normal equations are:
Y  n  bX  cX 2
XY  X  bX 2  cX 3
X 2Y  X 2  bX 3  cX 4
By solving for these equation we obtain the values of a, b and c.
Certain non-linear equations can be transformed into linear equations by taking
logarithms. Finding out the optimum values of the parameters from the
transformed linear equations is the same as the process discussed in the previous
section. We give below some of the frequently used non-linear equations and the
respective transformed linear equations.
1) Y = a c bx
By taking natural log (ln), it can be written as
ln Y = ln a + bX
or Y’ =  + X’
Where, Y’ = lnY,  = ln a, X’ = X and  = b
2) Y = aX b
By taking logarithm (log), the equation can be transformed into
log Y = log a + b log X
or Y’ =  + X’
where, Y’ = log Y,  = log a,  = b and X’ = log X
1
3) Y=
a  bX
1
If we take Y’ = then
Y
Y’ = a + bX
4) Y=a+b X
If we take X’ = X then
Y = a + bX’
Once the non-linear equation is transformed, the fitting of a regression line is as
per the method discussed in the beginning of this Unit.

126
We derive the normal equations and substitute the values calculated from the Correlation and
Regression
observed data. From the transformed parameters, the actual parameters can be
obtained by making the reverse transformation.

Check Your Progress 4

1) Using the data on scores in Statistics and Economics of Table 5.1, compute
the regression of y on x and x on y and check that the two lines are different.
On the scatter diagram, plot both these regression lines. Check that the
product of the regression coefficients is the square of the correlation
coefficient.
…………………………………………………………………....………….....
………………………………………………………………….......…………..
………………………………………………………………….......……..……
………………………………………………………………….......………..…
………………………………………………………………….......……..……

2) Suppose that the least squares linear regression of family expenditure on

clothing (Rs. y) on family annual income (Rs. x) has been found to be y = 100
+ 0.09x, in the range 1000 < x <100000. Interpret this regression line. Predict
the expenditure on the clothing of a family with an annual income of Rs.
10,000. What about families with annual income of Rs. 100 and Rs.
10,00,000?
…………………………………………………………………....………….....
………………………………………………………………….......…………..
………………………………………………………………….......……..……
………………………………………………………………….......………..…
………………………………………………………………….......……..……

5.15 LET US SUM UP

In this Unit we discussed an important statistical tool, that is, regression. In
regression analysis we have two types of variables: dependent and independent.
The dependent variable is explained by independent variables. The relationship
between variable takes the form of a mathematical equation. Based on our logic,
understanding and purpose of analysis we categorise variables and identify the
equation form.
The regression coefficient enables us to make predictions for the dependent
variable given the values of the independent variable. However, prediction
remains more or less valid within the range of data used for analysis. If we
attempt to predict for far off values of the independent variable we may get
insensible values for the dependent variable.

127
Summarisation of
Bivariate and Multi- 5.16 ANSWERS/HINTS TO CHECK YOUR
variate Data PROGRESS EXERCISES
Check Your Progress 1

1) + 0.47

2) + 0.996

3) + 0.98

4) + 0.995

5) – 0.84

Check Your Progress 2

1) 2/3

2) + 0.64

3) – 0.21, + 0.64, –0.30

4) + 0.82

Check Your Progress 3

1) + 0.98 ; y = 0.64x + 0.54; 8.2
2) x = 0.95y  6.4 ; y = 0.95x + 7.25
3) x = 2.23y  12.70 ; y = 0.39x + 7.33
(i) 29.6 (ii) 18.9
4) y = 0.613x + 14.81 ; x = 1.360y  5.2
5) y = 3.99 + 0.103x ; 8.11 tons
Check Your Progress 4
1) (i) y = a + bx = 5.856 + 0.676x
(ii) x =  + y = 29.848 + 0.799y
(iii) r = 0.73
(iv) 0.676  0.799 = 0.54
2) Expenditure on clothing, when family income is Rs. 10,000, is Rs. 1,000. In
case of income below 1,000 or above 1,00,000 the regression line may not
hold good. In between both these figures, one rupee increase in income
increases expenditure on clothes by 9 paise.

128
UNIT 6 INDEX NUMBERS 
Structure
6.0 Objectives
6.1 Introduction
6.2 Steps in Construction of Index Numbers
6.2.1 Selection of Base Period
6.2.2 Choice of a Suitable Average
6.2.3 Selection of Items and their Numbers
6.2.4 Collection of Data
6.3 Method of Construction of Index Number
6.3.1 Relative Methods
6.3.2 Aggregative Methods
6.3.3 Quantity or Volume Index Numbers

6.4 Merits of the Various Aggregative Measures

6.5 Tests for Index Numbers
6.5.1 Time Reversal Test
6.5.2 Factor Reversal Test
6.5.3 Chain Index Number and Circular Test

6.6 Cost of Living Index Number (CLI) or Consumer Price Index Number (CPI)
6.7 Worked-Out Examples
6.8 Let Us Sum Up
6.9 Answers or Hints to Check Your Progress Exercises

6.0 OBJECTIVES
After going through this Unit, you will be able to:
 define index numbers; and
 Construct and calculate them.

6.1 INTRODUCTION
An “index” in the common sense of the word is an “indicator” and not anything
more than that. “Index numbers” or “indices” are forms of the plural, but they all
mean the same thing.
An index number represents the general level of magnitude of the changes
between two (or more) periods of time or places, in a number of variables taken
as a whole. In this definition, the world “variable” refers to numerical variables
which can be measured in quantity, such as the prices of commodities.


Adapted from IGNOU study material of EEC 13: Elementary Statistical Methods and Survey
Techniques, Unit 10 written by J Roy with modifications by Kaustuva Barik
Summarisation of For example, we may like to compare the price level of an article between 2010
Bivariate and Multi-
variate Data and 2020 or between Mumbai and Kolkata. Let us consider the yield of rice in
2015 and in 2020 as 50,000 and 60,000 tons respectively. The year 2015 is taken
as base for comparison of yields that is 2015 = 100. The corresponding figure for
,
2020 will be , × 100 = 120. This is a single-commodity index number in its
simplest form, being just a relative number. In practice, however, we deal usually
with a number of commodities for the construction of an index.
Index numbers are ratios that are usually expressed as percentage in order to
avoid awkward decimals. Thus if one commodity costs Rs. 45 in 2019 and Rs.
150 in 2020 the ratio would be or 3.33. If instead of this, we express the ratio
into a percentage × 3.33, we say that the index is 333, based on 2019, which
is 100.

6.2 STEPS IN CONSTRUCTION OF INDEX NUMBERS

Many government and private agencies are engaged in computation of index
numbers or indices as they are often required for the purpose of forecasting
business and economic conditions, providing general information, etc.
It is not always the case that the comparison should be over time, but most
common types of index numbers measure changes over time. Similarly, index
numbers may be constructed for studying changes in any variable such as
intelligence, aptitude, efficiency, production, etc. However, the time series of
prices is frequently used. Therefore, in the discussion below we will concentrate
mostly on prices of commodities. The principles of construction are, however,
quite general in nature, and may thus be applied to other areas of interest.
There are various uses of price index numbers. The wholesale price index
number indicates the price changes taking place n wholesale markets. In the other
hand, the consumer price index number or the cost of living index number tells us
about the changes in the prices faced by an individual consumer. Its major
application is in the calculation of dearness allowance so that real wage does not
decrease; or in comparing the cost of living in, say, different regions. It is also
used to measure change in purchasing power of money. The reciprocal of a
general price index is known as purchasing power of money with reference to the
base period. For example, if the price index number goes up to 150, it means that
the same amount of money will be able to purchase 100/150 = 0.67 times or 67%
of the volume of goods being purchased in the base period.
6.2.1 Selection of Base Period
Since index numbers measure relative changes, they are expressed with one
selected situation (e.g., period, place, etc.) as 100. This is called the base or the
starting point of the series of index numbers. For example, a date is first chosen
and all changes are measured from it. The base may be one day such as with
index or retail prices, the average of a year, or the average of a period. While
selecting a base period the following aspects should be taken into consideration.
130
 1) The base period must be “normal” in the sense that the data chosen are Index Numbers

not affected by any irregular or abnormal situations such as natural

calamities, war, etc. It is desirable to restrict comparisons to stable
periods for achieving accuracy.
2) It should not be too back-dated as the patterns of trade, imports or
consumer preferences may change considerably if the time-span is too
long. A five to ten year interval is likely to be suitable for one base date,
and after that the index becomes more and more outdated. Greater
accuracy is attained for moderate short-run indices than for those
covering greater span of time.
3) For indices dealing with economic data, the base period should have some
economic significance.
6.2.2 Choice of a Suitable Average
An index number is basically the result of averaging a series of data (e.g., price-
relatives of several commodities). There are, however, several ways of a
averaging a series: mean (i.e., arithmetic mean), mode median, geometric mean
and harmonic mean.
The question naturally arises as to which average to chose. The mode has the
merit of simplicity, but may be indefinite. The median suffers from the same
limitations. Moreover, neither of them takes into account the size of the items at
each end of a distribution. The harmonic mean has very little practical
applications to index numbers. As a result, mode, median and harmonic mean are
not generally used in the calculation of index numbers. Thus, the arithmetic mean
is most commonly used. However, the geometric mean is sometimes used despite
its slight difficulty in calculation.
6.2.3 Selection of Items and their Numbers
The number and kinds of commodities to be included in the construction of an
index numbers depend on the particular problem to be dealt with, economy and
ease of calculations. Various practical considerations determine the number and
kinds of items to be taken into account. For a wholesale price index, the number
of commodities should be as large as possible. On the other hand, for an indexed
meant to serve as a predictor of price movement rather than an indicator
movement rather than an indicator of changes over time, a much smaller number
of items may be adequate. Care should, however, be taken to ensure that items
chosen are not too few which make the index unrepresentative of the general
level. A fixed set of commodities need not also be used for a very long period as
some items lose their importance with the passage of time and some new items
gain in significance. In general, the commodities should be sensitive and
representative of the various elements in the price system.

131
Summarisation of 6.2.3 Collection of Data
Bivariate and Multi-
variate Data As prices often vary from market to market, they should be collected at regular
intervals from various representative markets. It is desirable to select shops
which are visited by a cross section of customers. The reliability of the index
depends greatly on the accuracy of the quotations given for each constituent item.

6.3 METHOD OF CONSTRUCTION OF INDEX

NUMBER
Various methods of construction of index numbers are as follows:

1) Relative methods

a) Simple average of relatives

b) Weighted average of relatives

2) Aggregative methods

a) Simple aggregative formula

b) Weighted aggregative formula

i) Lasperyres’ index

ii) Paasche’s index

iii) Edgeworth-Marshall’s index

iv) Fisher’s Ideal index.

6.3.1 Relative Methods
If we record prices of a variety of commodities at a given date and at a later date
record the prices of similar items, the change in price can be simply expressed as
a percentage of the new compared with the old for each commodity.
This provides us with price relatives and if weights are available the next step
will be to multiply the relatives by the weights. Finally, an index number can be
produced if we add together the weighted relatives and calculate an average. It is
unrealistic to assume that the consumption of each commodity has been equal. So
most indices take account of the proportions of each item actually used. This
method to weighting shows the relative importance of each in the series.
Given k commodities with base year prices of
Po1, Po2,. . . . Pok,
and current prices of
Po1, Po2,. . . . Pok,
the price relative for the ith commodity will be where i = 1, 2, ……, k and the
subscript 0 refers to the base year and subscript n refers to the current year.

132
a) Simple Average of Relatives Index Numbers

The arithmetic mean of the price relative is given by

Index = 100 ∑ …(6.1)

For simplicity we can omit the subscript ‘i' and write

index = 100 ∑

b) Weighted Average of Relatives

The most suitable weights to use are the value of each item, which is denoted by
wi for the i-th commodity. We may use the value of the base year quantities sold
at the base year prices (wi = poi qoi) or current year quantities sold at current
prices (w1i = p1i q1i)
If we omit the subscript ‘i’ for simplicity, a weighted arithmetic mean of price
relatives using base year values as weights is given by
∑ ×
index = ∑
× 100 …(6.2)

You should note that the base year weighting preserves continuity, but it loses
“up-to-dateness” in the course of time.
Example 6.1: The table below presents the average fares per railway journey.
Using 2010 average = 100, calculations are made according to base year weights.

Class of No. of passenger Fare Weights Price 𝑃. 𝑤

Ticket journeys in 2010 relatives
𝑤 =𝑝 𝑞
(in millions) 2010 2020 𝑃
P= . 100
𝑃

Full Fare 23 12 60 276 500 138000

Excursions 25 6 30 150 500 75000
Festival 20 4 15 80 375 30000
Season 32 5 14 160 280 44800
tickets
Total 666 287800

Applying formula (6.2), we get

Index for 2020 = = 432.13.

Using Current year values (𝒘𝒏 = 𝑷𝒏 𝒒𝒏 ) as weights, the index is given by

𝑃
∑
𝑃 ×𝑤
index = × 100 … (6.3)
∑𝑤

133
Summarisation of Example 6.2: The table below shows the average fares per railway journey.
Bivariate and Multi-
variate Data Using 2010 average = 100, calculation are made according to current year
weights.

Class of No. of Fare Weights Price 𝑃. 𝑤

Ticket passenger relatives
journeys in 2010 2020 𝑤 = 𝑝 𝑞 𝑃
P= . 100
2010 (in 𝑃
millions)

Full Fare 25 12 60 1500 500 750000

Excursions 26 6 30 780 500 390000
Festivals 9 4 15 135 375 50630
Season-
27 5 14 378 280 105800
tickets
Total 2793 1296430
Applying formula (6.3), we get
1296430
index = = 464.17
2793
6.3.2 Aggregative Methods
In this method, the aggregate (sum-total) of the prices of all commodities in the
current or given year is expressed as a percentage of the same in the base year.
Thus, in the case of simple aggregative index, we have:
aggregate prices in current year
index number = × 100
aggregate prices in the base year

𝑃 + 𝑃 + … … … .. 𝑃
= × 100
𝑃 + 𝑃 + … … … .. 𝑃
∑𝑃 ∑𝑃
× 100 = × 100 … (6.4)
∑𝑃 ∑𝑃
where the summation symbol ∑ extends over all the selected commodities
numbering k. On the other hand, in the case of weighted aggregative index we
have,

𝑝 + 𝑝 𝑞 + … … … .. 𝑝 𝑞
General index = × 100
𝑝 𝑞 + 𝑝 𝑞 + … … … .. 𝑝 𝑞

∑𝑝 𝑞
× 100
∑𝑝 𝑞
∑𝑝 𝑞
or simply = × 100 … (6.5)
∑𝑝 𝑞
The weights used should be actual quantities bought or sold, and these are kept
unchanged until such time as the index requires to be revised.
134
There are many formulae for weighted aggregative index, but depending on the Index Numbers
type of weights used, we discuss four indices which are commonly used.
a) Laspeyres’ index
If we use base period quantities (q0) as the weights in the general weighed
aggregative index formula (6.5), we get what is known as Laspeyre’s formula
(L).
∑
𝐿 = ∑
× 100 … (6.6)

It can be seen that this index has fixed base year quantity as weights (𝑞 ) and is
equivalent to arithmetic mean of price relatives given at formula (6.2). Thus, we
can also write (6.6) as
∑ ×
𝐿 = ∑
× 100

b) Paasche’s index
If we use current year quantities (𝑞 ) as weights in the general aggregative index
formula(6.5), we get what get what is known as Paasche’s formula(P).
∑
𝑃 = ∑
× 100 … (6.7)

Where 𝑞 (actually 𝑞𝑛1 , 𝑞𝑛2 , … . . 𝑞𝑛𝑘 )are the quantities bought or sold in the
current period.
c) Fisher’s Ideal Index
An index number obtained as geometric mean (i.e., square root of the product) of
indices obtained by Laspeyres’ and Paasche’s formulae, satisfies certain
important properties (to be discussed later), is known as the Fisher’s ideal
formula.
∑ 𝑝𝑛 𝑞0 ∑ 𝑝𝑛 𝑞𝑛
𝐹 = √𝐿 × 𝑃 = ∑ 𝑝0 𝑞0
×∑ × 100 … (6.8)
𝑝0 𝑞𝑛

d) Edgeworth-Marshall Index
If the mean of the base period and the current period quantities is used as weight,
i.e.,
𝑤 = (𝑞0 + 𝑞𝑛 ), we get a compromise formula of ‘Edgeworth-Marshall index’.
∑ 𝑝 (𝑞 + 𝑞 )/2
𝐼 = × 100
∑ 𝑝 (𝑞 + 𝑞 )/2

∑ ( )
= ∑
× 100 … (6.9)
( )

We take some hypothetical data and calculate the above indices from it (see
Table 6.1).

135
Summarisation of Table 6.1: Illustrative Calculation of Laspeyres’
Bivariate and Multi-
variate Data Edgeworth-Marshall]s and Fisher’s Indices

Item Base Year Current Year

(𝑝 𝑞 ) (𝑝 𝑞 ) (𝑝 𝑞 ) (𝑝 𝑞 )
Price Quantity Price Quantity
(𝑝 ) (𝑞 ) (𝑝 ) (𝑞 )

A 20 7 25 9 140 175 180 225

B 42 6 40 8 252 240 336 320
C 30 17 25 4 510 425 120 100
D 8 15 14 10 120 210 80 140
E 10 8 13 5 80 104 50 65
Total 1102 1154 766 850
From the above table we calculate various indices as follows:
∑ 𝑃𝑛 𝑞0
1) Lasperyres’ price index = ∑ × 100 = × 100 =
𝑃𝑛 𝑞0

104.72 = 105

∑ 𝑃𝑛 𝑞𝑛
2) Paasche’s price index = × 100 = × 100 =
𝑃0 𝑞𝑛

110.97 = 111

∑ 𝑃𝑛 𝑞0 +∑ 𝑃𝑛 𝑞𝑛
3) Edgewroth-Marshall’s price index = ∑ ×
𝑃0 𝑞0 +∑ 𝑃0 𝑞𝑛

100 = × 100 =

× 100 = 107.28 = 107

∑ 𝑃𝑛 𝑞0 ∑ 𝑃𝑛 𝑞𝑛
4) Fisher’s ideal index = ∑ 𝑃0 𝑞0 ∑ 𝑃0 𝑞𝑛
× 100 =

[(𝐿) × (𝑃)] =
(104.72 × 110.97) = 107.8 =
108
Note that for the same price change, different formulae provide different values.
Moreover, when prices are increasing, Laspeyres’ index gives the lowest value
while Paasche’s index gives the highest value. Therefore, it is often said that
Laspeyres’ index is an under-estimate while Paasche’s index is an over-estimate
of true price change.
6.3.3 Quantity or Volume Index Numbers
We can get a quantity or volume index number, which measures and permits
136
comparison of quantities of goods, from corresponding price index number Index Numbers
formulae simply by replacing p. by q and q and q by p.
1) Quantity relative = × 100

2) Arithmetic Mean (A.M.) of quantity of relatives = 100 ∑ /𝑘

3) Weighted A.M. of quantity relative index:

∑ 𝑞𝑛 /𝑞0
a) Base year weights: ∑
× 100 where 𝑤0 = 𝑝0 𝑞0
∑ 𝑞𝑛 /𝑞0 ×𝑤𝑛
b) Current year weights: ∑ 𝑤𝑛
× 100 where 𝑤𝑛 = 𝑝𝑛 𝑞𝑛
∑ 𝑞𝑛
4) Simple aggregative quantity index = ∑ × 100
𝑞0
∑ 𝑞𝑛 𝑃 0
5) Laspeyres’ quantity index = ∑ × 100
𝑞0 𝑃 0
∑ 𝑞𝑛 𝑃𝑛
6) Paasche’s quantity index = ∑ × 100
𝑞0 𝑃𝑛

∑ 𝑞𝑛 𝑃0 ∑ 𝑞𝑛 𝑃𝑛
7) Fisher’s ideal index = ∑ 𝑞0 𝑃0 ∑ 𝑞0 𝑃𝑛
× 100
∑ 𝑞𝑛 (𝑃0 +𝑃𝑛 )
8) Edgeworth-Marshall’s index = ∑ 𝑞0 (𝑃0 +𝑃𝑛 )
× 100

Check Your Progress 1

1) What do index numbers seek to measure?
………………………………………………………………………………....
……………………………………………………………………………....…
…………………………………………………………………………………
…………………………………………………………………………………
……………………………………………………………………………....…
..........................................................................................................................

2) Discuss the various problems involved in construction of index numbers with

particular reference to price indices.
………………………………………………………………………………....
……………………………………………………………………………....…
…………………………………………………………………………………
…………………………………………………………………………………
……………………………………………………………………………....…

3) The following are the prices of six different commodities for 2020 and 2021.
Compute the price index by (a) aggregative method, and (b) average of
price relatives method by using arithmetic mean.
137
Summarisation of
Commodities Price in 2020 (Rs.) Price in 2021 (Rs.)
Bivariate and Multi-
variate Data
A 40 50
B 50 60
C 20 30
D 50 70
E 80 80
F 100 110
………………………………………………………………………………....…
…………………………………………………………………………....………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………

4) Calculate Fisher’s Ideal Index Number from the following group of items.
Base Year Current Year
Item No. Price Quantity Price Quantity
(in Rs.) (in Kg.) (in Rs.) (in Kg.)
1 4 1.0 3 4
2 8 1.5 7 5
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………

5) Calculate Laspeyres’ and Paasche’s Index Number from the following data.

Base Year Current Year

Item Quantity Price Quantity Price

Bread 6.0 40 7.0 30

Meat 4.0 45 5.0 50

Tea 0.5 90 1.5 40

……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
……………………………………………………………………………………
138
 Index Numbers
6.4 MERITS OF THE VARIOUS AGGREGATIVE
MEASURES
The different index numbers serve different purposes and, therefore, the
appropriateness of a particular index number depends on the purpose at hadn.
The Laspeyres’ index calculation is simpler, since this uses the base period
quantities as weighs which are not difficult to get and the denominator needs
calculating only once. But in this index a rise in pries tends to overstated, since
it does not take into account corresponding falls in demand or changes in output.
Indices such as Paasche’s on the other hand, use current period quantities as
weights which are difficult o get and the weights need to be constructed afresh
for every year. Moreover, Paasche’s index tends to understate the rise in prices
because it uses current weights. The Laspeyres’ index is probably more
commonly used, since it is convenient to employ fixed weights. But with the
passage of time the weights are rendered out of date. For example, in 1995 the
number of mobile phones in Odisha was nil. In 2020 there are more mobile
phones than the number of land line connections. The Paasche’s index uses the
preferable current weights, but since-up-to-date information on quantity of goods
produced or consumed or marketed or distributed are not readily obtained, the
Laspeyres’ index has a great advantage.

6.5 TESTS FOR INDEX NUMBERS

A perfect index number, which measures the change in the level of a
phenomenon from one period to another, should satisfy certain tests. There are
three major tests of index numbers: (1) Time reversal test, (2) Factor reversal test,
and (3) Circular test.
6.5.1 The Time Reversal Test
According to this test, if we reverse the time subscripts (such as 0 and n) of a
price (or quantity) index the result should be the reciprocal of the original index.

Symbolically,

I0n × I0n = 1

where I0n = index number for period n with the base period 0.

In0 = index number for period 0 with the base period n.

If from 2010 to 2020 the price changes from Rs. 4 to Rs. 16, the price in 2020 are
400 percent of the price in 2010, and the price in 2010 is 25 percent of the price
in 2020. The product of the two price relatives is 4×0.25 = 1. The test is based on
the analogy that the principle, which hold good for a single commodity, should,
also be true for the index number as a whole.

139
Summarisation of There are five methods which do satisfy the time reversal test. These are:
Bivariate and Multi-
variate Data 1) Simple geometric mean of price relative
2) Aggregative indices with fixed weights
3) Edgeworth-Marshall formula
4) Weighted geometric mean of price relatives if fixed weights are used
5) Fisher’s ideal index
∑ 𝑃𝑛 𝑞0 ∑ 𝑃𝑛 𝑞𝑛
Fisher’s ideal index F = ∑ 𝑃0 𝑞0
× ∑ 𝑃0 𝑞𝑛

If time subscripts are reversed,

∑ 𝑃0 𝑞𝑛 ∑ 𝑃0 𝑞0
F’ = ∑ 𝑃𝑛 𝑞𝑛
× ∑ 𝑃𝑛 𝑞0

Since F × F’ = 1 the test is satisfied

6.5.2 The Factor Reversal Test

With the usual notations, a “value index” formula is given by

∑ 𝑃𝑛 𝑞𝑛
Iv = ∑
𝑃0 𝑞𝑛

Now, for example, Laspeyres’ index for prices and quantities are given
respectively by
∑ 𝑃𝑛 𝑞0 ∑ 𝑞𝑛 𝑃0
Ip = ∑ and Iq = ∑
𝑃0 𝑞𝑛 𝑞0 𝑃0

The factor reversal test desires that Ip. Iq = Iv

But for Laspeyres’ index

∑(𝑃𝑛 𝑞0 ) ∑ 𝑃𝑛 𝑞0
Ip.Iq = ∑(𝑃0 𝑞0 )2
= Iv

On the other hand, Fisher’s ideal index satisfies this test, as shown below.

∑ 𝑃𝑛 𝑞0 ∑ 𝑃𝑛 𝑞𝑛
Ip = ∑ 𝑃𝑛 𝑞0
×∑
𝑃0 𝑞𝑛

∑ 𝑞𝑛 𝑃0 ∑ 𝑞𝑛 𝑃𝑛
Iq = ∑ 𝑞𝑛 𝑃0
×∑
𝑞0 𝑃𝑛

∑ 𝑃𝑛 𝑞0 ∑ 𝑃𝑛 𝑞𝑛 ∑ 𝑞𝑛 𝑃0 ∑ 𝑞𝑛 𝑃𝑛
Ip.Iq = ∑ 𝑃0 𝑞0
×∑ ×∑ ×∑
𝑃0 𝑞𝑛 𝑞0 𝑃0 𝑞0 𝑃𝑛

∑ 𝑃𝑛 𝑞𝑛 ∑ 𝑃𝑛 𝑞𝑛 ∑ 𝑃𝑛 𝑞𝑛
= ∑ 𝑃0 𝑞0
×∑ =∑ = Iv
𝑞0 𝑃0 𝑃0 𝑞0

140
To understand this principle further, we take the following example. Index Numbers

If the price and quantity per unit of an item changed in 2020, as compared to
2010, from Rs. 16 to Rs. 32 and from 100 units to 200 units respectively, then the
price and quantity in 2020 would both be 200% or 2.00 times the price and
quantity in 2010. The values (product of price and quantity) would be Rs. 1600 in
2010 and Rs. 6400 in 2020, so that the value ratio is 6400/1600 = 4.00 Thus, we
verify that 2.00 × 2.00 = 4.00, that is, the product of price ratio and quantity ratio
is equal to the value ratio.

Only the Fisher’s ideal index satisfies this test.

Example 6.3 we show with the following data that the Fisher’s ideal index
satisfies the factor reversal test:

Item Base Year Current Year

(𝑝 𝑞 ) (𝑝 𝑞 ) (𝑝 𝑞 ) (𝑝 𝑞 )
Price Quantity Price Quantity
(𝑝 ) (𝑞 ) (𝑝 ) (𝑞 )

I 6 50 10 56 300 500 336 560

II 2 100 2 120 200 200 240 240

III 4 60 6 60 240 360 240 360

IV 10 30 12 24 300 360 240 288

V 8 40 12 36 320 480 288 432

Total 1360 1900 1344 1880

Let us calculate the following from the data given in the above table.

∑ 𝑃𝑛 𝑞0 ∑ 𝑃𝑛 𝑞𝑛
Price Ratio: Ip = ∑ 𝑃𝑛 𝑞0
×∑ = ×
𝑃0 𝑞𝑛

∑ 𝑞𝑛 𝑃0 ∑ 𝑞𝑛 𝑃𝑛
Quantity Ratio: Ip = ∑ 𝑞𝑛 𝑃0
×∑ = ×
𝑞0 𝑃𝑛

∑ 𝑃𝑛 𝑞𝑛
Value Ratio: Iv = ∑ =
𝑃0 𝑞0

Ip.Iq = × × × = × =

= Iv which shows that the test is satisfied.

141
Summarisation of 6.5.3 Chain Index Number and Circular Test
Bivariate and Multi-
variate Data Two types of base periods are used for the construction of index numbers,
namely, (a) fixed base, (b) chain base. Most commonly used indices use fixed
base method. This method cannot take into account any changes in price or
quantity in any other year. It fails to include new commodities gaining
importance at a later date or exclude commodities losing significance in course of
time. These problems can be overcome by chain index numbers.
Using a suitable index number formula (say, Laspeyres’ index), link indices,
defined as follows, are first calculated: Link index = Index number with previous
period as base. The chain index is obtained by multiplying link indices
progressively. Thus, the chain index number Ion for period n with base period 0 is
given by

I01 = I01

I02 = I01 × I12

I03 = I01 × I12 × I23 = I02× I23

……………………………..

…………………………….

I0n = I01 × I12 × I23 ……..× I(n-1)n = I(n-1) × I(n-1)n

Example 6.4 The calculation of chain index numbers is illustrated with reference
to the following data:
Year Link index Chain index (Base 2010 =100)

2010 100 100

2011 I01 = 80 100× = 80

2012 I12 = 120 80× = 96

2013 I23 = 75 96× = 72

Thus, the chain index numbers for the years 2011 to 2013 with 2010 as the base
are 80, 96 and 72 respectively.

Circular Test: The circular test is an extension of time reversal test over a
number of years. It states that the chain index for the year 2013, calculated above,
starting from the base year 2010 will be same as the index number directly
calculated with fixed base period of 2010. In symbols,

142
I01= I12 × ……..× I(n-1)n ×= In-0) = 1. (Notice that I0n = Index Numbers

Considering an aggregate index with fixed weights

∑ 𝑃1 𝑞
∑ 𝑃0 𝑞

We can illustrate the test as follows:

With base period 0, we can trace the above formula from 1 to 3 years:

∑ 𝑃1 𝑞 ∑ 𝑃2 𝑞 ∑ 𝑃3 𝑞 ∑ 𝑃0 𝑞
× × × =1
∑ 𝑃0 𝑞 ∑ 𝑃1 𝑞 ∑ 𝑃2 𝑞 ∑ 𝑃3 𝑞

The formulae satisfying the requirements of circular test are:

1) Simple aggregative index

2) Simple geometric mean of relatives

3) Weighted aggregative index (such as Laspeyres’ index with constant

weights)

4) Weighted geometric mean of relatives with constant weights.

Fisher’s ideal index does not satisfy this test. It has been proved that no index
satisfies both the factor reversal and the circular test.

Check your progress 2

1) Compute the chain index number with 2010 prices as the base from the
following table giving the average wholesale prices of commodities A, B
and C for years 2010-2014.

Commodity Average whole sale Price (in Rs.)

2010 2011 2012 2013 2014

A 20 16 28 35 21

B 25 30 24 36 45

C 20 25 30 24 30
…………………………………………………………………………...…
…………………………………………………………………………...…
…………………………………………………………………………...…
…………………………………………………………………………...…
……………………………………………………………………………..
…………………………………………………………..…....................…

143
Summarisation of 2) Construct Fisher’s ideal Index number from the following data and
Bivariate and Multi-
variate Data show that it satisfies Factor and Time Reversal Tests.
Base Year Current Year
Commodities Price Expenditure Price Expenditure
Per unit (Rs.) per unit (Rs.)
A 2 40 5 75
B 4 16 8 40
C 1 10 2 24
D 5 25 10 60
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………

6.6 COST OF LIVING INDEX (CLI) OR CONSUMER

PRICE INDEX (CPI)
This is an index of changes in the prices of goods and services commonly
consumed by a homogeneous group of people, such as families of industrial
workers. The major items of consumption that are considered for the construction
of CLI are:

1) Food

2) Fuel and Light

3) Clothing

4) House rent

5) Miscellaneous
The common method for obtaining the consumption basket is to conduct a family
living survey among the population group for which the index is to be
constructed. Prices of selected items are also collected from various retail
markets used by consumers in question. It may be noted that each of the above
broad groups contains several sub groups. Thus, ‘food’ includes cereals, pulses,
oils, meat, fish, egg, spices, vegetables, fruits, non-alcoholic beverages, etc.
‘Miscellaneous’ includes such items as medical care, education, transport,
recreation, gifts and many others. When more than one price quotation is
collected for a single commodity, a simple average is taken. Index number is
constructed for each of the five groups using weighted average of the price
group; the weighs used are proportional to the expenditure on the consumed item
buy an average family. Next, the overall index (CLI) is computed as an weighted

144
average of group indices, the weights being again the proportional expenditure on Index Numbers

different groups (e.g., 50 per cent on food).

Using Laspeyres’ formula
∑ ×
Cost of living Index: I = ∑
where w =∑

The CLI or consumer price index (CPI) numbers have significant practical
implications and extensive public use. Its use as a wage regulator is the most
important. The dearness allowance (DA) of employees is primarily determined
by this index. When wages or incomes are divided by corresponding CLI, the
effect of changes in prices (inflation) is eliminated. This is known as the process
of deflation, which is used to find ‘real wages’ or ‘real income’ As mentioned
earlier the reciprocal of CLI gives us the purchasing power of money.
Example 6.5: Construction of an index for food
Item Prices Weights
Pn P0 P = (Pn × P0) w P×w
Rice 50 40 125.0 30 3750.0
Wheat 45 30 150.0 20 3000.0
Pulses 60 40 150.0 10 1500.0
Sugar 40 20 200.0 5 1000.0
Oil 75 60 125.0 15 1875.0
Potato 60 50 120.0 15 1800.0
Fish 200 150 133.3 5 666.5
Total 100 13591.5

∑ ×( ) ∑
Index (food) = ∑
× 100 = ∑
× 100

.
= = 135.915 = 135.92

Example 6.6: Construction of a final Cost of Living index number.

Item Weight Index Weight × Index
(Percentage
Expenditure)
Food 45 130 5850
Clothing 15 140 2100
Housing 20 170 3400
Fuel 5 110 550
Misc. 15 125 1875
Total 100 13591.5
,
Cost of Living Index = = 137.75 = 138

145
Summarisation of Check your progress 3
Bivariate and Multi-
variate Data 1) Calculate a number which will indicate the percentage change in volume
of traffic from October 2019 to October 2020, when account is taken of
the relative values of the different types of traffic.
Type of traffic Tons(‘000) Receipts(Rs.’000)
Oct. 2019 Oct. 2020 Oct. 2019
Merchandise 1246 1206 776
Fuel 4794 4229 562
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
………………………………………………….................………..……

2) Compute Paasche’s price index number for 2020 with 2015 as the base
from the following data:
Commodity Unit Price(Rs.) per unit Quantities sold
2015 2020 2015 2020
A kg. 4 5 95 120
B kg. 60 70 118 13
C kg. 35 40 50 70

……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
…………………………………………………………………………...

3) From the following data, compute the Laspeyres’ price index number for
2021 with 2019 as base:
Item Price(Rs.) Total Value(Rs.)
2019 2021 2019
A 12.50 14.00 112.50
B 10.50 12.00 126.00
C 15.00 14.00 105.00
D 9.40 11.20 47.00
146
 …………………………………………………………………………… Index Numbers

……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………

4) Calculate Marshall-Edgeworth index number from the following data:

2010 2015
Commodity Price Quantity Price Quantity
Rice 9.3 100 4.5 90
Wheat 6.4 11 3.7 10
Jowar 5.1 5 2.7 3
…………………………………………………………………………...…
…………………………………………………………………………...…
…………………………………………………………………………...…
…………………………………………………………………………...…
.......................................................................................................................
.......................................................................................................................

6.7 WORKED OUT EXMAPLES

In this section, we shall provide worked out examples so as to further familiarize
you with the topic.
Example1.7: Construction of Price Index
Item Unit Price per unit(Rs.)
2010 (P0) 2020 (Pn) (Pn ÷P0)×100
Rice quintal 100 220.00 220
Wheat kg. 1.50 2.40 160
Fish kg. 15.00 28.00 187
Bread lb. 0.60 1.35 225
Milk litre 2.50 4.00 160
Total 119.60 255.75 952
a) Aggregative Method
Index number for 2020 (base 2010 =100)
Average price per unit in 2020
× 100
Average price unit in 2010

147
Summarisation of ∑ / .
Bivariate and Multi- =∑ × 100 = × 100 = 213.84
/ .
variate Data
b) Method of Price Relative
Index number for 2020 (base 2010 =100)
∑ ×
=

= 952 5 = 190.4

Example 6.8: Calculate price index numbers from the following information,
using (a) weighted aggregative formula, and (b) weighted arithmetic mean of
price relatives:
Item Unit Price per unit(Rs.)
Base year Current year Weight
A quintal 85 115 19
B kg. 15 15 25
C dozen 45 61 40
D litre 55 100 20
E Lb 17 23 21

Calculation for Index numbers

Item P0 Pn w P0w PnwI = ( Pn÷P0)×100 Iw

A 85 115 19 1615 2185 135.3 2570.7

B 15 20 25 375 500 133.3 3332.5
C 45 61 40 1800 2440 135.6 5424.0
D 55 100 20 1100 2000 181.8 3636.0
E 17 23 21 357 483 135.3 2841.3

Total 125 5247 7608 17804.5

∑
a) Weighted aggregative index = ∑ × 100 = × 100 = 144.99 =
145
b) Weighted arithmetic mean of price relatives
∑ .
= ∑
= = 142.44

Example 6.9: Given below are the data on prices of some consumer goods and
the weights attached to the various commodities. Calculate price index numbers

148
for the year 2021 (base 2020 = 100), using (a) simple average, and (b) weighted Index Numbers
average of price relatives.

Price (Rs.)
Commodity Unit 2020 2021 Weights
Wheat Kg. 0.05 0.75 2
Milk Litre 0.60 0.75 5
Egg Dozen 2.00 2.40 4
Sugar Kg. 1.80 2.10 8
Shoes Pair 8.00 10.00 1
Calculations for price relative index.

Commodity Unit P0 Pn I=( Pn÷P0)×100 w I

Wheat kg. 0.50 0.75 150 2 300
Milk Litre 0.60 0.75 125 5 625
Egg Dozen 2.00 2.40 120 4 480
Sugar kg. 1.80 2.10 117 8 936
Shoes Pair 8.00 10.00 125 1 125

Total 637 20 2446

∑ ×
a) Simple average of price relative index = = = 127.4
∑
b) Weighted average of price relative index = ∑
= = 123.3

Example 6.10: On the basis of the following data, calculate the wholesale price
index number for the five groups combined (Base: 2015-16 = 100).
Group Weight Index no. for
the week ending 31.01.2021

Food 50 241

Liquor and tobacco 2 221

Fuel, power, light and lubricants 3 204

Industrial raw materials 16 256

Manufactured commodities 29 179

∑
We compute: General index =
∑

where I = Group index, and w = Group weight

149
Summarisation of
Bivariate and Multi- Group weight (w) Group Index(I) I×w
variate Data

Food 50 241 12050

Liquor and tobacco 2 221 442

Fuel, power, light and lubricants 3 204 612

Industrial raw materials 16 256 4096

Manufactured commodities 29 179 5191

Total 100 22391

Index number of wholesale price = = 223.91

Example 6.11: Annual production (in million tons) of four commodities are
given below:
Commodity Production Weight
2015 2019 2020

A 160 200 216 20

B 24 42 45 30
C 50 72 68 13
D 120 168 156 17

Calculate quantity index numbers for the years 2019 and 2020 with 2015 as base
year, using (a) simple arithmetic mean, and (b) weighted arithmetic mean of the
relatives.
Quantity relatives for 2019 with base year 2015 (=100)
𝑞 𝑞
𝐼= 𝑞 × 100 = 𝑞 × 100
Commodity A: × 100 = 125

Commodity B: × 100 = 175

Commodity C: × 100 = 144

Commodity D: × 100 = 140

Quantity relatives for 2020 with 2015 = 100
𝑞
𝐼= 𝑞 × 100
Commodity A: × 100 = 135

Commodity B: × 100 = 187.5

Commodity C: × 100 = 136

150
Commodity D: × 100 = 130 Index Numbers

Commodity Quantity relatives(I) Weight I×w

2019 2020 w 2019 2010
A 125 135.0 20 2500 2700
B 175 187.5 30 5250 5625
C 144 136.0 13 1872 1768
D 140 130.0 17 2380 2210
Total 584 588.5 80 12002 12303
∑
a) Simple arithmetic mean of quantity relatives = × 100

(where k = number of commodities)

Index number for 2019 = = 146
.
Index number for 2020 = = 147
∑
b) Weighted arithmetic mean of quantity relatives
∑

Index number for 2019 = = 150

Index number for 2020 = = 154

Example 6.12: From the following price(p) and quantity (q) data, compute
Fisher’s ideal index number.
Commodity 2015 (Base Year) 2020 (Current Year)
Price Quantity Price Quantity
A 12 10 17 10
B 14 9 16 11
C 11 12 13 10

Calculations for Fisher’s ideal index:

Commodity P0 q0 Pn qn P0 q0 Pnqn P0qn Pnqn

A 12 10 17 10 120 170 120 170

B 14 9 16 11 126 144 154 176
C 11 12 13 10 132 156 110 130

Total 378 470 384 476

∑ 𝑃𝑛 𝑞0
Laspeyres’ price index = ∑ × 100 = × 100 = 124.34 = 124
𝑃𝑛 𝑞0

∑ 𝑃𝑛 𝑞𝑛
Paasche’s price index = ∑ × 100 = × 100 = 123.96 = 124
𝑃0 𝑞𝑛

Fisher’s ideal index = √𝐿 × 𝑃 = √124 × 124 = 124.

151
Summarisation of
Bivariate and Multi- 6.8 LET US SUM UP
variate Data
In this unit you have been introduced to the concepts and methods involved in the
construction of Index Numbers. You have been shown how to use the
Laspeyres’, Paasche’s and Fisher’s formulae for calculating price as well as
quantity indices. You also know now how to measure changes in consumer price
or cost of living.

6.9 ANSWERS AND HINTS TO CHECK YOUR

PROGRESS EXERCISES
Check Your Progress 1
1) and (2): Do it yourself.
3) Simple Aggregative Index Number = 117.14
Average of Price Relative Method = 122.9
4) 84.2
5) Laspeyres’ Index Number = 86.02
Paasche’s Index Number = 81.25
Check Your Progress 2
1) 108.33. 135.41, 160.23, 165.56
2) Do it yourself
Check Your Progress 3
1) We find quantity for Oct. 2020 with Oct. 2019 as base. The required index
may be obtained as the weighted arithmetic mean of quantity relatives, using the
receipts in 2019 as weights.
Type of traffic q0 qn Weight Quantity (4)×(5)
(w) (Pn÷ q0)×100
(1) (2) (3) (4) (5) (6)
Merchandise 1246 1206 776 97 75272
Minerals 1125 981 252 87 21924
Fuel 4794 4229 562 88 49456

Total 1590 146652

∑ × ×
Quantity index = ∑
= = 92

2) Calculation for Paasch’s price index

152
Commodity P0 Pn q0 qn P0 qn Pn qn Index Numbers

A 4 5 95 120 480 600

B 60 70 118 130 7800 -----
C 35 40 50 70 2450 2800
Total 1590 146652
∑ 𝑃𝑛 𝑞𝑛
Paasche’s Price index = ∑ × 100 = × 100 = 116
𝑃0 𝑞𝑛

3) We are given the base price(P0), current price (Pn) and value in the base year
(P0q0). To find base year quantity (q0), we can use the relation
𝑃𝑞
𝑞 =
𝑃
Using P0, Pn and q0, we can find Laspeyres’ index as
∑ 𝑃𝑛 𝑞0
𝐿= × 100
∑ 𝑃0 𝑞0
Calculation for Laspeyres’ price index
Item P0 Pn P0 q0 Pn Pn q0

A 12.50 14.00 112.50 9 126.00

B 10.50 12.00 126.00 12 144.00
C 15.00 14.00 105.00 70 7 98.00
D 9.40 11.20 47.00 5 56.00
Total 390.50 424.00
∑ 𝑃𝑛 𝑞0 .
Laspeyres’ price index = × 100 = × 100 = 109.
𝑃0 𝑞0 .

∑ 𝑃𝑛 𝑞0 +𝑞𝑛
4) Marshall-Edgeworth index = ∑ × 100
𝑃0 𝑞0 +𝑞𝑛

∑ 𝑃𝑛 𝑞0 +∑ 𝑃𝑛 𝑞𝑛
=∑ × 100
𝑃0 𝑞0 +∑ 𝑃0 𝑞𝑛

Let us take 2010 as base and 2015 as current year.

Commodity P0 q0 Pn qn P0 q0 P0 qn Pn q0 Pn qn
A 9.3 100 4.5 90 930.0 837.0 450.0 405.0
B 6.4 11 3.7 10 70.4 64.0 40.7 37.0
C 5.1 5 2.7 3 25.5 15.3 13.5 8.1
Total 1025.9 916.3 504.2 450.1
504.2+450.1
Required Index = × 100 = 49.1.
1025.9+916.3

153
UNIT 7 DETERMINISTIC TIME SERIES AND
FORECASTING
Structure
7.0 Objectives
7.1 Introduction
7.2 Problems and Objects of Study of Time Series Data
7.2.1 Components of Time Series
7.2.2 Construction of Time Series: An Example
7.3 Measurement of Trend
7.3.1 Moving Average Method
7.3.2 Suitability of Moving Averages
7.3.3 Examples of Moving Averages
7.4 Method of Fitting Polynomials
7.4.1 Suitability of Least Squares Method
7.4.2 Examples of Least Squares Method
7.5 Monthly or Quarterly Trend Values from Annual Data
7.6 Measurement of Seasonal Variations
7.6.1 Method of Simple Average
7.6.2 Ratio to Trend Method
7.6.3 Ratio to Moving Average Method
7.7 Let Us Sum Up
7.8 Answers or Hints to Check Your Progress Exercises

7.0 OBJECTIVES
After going through this Unit, you will be able to
 construct a trend line for time series data;
 compute moving averages; and
 calculate various measures of seasonal variations.

7.1 INTRODUCTION
A time series is a set of observations on a variable measured at successive points
of time. Usually the variable values are recorded over equal time intervals-yearly,
quarterly monthly, etc. Generally the term ‘time series’ refers to economic data,
but it equally applies to quantitative data collected in other fields also. The time
series of National Income, Agricultural Income, and Agricultural Production are
based on annual observations.

Adapted from IGNOU study material of EEC 13: Elementary Statistical Methods and Survey Techniques,
Unit 11 written by S Bandopadhyay with modifications by K. Barik
Other examples of time series are yield of a crop in different years, population of Deterministic Time
Series and Forecasting
a country over different points of time, sales of a departmental store during
different seasons of the year, quarterly exports of tea, etc. For these types of data
one of the variables is time, denoted by tj and the other which is dependent on
time (such as yield, population, sale or export) is represented by yt. We will
analyse some of these series with the help of the methodology to be developed in
this Unit.

7.2 PROBLEMS AND OBJECTS OF STUDY OF

TIME SERIES DATA
A study of time series data reveals that in general the observed values of the
dependent variable (yt) change over time. These changes are due to interaction of
several forces such as increase in population, change in production techniques,
change of tastes and habits of people, variations in climate, etc. Part of these
changes is long term while others may be seasonal or cyclical. One of the main
objectives of study of time series data is to isolate and measure the effects of
various components. This analysis helps us in understanding the past behavior
and predicting the future. Such predication is of utmost importance to an
economist or a producer who can plan his production much ahead of sales.
7.2.1 Components of Time Series
A graphical representation of time series data reveals changes over time (in rather
exceptional cases the series exhibits no change during the period of observation).
However, these changes are not totally haphazard or random and at least a part of
it can be explained. Some of the movements are periodic in nature while some
others show persistent growth or decline. Along with these some unpredictable
movements, random in nature, are also found to be mixed up. Again, not every
series shows all the movements. It is assumed that the general series has four
important components.

i) Secular or long-term trend(T)

ii) Seasonal variation (S)

iii) Cyclical fluctuation (C)

iv) Irregular or random movement (I)

In the classical approach, it is assumed that the observed value yt may be

represented either as the product of the above components
i.e., yt = T × S ×C × I (multiplicative model)
or, as the sum of components
yt = T + S +C + I (additive model)

155
Summarisation of Although the additive model facilitates easier calculation, the multiplicative
Bivariate and Multi-
variate Data model has been most widely used in analysis of time series.
a) Secular Trend
By secular trend we mean the smooth, regular, long-term changes in the series
when observed over a period of time. Some series may exhibit an upward trend,
some series a downward trend while some others may remain more or less
constant over time. The upward trend of a series may be caused by factors such
as increase in population and improvement in techniques of production. For
example, the pattern of growth of many industries follows closely that of
population growth of the country. Again the advances in technology may give
rise to upward movement of most of the economics time series. But not all time
series will exhibit growth. Some may show decline while some others may show
fluctuations. The time series of crude death rates of a country is likely to show a
declining trend.
b) Seasonal Variations
The graphs of most of the time series reveal that a large number of fluctuations
are imposed on the trend. By seasonal variation we mean the periodic movement
in a time series where the period is not longer than one year. A periodic
movement is that which repeats at regular intervals or periods of time. For
example, the sales of cold drinks increase during summer and decrease during
winter, sales of garments are maximum during some seasons of the year, say
during May or festivals, the number of passengers carried by buses has a peak
during office hours, the number of books borrowed from a library has a peak
during some days of the week, etc. The factors which contribute to this type of
fluctuations are the climatic changes of different seasons, customs and habits
which people follow at different times.
C) Cyclical Fluctuations
By cyclical fluctuations we mean oscillatory movements of a time series, where
the period of oscillation, called the cycle, is more than a year. It includes those
factors leading to alternating periods of expansion and contraction that
characterize most economic and business series. Sometimes these fluctuations are
highly irregular with respect to their Sometimes these fluctuations are highly
irregular with respect to their shape, amplitude and direction. But the phenomena
they reflect – the periods of depression, recovery, boom and collapse-have been
observed in virtually all time series dealing with business and economics data.
d) Irregular Movement
The irregular movement includes component all factors not classifiable
elsewhere. Thus factors such as work stoppage, elections, wars, fire may affect a
particular time series; this category of movement includes all types of variations
not accounted for by secular trend, seasonal or cyclical fluctuations.
Unfortunately, factors of these kinds are frequently indistinguishable from

156
cyclical factors and as such in some discussions cyclical and irregular Deterministic Time
Series and Forecasting
components are combined together.
7.2.2 Construction of Time Series: An Example
As an illustration we prepare a time series according to the multiplicative model.
Table 7.1 presents trend, seasonal and cyclical-irregular components of a
hypothetical series.
Table 7.1: Hypothetical Time Series and its Components (Quarterly)
Year Components
Quarter Series Trend Seasonal Cyclical- Irregular
(yt) (T) (100S) (100CI)
I 79 80 120 82
1
II 58 85 80 85
III 84 90 92 102
IV 107 95 108 105

I 130 100 120 108

2
II 93 105 80 132
III 121 110 92 120
IV 161 115 108 130

I 216 120 120 150

3
II 132 125 80 132
III 150 130 92 125
IV 163 135 108 112

I 176 140 120 105

4
II 112 145 80 97
III 128 150 92 93
IV 142 155 108 85

I 134 160 120 70

5
II 86 165 80 65
III 94 170 92 60
IV 104 175 108 55

In Table 7.1 the series is represented by a multiplicative model, such that

yt = T × S ×C × I.

157
Summarisation of
Bivariate and Multi-
Thus the observation 79 (of I quarter of 1st year) = 80 ×
120 82
variate Data × .
100 100

Similarly, 112 (of II quarter of 4th year) = 145 ×

80 97
× .
100 100

Thus, each quarterly figure (yt) is the product of the secular trend (T), the
seasonal index (S), cyclical and the irregular component (CI). Such a synthetic
composition looks very much like an actual time series and has encouraged use
of the model as the basis for the analysis of time series data.

7.3 MEASUREMENT OF TREND

At times we are interested to know the trend movement in a time series. In such
circumstances, we have to eliminate the effects of other components (seasonal,
cyclical and irregular) from the series.
Two important methods of measuring trend are the ‘Moving Averages Method’,
and the ‘Method of Fitting Polynomials’. In moving average method, secular
trend is obtained by smoothing out fluctuations by the process of averaging. In
the latter, a polynomial of suitable degree is chosen either for the original
variables or for its transformed variable and its constants are determined by the
method of least squares. The choice of the degree of the polynomial can be made
by plotting the data on a graph paper where different scales, arithmetic, semi-
logarithmic or double-logarithmic scales may be used. Measurement of trend is
necessary for studying the behavior of the time series and for forecasting the
future.
7.3.1 Moving Averages Method
This is a simple method of smoothing out fluctuations of a series by calculating a
number of averages covering overlapping periods of the series. The first step
consists in selecting proper period of the moving average. If the period chosen
is 3 years, the moving averages are obtained by calculating a series of mean
values of three consecutive values covering overlapping periods of the series.
Denoting the original series by y1, y2, y3,….., the mean of the first three values,
given by (y1+ y2+ y3)/3, is placed at the midpoint of the period covering first
three years. This is the first moving average value. The second moving average
value is obtained by calculating the mean of the values covering the period from
second to fourth year. This is given by (y2+ y3+ y4)/3 and is placed at the mid-
p0int of the period covering second to fourth year. This process is repeated. It is
clear that some of the values for the years at the beginning as well as at the end
cannot be obtained by this method.
Two cases may be distinguished, viz., when the period of moving average is odd
and when it is even. If the period is odd (for example, if the period is three years),
the first moving average is placed at the second year, the second moving average
is placed at the third year and so on.
158
 If, however, the period is even (for example four years), the moving average Deterministic Time
Series and Forecasting
value fall between two consecutive years and ‘centering’ is necessary for getting
trend values for various years.
As an illustration, let us consider a schematic representation for the calculation of
centered 4-year moving averages. Here we will present two methods- the direct
method (Table 7.2), as well as the short-cut method (Table 7.3).
Table 7.2 Calculation of centered 4-year moving averages (Direct Method)

Year y1 4-year Moving Total 4-year Moving Centered Centered 4-

Average Moving year moving
Total

(1) (2) (3) (4) (5) (6)

1 y1 - -

2 y2 - -

y1 + y1 + y1 + y1 = T1 T1 / 4

3 y3 (T1 + T2)/4 (T1 + T2)/8

y2 + y3+ y4 + y5 = T2 T2 / 4

4 y4 (T2 + T3)/4 (T2 + T3)/8

y3 + y4+ y5 + y6 = T3
T3 / 4

5 y5 (T3 + T4)/4 (T3+ T4)/8

y4 + y5+ y6 + y7 = T4 T4 / 4

6 y6 - -

7 y7 - -

In the above illustration, the period of moving averages is 4 years. Both in the
direct and in the short-cut method col. 3 shows the 4-year moving totals. The first
value (T1) is placed between the second and the third year, the second moving
total (T2) is placed between the third and the fourth year and so on. The centered
4-year moving averages are placed at the third year, fourth year, by taking a
further 2 item moving average in the direct method (Table 7.2.) In the short cut
method (Table 7.3), the calculation of the 4-year moving average is omitted (as
shown in col. 4 of Table 7.2 in the direct method). Instead, the 2-item moving
totals of the 4-year moving averages are obtained (col. 4 and 5).
159
Summarisation of You should note that for a 4-year moving average, the procedure for centering
Bivariate and Multi-
variate Data leaves out 4/2 = 2 years at the end of the series each.
Table 7.3 Calculation of centered 4-year moving averages (Short Method)

Year y1 4-year Moving Total 2-item moving Centered 4-year

(M.T.) total (M.T.) moving average
(M.A.)
(1) (2) (3) (4) (5)
1 y1 - -
2 y2 - -
y1 + y2 + y3 + y4 = T1
3 y3 (T1 + T2) (T1 + T2)/8
y2 + y3+ y4 + y5 = T2
4 y4 (T2 + T3) (T2 + T3)/8
y3 + y4+ y5 + y6 = T3
5 y5 (T3 + T4) (T3+ T4)/8
y4 + y5+ y6 + y7 = T4
6 y6 - -
7 y7 - -

7.3.2 Suitability of Moving Averages

Moving average method is simple to apply but the success of this method
depends on the proper choice of the period. Moving average with a period
exactly equal to or a multiple of the period exactly equal to or a multiple of the
period of the cycle present in the series will completely eliminate the cyclical
component and give an estimate of the trend. This method is flexible but some
trend values at the beginning and at the end of the series have to be left out and
their number increases with increase in the period of the moving average. Again
as moving average assumes no law of change; the method cannot be used for
forecasting future trend.
7.3.3 Examples of Moving Averages
Example 7.3.1: Calculate the three and five year moving averages of the
following data:

Year 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Production
18 19 20 22 20 19 22 24 25 24 25 26
(‘000 tons)

160
Steps of Calculation Deterministic Time
Series and Forecasting
1) In Table 7.3.1 the figures in col. 3 are obtained as the sum of three
consecutive values of col. 2. Thus the first moving total (M.T.) is 57 = 18
+ 19 + 20 and is placed against 2001. The second moving total 61 = 19 +
20 + 22 is placed against 2002.

2) The three-year moving average (M.A.) in col. 4 is obtained by dividing

the corresponding three-year moving total in col. 3 by 3, the period of the
moving average. Thus 57 ÷ 3 = 19, 61 ÷ 3 = 20.3, etc.

3) The five-year moving totals in col.5 are obtained as the sum of five
consecutive values in col.2. Thus the first moving total against the year
2002 is 99 = 18 + 19 + 20 + 22 + 20.

4) The five-year moving total in col. 5 by 5. Thus moving average for 2005
is 107 ÷ 5 = 21.4.

Table 7.3.1: Calculation of (I) 3-year Average (II) 5-year Moving Average

Year Production 3-year 3-year 5-year 5-year

M.T. M.A. M.T. M.A.

(1) (2) (3) (4) (5) (6)

2000 18 - - - -

2001 19 57 19.0 - -

2002 20 61 20.3 99 19.8

2003 22 62 20.77 100 20.6

2004 20 61 20.3 103 20.6

2005 19 61 20.3 107 21.4

2006 22 65 21.7 110 22.0

2007 24 71 23.7 114 22.8

2008 25 73 24.3 120 24.0

2009 22 74 24.7 124 24.8

2010 25 75 25.0 - -

2011 26 - - - -

161
Summarisation of
Bivariate and Multi- Note that for 3-year centered moving averages = 1 year, and for 5-year
variate Data
centered moving averages = 2 years, respectively, are left out both at the
beginning and the end of the series.

Example 7.3.2: Compute trend values for the following time series using 4-
yearly moving averages.

Year 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Yield 52 54 55 57 58 61 63 66 67 70
(qntls.)

Solution:
Table 7.3.2(a): Calculation of 4-year moving average (Direct Method)

Year Yield 4-year 3-year 2 item M.T. of (Centered)

M.T. M.A. col.4(centered) 4-year
M.A.

(1) (2) (3) (4) (5) (6)

2009 52 - - - -

2010 54 218 54.50 - -

2011 55 224 56.00 110.50 55.250

2012 57 231 57.75 113.75 56.875

2013 58 239 59.75 1177.50 58.750

2014 61 248 62.00 121.75 60.875

2015 63 257 64.25 126.25 63.125

2016 66 266 66.50 130.75 65.375

2017 67 - - - -

2018 70 - - - -

162
 Table 7.3.2(b): Calculation of 4-year Moving Average (Shortcut Method) Deterministic Time
Series and Forecasting
Year Yield 4-year 2-item (Centered)
M.T. M.A 4-year M.A

(1) (2) (3) (4) (5)

2009 52 - - -
2010 54 218 - -
2011 55 224 442 55.250
2012 57 231 455 56.875
2013 58 239 470 58.750
2014 61 248 487 60.875
2015 63 257 505 63.125
2016 66 266 523 65.375
2017 67 - - -
2018 70 - - -

Steps of Calculation (Direct Method) – see Table 7.3.2(a)

1) Col. 3 is the sum of four consecutive values in col.2

Thus, 52 + 54 + 55 + 57 =218, 54 + 55 + 57 + 58 = 224.

2) Col. 4 = col. 3÷4. Thus 218 ÷ 4 = 54.5, 224 ÷ 4 = 56.

3) Col. 5 = Sum of two consecutive values in col.4.

Thus, 54.5 + 56.0 = 110.5, 5.00 + 57.75 = 113.75.

4) Col. 6 = col. 5÷2. Thus 110.5÷2 = 55.25.

Steps of calculation (shortcut method) – see Table 7.3.2 (b)

1) Col. 4 is the sum of two consecutive values in col.3.

Thus, 218 + 224 = 442, 224 + 231 = 455.

2) Col.5 = col.4 ÷ 8. Thus 442 ÷ 8 = 55.25

Example 7.3.3
Find trend values for the following series using a 3-year weighted moving
average with weights 1, 2, 1.

Year 1 2 3 4 5 6

Value 2 3 5 6 8 11

163
Summarisation of Solution:
Bivariate and Multi-
variate Data Table 7.3.3: Calculation of 3-year weighted moving average

Year Value 3-year weighted 3-year weighted

moving total moving average
(M.T.) (M.A)

(1) (2) (3) (4)

1 2 - -

2 3 13 3.25

3 5 19 4.75

4 6 25 6.25

5 8 33 8.25

6 11 - -

Step of calculation
1) Col. 3 figures are the weighted moving totals of col.2 figures with weights 1,
2, 1.
Thus 1 × 2 + 2 ×3+ 1×5 =13
1 × 3 + 2 × 5 + 1×6 =19
2) Col.4 = col.3 ÷ (sum of weights, i.e., 4)
Thus 13 ÷ 4 = 3.25, 19 ÷ 4 = 4.75

Example 7.3.4: Calculate the 4-quarter moving average for the following time
series data

Year
Quarter
2015 2016 2017 2018

1 62 66 72 79

2 58 60 67 74

3 72 74 80 88

4 60 64 69 77

164
Solution: Deterministic Time
Series and Forecasting
Year Quarter Value 4-quarter Centered 4-quarter
(M.T.) (M.T.) (M.A.)

(1) (2) (3) (4) (5) (6)

2015 1 62 - - -
2 58 - - -
259 - -
3 72 508 63.50
256
4 60 514 64.25
258

2016 1 66 518 64.75

260
2 60 524 65.50
264
3 74 534 66.75
270
4 64 547 68.38
277
2017 1 72 560 70.00
283
2 67 571 71.38
288
3 80 583 72.88
295
4 69 597 74.63
302
2018 1 79 612 76.50
310
2 74 628 78.50
318 - -
3 88 - - -
4 77 - - -

165
Summarisation of Check Your Progress 1
Bivariate and Multi-
variate Data 1) Given below is data on index of production for the period 2011 to 2020.

Year 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020

Index of 109.2 119.8 129.7 140.8 153.8 152.2 152.6 163.0 175.3 184.3
Production

1) Fit the trend line and predict the index of production for the year 2012 by
3-year moving averages method.
…………………………………………………………………………...…
………....………………………………………………………………..….
……..…………………………………………………………………….…
……………………...………………………………………………………
………………………………...……………………………………………

2) Bring out the advantages and disadvantages of moving averages method.

…………………………………………………………………………...…
………....………………………………………………………………..….
……..…………………………………………………………………….…
……………………...………………………………………………………
………………………………...……………………………………………

7.4 METHOD OF FITTING POLYNOMIALS

Method of fitting polynomials is perhaps the best and the most objective method
of determining trend. Here an appropriate type of polynomial is selected for trend
and the constants appearing in the trend equation are determined on the basis of
the given time series data.
The choice of an appropriate polynomial is facilitated by a graphical
representation of the data for which, apart from the usual arithmetic scales, semi-
logarithmic or doubly-logarithmic scales may be used. If the plotted data show
approximately a straight line tendency on an ordinary graph paper, the used is
Y = a + bx (straight line or first degree polynomial).
If they show a straight line on a semi-logarithmic graph paper, the equation is log
Y = a + bx, which is obtained by taking logarithm of Y = A.Bx (exponential
function). Note that a = log A and b = log b.
Some times a second or a third degree polynomial may also be fitted.
Y = a +bx +cx2 (second degree polynomial or parabola)

166
Y = a +bx +cx2 +dx3 (third degree polynomial) Deterministic Time
Series and Forecasting
The constants appearing in the above equations (such as a,b,c…) are obtained by
applying the principle of “least squares”, as in regressions (see unit 5). This
states that the values of the constants will be such as to make the sum of squares
of the deviations
∑(𝑦 − 𝑌) minimum,

where y = observed value,

Y = expected value obtained from the trend equation of the type

Y = a +bx

Or,

Y = a +bx +cx2 etc., and the summation is taken over all the observations.

In the case of straight line fitted by the method of “least squares”, the constants a
and b are determined from the following normal equations:
∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥 and

∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥

where n is the number of years covered.

Similarly in the case of parabola or second degree polynomial the constants a, b

and c are determined from the three normal equations
∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥 + 𝑐 ∑ 𝑥

∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥 + 𝑐 ∑ 𝑥

∑𝑥 𝑦 = 𝑎∑𝑥 + 𝑏∑𝑥 + 𝑐∑𝑥

Rule for writing down the normal equations

To get the first normal equation, multiply each observation by coefficient of a in
that equation and take sum over all the n observations.

Thus for straight line y = a + bx, as the coefficient of a is 1, the first normal
equation is ∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥.

For the second normal equation, multiply each observation by the coefficient of b
in that equation and take sum over all the n observations. In the case of straight
line, coefficient of b is x. So, the second normal equation is ∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 +
𝑏∑𝑥 .
Now we will consider trend fitting for periods covering odd (Table 7.4) and even
(Table 7.5) number of years taking the first degree polynomial.

167
Summarisation of
Bivariate and Multi-
Case I: Odd number of years (n = 5)
variate Data Table 7.4

Year y x x2 xy

(1) (2) (3) (4) (5)

1 y1 -2 4
2 y2 -1 1
3 y3 0 0
4 y4 1 1
5 y5 2 4
total Sy 0 10 Sxy

The normal equations are:

∑ 𝑦 = 5𝑎 + 𝑏 ∑ 𝑥 = 5𝑎 (Since ∑ 𝑥 = 0)
∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥 = 10𝑏
∑ ∑
∴𝑎 = , 𝑏=

where the origin (x = 0) is taken at the midpoint of the interval covered by 5

years, i.e., at the third year and unit of time = 1 year. In real life situations the
actual values of y’1 s are considered. Hence ∑ 𝑥 and ∑ 𝑥𝑦 will be known.

Case II: Even number of years (n = 6)

Table 7.5
Year y x x2 xy
(1) (2) (3) (4) (5)
1 y1 -5 25
2 y2 -3 9
3 y3 -1 1
4 y4 1 1
5 y5 3 9
6 Y6 5 25
total ∑𝑦 0 70 𝑥𝑦

The constants a and b will be obtained from the following equations

∑ 𝑥 = 6𝑎

168
∑ 𝑥𝑦 = 70𝑏 Deterministic Time
Series and Forecasting
Here the origin (x = 0) will be in the middle of 3rd and 4th year and the unit of x
=6 months.
7.4.1 Suitability of Least Squares Method
Trend lines are used for description of the growth or decline of the time series
and as an aid to the study of the long-term trend of the economy. The method of
fitting polynomials completely eliminates personal bias and trend values for all
the given periods can be obtained. This is, however, not possible with moving
average method. But the choice of the type of the polynomial curve is arbitrary
and one cannot be sure whether a linear or parabolic curve will represent the
trend best. The choice of the trend equation may itself lead to a bias. It is,
however, possible to get some idea of the pattern of trend from the scatter
diagram of the data.
7.4.2 Examples of Least Squares Method
Example 7.4.1
Fit a straight line trend by the method of least squares to the following data:
1) The data given below give the index of industrial production from 1961 to
1970

years 2005 2006 2007 2008 2009 2010 2011

Production 81 92 100 105 112 120 120

Estimate the production for 2012.

Solution:

Here the number of years is odd (n = 7) Let y = a+bx be the equation of the
straight line trend with origin (x = 0) at 2008 and one unit of x = 1 year. The least
squares normal equations are (see unit 5)

∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥

∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥

Thus, substituting the values of ∑ 𝑦, ∑ 𝑥 y, ∑ 𝑥 and ∑ 𝑥 from the above table in

the normal equations, we get

7a = 736, so a = 105.1

28b = 203, so b = 7.21

The trend equation is
Y = 105.1+7.21x, with origin at 2008 and unit of x =1 year. The value of x for
2012 would be 4.
169
Summarisation of Table 7.4.1: Fitting straight Line Trend
Bivariate and Multi-
variate Data
Year Production (y) x x2 xy

(1) (2) (3) (4) (5)

2005 81 -3 9 -243

2006 92 -2 4 -184

2007 100 -1 1 -100

2008 105 0 0 0

2009 112 1 1 112

2010 120 2 4 240

2011 126 3 9 378

Total 736 0 28 203

Hence, using the trend equation the estimate for 2012 is Y = 105.1 + 4 × 7.21 =
133.94.

Example: 7.4.2
Fit a straight line trend to the following time series data:

Year 2010 2011 2012 2013 2014 2015

Profits: 3.1 3.3 3.6 3.2 3.7 3.9

Estimate the profit for 2016.

Solution:

Here the number of years is even (n = 6). Let y = a+bx be the trend equation
with origin.

(n = 0) mid-way between 2012 and 2013 and unit of x = 6 months.

The normal equations are

∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥

∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥

170
 Table 7.4.2: Fitting straight Line Trend Deterministic Time
Series and Forecasting
Year Profit (y) x x2 xy
(Rs. lakhs)

(1) (2) (3) (4) (5)

2010 3.1 -5 25 -15.5
2011 3.3 -3 9 -9.9
2012 3.6 -1 1 -3.6
2013 3.2 1 1 0.0
2014 3.7 3 9 11.1
2015 3.9 5 25 19.5
Total 20.8 0 70 4.8

So, substituting the values of ∑ 𝑦 , ∑ 𝑥𝑦, ∑ 𝑥, and ∑ 𝑥 from the above table in
the normal equations, we get
6a = 20.8, or a = 3.47
70b = 4.8, or b = 0.07
The trend equation is
Y = 3.47 + 0.07x, with origin at the middle of 2012 and 2013 and unit of x = 6
months.
For 2016, x would be 7.
So, estimate for 2016 is
Y = 3.47 + 0.07 × 7 = 3.47 + 0.49 =3.96
Hence the estimated profit for 2016 is Rs. 3.96 lakhs.
Example: 7.4.3
The sales of a company (in thousands of rupees) for the year 2010 to 2016 are
given in the following table. Fit an exponential trend (Y = A,Bx) and estimate the
sales for 2017.

Year 2010 2011 2012 2013 2014 2015 2016

Sales 32 47 65 92 132 190 275

Solution:
Here the number of years is odd (n = 7). Taking log of both sides of the given
equation, we can write log Y = logA + x logB. Let a = logA and b= logB. Thus
we can write

171
Summarisation of logY = a + bx.
Bivariate and Multi-
variate Data Further, we take origin (x = 0) at 2013 and one unit of x = 1 year. The least
squares normal equations are:
∑ log𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥
∑ 𝑥log𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥
Table 7.4.3: Fitting Straight Line Trend

Year Sales (y) x x2 logy x. logy

2010 32 -3 9 1.5051 4.5153

2011 47 -2 4 1.6721 3.3442

2012 65 -1 1 1.8129 1.8129

2013 92 0 0 1.9638 0

2014 132 1 1 2.1206 2.1206

2015 190 2 4 2.2788 4.5576

2016 275 3 9 2.4398 7.3119

Total 833 0 28 13.7931 4.3237

So, substituting the values of ∑ log𝑦 , ∑ 𝑥. log𝑦, ∑ 𝑥, and ∑ 𝑥 from the above
table in the normal equations, we get
7a = 13.7931, or a = 1.97
28b = 4.3237, or b = 0.154
Thus, the fitted function is logy = 1.97 + 0.154x or Y = antilog(1.97 + 0.154x).
For 2017, x would be 4.
Thus, the estimated value for 2017 is
Y = antilog (1.97+0.154×4) = antilog 2.586 = 385.48.
A case with even number of years can be attempted as in the fitting of a straight
line (see Example 7.4.2).

Example: 7.4.4
The following table shows the production of cement in India during 2002 to
2008.
Fit a second degree polynomial to the data.
172
 Deterministic Time
Series and Forecasting
Year 2012 2013 2014 2015 2016 2017 2018

Production 23.7 27.1 30.2 33.1 36.4 39.3 45.0

Solution:
Here the number of years is odd (n = 7). Let y = a + bx+cx2 be the trend
equation with origin (x = 0) at 2015 and unit of x = 1year. The normal equation
is:
∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥 + 𝑐 ∑ 𝑥
∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥 + 𝑐 ∑ 𝑥
∑𝑥 𝑦 = 𝑎∑𝑥 + 𝑏∑𝑥 + 𝑐∑𝑥
Table 7.4.4: Fitting Second Degree Polynomial

Year y x x2 x3 x4 xy x2 y

2012 23.7 -3 9 -27 81 -71.1 213.3

2013 27.1 -2 4 -8 16 -54.2 108.4

2014 30.2 -1 1 -1 1 -30.2 30.2

2015 33.1 0 0 0 0 0.0 0.0

2016 36.4 1 1 1 1 36.4 36.4

2017 39.3 2 4 8 16 78.6 157.2

2018 45.0 3 9 27 81 135.0 405.0

Total 234.8 0 28 0 196 94.5 950.5

Substituting the values from the table in the normal equations

7a + 28c = 234.8
28b = 94.5
28a + 196c = 950.5
Solving these three equations, simultaneously, we get
a = 33
b = 3.37
c = 0.134
Hence, the second degree polynomial is

173
Summarisation of Y = 33 + 3.37x + 0.134x2,
Bivariate and Multi-
variate Data with origin (x = 0) at 2015 and unit of x = 1 year.
Example: 7.4.5
Fit a second degree polynomial to the following data. Estimate the trend value for
2012.

Year 2006 2007 2008 2009 2010 2011

Annual Indian 23.7 27.1 30.2 33.1 36.4 39.3

Imports (108 Rs.)

Solution:
Here the number of years is even (n = 6). Let y = a + bx+cx2 be the trend
equation with origin (x = 0) mid-way between 2008 and 2009 and unit of x = 6
months. The normal equations are ∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥 + 𝑐 ∑ 𝑥
∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥 + 𝑐 ∑ 𝑥
∑𝑥 𝑦 = 𝑎∑𝑥 + 𝑏∑𝑥 + 𝑐∑𝑥
Table 7.4.5: Fitting Second Degree Polynomial

Years y x x2 x3 x4 xy x2y

2006 507 -5 25 -125 625 -2535 12675

2007 602 -3 9 -27 81 -1806 5418

2008 681 -1 1 -1 1 -681 681

2009 914 1 1 1 1 914 914

2010 1255 3 9 27 81 3765 11295

2011 1361 5 25 125 625 6805 34025

Total 5320 0 70 0 1414 6462 65008

Substituting the values from the table in the normal equations

6a + 70c = 5320
70b = 6462
70a + 1414c = 65008
Solving the above three equations simultaneously, we get
a = 829.2, b = 92.31 and c = 4.924.
The second degree polynomials is

174
Y = 829.2 + 92.31x + 4.924x2, Deterministic Time
Series and Forecasting
with origin (x = 0) mid-way between 2008 and 2009 and unit of x = 6 months.
For 2012, x would be 7.
Therefore, estimate for 2012 is
Y = 829.2 + 92.31 × 7 = 3.47 + 4.924 × (7)2
= 829.2+646.17+241.28 = 1716.65.

7.5 MONTHLY OR QUATRELY TREND VALUES

FROM ANNUAL DATA
In a time series, annual data may be available in different forms such as (1)
monthly or quarterly averages fo02r each year, and (ii) annual totals.
If the trend equation is fitted to the monthly or quarterly data, there will be no
difficulty in obtaining monthly or quarterly values. However, it is not advisable
to fit a trend line by the method of least square to the monthly or quarterly data.
The trend line is usually fitted to the annual data and then the trend values may
be obtained for different months (or quarters). The method of obtaining monthly
(or quarterly) trend equation is discussed below.
Let y = a+bx be an annual trend equation. If we divide both sides of this equation
by 12, we get a monthly average equation. Thus = + 𝑥 is a monthly
average equation. Denoting by = ,𝐴 = 𝑎𝑛𝑑 𝐵 = , we can write the
monthly average equation as Y =A+Bx. Here, Y is monthly and average and B
denotes change in monthly average per unit change in x,i.e., 1 year.
To get a monthly equation, we have to determine the corresponding rate of
change of Y. since B is the average monthly change in Y per year, will denote
average change per month. Thus, the monthly equation can be written as
𝑌 = 𝐴 + 𝑥 or 𝑌 = + 𝑥, where x denotes month rather than year.

Similarly, we can write 𝑌 = + 𝑥 as the quarterly average equation, where a

unit of x denotes one year, and 𝑌 = + 𝑥 as the quarterly average equation,
where a unit of x denotes one quarter.
Shifting of Origin
We define a, in the equation Y = a + bx, as the value of trend in the year of
origin. Thus, with the shifting of origin the value of a changes. Let us assume
that the year of origin. (i.e., x = 0) is 2015 and we want to change it to 2018.
We note that x = 3 for 2018, therefore, trend for 2018 = a +3b. Treating this as
constant term in the trend equation, y = (a + 3b) + bx becomes the new trend
equation with 2018 as origin.

175
Summarisation of Example 7.5.1
Bivariate and Multi-
variate Data The trend equation for certain production data is Y = 150 +24x (y = annual
production in thousand tons and x = time with origin at 2008, unit of x = 1 year).
Estimate the trend value for May 2013.
Solution: The monthly trend equation is
𝑌= + 𝑥 = 12.5 + 0.167𝑥,

where Y = monthly production, unit of x = 1 month and origin at 2008, i.e., 30th
June 2008.
To estimate the trend for May 2013, we substitute x = 58.5 in the above
equation. Thus, we get Y = 12.5 + 0.167 × 58.5 = 22.25 (‘000 tons)
Example 7.5.2
The trend equation fitted to quarterly average sales for 7 years is given by y =
250 + 20x (unit of x = 1 year, origin = 30th June 2010). Estimate the trend value
for the first quarter of 2013 (January-March).
Solution: Here the quarterly average refers to average per quarter for each year.
The quarterly trend equation is given by 𝑌 = 250 + 𝑥, where Y = quarterly
sales, x = 1 quarter and origin at 30th June 2010.
The interval between 30th June 2010 and the 1st quarter of 2013 are 10.5 quarters.
Thus, to obtain the trend for 1st quarter of 2013, we substitute x = 10.5 in the
above equation.
Hence, the required trend is Y = 250 + 5× 10.5 = 302.5.
Check Your Progress 2
1) Fit a straight line trend to the following data and show how to obtain the
monthly trend values from the trend line fitted to the given time series.
Obtain two such monthly values.

Year 2010 2011 2012 2013 2014

Monthly Production: 38 40 41 45 47
(in ‘000 tons)

…………………………………………………………………………...…
………....………………………………………………………………..….
……..…………………………………………………………………….…
……………………...………………………………………………………
……………………...………………………………………………………
…………………...…………………………………………………………
……………………………...………………………………………………

176
2) The trend equation for certain production data is y = 240 + 48x (y = annual Deterministic Time
Series and Forecasting
production in tons, x = time with origin at 2010, unit of x = 1 year).
Estimate the trend for October 2016.
…………………………………………………………………………...…
………....………………………………………………………………..….
……..…………………………………………………………………….…
……………………...………………………………………………………
………………………………...……………………………………………
3) The trend equation fitted to quarterly average sales data is given by y =
60 + 8x (unit of x = 1year, origin = 30th June, 2018). Estimate the trend
value for first quarter (Jan-March.) of 2020.

…………………………………………………………………………...…
………....………………………………………………………………..….
……..…………………………………………………………………….…
……………………...………………………………………………………
……………………...………………………………………………………
………………………………...……………………………………………

7.6 MEASUREMENT OF SEASONAL VARIATIONS

There are a number of methods of measuring seasonal variations in time series,
depending on how the other components such as cyclical, trend and irregular
movements are present in it. For simplicity, we shall consider seasonal variations
in monthly or quarterly data only, but the procedure for weekly or daily data will
be similar. It may be mentioned here that annual data contains no seasonal
variation. The application of the methods, to be discussed below, will give us
4(or 12) numbers for quarterly (or monthly) data. These will be termed as
seasonal indices and are normally expressed as percentages. A figure of a
particular quarter (or month) indicates whether that quarter is above or below the
normal quarter. For example, a value of 80 for particular quarter indicates that the
business for exports or sales (say) during that quarter is slack and it is below the
normal quarter by 20%. We will consider only the multiplicative model for the
measurement of seasonal variations. The main methods for the measurement of
seasonal variation are given below.

1) Method of Simple Average

2) Ratio to Trend Method

3) Ratio to Moving Average Method

177
Summarisation of 7.6.1 Method of Simple Average
Bivariate and Multi-
variate Data This method assumes that the time series variable y is made up of only two
components, viz., seasonal (S) and irregular or random component (I). Thus, we
can write
Y = S.I
When we take average of y values for each month or quarter of all the years, the
irregular component gets eliminated and we are left with seasonal component.
We will illustrate this method in Table 7.6.
Table 7.6 Illustration of the Method of Simple Average

Years Quarters

I II II IV

1 y1 Y2 Y3 Y4

2 Y5 Y6 Y7 Y8

3 Y9 y10 y11 y12

4 y13 y14 y15 y16

5 y17 y18 y19 Y20

Total T1 T2 T3 T4

Average A1 A2 A3 A4

S.I. s1 s2 s3 s4

S.I. (adjusted) S1 S2 S3 S4

Explanatory notes:
a) T1 = y1 + y5+ y9 + y13 + y17 is the total of y values of first quarter of each year.
Similarly, T2, T3 and T4, are the totals of second, third and fourth quarters of
each year respectively.
b) Ai is the ith quarter average = , where i = 1, 2, 3, 4, and n denotes the number
of years.
∑
c) G is defined as the grand average = .

d) 𝑠 = × 100, 𝑖 = 1, 2, 3, 4.

e) s = s1 + s2+ s3 + s4

178
f) S1, S2, S3, and S4, are the seasonal indices for the first, second, third and the Deterministic Time
Series and Forecasting
fourth quarters respectively, where 𝑆 = × 400, 𝑖 = 1, 2, 3, 4. Note that the
sum of these 4 index numbers must be equal to 400. Further, 𝑆 = 𝑠 if s =
400.
g) For a time series with monthly data, the sum of 12 seasonal indices, one for
each month, must be equal to 1200.
Example: 7.6.1
Compute seasonal indices for the following data by the Method of Simple
Average.

Years Quarters

I II III IV

1992 72 68 80 70

1993 76 70 82 74

1994 74 66 84 80

1995 76 74 84 78

1996 78 74 86 82

Solution:
Table 7.6.1: Calculation of Seasonal Indices

Years Quarters

I II III IV

1992 72 68 80 70

1993 76 70 82 74

1994 74 66 84 80

1995 76 74 84 78

1996 78 74 86 82

Total 3776 352 416 384

Average 75.2 70.4 83.2 76.8

S.I. 43 92.15 108.90 100.52

179
Summarisation of Explanatory Notes:
Bivariate and Multi-
variate Data . . . .
Grand Average 𝐺 = = = 76.4

Seasonal Index for Quarter I, i.e., S1 = 98.43

Seasonal Index for Quarter II, i.e., S2 = 92.15

Seasonal Index for Quarter III, i.e., S3 = 108.90

Seasonal Index for Quarter IV, i.e., S4 = 100.52

Since the sum of these indices = 400, no adjustment is needed.

7.6.2 Ratio to Trend Method
If the data contain trend to an appreciable extent, we first find an appropriate
trend equation to determine the trend for various quarters or months. Usually the
monthly or quarterly trend values are obtained from the quarterly (or monthly)
average trend equation. The trend is then eliminated by expressing the original y
values as percentages of the corresponding trend values. This method is based
upon the assumption that cyclical variations are either not market or completely
absent.
Symbolically, we can write
× 100 = × 100 = 𝑆𝐼 × 100

From this, the irregular component can be eliminated by the use of Simple
Average Method.
Example: 7.6.2
The following table shows the sales (9n’000 Rs.) in a departmental store for five
different years. Obtain the seasonal indices by Ratio to Trend Method.

Years Quarters

I II III IV

2000 502 1632 605 362

2001 526 1700 680 390

2002 556 1820 780 422

2003 590 1955 888 464

2004 632 2110 1002 515

180
Solution: Deterministic Time
Series and Forecasting
Let us fit a straight line trend to the data on quarterly averages. The trend
equation fitted to quarterly averages y = a + bx, where y denotes quarterly
average of the year and the unit of x = 1 year. The table below has been
constructed from the given data by computing the averages of 4 quarters of each
year.
Table 7.6.2(a): Fitting Linear Trend

Years Quarters

y x x2 xy

2000 775 -2 4 -1550

2001 824 -1 4 -824

2002 894 0 0 0

2003 974 1 1 9774

2004 1065 2 4 2130

Total 4532 0 10 730

The normal equations are

∑ 𝑦 = 𝑛𝑎 + 𝑏 ∑ 𝑥
∑ 𝑥𝑦 = 𝑎 ∑ 𝑥 + 𝑏 ∑ 𝑥
Substituting values from the above table into the normal equations for linear
trend, we get
5a = 4532 or a = 906.4
10b = 730 or b = 73.
Thus, the quarterly average trend equation is T = 906.4 + 73x (origin: 2002, unit
of x = 1 year)
Note that we are using T instead of Y (used earlier in fitting of trends) From this,
we can write the quarterly trend equation as
𝑇 = 906.4 + . 𝑥 = 906.4 + 18.25𝑥

(origin: 30th June, 2002 unit of x = 1 quarter)

Shifting the origin to 3rd quarter (mid-point of third quarter) of 2002, the
quarterly trend equation becomes
T = 906.4 + 18.25 (x + 0.) = 906.4 + 9.125 + 18.25x
= 9.125 + 18.25x.
181
Summarisation of Putting appropriate values of x, we can get the trend values (T) for various
Bivariate and Multi-
variate Data quarters. The next step is to express the original values (y) as percentage of trend
i.e., (y ÷ T)×100, giving “trend ratios”. The trend values along with the trend
ratios are shown in Table 7.6.2(b).
Table 7.6.2(b): Calculation of Trend Ratios

Year Quarter x T = y (y ÷
915.525 + T)×100
18.25x

2000 I -10 733.0 502 68

II -9 7751.3 1632 217

III -8 769.5 605 779

IV -7 787.8 362 46

2001 I -6 806.0 526 65

II -5 824.3 1700 206

III -4 842.5 680 81

IV -3 860.8 390 45

2002 I -2 879.0 556 63

II -1 897.3 1820 203

III 0 915.5 780 85

IV 1 933.8 422 45

2003 I 2 952.0 590 62

II 3 970.3 1955 201

III 4 988.5 888 90

IV 5 1006.8 464 46

2004 I 6 1025.0 632 62

II 7 1943.3 2110 202

III 8 1061.5 1002 94

IV 9 10779.8 515 48

182
The trend ratios are now arranged by quarters and the seasonal indices are Deterministic Time
Series and Forecasting
calculated by the method of simple averages.

Table 7.6.2(c) : Calculation of Seasonal Indices

Years Quarters
I II III IV
2000 68 217 79 46
2001 65 206 81 45
2002 63 203 85 45
2003 62 201 90 46
2004 62 202 94 48
Total 320 1029 429 230
Average 64.0 205.8 85.8 46.0
S.I. 63.74 209.98 85.46 45.82

7.6.3 Ratio to Moving Average Method

This method is used when the time series data is composed of all the four
components. We know that moving averages, with period equal to the periodic
variations, and completely eliminates those variations. Thus, if we take 4 period
moving average (M) of quarterly data (or 12 period moving averages of monthly
data), the seasonal variations (and some irregular movements) are eliminated
from the original (y) values. Further, by expressing y as a percentage to moving
average, i.e., (y ÷ M)× 100, we get values consisting of seasonal and irregular
components. The seasonal component is, then, isolated from irregular component
by the use of the Method of Simple Average.

Symbolically, we can write

× 100 = = 𝑆𝐼"
′

Example: 7.6.3
Use the Ratio to Moving Average Method to calculate seasonal indices for the
following data.

Year Summer Monsoon Autumn Winter

2009 30 81 62 119
2010 33 104 86 171
2011 42 153 99 221
2012 56 172 129 235
2013 67 201 136 302

183
Summarisation of Solution:
Bivariate and Multi-
variate Data Table 7.6.3: Calculation of Seasonal Indices by Ratio to Moving Average Method

Year Quarter y 4-period M.T. Centered Total 4-period M.A. (M) ( y ÷ M)×100

2009 Sum 30 - - - -
Mon 81 -- - -- -
292 - - -
Aut 62 597 73.38 84.50
295
Win 119 613 76.63 155.30
318
2010 Sum 33 660 82.50 40.00
342
Mon 104 736 92.00 113.04
394
Aut 86 796 99.63 86.32
403
Win 171 855 106.88 160.00
452
2011 Sum 42 917 114.63 36.64
465
Mon 153 980 122.50 124.90
515
Aut 99 1044 130.50 75.86
529
Win 221 1077 134.63 164.16
548
2012 Sum 56 1126 140.75 39.79
578
Mon 172 1170 146.25 117.61
592
Aut 129 1195 149.38 86.36
603
Win 235 1235 154.38 152.23
632
2013 Sum 67 1271 158.88 42.17
639
Mon 201 1345 168.13 119.55
706 - - -
Aut 136 - - - -
Win 302 - - - -

184
The moving ratios are now arranged by quarters and the seasonal indices are Deterministic Time
Series and Forecasting
calculated by the method of simple averages.

Year Quarters

Summer Monsoon Autumn Winter

2009 - - 84.50 155.30

2010 40.00 113.04 86.32 160.00

2011 36.64 124.90 75.86 164.16

2012 39.79 117.61 86.36 152.23

2013 42.17 119.55 - -

Total 158.60 475.10 333.04 631.69

Average 39.65 118.78 83.26 157.92

S.I. 36.69 118.89 83.34 158.08

Check Your Progress 1

1) The following data represent the production of finished steel tins for the
years 1992 to 1995:
Production of Finished Steel Tins (000 tons)

Year Jan. Feb. Mar Apr. May. Jun. Jul. Aug. Sep. Oct. Nov. Dec.

1992 420 414 502 365 368 332 390 396 429 417 422 496

1993 491 466 516 337 342 360 409 402 372 391 394 446

1994 463 465 478 310 325 406 415 437 438 445 430 416

1995 502 487 536 404 418 429 489 492 475 456 476 476

Compute the seasonal indices by the Method of Simple Averages.

…………………………………………………………………………...…………
....………………………………………………………………..….……..….….…
…………………………………………………………….………………………..
.………………………………………………………………………………..……
…...……………………………………………………………………………....…
………………………………………..................................................................….
185
Summarisation of 2) The following table gives the production of steel in India from 1992 to
Bivariate and Multi-
variate Data 1995 (in ’000) over the different quarters.

Year 1st 2nd 3rd 4th

quarter quarter quarter quarter

1992 1336 1065 1215 1335

1993 1463 1039 1183 1161

1994 1306 1041 1290 1321

1995 1525 1251 1456 1408

Obtain seasonal indices by the method of Ratio to Trend, assuming linear trend.
…………………………………………………………………………...…………
....………………………………………………………………..….……..….….…
…………………………………………………………….………………………..
.………………………………………………………………………………..……
…...……………………………………………………………………………....…
………………………………………..................................................................…

3) Given the following quarterly sales figures in thousands of rupees for the
years 2006 to 2009. Find the specific seasonal by the method of moving
averages.

Years Quarters

I II III IV

2006 290 280 285 310

2007 320 305 310 330

2008 340 321 320 340

2009 370 360 362 380

…………………………………………………………………………...…………
....………………………………………………………………..….……..….….…
…………………………………………………………….………………………..
.………………………………………………………………………………..……
…...……………………………………………………………………………....…
………………………………………..................................................................…

186
4) The seasonal indices for the sales of garments of a particular type in a certain Deterministic Time
Series and Forecasting
shop are given below:
Quarter Seasonal Index
Jan-Mar 97
Apr-Jun 85
Jul-Sep 83
Oct-Dec 135
If the total sales in the first quarter of a year are Rs. 15,000, determine how
much worth of garments of this type should be kept in stocky by the shop
owner to meet the demand for each of the other three quarters of the year?

…………………………………………………………………………...…………
....………………………………………………………………..….……..….….…
…………………………………………………………….………………………..
.………………………………………………………………………………..……
…………………………………………………………………………...…………

7.7 LET US SUM UP

Time series is a set of observation on a variable recorded over time – usually at
equal intervals. The change in the variable concerned can be explained to some
extent on the basis of components of time series. These components are trend,
seasonal variations, cyclical fluctuations and random movements. The observed
value of the variable may be represented either as the product of the aforesaid
components (multiplicative model) or as the sum of the components (additive
model).
Trend can be measured by the method of moving averages and fitting equations
by the method of least squares. Once we estimate the trend, we can predict future
values and also estimate the monthly or quarterly values from the annual trend.
Seasonal variation can be estimated by three methods: Method of Simple
Averages, Ratio to Trend Method and Ratio to Moving Average Method. The
method of simple average is used to average out the irregular component. The
ratio to trend method can be used if cyclical variations are supposed to be absent
while the ratio to moving average method is recommended when the time series
variable is composed of all the four components.

7.10 ANSWERS OR HINTS OT CHECK YOUR

PROGRESS EXERCISES
Check Your Progress 1
1) Read Example 7.3.1 and answer.
2) Read Example 7.3.2 and answer.
187
Summarisation of Check Your Progress 2
Bivariate and Multi-
variate Data 1) y = 42.2 + 2.3x, origin: 2012, unit of x = 1 year (monthly average equation)
y = 42.3 + 0.19x, origin: July, 2012, unit of x = 1 month (monthly equation)
Estimate for March 2011 (x = -16) = 39.23
Estimate for September 2013 (x = 14) = 44.98
2) 45.17 tons
3) 73
Check Your Progress 3

1) 10957, 107.00, 118.69, 82.71, 84.87, 89.19, 99.47, 100.87, 100.11, 99.82,
100.58, 107.12

2) 112.27, 86.62, 100.21, 100.90

3) 104.2, 97.9, 96.5, 101.4

4) Rs. 13144; 12835; 20876.

188
UNIT 8 VITAL STATISTICS*
Structure
8.0 Objectives
8.1 Introduction
8.2 Data Sources
8.3 Uses of Vital Statistics
8.4 Measurement of Population
8.4.1 Linear Interpolation Method
8.4.2 Using Compound Growth Rate Formula
8.4.3 Natural Increase and Net Migration Method
8.5 Vital Rates
8.5.1 Crude Birth Rate
8.5.2 Crude Death Rate
8.5.3 Crude Rate of Natural Increase
8.5.4 Rate of Net Migration
8.5.5 Rate of Total Increase
8.5.6 Infant Mortality Rate

8.6 Life Tables

8.7 Applications of Life Tables
8.7.1 Calculation of Probability of Surviving and dying
8.7.2 Uses in Actuarial Science
8.7.3 Other Applications of Life Tables
8.7.4 Limitations of Life Tables

8.8 Let Us Sum Up

8.9 Key Words
8.10 Some Useful Books
8.11 Answers or Hints to Check Your Progress Exercises

8.0 OBJECTIVES
After going through this Unit, you will able to
 explain the sources of data in vital statistics;
 calculate various vital rates;
 explain the procedure of construction of life tables; and
 appreciate the application and limitations of life tables.

*
Adapted from IGNOU study material of EEC 13: Elementary Statistical Methods and Survey Techniques,
Unit 12 written by C G Naidu with modifications by Kaustuva Barik.
Summarisation of
Bivariate and
8.1 INTRODUCTION
Multivariate Data
Vital statistics is mainly concerned with the factors contributing to population
growth. Some of these factors are birth rates, death rates, expectancy of life, and
migration. As you go through this Unit you will be in a position to appreciate the
importance and applications of vital statistics in economics. The main objectives
of this Unit are to introduce some of the basic concepts of vital statistics, the data
sources, how to measure various ratios, and what are the applications of these
ratios in projecting the population, calculating life expectancy, uses in actuarial
science, etc.

8.2 DATA SOURCES

The data for vital statistics are usually collected through the following four
methods; viz., Registration, Census, Survey and Sample Registration System. We
discuss these methods below:

i) Registration Method: This method consists of continuous and permanent

recording of births, deaths, marriages, migration, etc. Many countries
including India have made registration of births and deaths compulsory under
the law. The registration office issues a certificate on registration of a birth or
death. Although, the registration method in simple and effective it suffers
from the problem that all the births and deaths that are occurring are not
registered. This is because the law has not been enforced strictly, particularly,
in rural India.
ii) Census: Almost all the countries in the world conduct census periodically to
enumerate their population. The census provides the vital statistics
information such as age, sex, marital status, education level occupation,
religion, etc. However, these information pertain to the census years only
(once in ten years). The data for the years other than census years are not
available.
iii) Surveys: Surveys are conducted in areas where the registration method is not
effective or not functioning properly. The surveys enable us to have required
vital statistics of these regions.
iv) Sample Registration System (SRS): This is a large scale demographic
survey conducted in India for providing reliable annual estimates of birth
rate, death rate and other fertility and mortality indicators at the national and
sub-national levels. The field investigation consists of continuous
enumeration of births and deaths by a resident part-time enumerator,
generally a school teacher followed by an independent survey every six
months by an official. The data obtained through these operations are
matched. The unmatched and partially matched events are re-verified in the
field and thereafter an unduplicated count of births and deaths is obtained.
The SRS was initiated by the office of the Registration General, India on a
pilot basis in a few selected states in 1964-65. It became fully operational
during 1969-70 covering about 3700 sample units. Thereafter the sample size

190
 has been periodically increased. The frame was recently updated in 2014 Vital Statistics
comprising 8861 sample units.

8.3 USES IF VITAL STATISTICS

Vital statistics are useful in many spheres of human activity. Some important
uses of vital statistics are as follow.
1) The vital statistics help us in understanding how the population profile of a
country or a region within the country is changing. The profile of the
population is in terms of age, sex, religion, births, deaths, migration,
marriages, etc. these statistics help us in predicting the future population
structure of a country or region.
2) The estimation of population trends and projections help the policy planners
and administrators for better planning and evaluation of economic and social
development programmes. For example, the transportation infrastructure is
influenced directly by the number of people in an area.
3) The mortality statistics help us to improve the health, conditions of the
communities. For example, statistics on communicable diseases help the
health authorities to improve the sanitary conditions of the area affected and
improve medical facilities.
4) Actuarial science including life insurance is based on vital statistics. You will
learn more about it in section 12.4 of this Unit.

8.4 MEASUREMENT OF POPULATION

The total population of a country is usually expressed at a particular point of
time. For example, the latest census in India was conducted in 2011 and the total
population of the country was found to be 1.21 billion on 31.3.2011. The total
population measured at a census is usually considered as accurate. As you may
be aware, the census in India are conducted once in 10 years. The inter-censal
data are estimated using the following methods.

8.4.1 Linear Interpolation Method

The estimation of total population for a given inter-censal year can be calculated
using the following formula:
n
Pt  P0  (P1  P0 ) … (8.1)
N
where, P1 is estimated population at a given inter-censal year t
P0 is population in the previous census

P1 is population in the succeeding census

n is the number of years between the given year and the previous census
year
N is the number of years between the two census years

191
Summarisation of The above method provides us a good estimate at a constant rate between the
Bivariate and inter-censal years.
Multivariate Data
Example 8.1: The total population of India in 1991 census was 846 million and
in 2001 census was 1027 million. Calculate the total population of India in 1996.
Here, P0  846, P1  1027, N  10, n  5
5
Therefore, P1996  846  (1027  846)  936.5 million.
10
The limitation of the above method is that is that we can estimate the population
only for the years between two census years. We cannot have the estimates for
the future years.

8.4.2 Using Compound Growth Rate Formula

Normally it was observed that the population growth takes place in a geometrical
progression. In case the base year population and the population compound
growth rate (between the base census year and succeeding census year) are
known, we can estimate the total population for a given year using the following
formula.
P1  P0 (1  r)n
where, r is the compound growth rate (between the base census year and
succeeding census year)
n is the number of years from the base year (usually previous census year)
P0 is the base year (usually previous census year
Pt is the estimated population at a given year t from the base year

Example 8.2: The population of a small town in 2001 was 50500. The compound
growth rate of the population of that town between 2001 and 2011 was 0.025.
Estimate the population of the town for the year 2015 (assuming that the
population growth rate will be the same beyond 2011).
Here, we are given P0  50500, r  0.025, and n = 14 (since 2015 – 2001 = 14)
Therefore, 𝑃 = 50500(1 + 0.025) = 71355

8.4.3 Natural Increase and Net Migration Method

You know that the census gives us the total population. Similarly, the total
number of births, deaths, and migration statistics are obtained from registrars.
The population of an area increase by:
i) Natural increase (that is, total number of births – total number of deaths)
ii) Ne migration (that is, total number of people immigrated to the area – total
number of people emigrated out of the area).
The population for a given period is calculated using the following formula.
Pt  P0  (B  D)  (I  E)
Where, Pt is the estimated population at a given year t form the base year
(usually previous census year)

192
 P0 is the base year (usually previous census year) Vital Statistics

B and D are the total number of births and deaths respectively during
the base year to the year t.
I and E are the total number of immigrants and emigrants respectively
during the base year to the year t.
Example 8.3: The population of a small town in 2011 census was 22000. From
2011 to 2013 the number of births, deaths, immigrants and emigrants are 800,
150, 2500 and 1500 respectively. Find the total population of the town in 2013.
Here, P0  22000, B = 800, D = 150, I = 2500, E = 1500
Therefore, 𝑃 = 22000 + (800 – 150) + (2500 – 1500)
= 23650
Check Your Progress I
The following table gives information on mid-year total population of India and
annual compound growth rates of population.

Year Population (Crores) Period Compound

growth rate (%)
1950 36.99 1950-60 1.9
1960 44.59 1960-70 2.2
1970 55.50 1970-80 2.1
1980 68.70 1980-90 2.0
1990 84.17 1990-2000 1.8
2000 100.27 2000-2010 1.4
2010 113.17 2010-20 1.1
2020 132.61
Sources: US Census Bureau: IDB Summary Demographic Data for India, 2020

Note that the compound growth rates are in terms of percentage. Divide it by 100
to get the required r. For example, for the period 1950-60 the compound growth
rate is 1.9%. Therefore, r = 1.9/100 = 0.019.
On the basis of the above table answer the questions below:
1) Find the mid-year population for the following years using linear
interpolation method.
Year Mid-year population
1954
1966
1973
1985
1998
2005
2018

193
Summarisation of 2) Find the mid-year population for the following years using compound growth
Bivariate and rate method.
Multivariate Data
Year Mid-year population
1954
1966
1973
1985
1998
2005
2018

3) Compare the above two methods and draw your conclusions.

……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………

4) Briefly explain different data sources for vital statistics.

……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………

5) What are the important uses of vital statistics?

……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………
……………………………………………………………………………

8.5 VITAL RATES

Normally, the data on vital statistics are available in the form of number of births,
number of deaths, etc. In order to have a meaningful utility of these data we
generally transform this data into some vital rates or ratios. Number of births or
deaths in year per 100 persons is usually low and would result in small fractions.
The changes in these ratios would also be not perceptible. In order to avoid this

194
problem, vital rates are expressed on the basis of per thousand persons. In this Vital Statistics
section you will learn some important vital rates.

8.5.1 Crude Birth Rate

The crude birth rate is defined as the number of births per 1000 population in a
specific community or region. To calculate the crude birth rate we use the
following formula:
Annual Number births (in a community or region)
Crude birth rate   1000
Annual mid year population (of the community or region)

The crude birth rate tells us at what rate the births are occurring in a region or
community.
Example 8.4: The mid-year population and number of births occurred of a tribal
community in Madhya Pradesh in 2020 are 40,000 and 1200 respectively. Find
the crude birth rate.
Here, we have 2020 mid-year population = 40000 and the 2020 number of births
= 1200
1200
Crude birth rate   1000
40000
= 30 per 1000 persons per annum

8.5.2 Crude Death Rate

The crude death rate is defined as the number of deaths per 1000 population in a
specific age group or sex group or community or region. To calculate the crude
death rate we use the following formula.

Annual number of deaths (in a specific age group

or sex group or community or region)
Crude Death Rate   100
Annual mid year population (of thespecific age group
or sex group or community or region)

The crude death rate tells us at what rate the deaths are happening in a age group,
sex group or region or community.
Example 8.5: The mid-year population and the number of deaths registered in
2020 for a town in Maharashtra among females are 25000 and 245 respectively.
Find the crude death rate.
Here, we have 2020 mid year female population = 25000 and the number of
deaths in 2020 = 245.
245
Crude death rate (females)   1000
25000

= 9.8 per 1000 persons per annum among females.

8.5.3 Crude Rate of Natural Increase

The crude rate of natural increase is defined as the rate at which a population
increases in a given year because of a surplus of births over deaths expressed as
per 1000 persons.

195
Summarisation of The annual natural increase is measured as: annual number of births-annual
Bivariate and number of deaths.
Multivariate Data
The formula for calculating the crude rate of natural increase is
Annual naturalincrease
Crude rate of natural increase  1000
Annual mid year population

= crude birth rate – crude death rate

The crude rate of natural increase for a given year tells us at what rate natural
increase has added the population over the year.
Example 8.6: In India the crude birth rate and crude death rates in 2016 are 21.4
and 7.5 respectively. Find the crude rate of natural increase.
Crude rate of natural increase = 21.4 – 7.5
= 13.9 per 1000 per annum
8.5.4 Rate of Net Migration
Migration is defined as the movement of people across a specific boundary of a
region for the purpose of establishing a new or sent permanent residence.
Immigrants are those who have come into the region and eminigrants are those
who have moved out of the region.
The annual net migration is defined as: Annual number of immigrants – annual
number of emigrants.
The formula for calculating the annual rate of net migration is:
Annua net migration
Annual rate of net migration   100
Annual mid year population

The annual rate of net migration tells us at what rate the net migration has added
to the population over the course of the year.
Example 8.7: In 2020 for a region the annual number of immigrants, emigrants,
and mid-year population are given as 6500, 5200 and 667700 respectively. Find
the annual rate of net migration.
Here, we have the number of immigrants = 6500
the number of emigrants = 5200
mid-year population = 66700
Annual net migration = 6500 – 5200 = 1300
1300
Annual rate of net migration  1000 = 19.7 per 1000 per annum
66700

8.5.5 Rate of Total Increase

The total increase in population in measured as:
Annual natrual increase  annual net migration.

Annual totalincrease
Rate of total increase   100
Annual mid year population

= crude rate of natural increase + rate of net migration.

196
The rate of total increase for a given year tells us the rate at which the population Vital Statistics
has increased over the year.
Example 8.8: The annual natural increase, annual net migration, and annual mid-
year population in 2018 for a region are recorded as 1500, 500 and 50000
respectively. Find the rate of total increase.
Here, we have
Annual natural increase = 1500
Annual net migration = 500
Mid-year population = 50000
Annual total increase = 1500 + 500 = 2000
2000
Rate of totalincrease   1000
50000
= 40 per 1000 per annum.
8.5.6 Infant Mortality Rate
The infant mortality rate is defined as the member of deaths of infants (less than
one year old) per 1000 live births in a given year. The formula to calculate the
infant mortality rate is given as:
Annual infant deaths (of males or females or total)
Infant mortality rate  100
(Annual live births (of males or females or total)

The infant mortality rate tells us for a given year the chances of a birth failing to
survive one year life. The infant mortality rates can be calculated separately for
males and females.
Example 8.9: In 2019 for a small town the total number of live births and infant
deaths among females are recorded as 3000 and 25 respectively. Find the infant
mortality rate among females.
Here, we have
Annual live female births = 3000
Annual infant deaths = 25
15
Infant mortality rate   1000
3000

= 8.33 per 1000 per annum

Check Your Progress 2
The provisional estimates of crude birth rate, crude death rate, natural growth rate
and infant mortality rate in India for the year 2018 are as follows:
Vital rate Total Rural Urban
Birth rate 20.0 21.6 16.7
Death rate 6.2 6.7 5.1
Natural growth rate 13.8 14.9 11.6
Infant mortality rate 32 36 23

Sources: Sample Registration System Bulletin, May 2020

197
Summarisation of Observe that all the vital rates are higher in rural areas than in urban areas. Write
Bivariate and one most significant reason for each of the following:
Multivariate Data
1) The birth rate in rural areas is high because:
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
2) The death rate in urban areas is low because:
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
3) The infant mortality rate in rural areas is high because:
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................

8.6 LIFE TABLES

The life expectancy is defined as the average number of additional years a person
could expect to live if the current mortality trends continue for the rest of that
person’s life. A life table is a tabular display of life expectancy and probability of
dying at each age or age group for a given population, according to the age-
specific death rates prevailing at that time.
The life table gives us an organized complete picture of a population’s mortality.
We can explain it with an example. We start with a group (usually called
‘cohort’) of 100,000 female births and estimate the number which will survive to
every age or age group, if they are subjected to the existing mortality condition.
We can say, for example, that out of 100,000 initial female births 95,000 will
reach the age of 15 years, 92,500 the age of 25 years, and so on, and the mean
age at which all 100,000 will die is 72 years.
The construction of a life table is a simple process. It involves the following steps
that are repeated for each age group.
i) Age interval (x to x + n): The period of life between two exact ages. The
exact age (x) represents the lower limit of each age interval, beginning with
0 and incrementing to 1, 5, 10, 15 and so on upto 100+ (which is an open
interval). The first and second age groups are usually ‘<1’ and ‘1-4’ and the
last age group is ‘100+’ whereas the rest of the age groups are of equal size,
like ‘5-9’, ‘10-14’, ‘15-19’, …, ‘ 95-99’.
ii) Width of the age interval (nx): This is the number of years in the age
interval (x to x+n). Usually, the first value is 1 (interval <1), the second is 4

198
 (i.e., 1-4) and the remaining values are 5 (namely, 5-9, 10-14, … 95-99). Vital Statistics
The last value is an exception, which is again taken as 1 (100+).
iii) Number of deaths recorded in the age interval (dx): This column
presents the number of persons dying in that age group during the year
corresponding to the life table.
iv) Number of persons in the age interval (Px): This column indicates the
number of persons in the age interval during the year corresponding to the
life table.
v) Separation factor ( n ax ) : This represents the average number of years
lived by those who die between age x and (x+n). Although, it is necessary
in calculations, this factor is not typically presented as a column of the life
table. Each person living in the interval (x to x+n). In a complete life table,
a value 0.5 (that is, half of one year) is valid from the age of 5 years. For a
simpler calculation, it is assumed that those who die in the 5 year age
intervals of a life table live on average 2.5 years. However, remember that
the value of the fraction depends on the mortality pattern over the entire
interval and not the mortality rate for any single year. In addition, since a
large portion of infant deaths occur in the first few weeks of life, this value
is much smaller in the <1 and 1-4 age groups.
Similarly, the death rates in the last three groups (namely, 91-94, 95-99, and
100+) are very high. Therefore, the value of the separation factor is small in the
age group 91-94 and 95-99. In the last age group (100+) since the death is certain
we have taken the separation factor as 1.
Calculation of the separation factor is easy if the date of birth and the date of
death are available. For the purpose of constructing a life table the separation
factor will be given in the table. When they are not, values from model life
tables, such as those tabulated by Coale and Demney shown in Table 8.1 can be
utilised for and the rest are taken as 0.5 years for every year in the group interval
(that is 2.5 in year interval).
Table 8.1: Separation Factors for Ages < 1 and 1-4

Separation factor for age < 1 Separation factor for ages 1-4
Zones Men Women Both Mean Women Both
sexes sexes
Infant North (1) 0.33 0.35 0.3500 1.558 1.570 1.5700
Mortality East (2) 0.29 0.31 0.3100 1.313 1.324 1.3240
Rate South (3) 0.33 035 0.3500 1.240 1.239 1.2390
>0.100 West (4) 0.33 0.35 0.3500 1.352 1.361 1.3610
Infant North (1) 0.0425 0.05 0.05 1.859 1.733 1.7330
Mortality East (2) 0.0025 0.01 0.01 1.614 1.487 1.4870
Rate South (3) 0.0425 0.05 0.05 1.541 1.402 1.4020
<0.100 West (4) 0.0425 0.05 0.05 1.653 1.524 1.5240

Source: Coale, Ansley J. and Demeny P. (1966) Regional Model Life Tables and Stable
Population, Princeton University Press.
Notes: (1) Iceland, Norway and Switzerland; (2) Austria, Czechoslovakia, North-central Italy,
Poland and Hungary; (3) South Italy, Portugal and Spain; (4) Rest, of the World.

199
Summarisation of vi) Central Mortality ( n M x ) : This column results from dividing the number
Bivariate and
Multivariate Data of deaths in the age interval x to x+n (column dx) by the number of people
in this age group (column Px).
dx
n Mx 
Px

vii) Probability of dying between the ages x and x+n  n qx  : The

probabilities of dying are calculated based on the age -specific mortality
rates for each age group. This column is interpreted as the probability of
dying between the ages for the person who has survived up to age x. For the
last age group of the table, wwhere
here death is unavoidable, the probability of
dying is 1. For other age groups, the calculation is more complicated. The
formula for calculation is given below:

viii) Probability of survival between the ages x and x+n  n p x  : It is

interprete
interpreted
d as the probability of a person who reaches age x to reach the
exact age x + n alive. The formula for calculation is given below:

n px  1  n q x

Since, it is 1  n q x , we normally do not show this as a separate column in

the life table.
ix) Survivors to exact age x( n 1x ) : This column indicates the number of
persons living in the age group x to x + n out of the initial cohort which is
usually taken as 100,000.
x) Deaths between the exact ages x and x+n ( x d x ) : This is calculated using
the following formula.

n d x  n1x  n q x

xi) Number of years lived by the total of the cohort of 100, 000 births in
the interval x to x + n ( n Lx ) : Each member of the cohort who survives
the inte
interval
rval x to x+n contributes n years to L, while each member who dies
in the interval x and x+n contributes the average number of years lived by
those who die in this period (that is, the separation factor of deaths n a x ).
The n L x is calculated using the following formula.

n Lx  n x  n1x  n  n a x  n d x

where, n l x  n  n lx  n p x

or,

l
n xn  n lx  n d x
xii) Total years lived after exact age x( nTx ) : This number is essential for the Vital Statistics
calculation of life expectancy. It indicates the total number of years lived by
the survivor n lx between the anniversary x and the extinction of the whole
generation. The value of the first row of nTx is the total number of years
lived by the cohort until death of its last component.
T  Sumof n Lx (from last row of n Lx to the current row of n Lx )
n x

xiii) Life expectancy at age x( n ex ) : Among all the indicators provided by the
life table, the most widely used is the life expectancy ( n ex ) which
represents the average number of years lived by a generation of newborns
under given mortality conditions.
Table 8.2 below provides the basic information required for construction of a life
table. The data pertains to Indian females in 2000. Let us construct the life table.

Table 8.2: Basic Information

Age Number of Number of nx Separation
deaths (dx) people (px) factor
( n ax )
<1 788471 11655599 1 0.1
1-4 430704 44728827 4 1.6
5-9 137870 54725561 5 2.5
10-14 69159 52128201 5 2.5
15-19 100055 48475620 5 2.5
20-24 119360 42745630 5 2.5
25-29 116085 39848328 5 2.5
30-34 109226 35983667 5 2.5
35-39 102540 31934500 5 2.5
40-44 124848 27744053 5 2.5
45-49 150315 23125487 5 25
50-54 172910 19212249 5 2.5
55-59 226553 16258203 5 2.5
60-64 288036 13715985 5 2.5
65-69 354148 10813430 5 2.5
70-74 368365 7554310 5 2.5
75-79 335430 4615527 5 2.5
80-84 252665 2332329 5 2.5
85-89 130278 817817 5 2.5
90-94 42440 183658 5 2
95-99 8199 24796 5 2
100+ 915 1961 1 +
All ages 4428572 488625738
Source: World Health Organisation
Using the formulas given earlier the following life table is constructed.

201
Summarisation of Table 8.3: Life Table
Bivariate and
Multivariate Data Age nx n ax n Mx n qx n px n lx n dx n Lx nTx n ex

<1 1 0.1 0.06765 0.6377 0.93623 100000 6376.52 94261.1 6268416 62.6842

1 to 4 4 1.6 0.0063 0.3765 0.96235 93623.5 3524.63 366.35 6174155 65.9467

5 to 9 5 2.5 0.00252 0.1252 0.98748 90098.8 1127.83 447675 5808120 64.4639

10 to 14 5 2.5 0.00133 0.00661 0.99339 88971 588.245 443384 5360446 60.2493

15 to 19 5 2.5 0.00206 0.01027 0.98973 88382.8 907.437 439645 4917061 55.6337

20 to 24 5 2.5 0.00279 0.1386 0.98614 87475.3 1212.83 434345 4477416 51.1849

25 to 29 5 2.5 0.00291 0.01446 0.98554 86262.5 1247.4 428194 4043071 46.8694

30 to 34 5 2.5 0.00304 0.01506 0.98494 85015.1 1280.57 421874 3614877 42.5204

35 to 39 5 2.5 0.00321 0.01593 0.98407 83734.5 1333.63 415339 3193003 38.1324

40 to 44 5 2.5 0.0045 0.02225 0.97775 82400.9 1833.39 407421 2777655 33.7092

45 to 49 5 2.5 0.0065 0.03198 0.96802 80567.5 2576.57 396396 2370243 29.4193

50 to 54 5 2.5 0.0090 0.04401 0.95599 77990.9 3432.36 381374 1973847 25.3087

55 to 59 5 2.5 0.01393 0.06733 0.93267 74558.6 5019.87 360243 1592474 21.3587

60 to 64 5 2.5 0.0210 0.09976 0.90024 69538.7 6937.35 330350 1232230 17.7201

65 to 69 5 2.5 0.03275 0.15136 0.84864 62601.3 9476.39 289318 901880 14.4067

70 to 74 5 2.5 0.04876 0.21732 0.78268 53126 11545.3 236767 612562 11.5304

75 to 79 5 2.5 0.07267 0.3075 0.6925 41580.7 12786.2 175938 375795 9.03774

80 to 84 5 2.5 0.10833 0.42622 0.57378 28794.5 12272.9 1132900 199857 6.94081

85 to 89 5 2.5 0.1593 0.56964 0.43036 16521.6 9411.37 59079.6 86567 5.23963

90 to 94 5 2 0.23108 0.68236 0.31764 7110.23 4851.74 20996 27487.4 3.8659

95 t0 99 5 2 0.33067 0.82999 0.17001 2258.5 1874.53 5668.9 6491.49 2.87425

100+ 1 + 0.46678 1 0 383.969 383.969 822.593 822.593 2.14235

Life expectancy always decreases from the first row of the table to the last row,
with the exception of the second row and sometimes the third row (age group/5-
9), which can be greater than the first row (age group/<1) in countries with high
infant mortality. It is generally observed that for a given population, life
expectancy is greater in women than in men and overall life expectancy should
be approximately between the two. However, in countries where the maternal
mortality is high the general living conditions of women are worse, life
expectancy among women is lower than men.

8.7 APPLICATIONS OF LIFE TABLES

The life table is widely used in demographic, actuarial, social and health studies.
The principal objective of a life table is to calculate life expectancy at birth and at
other ages. However, life table provides interesting demographic data which have
various applications. In this section you will learn the applications of the life
table.

202
8.7.1 Calculation of Probability of Surviving and Dying Vital Statistics

While constructing life table you have learnt that n q x is the probability of dying
between the two ages (x, x+n) for the person who has survived up to age x. For
example, let us consider the row corresponding to age group 30-34 years in Table
8.3. The probability of dying (females) between 30 to 34 years of age, for those
who have survived up to 30 years of age, is 0.01506 ( n q x ). It means that out of
every 100,000 Indian females who have survived the age of 30 years, 1506 ( =
100, 000 ×0.01506) will die between the age 30 and 34 years. Secondly, n p x tells
us the probability of living between the two ages (x, 30-34 years/ x + n) of
survival is (1 – 0.01506) = 0.98494 ( n q x ). That means out of every 100,000
Indian females who have survived the age of 30 years, 98494 will survive in the
age group 30-34 years.

Thirdly, we can calculate the probability at birth of a person dying between ages
0-4 years. This is given by the number of original births dying ( n d x ) between the
ages 0-4 years, divided by the number of original births (usually 100000). In our
example, n d x  1281 and the probability is 0.01281 (= 1281/100000). This
probability tells us that on and average out of every 100,000 female births in
India (subject to mortality in 2000), 1281 females will die between the ages 0-4
years.
8.7.2 Uses in Actuarial Science
Life tables have important applications in actuarial science especially in the field
of life assurance. Life tables form the basis for determining the rates of premiums
necessary to various amount of life assurance. Life tables provide the actuarial
science with a sound foundation, converting the insurance business from a mere
gambling in the human lives to the ability to offer well calculated safeguard in
the event of death.
Actually, the calculations involved in the fixation of premium amounts in Life
assurance are very complex, but the underlying principles are simple. Let us
consider a few examples.
Example 8.10: According to mortality conditions in India for the year 2000,
what annual premium would an Indian female have to pay on a whole life policy
worth Rs. 100,000 if this life was assured at birth, assuming that the assurance
office earns no income on its funds?
Let the premium be Rs. X per annum. Since a female on the average can be
expected to live 62.7 years, over her life time she will have paid Rs. x × 62.7 in
premiums. This will have to be equal to the value of the policy Rs. 100000.
Therefore, Rs. x × 62.7 = 100000 and x = 100000/62.7 = Rs. 1594.90.
Example 8.11: In the above example if the policy was taken at the age of 25
years, then find the annual premium.
If the policy was taken at age 25 then the total premium paid will be Rs. x × 46.9
for 46.9 years expectation of life at 25 years age. Then the annual premium must
be x = 100000/46.9 = Rs. 2132.20.

203
Summarisation of Example 8.12: In example 8.10 if the policy is an endowment policy, taken at 30
Bivariate and
Multivariate Data
years of age and payable up to 50 years of age or prior deaths. What is the annual
premium to be paid?

If the policy is an endowment policy, taken out say at 30 years payable up to 50

years or prior death, we should proceed on a somewhat different method. From
Table 8.3 we know that 850155 ( n lx ) survivors at age 30 live 1600530 years (=
415339 + 407421 + 396396 + 381374) ( n Lx ) years between ages of 30 and 50.
Consequently, on the average a total of Rs. x × (1600530/85015) premiums will
be collected and hence the annual premium must be Rs. 100000 / 18.83 = Rs.
5311.67.
8.7.3 Other Applications of Life Tables
Apart from its uses in insurance life tables is useful in undertaking comparative
analysis of mortality conditions across countries or region. We discuss some of
the applications of life table below:
i) Calculation of mortality due to specific causes: Life tables for different
groups of population such as sex (male/female), age distribution (different
age groups), and religion are calculated for comparisons. The mortality
statistics may prompt us to find the specific causes of deaths in different
groups of population.
Table 8.4: Life Expectancy at Birth: Select Countries – 2019
Males Females
years years
Australia(a) 80.7 84.9
Canada 79.9 84.1
China 74.5 79.5
France 79.7 85.9
Germany 78.6 83.3
Hong Kong (SAR of China) 82.3 87.7
India 68.2 70.7
Indonesia 69.4 73.7
Italy 81.2 85.6
Japan 81.3 87.3
Korea, Republic of 79.7 85.7
Netherlands 80.3 83.4
New Zealand 80.2 83.6
Papua New Guinea 63.0 65.6
Singapore 81.0 85.4
United Kingdom 79.5 83.1
United States of America 76.1 81.1
Source: www.worlddata.info

ii) Comparison of mortality conditions: The life expectancy at birth and

other ages are the best indices of mortality. These indices considerably vary
from place to place and time to time. Over time, in most countries, life
expectancy has increased steadily due to improved health facilities. As
mentioned earlier, female life expectancy is higher than male life
expectancy expect even though female maternal mortality is high. Table 8.4

204
 above explains the life expectancy for males and females in some selected Vital Statistics
countries.
iii) Population projections: Life tables have also been used in preparation of
population projections by age and sex. That is, in estimating what the size
of the population will be at some future date.

8.7.4 Limitations of Life Tables

Life tables are based on demographic data collected from sources such as census
and SRS. Therefore, life tale estimates have all the disadvantages of any
statistical measure based on population censuses and vital records. Data on ages
and mortality registration may be incomplete or biased. Infant mortality weight
heavily on life expectancy, which means that under-reporting of this indicator, a
habitual fact in many countries can have an important effect on the result of the
tables. Also, important differences in specific age/sex groups with high mortality
may be overlooked, since this would have little effect on the overall life
expectancy.
Constructing life tables for small populations, at the local or sub-regional level, is
generally not recommended, since migratory movements affect the population
structure more than at the regional or national levels. In these cases, a very small
number of deaths can be obtained, which may produce imprecise calculations of
the table’s columns.

Check Your Progress 3

Read the Life Table given in Table 8.3 in the text. Now interpret the values in the
life table by answering the following questions.
1) What is the probability of a female child in India in 2000 would die before
reaching 1 year of age?
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
2) How many years is a female born in 2000 in India expected to live?
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
3) What is the probability of dying of a female between 15 and 20 years of age?
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................

205
Summarisation of 4) What is the mortality rate between 15 and 20 years of age?
Bivariate and
.............................................................................................................................
Multivariate Data
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
5) What is the probability that a female reaching 15 years of age reaches 20?
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................
6) How many additional years is a female between 15 and 20 years of age in
2000 in India expected to live?
.............................................................................................................................
.............................................................................................................................
.............................................................................................................................

8.8 LET US SUM UP

Vital statistics is mainly concerned with births and deaths. The reliability of vital
rates depends upon the effectiveness of the registration system. Incompleteness
of registration of births and deaths, in spite of the laws, has made it difficult to
give a correct picture of birth and death rates.
Life tables present the mortality and survival experience of a whole population
and permit evaluation of its affect on specific groups and over different periods.
It is a simple instrument that is easily constructed with data collected routinely.
It is important to keep in mind that life tables are constructed based on population
data from censuses and mortality registries. Therefore, the quality of the data
affects the validity of the life table.

8.11 ANSWERS OR HINTS TO CHECK YOUR

PROGRESS EXERCISES
Check Your Progress 1
1)
Year Mid-year population
1950 40.03
1966 51.14
1973 59.46
1985 76.44
1998 97.05
2005 106.72
2018 128.72

206
2) Vital Statistics

Year Mid-year population

1954 39.88
1966 50.81
1973 59.07
1985 75.85
1998 97.08
2005 107.49
2018 122.55

3) The estimated mid-year populations using linear interpolation method are

slightly less than the method using compound growth rate method.

4) See Section 8.2 and answer.

5) See Section 8.3 and answer.

Check Your Progress 2

1) The birth rate in rural areas is high because of the lack of awareness among
the people on the family planning methods and its need.

2) The death rate in urban areas is low because of the improved health facilities
in towns and cities.

3) The infant mortality rate in rural areas is high because of the lack of health
facilities in rural areas and malnutrition among mothers.

Check Your Progress 3

1) The Probability for a female under 1 to die in India in 2000  1 q 0  is 0.06377.

2) The number of years that a female born in 2000 in India expected to live
 1 e0  is 62.68 years.
3) The probability of a female dying between 15 and 20 years of age group
 5 q15  is 0.01027.
4) The mortality rate between 15 and 20 years age group  5 M15  is 0.00206.
5) The probability that a female in the 15-19 age group reaches 20-24 years age
group  5 q15  is 0.98973.
6) The life expectancy of a female in the age group 15-20 years in 2000 in India
 15 e15  is 55.63 years.

207