INDEXNUMBERS

10-11
Example 10-2, The prices of agricultural commodities for 1946-47 and for the month of December
1950are given below along with the value of the output of these commodities in 1946-47:
Prices Value of output
Commodities Unit
inmillion (Rs.)
maund
1946-47 (Rs.) Dec. 1950 (Rs.) (1946-47)
Rice 13-75 13-75 8364
maund 9.70
Wheat 9.70 2207
maund 6-03
Jowar 800 876
Cotton(raw) lbs.
466-00 433-00 701
lbs. 1-25
Tea 1-75 534
Clculate the weighted index number of prices of these commodities for December 1950
1946-47as base. taking
Salution. Since the quantities consumed in the current period (Dec. 1950) are not known we can only
use Laspeyre'sformula for obtaining price index number.
TABLE 10-2. COMPUTATION OF LASPEYRE'S INDEX
Po Pi Po g0
Commodities
(1) (2) (3) (4)= (3) + () (2) x (4)
Rice 13-75 13-75 8364 608-29 8363-99
Wheat 9.70 9.70 2207 227-53 2207-00
Jowar 6-03 8-00 876 145-27 1162-16
Cotton (Raw) 466-00 433-00 701 1-50 649-50
Tea 1-25 1-75 534 427-20 747-60
E po go = 12682 }p, qo = 13130-25
Laspeyre's Price Index is given by:
Pola = Zp1 90 13130-25
X 100= 19682.00 X 100= 103-53.
po q0
Example 10-3. The table below gives details of price and consumption of 5 commodities for 1995 and
1997. Using an appropriate formula arrive at an index number for 1997 prices with 1995 as base.
Commodities Price per unit 1995 (Rs.) Price per unit 1997 (Rs.) Consumption value 1995 (RS.)
Rice 40 48 800
Wheat 25 27 400
Oil 95 105 760
Fish 110 120 I100
Milk 80 100 480
[.C.W.A. (Intermediate), Dec. 1998]
Solution. Since we are given the base year (1995) consumption values (Po4o) and current year
quantities (gi) are not given, the appropriate formula for index number is Laspeyre's price index.
INDEX
TABLE 10-3. CALCULATIONS FOR LASPEYRE'S PRICE
Base Year 1995 Current Year 1997
Commodity
Price per unit (p)Consumption Value Price per nit P)
(Rs.)

ET
(Rs.) (Rs.)Po40 P1o
() (2) (3) (4)
Rice 40 800 48 20 960
Wheat 25 400 27 16 432
Oil 95 760 105 840
Fish 110 1100 120 1,200
Milk 80 480 I00 6 600
E Po 4o =3,540 }p,qo = 4,032
 FUNDAMENTALS OF STATISTA
10-16
|Again, if (**) holds then
Pola = x 100 = x 100
x 100 = Ep; (g/k) x 100
}po 40 (Using (")] Popo 40 Epo(qlk)
[Prom (t
cZpo90 x 100 = 100 c
x100 = Epo41 x 100= Po
Po,Pa Zpi91 x 100 = x 100
k
po91 Epo91 [Using (")) Hence, if (**) holds, we again have Po =Pis
cEpo91 X 100 = 100 c

Hence if (*) holds then Po,la = Poifa

This proves the statement by
Karmal.
Laspeyre's price index is greater than Paasche's. When wosl
Example 10-11. Show that, in general, [DelhiUniv. B.A. Econ. (Hons.), 1996. 10
this not be so ?
price-relatives and quantity-Telatives respectively. and
denote the
Solution. Let the variables x and y
be the value in the base year so that
and w= value in the base year= Po o
y). The
Po
frequency distribution where w is the weight (frequency) of the pair (z,
Consider the bivariate given by :
coefficient between x (price-relatives) and y (quantity-relatives) is
the correlation
Covariance (x, y), Cov (x, y) ro,o, = Cov (r, y)
..

r=

(weighted) distribution
For a bivariate frequency
Efu-)(-) }w(r- ) (-y)
Cov (r y) = Sw

Z(2)()
Sw
we get
Substituting the values of x, y and w from (),

Cov (x, y) =
zp)()] X
£po 90
Lpo90
}poqo
Epog1
Epogo
Epofo
Po,a [From (i))
Po,Pa

where Vo po 90 index numbers respectively

is the value index and Po and
Po" are Laspeyre's and Paasche's price
Po,la
=
1 PoPu
Po,Pa Vo
Po,e
=l
Po, Pa Vo1
 10-17
under
normal economic conditions, the
practice,
aivey
Ia quantity consumed of any commodity will be
lesSWhen prices are higher i.e., the variables x(price-relatives) and y (quantity-relatives) are
asatiely
coTelatedsothat ry<l. Further since o, and o, are non-negative and Vo1 is always positive, we

fom
(an)

Po,la
=1+Some non-negative quantity
Pola
>1
Po,Pa Pola Poe,
Hence,ingeneral, Laspeyre's price index is greater than Paasche's price index. In other words.
spriceindexhas an upward bias in general, with rise in price.
thenfrom (i), we get However,in rare situations when
Poa
Po, Pa - Some positive quantity
Pola
<1
Po,Pa Pole< Pou
it,ifthe correlation coefficients between the price relatives and quantity relatives is positive (a
tenomenon which is rarely observed in economics), then
miceindex,
Laspeyre's price index is less than Paasche's
Burther from (i) we see that Po =PoPa, if
(a) either ry,=0
(b) or o, =0 =Constant for each commodity,
ie., the price movements are same for all the commodities.

(c) or g, =0 =Constant for each commodity,

ie., the quantity movements are same for all the commoditis.
Example 1012. Prove that Fisher's ideal index number lies between Laspeyre's and Paasche 's index
TAmbers.
Solution. Let us consider two real positive numbers a and b such that a>0 and b>0.
Let a<b Also a <b
a' < ab [Multiplying by a> 0] ab <b? [Multiplying by b> 0]
a<Vab Vab <b (6>0)
(:: a>0, negative sign is rejected)|
Hence a<b a<Vab<h ...(*)
S, the geometric mean of two real positive numbers lies between them. Hence, Fisher's ideal index
She geometric mean of Laspeyre's and Paasche's index numbers lies between them. More
Mecisely,
if Po, la < Po, Pa then Poie< Po<Poe
and
if Po,Pa <Poa then PoPe< Pof<Pola
Remarks 1. In particular if Po, a= Po,then all the three indices are equal.
2. In
have : fact, Marshall-Edgeworth index also lies between Laspeyre's and Paasche's indices. Rather we

and
if Po la < Po Pa then Po,le< Po<PoME< Po,Pa
if PoPa <Pola then PoPa< Po<PoME< Po,la
 FUNDAMENTALS OF STATISTICA
10:18
relatives.
this method, first of all we obtain the price
1053. Simple Average of Price Relatives. Inby expressing the price of the commodity in the Curre
each commodity. The price relatives are obtainedi. e.,
year,
year as a percentage of its price in the base
P x 100
P =Price Relative for a commodity = Po ...(1017)
numbers for each commodity. The price index for th.
Price-relatives are the simplest form of the index of centi
price-relatives by using some suitable measure
composite group is obtaincd on averaging these geometric mean (G.M.). Price index using simple
arithmeti
tendency, usually arithmetic mean (A.M.)or
mean of the relatives is given by :
P(A.M) = x 100)-xp ..(1018
where n is the number of commodities in the group.
relatives, the price index is given by :
Using simple geometric mean of the price
Po1 (G.M.) -n10)"-[nP]" ...(1019)
(10-19), we le
the product of the price-relatives for the n commodities. To evaluate
where II denotes
logarithms. Takinglogarithm of both sides in (10-19), we get

log Po,(GM.) logx100)=log P n

...(10-19)
Po1 (G.M.) =AntilogElog P|
overcomes
onthe simple average of the price-relatives
Merits and Demerits. The index number based
aggregate method', viz.,
some of the drawbacks of the 'simple hence the index
Price-relatives are pure numbers independent of the units of measurement and
(i)) the units in which the prices are quoted.
number based on their average is not affected by
observations (large and small price quotations) do not influence the index unduly. I
(ii) The extreme observations.
gives equal importance to all thus neglecs
drawback of this method is that it gives equal weights to all the commodities andaverage of the
The drawback is removed by taking the weighted
their relative importance in the group. This
price-relatives as discussed in § 10-5-4, page 10-20.
of this method is the choice of the average to be used. As already discussed, [Link]
Another limitation However, because of
theoretically a better average than [Link] advocated the use
though difficult to compute, is economists, notably F.Y.
computational ease, A.M. is used in practice. Somebut it did not find favour with others and is seldom useu
price-relatives
of harmonicmean for averaging the positively skewed and the skewnes
Remark. The distribution of the price-relatives is found to be
away from the given year.
increases as the base 1s shifted more and more
year fron the following average annual wholesae
Example 10-13. Construct Index Number for each
:
prices ofcotton with 1993 as base Year
Wholesale Prices (Rs.)
Year
Wholesale Prices (Rs.)
1998 70
1993 75
1999 69
1994 50
2000 75
1995 65
60 2001 84
1996 80
72 2002 yearas
1997
prices in the current
Solution. The index numbers for each year are obtained by expressing the
1993.
apercentage of the price in the base year
OEX
NUMBERS
10.19
TABLE 10-10. COMPUTATION OF PRICE INDEX
NUMBERS
Year Wholesale Prices Index Number
(Rs.) (Base: 1993 = 100)
1993 15 100
1994 50 50
5 X100 = 66-67
1995 65 565 X100 86-67
1996 60 560 X100= 80-00
1997 72 72
75 x100= 96-00
1998 70 70
75 x 100= 93-33
1999 69 69 X 100 = 92-00
75

2000 75 75 x 100= 100-00

84
112-00
2001 84 75 x 100 =
80 x 100= 106-67
2002 80

Example 10:14. The following are the prices (in Rs.) of commodities in 1995 and 2000. Calculate a
well as
mean as geometric mean.
nrice index based on price-relatives using the arithmetic
Commodity
Year
B C D
A 120
50 85
45 60 20
1995 75 90 130
70 30
2000 55
Solution. BASED ON A.M. AND GM.
CALCULATIONS FOR PRICE INDEX
TABLE 10-I1. Price Relative
Price log P
Commodity
In 2000 (p) x100
In 1995 (p) 2-0871
122-22
5 2-0667
45 116-67
A 70
60 2-1761
B 150-00
30 2.1761
20 150-00
C 75 2-0246
50 105-88
D 90 2-0347
85 108-33
E 130
XP= 753-1 X log P= 125653
120

Arithmetic Meanis:
Index Number based on 753 125-517
=) x 100 =~P=%
Po,(A.M) \Po
by :
n

given
Geometric Mean is
Number basedon 2-0942
Index P= x 12:5653=
Jog Po (G.M.) =Zlog
Antilog (2-0942) = 124-3.
Pot (G.M.)=
10-20 FUNDAMENTALS OF STATISTICS

as base year from the

Example 10-15. Calculate Price Index for 1995 and 1996 using 1990 fol owing
data :
Prices (Rs. per unit)
Commodity
1990 1995 1996
4
A 5
10 7
B 7
12 6
C 8
20 17 16
D
550 540
S00
Solution. INDICES
TABLE 10-12, CALCULATIONS FOR PRICE
Prices Price Relatives
1995 1996 For 1995 For 1996
Comnodity 1990
P (p;/po)x 100 (P2/Po) x100
Po
6 4 120-00 80-00
A
7 10 7 142-86 100-00
B
8 12 6 150-00 75-00
C
20 17 16 85-00 80-00
D
550 540 110-00 108-00
500
Total
607-86 443-00

=x10)=
Priceindex for 1995 Po
= 121-57

P =88-6
Price index for 1996=Xx10044300
Po
10-5-4. Weighted Average of Price Relatives. The shortcoming of Simple Averagemethod of Relatives
in this which
Method which assumes that all the relatives are equally important is overcome
importance theof
consists in assigning appropriate weights to the relatives according to the relative weighted
different commodities in the group. Thus, the index for the whole group is obtained on taking the
on weighted A.M., the price index is
average, usually A.M. or G.M. of the price-relatives. Thus, based
given by :

WP ...(10-20)
Po1 (A.M.) =

where Wis the weight attached to the price-relative P.

STEPS FOR COMPUTING Po1 (A.M.) IN (10-20)
1. Find the price-relatives (P) for each commodity, ie., compute P = Po
X 100

2. Multiply the price-relatives in Step 1 by the corresponding weights (W) assigned to get the product
WP.
[Link] the sum of products obtained in Step 2for all the commodities to get EWP.
4. Divide the sum in Step 3 by XW, the total of the weights assigned.
The resulting figure gives the price index based on the weighted average of price-relatives.
The price index based on the weighted geometric mean of price-relatives is given by
..(1021)
Po1 (weighted G.M.)
NDEN
NUMBERS 10-21

logarithmof both sides, we get

Toking
W
log P
\Po

Po (weighted G.M.) =Antilog W

log P) ...(10-22)
W
Forcomputationalpurposes, formula (10-22) is used and requires the following steps.

FOR COMPUTING Po1 (G.M.) IN (10-22)

STEPS
P1
Computethe price-relativessP= Po x 100, for each commodity.
1.
[Link] of all the price-relatives. This gives us log Pvalues.
3. Multiply log P values for each commodity by the corresponding weights (W) assigned. This will
values.
give (W. log P)
Find the sum of the values (W. log P) in Step 3 over all the commodities to get Wlog P.
5. Divide the sum obtained in Step 4 by W, the sum of weights.
6. Antilog of the value obtained in Step 5 gives required price index.
Remarks 1. Since price-relatives are the simplest form of the index numbers, we may also use the
otetion I for P, i.e., we may write
x 100 ...(10-23)
Po
2. In the usual notations, we have :

Po,le }p190 X 100 = E PW

...()
EW

where P Pi x 100, is the price-relative of the commodity

Po
and W = Po qo, is the value of the commodity in the base year,
the summation being taken over different commodities.
From (*), we conclude that Laspeyre's price index is the weighted mean of the price-relatives, the
corresponding weights being the values of the commodities in the base year.
Similarly, Paasche's price index number is given by :

X WP
Po, Pa X 100 = ...(**)
}po91
where P= X 100 and W= Poq1.
Po
Hence, Paasche's price index can be expressed as the weighted mean of the price-relatives, the
corresponding weights beingw= Pogi i.e., the values obtained on taking the current year quantities at the
base year prices.
Example 10-16. The following table gives the prices of some food items in the base year and current
Vear and the quantities sold in the base year. Calculate the weighted index number by using the weighted
aVerage of price-relatives.
10-22 FUNDAMENTALS OF STATISTICS
Items Base Year Quantities Base Year Price Current Year Price
(Units) (in Rs.) (in Rs.)
A 7 18-00 21-60
B 6 3-00 4-65
16 7-50 9-00
D 21 2-50 2-25
(C.A. (Foundation), May 1999)
Solution.
TABLE 10-13. CALCULATIONS FOR PRICE I NO, USING PRICE RELATIVES
Item Base Year Quantities Prices in Rupees Price Relatives
Weights (W)
Base Year Current Year WP
P= X100
(P) (p)
A 7 18-00 21-60 120 840
B 6 3-00 4-65 155 930
C 16 7-50 9.00 120 1920

D 21 2-50 2-25 90 1890

E W= 50 WP=5580

Price index number for the current year (Base Year = 100), using weighted average of price-relatives is
given by :
E WP 5580
Poi (Weighted A.M.)= 50 111-6.

Example 10-17. A departmental store sells stereo systems, television sets and radios. The percentage
distribution of the total sales volume (in rupees) is estimated as 30per cent stereos, 50per cent televisions
20,000,
and 20 per cent radios. The prices of one stereo, one television and one radio in 1990 were Rs.
Rs. 15,000 and Rs. 500 respectively while their respective prices in 1995 were Rs. 25,000, Rs. 20,000 and
Rs. 800.
Aweighted price index for 1995 with base 1990 is to be computed.
(i) Which index number formula is appropriate, why ?
(ii) Compute the index. (Delhi Univ. B.A. (Econ. Hons.), 1997]
Solution. (i)) The most appropriate index number formula in this case is the weighted average of the
price-relatives, the weights being the percentage distribution of total sales volume (in Rs.) for different
commodities. For the calculation of this index number, we need only the prices of the various commodities
(stereos, televisions and radios in this case) in different years. This approach also gives due weightage to
each commodity, assuming that the sectoral distribution did not change during this period.
TABLE 10-14. CALCULATIONS FOR INDEX NUMBER
(i)
Price (in Rs.) per unit Weights (W) Price Relative
Commodity
P WP
1990 (P) 1995(P) (Share in total sales) P= x100
Stereo 20,000 25,000 30 125-00 3,750
50 133-33 6666-66
Television 15,000 20,000
500 800 20 160-00 3200
Radio
WP= 13616-66
W= 100
Hence, the weighted price index for 1995 with base 1990, is given by :
WP 13616-66 = 136-17,
Pon = EW 100
INDEXNUMBERS 10-23

Example 10-18, Calculate index number of prices for 1995 on the basis of 1990from the data given
below:
Commodity Weight Prce per unit 1990(Rs.) Price per unit 1995 (Rs.)
A 40
16 20
B 25
40
20
12 15
D 15
2 3
Ifthe weights offcommodities A, B, C, Dare increased in the ratio l : [Link], what will be increase in
indexnumber?
[1.C.W.A. (Intermediate), June 1998]
Solution.
TABLE 10-15. CALCULATIONS FOR INDEX
Price per unit in
NUMBERS
Price Increased
Commodity Weight Rupees Relative WP W, P
(W) Weight
P, (W)*
1990 (Po) 1995 (P) P=x100
16 20 40
40 125
A S000 40 +10 =44 5500

B 25 40 50 125 3125 2542x25=30 3750

10

20 12 15 125 2500 20 + 3x 20
10 = 26 3250
C

15 2 3 150 2250 15 + 2 1 3150

D 10

EW= 100 E WP= 12875 E W, = 121 W,P= 15650

:2:3:4, (Total =10), the increase
* Since the weights of the commodities are increased in the ratio 1
in weights are :
3 4
2 X 25 = 5. (C):0 x 20=6, (D):0 x 15=6
(A): 0 x 40 = 4, (B) :10
WP 12875
100=128-75
Original Index Number () = W
X WP_ 15650 = 129-34
New Index Number (1) =s W, 121
Increase in the Index Number =I-l=0-59.
the budgets of middle class families in afamily gave the following
xample 10-19. An enquiry into
information : Clothing Fuel Others
Rent
Expenses on Food
15% 20% 10% 25%
30% 20 40
70
Prices (in Rs.)in 1997 100 20 55
60 15
Prices (in Rs.) in 1998 90 20

Compute ihe price index number using : ofprice-relatives.

(i) Weighted G.M.
() Weighted A. M. of price-relatives.
10-24
FUNDAMENTALS OF STATISTICA
Solution. TABLE [Link] OF PRICE INDEX USING A.M. & G.M
Price
Groups Weights 1997 1998 PriceRelatives WP log P
(W)
Wlog P
(P) (P:)
P=x 00
Food 30 100 90 90-0 2700 1-9542
Rent 15 20 20 100-0 1500 2-0000 58-626
Clothing 20 70 60 85.7 1714 1-9330
30-000
Fuel 10 20 15 75-0 750 1-8751 38-660
Others 25 40 55 137-5 3437-5 2-1383
18-751
) WP = 10101-5
S3-457
W
log P=
() Index Number for 1998 w.r.t base year 1997, based on the weighted arithmetic 199-494
relatives is given by mean of price.
WP 10101-5
Po (A.M.) = 100 =101-015
(ii) Index Number for 1998 w.r.t base year 1997, based on the
relatives is given by weighted geometric mean of Drica

log Po1 (G.M.) = 2Wlog P_199.494

100 = 1-9949

Po (G.M.) = Antilog (1-9949) = 98-83.

EXERCISE 10-1
1. (a) What is an index number ? Explain the various problems
involved in the construction of index numbers
(b) Discuss the problems faced while constructing an index number.
(c) What are the important points which have to be considered in the
(Delhi Univ. B. Com. (Pass), 2000
construction of index numbers ?
2. "For constructing index numbers the best method on
[CA. (Foundation), May 1997
point of view." Discuss. theoretical ground is not the best method from practical
(Delhi Univ. B. Com. (Hons.), I99]
3. It has been stated that the technique of index
number construction involves four major factors :
(a) choice of items ; (b) base period (c) form of average
Do you agree with this view ? If so, explain these four factors and
(d) weighting system.
youdo not agree, give your main views on the main problems involved in discuss the problems to which they give rise. Ii
4. "An index number is a special type of
index number construction.
average". Discuss.
5. In the construction of index numbers, the
mean". Discuss. advantages of geometric mean are greater than those of arithmetc
6. What are index numbers ? Why are they called
economic barometers ? [Delhi Univ. B. Com. (Pass), 2002]
7. (a) What is implied by "weighting" in the process of
index number construction ? Why is it necessary ? What
are the commonly proposed weighting schemes ?
(b) What is the importance of weighting in the
and geometric mean--will you prefer incalculating construction
of
index numbers index numbers
and why ? ? Which of the three-mean, medan
8. (a) What are Index Numbers ? How are they
general Price Index Numbers.
constructed ? Explain the role of weights in the constructon o
(b) Distinguish belween unweighted and weighted index
numbers. Enumerate some of the important methods o
weighting a price index and discuss their relative merits and demerits
9. (a) Explain that the Laspeyre's Method has an
Point out, under what conditions
upward bias while the PaaSche's Method has a downward bi
) they give equal results, (i) Laspeyre's method gives a result lesser than
Paasche's method.