Preview

INTERNATIONAL LABOUR OFFICE

COST-OF-LIVING STATISTICS
Methods and Techniques for the Post-War Period

Report prepared for the Sixth International Conference
of Labour

Statisticians

(Montreal, 4-12 August 1947)

GENEVA
1947

STUDIES AND REPORTS
New Series, No. 7
(Part 2)

P U B L I S H E D B Y T H E INTERNATIONAL L A B O U R

OFFICE

Geneva, Switzerland

Published in the United Kingdom for the INTERNATIONAL LABOUR
by Staples Press Limited, London

OFFICE

Distributed in the United States by the INTERNATIONAL LABOUR OFFICE,
Washington Branch, 1825 Jefferson Place, Washington, 6, D.C.

PBINTED BY ATAR,

GENEVA

CONTENTS
Page
PREFACE

INTRODUCTION

Discussions on Cost of Living a t the Second International
Conference of Labour Statisticians
The Task of the Sixth Conference
Clarification of t h e Concept of "Cost of Living"
New Functions of the Cost-pf-Living Index
Objectives
CHAPTER I. Purposes of Cost-of-Living Indices
Measuring Changes over Time in Retail Prices of Consumer
Goods
Measuring Place-to-Place Differences in Costs of Living at a
Given Time
Groups which should, if Feasible, be Covered by Cost-of-Living
Indices
Wage-Earners and Clerical Workers
Farmers
Other Groups
Total Population
CHAPTER II. Techniques for Obtaining and Utilising Representative
Price Data
Selection of a Sample of Representative Outlets
Methods of Determining Goods of the Same or Equivalent Quality
Training of Personnel
Administration of Price Collection Programmes
Central Control of Price Collection
Studies and Correction of Biases in Reporting
Special Problems of Illegal and Black Market Prices . . . .
Criteria to be Used in Choosing Sample of Communities in
which Prices are to be Obtained
CHAPTER I I I . Weighting Diagrams
The Items to be Priced: Criteria for Selection
Determination of Weights for Items Priced
Imputed Weights
Methods of Handling Seasonal Items
Fresh Fruits and Vegetables
Clothing
Fuel

1
1
3
3
4
6
6
13
14
14
16
16
17

19
19
20
23
24
24
26
27
30
33
33
35
35
36
36
40
41

CONTENTS
Page

Need for New Studies of Consumer Purchases
Changes in Consumption Patterns
Criteria for Determining the Need for New Studies of Consumer Purchases
Use of Small Sample Surveys
Need for Full-Scale Studies of Consumption
Special Studies of Conditions of Family Living in Less WellDeveloped Areas
Methods of Introducing New Weighting Diagrams
CHAPTER IV. Proposed Resolutions

42
42
44
45
46
46
47
49

APPENDICES

I. Resolutions of the Second International Conference of Labour
Statisticians, Geneva, April 1925

II. United Kingdom: Cost of Living Advisory Committee, Interim
Report

PREFACE
The Sixth International Conference of Labour Statisticians
was held in Montreal from 4 to 12 August 1947 under the auspices
of the International Labour Organisation and adopted Resolutions setting international standards for the compilation and
publication of statistics of employment, unemployment and the
labour force, cost of living and industrial injuries.
The present report on cost-of-living statistics, including
proposed Resolutions, was printed and circulated in advance
to the Governments. As the report, although prepared primarily
for the Conference, may be of general interest as a study of
the subject, it has been reprinted separately from the account
of the proceedings of the Conference itself. The present report
is as submitted to the Sixth International Conference of Labour
Statisticians except for a few minor corrections.
The Resolutions as finally adopted, together with a résumé
of the discussions in the several Committees dealing with the
different topics on the agenda of the Conference, will be found
in the General Report of the Conference, which is issued in a
separate publication.1

Studies and Reports, New Series, No. 7, Part 4.

INTRODUCTION

Discussions on Cost of Living
at the Second International Conference of Labour Statisticians
The subject of cost of living is not new to the International
Conference of Labour Statisticians, since it was on the agenda
of the Second Conference held in April 1925, at which a series of
Resolutions touching various aspects of the calculation of
index numbers was adopted. These are reproduced in Appendix I.
For this Second Conference, the Office prepared a report x
on methods of compiling cost-of-living index numbers, including
a statement of their purposes and nature, an examination of
methods of construction (items, weights); a description of
agents and methods of pricing (coverage of towns, selection of
retail outlets, techniques of quality control); and methods of
calculation. In an appendix a short description of the costof-living index numbers available in the different countries was
given. a
The Governing Body of the International Labour Office, at
its 100th Session (Montreal, October 1946), authorised the
Office to convene the Sixth International Conference of Labour
Statisticians in Montreal in August 1947, and to include on the
agenda, as its third item: "Cost-of-living statistics; wartime
and post-war developments in methods and techniques".
The Task of the Sixth Conference
Attention is being focused upon post-war questions and
issues in respect of cost of Jiving. A series of new problems has
emerged and old problems have appeared with new emphasis.
1
Methods of Compiling Cost-of-Living Index Numbers, Studies and
Reports, Series N, No. 6 (Geneva, 1925).
2
For the proceedings of the Conference see The Second International
Conference of Labour Statisticians, Studies and Reports, Series N, No. 8
(Geneva, 1925).

COST-OF-LIVING STATISTICS

New techniques and methods have been developed for improving
cost-of-living index numbers in the post-war world. These new
developments should be taken into account in restating and
supplementing the Resolutions of the Second Conference and
in formulating new resolutions on cost-of-living indices.
The present Conference, then, will consider the general
problems of improving cost-of-living indices. The discussions
will cover the following topics:
I. Purposes which may be served by different types of
cost-of-living indices in the post-war period:
A. Measuring changes over time in retail prices of consumer
goods: (1) the pre-war consumption level of a given economic
group in specified types of communities; (2) the current consumption level of a given economic group in specified types of
communities; (3) a consumption level meeting a given standard
of adequacy but taking account of price-induced substitutions
which may be made without altering the level of satisfactions
provided;
B. Measuring place-to-place differences in living costs at a
given time;
C. Groups which should, if feasible, be covered by cost-ofliving indices: (1) wage-earners and clerical workers (moderate
income families) in large cities, in small cities, and in rural
communities; (2) farmers; (3) other groups; (4) total population
(weighted by total consumer purchases).
II. Techniques for obtaining and utilising representative
price data:
A. Selection of a sample of representative outlets;
B. Methods of determining goods of the same or equivalent
quality;
C. Training of personnel;
D. Administration of the price collection programme;
E. Special problems of illegal and black market prices;
F. Criteria for choosing sample of communities for price
collection.
III. Weighting diagrams:
A. Criteria for selecting items to be priced;
B. Determination of weights for items priced;
C. Need for new studies of consumer purchases;
D. Methods of introducing new weights.

INTRODUCTION

Before proceeding to a discussion of these various topics,
three points may be briefly considered.
Clarification of the Concept of " Cost of Living "
At the outset, there is need for clarification of the concept
of the cost-of-living index. In the United States, for example,
the term "consumers' price index", or index number of retail
prices of consumers' goods, has been substituted for the designation "cost-of-living index number". This was done on the
ground that the term "cost of living" was often interpreted
to include changes of living costs due to other causes than price
changes. Changes in costs of living of workers due to their
transfer to new mushroom towns, where living costs were
extremely high in comparison with costs in the places from
which the workers had come, are not included in the usual
concept of cost-of-living index; and such indices are not designed
to measure such costs. It was considered important that the
index should not be misjudged, hence the change of name.
This clarification is of course in line with the general concept
of cost-of-living index numbers adopted by most countries;
the index number is designed to measure, by means of appropriate weighting, changes in retail prices of goods and services
consumed in the particular groups to which the index relates.
On this point, the Office suggests the following:
In order to promote understanding of the nature and uses of indices
of retail prices charged a particular group, the term "cost-of-living
index" should be replaced in respect of such indices by the term
"retail price index" or "consumer prices index"1 specifying the particular group to which the index applies (I, 12).
New Functions of the Cost-of-Living Index
In the second place, the war brought changes affecting the
role and function of the cost-of-living index. In some countries,
at least, attempts were made to control price movements, and
to substitute rationing, allocation of resources, export permits,
price ceilings, etc., for the control exercised in a free economy
by the mechanism of price changes. As a result, price changes
affecting the cost of living of wage-earners were, in a number of
countries, subjected to control and in these countries, therefore,
1
Throughout the report, the roman numeral refers to the number
of the Resolution, and the arabic numerals to the paragraphs and subparagraphs.

COST-OF-LIVING STATISTICS

the cost-of-living index had, in addition to its function of
measuring price changes, a derivative usefulness in indicating
the success or failure of price control. Successful price control
meant that there were no important changes in cost of living.
The indices were watched to see that they did not move. If
the indices did not move, in so far as changes in the cost of
living were concerned, wages would not need to be readjusted,
war costs would be stable, and the war economy would tend to
function smoothly. This attitude might imply a dangerous
complacency if the index showed no upward trend; and with
regard to wage adjustments the lack of movement of the index
might easily satisfy those with administrative responsibilities
that no action was required.
In some cases, furthermore, especially where wage adjustments were made dependent upon the movements of the cost-ofliving index, administrative action was taken with a view to
preventing the cost of living from rising. For example, subsidies were paid to farmers and others to prevent a rise in
retail prices of certain foodstuffs to the consumer. In this connection it has sometimes been alleged that these subsidies were
paid particularly in respect of those food items that were
included in the index, to the neglect of other food items not so
included: with the result that the index ceased to reflect the
real movement of all food prices, but actually showed the
movement mainly of those food items that figured in the index.
If was, of course, true that the items in the index had been
originally selected on the basis of surveys to represent food
consumption. Other illustrations might be cited to show the
index combining the function of a measure of price changes
with that of a tool of administrative control.
Both these points reinforce the need for accuracy in the
indices.
(objectives
The objectives of the revision of cost-of-living indices may
be summed up in terms of increased accuracy.
In some countries much attention has been paid to the
extensive changes in consumption habits that have taken place
during and since the war, including especially those changes
growing out of the immediate needs for reconstruction and for
the rehabilitation of the national economy. In other countries,
price control techniques for the stabilisation of cost of living

INTRODUCTION

have occupied the forefront of interest, and, with the removal
of controls, rapid price changes are occurring with consequent
effects upon the costs and standards of living of working groups.
These must be accurately measured. A series of special problems in methods and techniques is suggested by such points
as the inclusion of seasonal items, quality specifications, training
of personnel, administration of price collection, selection and
weighting of items and construction of a national index. Finally,
there is particular need for clarity in defining the objective, in
order not only that the procedures should be accurate but also
that they should be adapted to the specific purposes to be
served. In a word, increased accuracy in methods is necessary
for the achievement of specific goals.
*

A draft of the present report was submitted to a small group
of statistical experts, at a preparatory meeting authorised by the
Governing Body of the International Labour Office, for a preliminary discussion, with especial reference to the definition
of the questions to be included. The report as presented has
been revised in the light of the discussions at this preparatory
meeting, and the Office would like to express its thanks to the
experts who participated for their most helpful criticisms and
suggestions. It should, however, be pointed out that the Office
assumes full responsibility for the proposals submitted in this
report to the Sixth International Conference of Labour Statisticians.

CHAPTER I

PURPOSES OF COST-OF-LIVING INDICES
The purposes which may be served by different kinds of
cost-of-living indices in the post-war period comprise two
general types—those comparing living costs in different times
and those comparing living costs in different places. The
groups for which indices are calculated may be wage-earners,
salaried employees, farmers or the whole population. Indices
may be calculated for particular cities, for groups of cities, or
for the country as a whole. The objective is specified as relating
to retail price changes or differences—not to differences in
standards or levels of living. 1 As suggested earlier, therefore,
the emphasis is placed upon changes in retail prices of consumer
goods, the latter representing the list of goods and services which
constitute or embody the particular level of living of the groups
concerned in the particular community or country for which
the index is constructed.
MEASURING CHANGES OVER TIME IN RETAIL PRICES OF
CONSUMER GOODS

The classical method of measuring changes over time in
retail prices is to take a list of commodities and services and
price them at different dates; an index number for any year can
be found by dividing the cost in that year by the cost in the
base year, and multiplying the result by 100.
The question of how to determine the list of goods and
services to be priced is a crucial one. In principle, the goods
and services should represent the consumption level of a given
economic group in specified kinds of communities. This is
found, in general, by means of family-living studies. On the
basis of these studies a list of goods and services representing
the level of living of the group can be drawn up. This consump1
Indices to measure differences in standards or levels of living would
form an entirely different topic : the present report does not cover this
aspect of the subject.

PURPOSES OF COST-OF-LIVING INDICES

tion level may be taken as the pre-war level, the current consumption level, or a consumption level to meet a given standard
of adequacy, taking account of price-induced substitutions
which may be made without altering the level of satisfactions
provided. The purposes served by the index numbers based
on these lists of goods and services vary with the manner in
which the consumption level is defined.
The pre-war level of consumption is one of the principal
ways in which the level of living is defined. The list of goods
and services is then based on studies of pre-war family expenditures. A cost-of-living index may be established to measure
changes in the cost of this pre-war consumption level by pricing
this list of goods and services. This technique utilises a fixed
budget corresponding to the pre-war base period.
A somewhat different formulation of purpose is to find the
cost of the current consumption level of a given economic group
in specified types of communities. This implies that familyliving studies are made either recently, for example during the
post-war period, or are made by means of special sample studies
so as to show current trends in consumption of goods and
services. In principle, the more closely the list of goods and
services priced corresponds to the actual consumption of a
given economic group, the more faithfully it will reflect current
price changes.
This suggests two different techniques. In the first place,
the fixed budget of goods and services which represents the
current consumption level can be priced over the period before
and after the current year to show changes in the cost of the
post-war level of consumption; or in the second place, the budget
of goods and services may be changed from time to time, so
as to correspond always as closely as possible with current
consumption. In the latter case, the technique of chain indices
and link relatives should be employed to construct the index.
Within each link the price ratio between the beginning and the
end of the period is based upon a fixed budget of goods and
services representing the level of current consumption; but this
fixed budget is modified from period to period to correspond
with changes in current consumption.
With either of these techniques, the cost-of-living index
is kept tied to the goods and services that reflect the level of
current consumption. In the one case, when the fixed base
system is utilised, the method requires in effect that the index

COST-OF-LIVING STATISTICS

number series be recomputed each year on the new consumption
base. In the other, however, where the changing base system
is used, the method links the indices together in a chain so as to
give a trend in which each link is based upon the goods and
services included in current consumption. In either case, the costof-living index is based upon weights kept on a current basis.
This requires constantly renewed consumption studies; but the
method yields an index on the basis of current weights which
avoids the bias associated with a fixed base that is out of date.
Finally, a third purpose may be served by pricing the goods
and services on a consumption level that meets a given standard
of adequacy in which, from time to time, price-induced substitutions may be taken into account, as far as possible, without
altering the level of satisfactions provided. This method
involves determining the scope of allowable price-induced
substitutions. It allows cost comparisons to be made between
different goods and services, where the substitutions conform
to the conditions specified. The important point is that the
standard of adequacy should not be impaired.
In referring the consumption level to a given standard of
adequacy, however this is defined, the definition is originally
embodied in a list of goods and services as of some definite
time, for example the pre-war level. Thereafter from time
to time substitutions can be made in this list of goods and
services so that in terms of the standard of adequacy the level
of consumption is not changed. Each important purpose which
is served by the different goods and services must continue to
be served to the same degree of adequacy. If some particular
food product, for example, becomes extremely high in price,
it may be replaced by an equivalent amount of a substitute
at a lower price that will yield the same calories, vitamins and
minerals and otherwise satisfy the same food requirements as
the food product originally priced for the index. To take
clothing as an example, an item meeting the same general
specifications and serving the same purpose as the item originally
in the index but available at a lower price than the original
item can be substituted for it, with the result that the standard
of adequacy is maintained.
It is obvious that such a procedure requires careful consideration of the conditions and circumstances under which substitutions are to be allowed. In general, substitutions are induced
by price advantages — where equivalent items can be pur-

P U R P O S E S OF COST-OF-LIVING INDICES

chased more cheaply without lowering the standard of living.
On these questions the Office proposes the following:
Once the purpose of a cost-of-living index has been determined,
every possible safeguard should be taken to ensure that its preparation
reflects actual market price. The public should be informed both of
the purpose of the index and of these safeguards (1,1).
The purposes which may be served by different types of cost-ofliving indices in the post-war period are the measuring of changes over
time in retail prices currently charged consumers for:
(a) the pre-war consumption level of a given economic group in
specified types of communities;
(b) the current consumption level of a given economic group in specified
types of communities;
(c) a consumption level meeting a given standard of adequacy taking
account of price-induced substitutions which may be made without
altering the level of satisfactions provided (I, 2 (1)).
The purpose of an index for measuring changes in retail prices
currently charged consumers for the current consumption level of a
given consumers' group in specified types of communities may be:
(a) to measure the variations in the cost of maintaining unchanged
the pattern of consumption in a specific post-war year; or
(6) to measure price changes using the current pattern of consumption as a weighting diagram adjusted as necessary year by year to
conform to changes in the consumption pattern (I, 2 (2)).
The Conference may wish to place on record an expression
of opinion on the subject in the following terms:
f Where the purpose of the index is to measure changes in retail
prices charged for a post-war consumption pattern, the weighting
diagram should correspond to the current consumption pattern and
should be adjusted as necessary year by year, the index being constructed on the link method (I, 2 (3)).
Where circumstances are such as to render it desirable to measure
the variation in the cost of maintaining unchanged the pattern of
consumption in a specific year, the pattern of consumption should be
examined, and the weighting diagram adjusted, if necessary, at
intervals of five or ten years to correspond with the changes in the
consumption pattern (I, 2 (4)).
In countries where it is desirable in the immediate future to introduce a new weighting diagram representing current consumption:
(a) the index based on the new weights should be carried as far back
into 1946 or 1947 as the availability of representative price quotations permits;
(ô) consideration might be given to splicing the new indices by groups
of items to the indices by groups of items as previously computed,
and a new all-items index computed back to VJ Day (I, 3).
In further comment on these Resolutions the determination
of the purpose of the index is of course a question of fundamental
policy. This purpose may be defined, as suggested, in terms
of the particular level of consumption; or it may be defined in

COST-OF-LIVING

STATISTICS

more general terms, such as to measure as accurately as possible
the changes, due to price changes only, in the cost of a given
level of living, leaving to the statistical agency not only the
technical methods of determining the price changes but also
the choice of base as well as the method of defining the level
of living. In the latter case the statistical office may be called
upon to decide whether the purpose of establishing an adequate
measure of cost of living is best served by taking the pre-war
consumption level of a given economic group, or the actual
consumption level of a given economic group, or some combination of these.
It will be for the Conference to decide whether it wishes
to adopt any further directives than those envisaged in the
Resolutions outlined. In any case, once the purpose is determined, the objective is clear: every precaution should be taken
to ensure that the data obtained reflect actual market prices.
One point may be worth further comment, namely the use
of the link relatives or chain indices when the quantity weights
are changed every year. The purpose of the change of base is
to increase the accuracy of measuring price changes, by using
the weights that are most applicable, namely the weights of
the current year. Since in practice the weights of the current
year can be obtained only with some time lag, the weights
actually used will usually relate to the preceding year. As the
weights used are no longer the fixed weights of the base year but
vary according to changes in current consumption, the method
of construction of the index must be replaced by the chain
index method, in which fixed current weights are used for the
intervals between changes, but the price changes in each such
interval are linked to the index numbers for the preceding
period. 1 Mention may be made of the adoption by Sweden
1
Provided that the individual price ratios are correct, the use oí
correct weights will give correct index numbers by the chain method.
Under these conditions, the objection to the chain index method, in
that it is more difficult to describe than the index based on fixed weights,,
does not seem to be important. A more serious difficulty arises in the
possibility that in some one or more years the prices, and hence the
price ratios, may be in error. In the chain index method an error
incorporated in the figures of a particular year will extend throughout
the series. However, on the other hand, such errors might be expected
to tend to cancel one another out so that the net error over a period
should be small. In contrast, an error in the fixed base index due to
errors in prices for a particular year affects only the index for that year
and does not affect the index for later years. In any case, the importance
of possible errors in relation to the base period can be reduced by keeping
the base of reference fairly recent.

PURPOSES OF COST-OF-LIVING INDICES

of this method of annual revision of weights and the chain
index method.
In this connection other types of consumption-goods price
indices may be considered. In particular, special indices have
been proposed for purposes of deflating national income figures.
In such indices a much more extended list of items and prices
is included than is customarily included in consumer price
indices.
It has sometimes been suggested that such calculations as
are made for purposes of deflating national income figures or
for available calculations of retail price movements might offer
points of value for purposes of checking the results of the usual
cost-of-living index numbers. At all events, the study of this
closely related problem of devising an index of price change for
purposes of deflating national income should lead to useful
results.
Certain other consumption-goods price indices which might
be studied include those which envisage changes in the level
of living on which price changes are measured, that is, the level
of living corresponding to the list of goods and services priced
at the end of the period may represent a different level from that
corresponding to the list of goods and services priced at the
beginning of the period. If the level of living of the wageearning population during a period such as that of the recent
war has actually changed, should such a change be taken into
account in the construction of index of price changes by appropriate changes in the list of goods and services priced at different
times ? It is not suggested that the price indices should attempt
to measure this change of level; but merely that a technique
might be devised to measure current changes in prices of the
goods and services which correspond to the cost of living,
say, of the average wage-earner family under the existing
conditions.
On these questions the Office proposes the following Resolution:
The Sixth International Conference of Labour Statisticians :
Recognising the importance of developing and appraising different
measures of consumption-goods price movements for different purposes ;
Requests the Governing Body of the International Labour Office
to direct the Office to undertake a study of a variety of measures of
consumption-goods price movements with a view to determining the
relative advantages of each for different purposes, having regard both
to the choice of weights and to the prices employed in each type, and
in particular, for the purposes of deflating the national income (III).

COST-OF-LIVING STATISTICS

Still another type of index number is that which measures not
price relatives or changes in retail prices, but the relative costs
of a given standard of living.
This suggestion has been made particularly in the field of
food costs. Where the quantities of food which can be obtained
under a ration system are limited, it has seemed inappropriate
to certain critics to accept the price ratio of these rationed items
as the sole evidence for price changes, since in the case of individuals it may be necessary to purchase supplementary foodstuffs, in addition to rationed foods up to the amount of the
ration, in order to supply an adequate total of calories. The
prices of these supplementary foodstuffs, if uncontrolled, may be
much higher than the prices of the corresponding equivalent
quantities of rationed foods which under normal conditions would
be purchased. As a result, the increase in actual cost of food
consumed by individuals or families may often have been greatly
in excess of that indicated by relative food prices. It is true that
the Government bureaux charged with cost-of-living investigations have usually not attempted to estimate the cost of supplementary foods. This has been left to private investigators. 1
Government bureaux are usually reluctant to set up comparable
budgets to show relative food costs in which arbitrary judgments
have wide scope. For the purpose of an official index, it is held
that this procedure is too arbitrary to be adopted as an approved
technique. At the same time, it remains true that an important
point of criticism of official cost-of-living indices has been this
failure to take into account these actual conditions of the market.
On this point, however, the Office does not make any specific
proposal except as noted in the next section. A somewhat
similar problem is presented in the study of comparative costs of
living in different places, and the Office, in considering the
question of comparative costs, will have to study the problem
in that setting. 2
1
See, for example, computations such as those published by the
Institute of Statistics, Oxford University: e.g., T. Schulz, "Human
Needs" Diet, November 1943; Institute of Statistics (Oxford) Bulletin,
Vol. 5, No. 17, 11 Dec. 1943, pp. 273-276.
* In this connection, mention should be made of the Resolution of
the Fourth International Conference of Labour Statisticians: "It is also
desirable that as regards certain articles of food for which direct comparison of prices as between one country and another is impracticable, the
International Labour Office should explore the question of supplementing
the present method of calculation by a method in which account would
be taken of the possibilities of comparisons of prices on a basis of nutritive value. " International Standardisation of Labour Statistics, Studies
and Reports, Series N, No. 25 (Montreal, 1943), p. 107 [284].

PURPOSES OF COST-OF-LIVING INDICES

MEASURING PLACE-TO-PLACE DIFFERENCES IN COSTS OF LIVING
AT A GIVEN TIME

The measurement of differential costs of living in different
communities, including that of changes in living costs in transfer
from one community to another, whether in the same country
or in different countries, presents a series of difficult problems,
which have been studied over a period of years by a number of
countries, e.g. the United States and Sweden, as well as by the
International Labour Office.
The technique requires, in general, the pricing in each
community of a standard set of commodities and services. In
practice, adjustments have to be made in the list of commodities
and services to take account of local conditions of supply and
local consumption patterns. In communities, for example,
where anthracite is used for house heating, the cost of fuel will
normally be based largely upon the price of anthracite. Where,
however, the fuel locally available is bituminous coal or briquettes, the list of items priced and the techniques employed
must be adjusted to take this difference in the local supply into
account. A further difficulty is that climatic differences may
affect fuel requirements for maintaining a given standard of
comfort. All these questions involve theoretical and practical
problems. Not least among them is the problem of specifying
qualities and grades of the specific items of commodities and
services so that the prices obtained in different communities
relate to goods and services of the same quality.
The topic of international comparisons of real wages and of
costs of living was considered by the Fourth International
Conference of Labour Statisticians, which recommended inter
alia that the Office set up a committee of statistical experts to
consider the problems further. x The Office has published a
number of reports on the subject. a
In view of the fact that this problem is being studied by the
International Labour Office, and that a report is being prepared
for submission to the International Labour Office Committee
of Statistical Experts, the Office suggests that consideration of
the subject by the International Conference of Labour Statisticians be postponed, and proposes a Resolution in the following
terms :
1
2

Ibid., pp. 104-107, especially [286].
Ibid., pp. 24-30, and references there cited.

COST-OF-LIVING STATISTICS

The Sixth International Conference of Labour Statisticians:
Noting the importance of developing accurate methods for international, inter-regional and inter-city comparisons of costs of living
and recognising the theoretical and practical difficulties in the way of
obtaining satisfactory results;
Requests the Governing Body of the International Labour Office
to direct the Office to continue its studies of these problems, and
to place the subject on the agenda of a future International Conference of Labour Statisticians when appropriate methods and
techniques have been further developed and tested (II).

GROUPS WHICH SHOULD, IF FEASIBLE, BE COVERED BY
COST-OF-LIVING INDICES

Wage-Earners and Clerical Workers
The group for which cost-of-living indices are most commonly
calculated is the group of wage-earners and clerical workers, that
is, moderate income families. Index numbers of the changes
in living costs of this group are of the greatest importance. In
effect, a primary function of the cost-of-living index is to
measure the increase in cost of living with reference to interpreting the movement of real wages and to the need for changes
in wages for maintaining purchasing power.
This group of the population, however, is subject to different
conditions in different circumstances. Additional detail is
needed to show the movement of the cost of living in different
communities. In certain countries cost-of-living indices are,
in general, provided for each major city; in other countries
only for the capital or chief city. Specific issues of wage negotiations arise in regard to particular cities. In order to throw
light upon the changes in living costs separate index numbers
of cost of living are likely to be required, therefore, for each
major city.
As an alternative to showing separate indices for individual
cities, an index may be set up for groups of cities. If the index
numbers for the large cities show common trends, it would be
an economy, of course, to base the calculation on a sample of
large cities that would show adequately the movements of
living costs in this group. Whether to show a single index for
all large cities together, or to show them for each major city,
depends on the one hand upon the funds available, and on the
other, upon evidence indicating that an index for the group of

PURPOSES OF COST-OF-LIVING INDICES

cities is sufficiently valid for the individual cities included. One
particular question is whether the group of large cities should be
subdivided to show subgroups for the principal geographical
areas. "Where a country is large and extends over widelydifferent areas, and where price policies and supply conditions
vary greatly in different parts of the country, as in the case
of the United States and Canada, the addition of a geographical
classification may be desirable.
Another important objective requiring a subdivision of the
material is to show the movement of the cost-of-living index
in the small cities. This may differ considerably from that
in the large cities, both on account of different weights assigned
to certain elements in the cost of living and the differences in
price movements of important groups of commodities arising
from differences in supply conditions. In general, however,
it should not be necessary to construct cost-of-living numbers
for each small city, but a cost-of-living index number for small
cities as a group should suffice. The problem of selecting the
cities to be included in an adequate sample will be discussed
later. A geographical classification may be advisable where
conditions as to supply and price vary considerably in different
parts of the country.
A third group for which a separate cost-of-living index should
be constructed is that of wage-earners and clerical workers
in rural communities. These communities have special conditions of price and supply with reference to food products, on the
one hand, and manufactured goods, on the other. Their food
products are likely to be obtainable at lower prices and to be
subject to smaller variations than in large cities, because of
conditions of supply in the local area. Manufactured goods,
on the other hand, produced outside the rural communities
are subject to the marketing costs and conditions incident to
distribution in small communities separated by considerable
distances from the point of production.
The need for special indices of cost of living for wage-earners
and clerical workers in rural communities is perhaps not so great
as in the case of either of the other groups. They are less
likely to be needed in important wage negotiations on account
of the relatively smaller number of such groups in these communities. Nevertheless, the movements of retail prices in these
communities are important from the point of view of the welfare
of wage-earners and clerical workers who live in them.

COST-OF-LIVING

STATISTICS

Farmers
A second important population group for which index
numbers of cost of living should be obtained is that of the farm
population, comprising not only farmers and farm families
but also farm labourers and farm tenants. Special problems
attach to the calculation of cost-of-living indices for this group.
Since the families concerned live upon farms, it is customary,
if not universal, for them to grow part of their own food. The
problem of determining the cost of their food, therefore, in so
far as it is home-grown, presents very serious difficulties. In
practice, where farm cost-of-living index numbers are calculated,
they are limited to items for which cash outlays are required,
and the indices therefore are indices of cash elements in the
cost of living. A second important group of items difficult to
appraise is that of rent or housing costs. For farm families
housing is usually either owned by the farmer or included
as part of the rent contract of the farm. In many cases, lodging
is included as part of the wage paid to farm labour. As a result
it is extremely difficult to assess this item as part of the cost
of living of the farm group. Again, the usual technique limits
the cost-of-living index to the items of cash outlay. In any
case, if the index is limited to cash outlays, it is difficult to
compare cost-of-living trends in this group of the population
with other groups.
Special attention has been devoted to the problems of
measuring the changes in cost of living of farmers in connection
with the establishment of so-called parity prices, a principle
according to which prices of farm products should be regulated
with reference to the prices of goods which farmers buy, including, in this latter concept, not only consumption goods consumed by farm families, but also manufactured goods which are
used in farm production. The prices of production goods used
in farm operations should not be included in computing cost-ofliving indices for the farm population.
Other Groups
Special index numbers of the cost of living of other groups
should also be considered. Chief among them is the group of
salaried employees or officials. Such calculations are made, in
many cases, where European groups are living in areas such
as India, China, and Indonesia, where the standards of living

PURPOSES OF COST-OF-LIVING INDICES

of the European group are essentially different from those of the
local populations. Special index numbers for European populations are found, for example, in the Netherlands Indies and in
Shanghai. In a number of countries, furthermore, special
index numbers have been prepared for the group of salaried
employees, who usually have incomes higher than those of
wage-earners and clerical workers. The justification for such
special cost-of-living indices is that price movements of the
items purchased by these groups may vary substantially from
those of items purchased by wage-earners and clerical workers.
From time to time special groups may call for special costof-living index numbers. Teachers may be cited as an example.
The provision of index numbers for such special groups must,
in each case, be considered on its merits. For the calculation
of such index numbers family-living studies showing the budget
outlays of families of this special group concerned must be
available to provide satisfactory weighting diagrams for the
cost-of-living index. In general, to justify such a special
index, evidence should be available to show either that the
prices paid by such special groups, or the weighting diagrams
for such groups, differ from those customary in this group of
wage-earners or of clerical employees, to such a degree as to
make a special index desirable and necessary.
Total Population
In addition to the cost-of-living index numbers for these
groups, a comprehensive index should also be prepared for the
total population. The essential characteristic of this index is
that the prices are weighted by total consumer purchases.
For such an index adequate data must be available to show
the total of consumer purchases of the entire population, either
based on a comprehensive study of consumer purchases or on
consumption figures estimated from production, imports and
exports. Furthermore, special price series are needed for a
much wider list of items than for the usual cost-of-living index.
Luxury items, such as jewellery, silverware, etc., which do not
figure to any extent in the usual cost-of-living series, should
enter into a general cost-of-living index with their appropriate
weights.
The requirements for all these special index numbers must,
of course, be considered in relation to funds available, and the

COST-OF-LIVING

STATISTICS

importance of the index for these groups. Where it is found,
on analysis, that a cost-of-living index for one group or area
does not differ materially from that for another group or area,
one series may be dropped; on the other hand, where sharp
differences appear, and are maintained, the need for continuing
the different series is evident. The experience of the different
countries with such specialised index numbers can be drawn
upon, in order to show the need for, and the value of, these
specialised indices.
On these points the Office proposes the following * :
Separate price indices should be established to measure changes over
time in the cost of living for the most important groups in the city
and in the rural population, and for the total population, using appropriate price series and weighting diagrams (I, 4).

1
The Resolutions of the Second International Conference of Labour
Statisticians on these points were as follows:
"A series of national index numbers, calculated by using weights based
on the consumption of industrial workers generally, should be published.
In cases where desirable index numbers may also be calculated for other
classes of workers.
"In countries in which there are marked differences in economic
conditions in different districts or among different sections of the working
classes, it is desirable that separate index numbers should be published
for such areas and classes. In these cases the weights used should,
as far as possible, be based on working-class consumption in the districts
or among the sections of workers to which the index numbers relate."
The Second International Conference of Labour Statisticians, Studies and
Reports, Series N, No. 8 (Geneva, 1925), p. 69.

CHAPTER II

TECHNIQUES FOR OBTAINING AND
UTILISING REPRESENTATIVE PRICE DATA

SELECTION OF A SAMPLE OF REPRESENTATIVE OUTLETS

An important problem is that of obtaining a sample of representative outlets for obtaining price data.
In principle, in making up an average price for each commodity in each city, all the outlets at which goods are purchased
by the particular group of population concerned should be taken
into account and each type of outlet weighted in accordance
with its importance. Much attention has been paid to obtaining
prices from chain stores, department stores, suburban branches
of department stores, specialty shops, small individual retail
shops, etc. In order to determine the importance of each
type of outlet, special studies are required to survey the extent
to which the different outlets are in fact patronised by the
population in question.
With these objectives in mind, the particular outlets in
which prices are to be obtained must be selected. This is a
familiar type of problem to which sampling techniques should
be applied. Criteria of selection should be developed for the
special field of retail outlets in order that the best possible
results may be obtained with the minimum expenditure in time
and money by the pricing bureau.
One aspect of the problem is the ques.tion of possible bias
in the results arising from the selection of outlets. In other
words, would a different or better proportioned sample or a
better weighting of the results obtained lead to a substantially
different cost-of-living index number ?
This question was studied in its various aspects by the
Committee appointed to investigate the United States Bureau
of Labor Statistics Cost-of-Living Index. Additional samples

COST-OF-LIVING STATISTICS

of food prices in stores within city limits and in stores in suburban areas were examined with a view to determining the
influence of the selection of outlets on the results in terms of
price movements. 1
On this topic, the Office proposes the following:
In view of the importance of the proper selection of outlets for
purposes of pricing, a statement of methods followed in the preparation
of the cost-of-living indices should be published in connection with
such index numbers and should include a brief review of the different
types of outlets covered by the pricing techniques, the weights assigned
to each type, and the sampling methods employed in their selection
METHODS OF DETERMINING GOODS OF THE SAME OR EQUIVALENT
QUALITY

A point of fundamental importance in cost-of-living procedures is to ensure that the prices obtained at different times
—or in different places, where relative cost of living in placeto-place comparisons are sought—relate to goods of the same
or equal quality. This requirement is of course obvious; the
problem is how this objective is to be achieved.
The technique to be recommended is precise specification
of the goods to be priced. For purposes of cost-of-living enquiries where comparisons are made of prices over a period of
time, the goods must be so clearly described that the prices
obtained in successive months relate to the same goods.
The instructions given to the agents who collect the prices
should contain a sufficient description of the article so that each
month, as the prices are collected, goods of the same quality and
specification are priced. In detail, for example, this may call
for specification of the particular "brand" or make of article.
Where a number of brands are on sale, for example of canned
fruits, and the prices vary from brand to brand, it is customary
to obtain the price for a specific brand name. The instructions
may also specify definite quantities, since with such items the
price varies according to whether a large or a small quantity
is purchased. With many articles, a precise specification is
needed, as in the case of "dining-room suites" or "top coats",
etc. The essential requirement is that prices each month
should correspond with prices for the preceding month, that is,
that the same article is priced.
1
See O F F I C E OF ECONOMIC STABILIZATION ; Report of the
Committee on the Cost of Living, p p . 116-123.

President's

REPRESENTATIVE PRICE DATA

A special problem arises where the articles change from one
month to the next. In cases where the goods priced in the
preceding month are not available for pricing during the current
month, the common practice is to obtain prices for the preceding month as well as for the current month for a similar
item, so that the price changes reported relate to identical
articles.
Among the techniques of price collecting which bear upon
the success of the measures to ensure prices of identical items
are the type of collection agency, whether obtained by questionnaires or by special agents, the uniformity of supervision,
for example, whether by agents under the direction of a central
agency or by agents under the direction of local offices, etc.
These and similar points may affect the quality of the results
obtained.
Thus, the most detailed specifications are usually employed
in areas where agents of the statistical offices canvass retail
outlets in person and are in a position to compare the prices
quoted directly with the prices as posted or marked on merchandise. As a rule, less detailed specifications are used where
price collection is entrusted to local statistical offices, or where
reliance is placed on questionnaires mailed to retail dealers.
Of course, the detail and character of the specifications depend
upon the degree of standardisation or uniformity of consumer
buying habits, as well as on the nature of the particular items
priced.
The methods followed in different countries in regard to
these points vary considerably. In the United States, for
example, an elaborate set of detailed specifications has been
prepared for the use of the agents who collect the price
data.
In Canada, also, detailed specifications are provided for
many of the items which are priced directly by agents who
visit retail stores. Data obtained by means of questionnaires
are checked for discrepancies against those supplied by agents
who are furnished with quality specifications. Special provisions
are made in the lists of specifications for changes in quality
of clothing and home furnishings.
In France, local officials are relied upon to supply prices for
items of similar quality in successive pricing periods, the items
to be priced being specified in terms of a broad classification
of articles in the 34-items retail price index.

COST-OF-LIVING

STATISTICS

In Denmark, the Statistical Department supplies the
municipal offices responsible for price collecting with specifications for the items to be priced. Where a detailed description of the item is not possible, the instructions require the
pricing of that quality item which dealers consider to be most
popular with the buying public.
In Mexico, the Office of Economic Barometers relies on the
judgment of its representatives to price the same quality of
articles each week. These representatives make actual purchases
of all items in the Mexico City cost-of-living index.
In Switzerland, the agents of the local statistical offices are
furnished with quality specifications for foods by the Federal
Office in charge of pricing. But so far as food prices for various
localities are concerned, it appears that the dates of collection
and the quality standards differ from place to place.
Where identical qualities cannot be obtained, the next best
procedure is to obtain an approximate substitute. As already
noted, the usual procedure followed in cases of substitutions
is to obtain the price of the substitute article for the preceding
as well as for the current month, so as to be able to compare the
prices of an identical item during the pricing period.
Apart from the problem of specifying the quality for the
goods priced by means of specifications and tests, the question
should be considered of devising and adopting standard substitution procedures to be used when goods available in the
market change. During the war period, for example, inferior
qualities might often be substituted for standard articles. A manufacturer might introduce a change of specification in a standard
product — for example it might be offered in a slightly smaller
package or a slightly smaller amount be placed in each. Such
a change, however, given a reasonably efficient price and inspection service, would normally be discovered promptly.
Substitutions have been introduced during the war in respect
of many items. This process may be illustrated, for example, by
the tendency, when price ceilings are imposed, for goods in the
lowest price-class to disappear from the market and for articles
only slightly better than those in the lowest price-class to be
graded as if they were in a markedly superior class; a sort of
upgrading of goods above the class where they would normally
be placed. Low-priced shirts, for example, as obtainable before
the war may no longer be available. While price ceilings prevent
price increases for identical goods, a slight change sufficient to

REPRESENTATIVE PRICE DATA

make the goods no longer identical with the pre-war class is
made and the article is marketed at higher prices. In such a
situation, a common practice in most pricing offices has been to
compare the price of the item in the lowest price-class at any
given time with the price of the similar but not necessarily identical item in the lowest price-class before the war. This of
course ignores the change in the quality of the article.
Another illustration is found in the "deterioration of bread",
for example, by the compulsory admixture of rye, maize, etc.,
to wheat, adopted in certain countries as a war measure. In
collecting and comparing prices of bread, should account be
taken of the differences in quality over the period during which
such changes in specifications were introduced ? In practice,
in most countries, the prices were compared as if the articles
were of the same quality. In one or two cases, however,
where technical methods are available for determining, for
example, the effect of deterioration on the wearing qualities of
cloth, adjustments have been made to take deterioration into
account, so as to present prices in terms of an unchanged standard article.
In regard to these quality controls and, in particular, to
methods of taking deterioration of a quality into account in
price rates, the Office has at this time no definite proposals to
present. There appear to be no obvious or adequate methods
of taking such quality changes into account.
The Conference, however, may wish to go on record as
favouring a study of the whole problem by the Office in conjunction with methods of standardising qualities for place-toplace comparisons of costs of living. Accordingly, the following
Resolution is presented for the consideration of the Conference:
The Sixth International Conference of Labour Statisticians:
Recognising that the measurement of price changes over time or of
place-to-place differences in prices involves the establishment of
equivalence in the quality of goods priced at different times, or in
different places at the same time;
Requests the Governing Body of the International Labour Office
to direct the Office to make a study of the whole problem of quality
specifications in relation to the measurement of cost of living (IV).
TRAINING OF PERSONNEL

An important and almost self-evident aspect of good price
collecting is adequate attention to the proper training of the
personnel entrusted with the task of obtaining price quotations.

COST-OF-LIVING STATISTICS

But not only should the personnel selected be qualified, that is,
they should have sound, basic training in social and economic
enquiries; the staff should also receive instruction in the special
techniques required, as shown by experience, in the collection
of price data. They should be trained to apply and utilise
these techniques effectively. The training must, of course, be
given by the agency which collects the prices. The experience
of the agency is the best possible source for knowledge of special
difficulties in obtaining comparable price quotations. The training must include, for example, instruction in identifying brands,
grades and qualities of items to be priced. A broad knowledge
on the part of the personnel engaged in price collecting relating
to general specifications of items priced is naturally highly
useful. The training should include methods of interviewing
managers and others who furnish the quotations, of obtaining
price data from other sources, such as price lists, price tags,
etc. The agents should be familiar with the various pitfalls
which may affect the results. The experience of other price
collecting agencies may often be of great service in suggesting
techniques to follow and in throwing light upon the special
instructions needed. Among the obvious techniques to be
considered are the methods of checking price quotations obtained,
and either having them verified by another agent visiting the
same outlet, or testing them in purchases made by a buyer, for
the purpose of obtaining a price in an actual transaction,
without disclosing his connection with the pricing agency.
These techniques are directed partly to a review of the work
of the agents, and partly to a test of the method of obtaining
the most reliable quotations from the particular outlets.
On this point the Office proposes the following:
The most effective administration of the price collecting programme
requires the careful selection and training of price collection personnel
by the agency responsible for making the index. The periodic use of
"check pricing" or "purchase checking", in which price quotations
are verified by means of duplicate prices obtained by different agents,
or in actual purchases of the goods priced, is recommended (I, 6).
ADMINISTRATION OF PRICE COLLECTION PROGRAMMES

Central Control of Price Collection
An important phase of the administration of price collection
is centralised supervision or control. Different countries
employ different administrative machinery to obtain prices

R E P R E S E N T A T I V E PRICE DATA

used in the compilation of price indices. Some of them depend
on local administrative bodies or agents employed by local
offices. For instance, in Switzerland, the local statistical
offices collect prices for the cost-of-living index; the local
agents are furnished with quality specifications for foods, for
example, by the Federal Office of Industry, Handicrafts and
Labour. In other countries, the administration may be centralised, and agents employed directly and supervised by a central
statistical office. In general, a primary objective must be to
obtain effective uniformity of collection procedures. The
agency responsible for the results should have sufficient control
of the techniques and the selection and training of agents to be
able to obtain the necessary uniformity to ensure satisfactory
results.
On this point, the report on cost of living presented to the
Conference in 1925 gave illustrative examples of different
sources of price information.* A large variety of methods was
found and in some cases even within the same country different
methods were followed for different parts of the materials
collected for the cost-of-living index.
Without attempting here to review these differences in
administrative organisation, the important point to be considered
is the influence of the type of organisation upon the quality of
the results.
The use of specially trained agents by the central office for
direct surveys of prices in retail outlets generally results in more
reliable price data than questionnaires distributed to dealers.
This is especially true for pricing items where close attention
must be paid to quality specifications. The use of trained
representatives affords the possibility of checking the information given by the store managers with the price tags on the
merchandise. The effectiveness of the control depends on the
extent of centralisation of the statistical work on prices as well
as on the funds devoted to the purpose.
The Conference may wish to consider whether it desires to
adopt a Resolution on this topic either favouring centralised
supervision or control of price collection, or suggesting that
uniformity of techniques and methods throughout a given country
is desirable to ensure the best results. On this point, however,
the Office has formulated no specific proposal.
1
See Methods of Compiling Cost-of-Living Index Numbers,
a n d Reports, Series N, No. 6 (Geneva, 1925), pp. 33-35.

Studies

COST-OF-LIVING

STATISTICS

Studies and Correction of Biases in Reporting
Special problems are met with in the determination and
correction of biases in reporting prices. This is a highly technical matter; the first step is obviously to find out whether
biases exist. For this purpose special studies are essential.
The whole technique of price collection should be reviewed with
reference to the possibility of errors in reporting. Methods
should be devised to check the accuracy of reports and to
discover whether biases exist; studies should be made to determine the best methods of avoiding errors and biases.
One method is the use of check interviews. The price
collection technique may provide, as a general plan, that the
agent obtain price data from the store manager in the form
of a list of prices of items corresponding to the specifications.
Can such lists be accepted as giving a true picture of the prices
paid if, for example, all prices are subject to price ceilings?
In such a case the prices on the list might not correspond in all
respects to the prices actually charged. Prices of goods in
bargain sales may not be given adequate recognition in the
pricing programmes. If there are difficulties in the methods
followed, the use of duplicate interviews by different agents at
approximately the same time may help to test the accuracy and
the efficiency of the price techniques.
The actual purchase of goods by an agent not identifiable
as coming from the pricing bureau provides a practical check
on the fist of prices furnished to the agent. One difficulty
with the procedure of purchase of goods is that, to be effective,
the purchaser must be able to identify the items purchased in
terms of the specifications of the items priced, and further,
the agent who buys the goods may have to rely upon his memory
in regard to specifications which are not indicated in the purchase receipts.
Such tests of the pricing techniques have been made in one
or two countries. In the United States a series of tests of
pricing methods was made following criticisms of the price
indices calculated by the Bureau of Labor Statistics. 1
The chief point of interest here is not the actual results of
the analysis but the techniques followed in testing the methods
used. To test the sample of outlets, prices were obtained in
1
See Report of the President's Committee on the Cost of Livinq,
pp. 117-124.

REPRESENTATIVE PRICE DATA

additional stores to determine whether the particular selection
of stores might have influenced the results. One technique
was to test the differences in prices in the case of certain bestselling brands. Another test was made of prices from chain
store outlets. Since the pricing technique followed was to
obtain price scales for chain stores at headquarters, the food
prices as charged in the various chain outlets in different parts
of the cities surveyed were obtained and compared with the
price scales. The result showed small percentage differences—
1.6 per cent, below to 1.7 per cent, above—between the actual
prices charged and the prices obtained from the price scale.
Still another test was a comparison of prices secured by
agents at interviews with store managers with prices actually
paid by purchasers. This test indicated slight differences-—one
per cent.—but these differences may have been due in part to
slight divergences between the goods purchased and the goods
specified for pricing.
On this point also the Office presents no Resolution.
The development of appropriate administrative techniques as
already discussed will help in obtaining correct prices. It
goes without saying that every effort should be made to obtain
correct price data in order that the indices may present accurate
and reliable results; and the administrative agency must be
in a position to develop, introduce and enforce rules, techniques
and methods requisite to accomplish this purpose.
SPECIAL PROBLEMS OF ILLEGAL AND BLACK MARKET
PRICES

An important question of principle has been raised in a
number of countries, namely, whether the cost-of-living indices
should be based upon official prices or upon actual prices including both official and unofficial, illegal, or black market prices.
In some countries, for example, where official maximum prices
are published, cost-of-living index numbers are often based upon
these official prices; they may differ considerably from the
actual prices on the market. In other cases, a certain small
quantity of goods can be obtained at official prices up to the
limits of the rationed quantities, but any purchases beyond the
amount allotted in the ration must be made, if at all, at black
market prices. The use of the official prices in calculating the
cost of living doubtless saves much time and trouble. Do these

COST-OF-LIVING

STATISTICS

results give what is wanted ? The question has various aspects
which differ from country to country: (1) The sale of the item
in question may be illegal but no maximum price be fixed; in
this case should price quotations for such items be included in
the index ? (2) A maximum price may be fixed but be ignored
in the marketing of goods; here there may be an occasional sale
at the fixed price, or all sales may be above the maximum price
or the item may disappear entirely from the market. (3) A
list of maximum official prices may be published and accepted
by the compilers of the cost-of-living index without investigation as to the extent of sales at illegal prices or whether such
sales are in fact taking place. (4) There may be a recognised
and extensive black market in the items but no official method
of obtaining price quotations.
These problems may be illustrated by specific cases. In
Greece, for example, during the period of German occupation,
limited quantities of rationed goods were available at official
prices. These quantities in the ration, however, had to be
supplemented by the average family by goods purchased in the
black market. The problem of devising a suitable cost-ofliving index for such conditions includes the question of how
to combine the official and unofficial price quotations in the
index.
Another example concerns foodstuffs for Indians in the Fiji
Islands, for whom the staple item of consumption is rice. Rice,
in fact, was not available at the official price; it was available
only in the black market. The question involved was whether
the cost-of-living index on the basis of which the wages of the
Indians were to be regulated should include rice at the official
price or at the black market price at which it was actually
available. This obviously includes both a statistical question
as to what the prices actually were and a political question as
to whether the Government's statistical branch should recognise
in its black market price quotations the existence of violations
of law, for which another branch was responsible. Should the
statistics be compiled on the basis of prices which were actually
fictitious, and thus prevent an increase of wages that the actual
price of rice would have justified? Other examples may be
cited from the experience of the United States Bureau of Labor
Statistics.
In the first place a distinction should be made between
illegal prices in openly patronised outlets—where goods are

R E P R E S E N T A T I V E PRICE DATA

openly purchased by the persons in the groups to which the
indices relate—and black market prices in hideaway outlets.
In the former case, there is no reason why the illegal prices should
not be included; they are evidently effective prices and can be
obtained with reasonable accuracy since there is not sufficient
motive on the part of the seller for falsification. They belong
in the index in proportion as they actually enter into the cost
of living of the particular group.
Greater difficulties arise in the case of black market prices.
The criteria for determining when these should be used in costof-living índices are twofold : first, that the black market prices
are actually an important element in the cost of living of the
particular group to which the index refers, and second, that
they can be obtained with reasonable accuracy.
In theory, the best solution of these problems is to be
obtained by applying the principle that the cost of living should
measure accurately the existing state of prices. With regard to
black market prices, two questions are posed: (1) is any considerable fraction of the purchases of the item in question made on
the black market? and (2) can black market prices be obtained
with accuracy so as to show changes over a period of time?
On the first point, it is evident that if no considerable part of the
total of purchases is made on the black market, the importance
of including such prices is not great; if the average wage-earner
can purchase at normal prices up to the ration quota and if this
quota is reasonably sufficient there is no special reason for
including a study of black market transactions. The position
is naturally different if all or most of the transactions are in black
market operations. On the second point, black market transactions, by their very nature, do not lend themselves to accurate
price reporting, and prices may obviously be difficult to obtain
for use in official price indices.
The question was raised at some length in connection with
the criticisms levelled at the United States cost-of-living index.
The pricing methods of the Bureau of Labor Statistics included
recording and tabulating of both legal and illegal prices where
goods were obtained through the normal operations of purchasing. The mere fact that a price was above the maximum
allowed did not exclude it from consideration for the index.
For example, a chain store might offer goods at a price above
its regular price prevailing at the base date and therefore be
technically illegal. But if such a price was quoted to the price

COST-OF-LIVING STATISTICS

collectors, it was included, even though "illegal", in the tabulations on the ground that it related to a normal transaction.
Prices of black market items, which were not for sale except
clandestinely and contrary to law, were not included in the
pricing technique for two reasons: (1) they were considered to
be too unreliable and could not be secured on a trustworthy
basis; and (2) in general, the volume of such transactions was
small.
The Office proposes the following to cover these points:
In a period of price control:
(a) illegal prices charged in outlets regularly and openly patronised
by the groups to which the index applies should be regularly collected
and included in the index;
(b) black market prices charged for essential goods in hideaway
outlets should be included in so far as they can be obtained with
reasonable accuracy and are paid by the group to which the index
applies (I, 7).
CRITERIA TO BE USED IN CHOOSING SAMPLE OF COMMUNITIES IN
WHICH PRICES ARE TO BE OBTAINED

An important question which has received relatively little
attention is the problem of selecting the sample of communities
for which prices are to be obtained. The question can best be
put in concrete terms. In a country with one metropolitan
city and a large number of small cities, there will obviously be
need for an index of cost of living in the metropolitan city as
well as an index to represent price movements in the small
cities. In regard to the latter, should prices be collected in all
or in a few, and if in only a few, how are these few to be selected ?
If the price movements in all these communities are approximately similar, to collect prices in all might not only not be necessary but would represent a wasteful expenditure of funds. If the
prices are collected for a sample of communities only, however,
the question of the techniques to obtain a satisfactory sample
must be considered. The technical questions involved must of
course be passed by sampling experts. The point to be considered here is the desirability of recognising that the problem
involves the application of sampling techniques and that
technical advice is necessary in the application of these principles.
In general, three points should be considered. In the first
place, price movements in different types of communities in
different locations should be studied to determine the possibilities

REPRESENTATIVE PRICE DATA

of grouping according to the behaviour of their price trends.
The obvious groups are the large or metropolitan cities, the
medium-sized cities, the small cities, and the rural areas.
Geographical location may offer another basis of classification.
Where a country is large, geographical subdivisions may be
tested to discover whether characteristic differences in price
movements are found for different geographical areas within the
group, for example, of small cities.
The second point is, obviously, to determine, by means of
special studies, the existence of similarities or correlations in
price movements in different types of communities. Where
such correlations are found to exist, for example in the small
cities in a particular country, it will clearly not be necessary to
develop price series for each small city, but a sample of them will
yield a sufficient basis fer a satisfactory price index for the
groups of small cities. On the basis of the analysis of these
price trends, the decision can be taken as to whether to provide
separate series for individual cities or for groups of cities.
The third point to be discussed is the question of available
funds. Where funds are limited, the need for restricting the
number of series and for obtaining the best possible results for
the expenditure is the more obvious. In any case the selection
of the cities for which prices are to be obtained and the indices
to be calculated should be made in the light of the value of the
results in relation to costs.
As a final result, it should be possible to reduce to a minimum
the number of price series to be maintained to secure satisfactory
results and, at the same time, to furnish materials for a national
average retail price index, which will be both adequate and
reliable.
On this point, the Office proposes the following :
Studies of the correlation of price changes in different communities
should be made to determine the number and identity of the communities required to be covered for a satisfactory national average index for
a given group (I, 8).
The question might be raised specifically whether in each
country the cost-of-living index should be limited to the chief
city or should in general be a national average. On this point,
practice diners from country to country. In the majority of
countries the first objective is usually to establish a cost-ofliving index for the chief or capital city. As funds are available,
the need for a more adequate coverage of large and small

COST-OF-LIVING

STATISTICS

cities other than the capital city is recognised, and separate
series are set up to furnish the necessary information for wage
adjustments and other purposes for these different localities.
Finally, a national average cost-of-living index may be computed,
based on a weighted combination of the separate cost-of-living
indices for the different localities in the country. Among the
countries which calculate a "national average" of one kind or
another are Norway, Sweden, Great Britain, Cuba, and the
United States, etc.
In regard to this question, two points may be raised. First,
where wages are determined in a national wage scale, the diversities of cost-of-living index movements are ignored: in such
cases, there is more need for a national average than where
wage negotiations are pursued separately for each locality.
Secondly, if a true national average is sought as a primary
objective, the question might properly be raised whether the
farm population should not also be covered, and the movement
of the index for this group of the population be given its due
weight in the average. This raises again with additional
emphasis the question whether the cities for which index numbers
are calculated represent the best selection for purposes of computing a national average.1

For a discussion of certain of these problems as applied in the
United States see Report of the President's Committee on the Cost of
Living, pp. 282-283.

CHAPTER III

WEIGHTING DIAGRAMS
T H E ITEMS TO BE PRICED : CRITERIA FOR SELECTION

An obvious criterion for the selection of items to be priced
is their importance. This applies especially to the common
"items", such as principal foods which are consumed by a large
proportion of families. Even in respect of these, there are
many specific items, quality specifications, differences in brands,
trade names, and minor differences that multiply the possible
variations of an apparently homogeneous product. For many
"items", the possible choice is so wide that principles or criteria
of selection need to be developed. In the case of furniture,
clothing, etc., the lists of specific goods that might be priced
is so extensive that, in practice, only a few can be included in the
pricing technique.
From the point of view of obtaining a satisfactory picture of
price movements the essential requirement is that there should
be represented in the index all types of items which have
characteristic price trends. This applies to the goods purchased by the group to which the index relates. The factors
that determine price trends of different items are numerous
and varied. They include such elements as source of supply,
variations in supply, differences in production, market conditions, and local or international determinants of supply or price.
For example, the price trends of different kinds of cheese will
vary according to whether the cheese is locally produced, when
local conditions of production will play an important part;
or imported, when conditions of production at the source,
factors concerning importation, tariff duties, elements of competition, etc., may affect price trends.
Studies should be made of the correlation of price movements
in previous periods between the different commodities, so as to

COST-OF-LIVING STATISTICS

be able to group together those commodities which, in fact,
exhibit similar price trends. Once this grouping has been
established on the basis of correlation of price trends and price
behaviour over a period, it becomes possible to proceed to the
selection, on the basis of a satisfactory technique, of the items
to be priced. Each group must be represented by a suitable
sample of items.
Secondly, during a period of selective price control, this
selection of goods and services, included in the index, may need
to be reviewed and altered in order to maintain the principle
that all groups with characteristic price movement must be
included in a suitable sample of items. The behaviour of
prices of the different items, in other words, may so change
during such a period that a corresponding change must be made
in the treatment of the items in the index.
Thirdly, constant vigilance must be maintained to ascertain
the existence and influence of special supply and distribution
factors that may alter these price relationships. When a
new factor enters into the picture, such as, for example, the
introduction of a new plastic that may affect the price of tortoiseshell combs or frames for spectacles, the pricing agency will
wish to ascertain whether this factor alters the behaviour of
the price trends. Conceivably, it may produce a sharp decline
in prices of these items of manufacture utilising the new techniques, and subject them to new trend movements. All
such new developments need to be studied not only with
reference to the selection of items, but also to the weighting
diagrams.
A special problem is the question of measuring the cost of
housing for home owners in the calculation of the trends of
rentals or of housing costs. When home ownership is, in fact,
a considerable element in the housing cost of the group to which
the index applies, it should obviously be included, with its due
weight, in the index. The problem of selecting particular items
of home ownership for inclusion raises special difficulties.
The extent to which home ownership is a factor in housing costs
in the group to which the index relates must be particularly
studied. Where home ownership is an important factor, for
example, in cost of living of salaried employees, or of wageearners in certain types of communities, the importance of this
factor should be determined by factual evidence based on surveys of living conditions in these groups.

WEIGHTING DIAGRAMS

DETERMINATION OF WEIGHTS FOR ITEMS PRICED

Imputed Weights
An important problem is that of weighting the items priced.
In general, the principle of weighting is in accordance with
the consumption expenditure on the item. This principle,
however, as applied to cost-of-living indices, leaves open the
question of what to do with the consumption expenditure on
items which are not priced, whether the weights corresponding
to these items are to be disregarded or are to be applied to
those items which are priced for the index. In practice, a
common method is to calculate weights for a group of items and
apply the combined weight of the group to the price movements
of those particular items which fall into the group. This method,
however, assumes that the price movements of the entire group
are fairly represented by the price movements of the specific
items priced. A more satisfactory procedure is to determine
from an analysis of price trends the items which move together,
so that it can be established how far and to what items the price
movements of a particular item can be extended, or in other
words, whether and to what extent the price trends of a particular item can be considered as representative of the price trends
of other items. The true principle should be to weight the items
priced by the weights corresponding to their consumption
expenditure, to which may be added the weights of items not
priced, provided that the price movements of the former are
representative of the price movements of the latter. If three
food items, for example, exhibit similar price trends and only
one is included in the index, this can logically be given the weight
corresponding to all three. On the other hand, it is inadmissible
to give a weight corresponding to the consumption expenditure
on all items of furniture, for example, to the price movements of
the two or three items in the furniture group that happen to
be priced, unless there is evidence that the price trends of the
other items in the furniture group are actually correlated with
the price movements of the two or three items priced. This point
is obviously related to the question above, namely, how to
select items to be priced. If the furniture group is to be represented adequately, it must include representation of all the
items that possess characteristic price behaviour. If all such
items are included in the pricing schedule, each can then be

COST-OF-LIVING

STATISTICS

given a weight to cover other items with similar price movements so that the whole group of furniture can be adequately
covered.
In regard to this point, the Office proposes the following :
In establishing the weighting diagram for a cost-of-living index
for a particular group, individual items priced should be assigned
weights corresponding to the consumption expenditures not only on
the individual items themselves but also on other articles not priced,
in accordance with the principle that the weights for items not priced
may be added to the weights for those priced, when the price movements of the latter are representative of the price movements of the
former. In this manner, all items purchased by the group covered
can be represented in the index, though not all are priced (I, 9).
Methods of Handling Seasonal Items
Fresh Fruits and Vegetables.
An important group of commodities, which in the past has
been relatively neglected in cost-of-living indices and has been
singled out for special comment in some of the criticisms levelled
against the indices in different countries, is that of fresh fruits
and vegetables. Previously the usual practice has been to
include only those items of this group which were available
throughout the season. As a result, the items included have
often been limited to potatoes, onions, dried peas and beans,
and the like. The reasons for this relative neglect have been in
part the difficulty of handling the problems of seasonal variations in supply and in part the difficulty of seasonal pricing (even
with respect to potatoes sharp changes in prices appear at the
time new potatoes come into the market). One reason for the
increased attention paid to this group during the recent period
has undoubtedly been the trend toward greater consumption
of these items, which are good sources of vitamins and minerals.
Fresh vegetables. The usual practice may be described as
the method of the fixed list. Only those items are included
which are available throughout the year, and which can be
priced throughout the year without difficulty. In effect, this
means the omission of most fresh vegetables, that is, so far as
they are subject to seasonal influences. Thus, in Canada, for
example, the vegetables priced have been limited to potatoes,
onions, beans, and canned peas, corn and tomatoes, all of which
can be obtained and priced throughout the year. On the other
hand, fresh vegetables from the United States or Mexico, which

W E I G H T I N G DIAGRAMS

are subject to large price variations, as well as home-grown
products available for only a few months of the year, have
been omitted. It is true, however, that during the war imports
were largely restricted.
The usual obvious device resorted to in efforts to make the
index more representative in respect of vegetables has been
merely to add to the list: if the items added are available
throughout the year the principle of a fixed quantity budget
priced each month throughout the year can still be adhered to
and at the same time a somewhat broader basis is provided for
the index. This has been done, for example, in the United
States, where the list of fresh vegetables obtained throughout
the year, owing to refrigerated transportation, is a fairly long
one. Under these conditions the method of a fixed list priced
throughout the year gives a reasonably satisfactory result.
The vegetables in the price series in the United States now
include green beans, cabbage, carrots, lettuce, onions, potatoes,
spinach, sweet potatoes, as well as canned corn, peas, tomatoes
and green beans. This method, however, may be open to
objection if the locally available vegetables include other
important items. The home-grown products which are not
included are more likely to fluctuate according to the different
patterns than those which have been included in the list.
In Canada and New Zealand, the list of vegetables has been
increased without altering the essential technique of fixed
quantities of a fixed list.
Other techniques, however, have been employed in certain
other countries. In Argentina, a special treatment of seasonal
vegetables and fruits is provided by fixing for each month the
quantities of the specific fruits and vegetables that are to be
priced for that month. Thus, in the case of vegetables, the
list ranges from 11 to 1.7 with varying quantities according to
the season. 1
In the case of Sweden, seasonal factors are taken into
account mainly by price adjustments. The technique followed
since March 1941 is as follows: the price during the month when
the new crop comes on the market is accepted as a current
price; for the next month the price used is the average of the
current price and the price for the preceding month; for the
third month the average of the three months, etc., up to the
1
See Revista de Economía Argentina, Vol. XLII, No. 298, Apr. 1943,
pp. 116-122.

COST-OF-LIVING

STATISTICS

sixth month when the price is the average for the six months
following the introduction of the new crop. Thereafter, the
price used in the index remains unchanged until the next new
crop comes in. The justification for this procedure is that the
housewife can either purchase the year's requirements at the
one time and preserve the vegetables and fruits for later use
as needed, or buy supplies from month to month for current
needs. The procedure for averaging prices just described
allows thrifty purchasing at the beginning of the season as
well as current purchasing later in the season each to influence
the cost of the item as it appears in the budget. By this
procedure price fluctuations are minimised in their effects upon
changes in costs. Each item appears in the budget with the
same weight each month. 1
In New Zealand, the new wartime prices index includes a
special group covering fresh fruits and vegetables, in addition
to potatoes and onions shown in the old cost-of-living index.5
In Germany, the basic quantities of "vegetables" for the costof-living budget remain constant, but the kinds vary with the
season; in each market, the price quotations of the three most
important vegetables in that market are utilised. The consequence is that the selection of vegetables varies not only as
between different areas but also as between the seasons, the
three most commonly used vegetables being priced in each
month even though they are not the same as in the preceding
month. The result is the cost in the assigned quantities of the
three vegetables which are in fact most used in the different
cities for which price quotations are obtained.
Another method sets up separate quantities for each month
of the year and makes a monthly comparison with the same
month of the base year. In Palestine, in the revision for the
new index the actual consumption of vegetables in Jewish and
Arab households was ascertained during the different months
of the base year (1942) and these figures were utilised to give for
each month the relative quantities of the different vegetables
consumed. The budget quantities of the different vegetables
consumed during each month of the base year were then so
fixed that the cost of these quantities at the prevailing prices
1
Finansdepartementet : Betankcm.de angdende Levnadskostnadsindex
Avgivet av särskilt tillkallade sakkunniga (Stockholm, 1943), pp. 94-103,
anda especially p. 98.
Official Year Book, 1943, p. 534.

W E I G H T I N G DIAGRAMS

would equal the monthly allotment for vegetables. If prices
had been constant throughout the base year, this monthly
allotment would have been constant. Since, however, prices
increased considerably from January of the base year to January
of the next, the monthly allotment was adjusted on the basis of
a straight-line progressive increase throughout the year. The
result gives a series of sets of quantities, one set for each month
in the year, the cost of which, when priced for the corresponding
months in a subsequent year, gives a relative cost of vegetables
which reflects the price increase of vegetables over the cost in
the base period.
Fresh fruits. As for fresh fruits, a similar tendency is to
be noted as in the case of fresh vegetables. Many countries
have been inclined to limit fruits to a relatively few items. In
Denmark, the index has included only apples and fruit juices;
in Switzerland, apples, pears and prunes; in Sweden, apples,
bananas, oranges and prunes; in New Zealand, prior to the
inauguration of the wartime index, fruits were limited to apples
and lemons. In Canada, the items priced included apples
(evaporated), currants, lemons, oranges, raisins, prunes and
canned peaches; in the United States, apples, bananas, lemons,
oranges and canned peaches, pears and pineapples. In Palestine,
with the new index, the fruits included vary with the season,
as is the case also with the Argentine index. 1
The variation in the number of items included is due in part
to local consumption habits; a larger variety of fruits is consumed, it seems probable, in the United States than in Denmark.
In part, however, the small number of fruit items included in the
indices may reflect disinclination of the statistical authorities
to concern themselves with the collection of prices on these
items which are subject to wide seasonal fluctuations in consumption as well as in prices.
To summarise, a number of countries have found it desirable
or necessary to expand the coverage of fresh fruits and vegetables
in their cost-of-living index numbers. This may partly be in
recognition of changes in consumption towards the inclusion
1
In Argentina, bananas appear each month in quantities varying
from three dozen to 6 % dozen; cherries appear in the months from
December to May, duraznos in December, mandarines in April and May,
oranges in April, May and J u n e ; oranges from Paraguay from May to
August, muscatel grapes in February and March, and black grapes in
J a n u a r y , March and April. Revista de Economía Argentina, Vol. X L I I ,
No. 298, Apr. 1943, p p . 116-122.

COST-OF-LIVING

STATISTICS

of a larger variety of such foods. At all events, the problems
involved in this particular element in the food budget are
especially difficult, and the movement towards increased accuracy of the indices has tended to emphasise the importance of
solving these problems as well as of including these items with a
satisfactory technique in the quantity budgets.
Clothing.
A problem of seasonal pricing or of seasonal variations in
quantities purchased is met with in the case of clothing. A
common practice is to keep the entire list of articles of both
summer and winter clothing in the index throughout the year.
Winter articles are priced during each of the winter months but
during the summer months they are carried forward at the
prices prevailing during the last month of the winter season;
the summer articles are priced during each of the summer
months, but during the winter months they are carried forward
at the prices prevailing during the last month of the summer
season. This method may produce a sudden jump in the price
movement at the time when the shift is made from the prices
prevailing at the end of the previous season to those prevailing
during the first month of the new season. However, this may
well be less serious than the sudden change in the price movement
that would appear if in the intervals between seasons the prices
of the out-of-season goods were carried forward on the basis of
the trend of prices of the in-season goods. In the United
States, for example, the first method was substituted for the
second so as to eliminate the abrupt variations produced at the
time of the change of seasons when the prices as carried forward
by the trend line were replaced by current prices. Apart from
this difficulty, the procedure introduces a damping effect on
price movements of in-season clothing because of the carrying
forward, unchanged, of the cost of out-of-season clothing in the
index. The question evidently requires a detailed study of the
best method of shifting from in-season to out-of-season goods
in the light of actual experience.
A possible alternative method would be to assign to each
month of the base period a list of clothing items to be priced
during that month and in subsequent years to compare the
cost of the same list of goods each month with the corresponding
month in the base year to form a clothing price index number.
Another method would be to shift the clothing budget from

W E I G H T I N G DIAGRAMS

summer to winter items at the change of the seasons, and to
compare the prices of these each year with the corresponding
prices in the base period. This method, however, might produce discontinuity at the shift of the seasons, when the cost of
the summer clothing list at the end of the summer season is
followed by the cost of the winter clothing list at the beginning
of the winter season. If a discontinuity appears, it means an
abrupt change in the movement of the index that does not
correspond to any actual price movements at the time of the
change of the seasons.
Fuel.
This offers another example of seasonal consumption. In
general, however, price quotations for fuel can be obtained each
month, since coal or wood can be purchased throughout the
year with but slight variations in price. Hence, carrying the
fuel item unchanged throughout the year, and pricing it each
month, offers no special difficulties; in fact, it allows for the
possibility that purchases of fuel may be made at various times
during the year and stored for use as required. Even though
this procedure may give too much weight to the possibility of
storage, the advantages of applying a simple technique of price
changes are sufficient to justify this procedure.
A special problem relates to the use of different kinds of
fuel in different localities and to the adjustments in the standard
list of goods priced to take account of the fuels actually used
in each locality. A similar problem applies in calculating
indices of the cost of living in different places. In practice,
the price trends in each locality should properly be based on the
items of fuel which are, in fact, used in that place. Thus, in
effect, an index of fuel may be based, for example, on the
movement of prices of anthracite in one community, but on
the movement of prices of bituminous coal or of natural gas in
another, depending upon customary consumption. 1
With regard to this question of seasonal goods the Office
proposes the following:
Every effort should be made to include a suitable representation of
fresh, canned, dried and frozen fruits and vegetables in the list of
goods priced, and, whenever necessary and feasible, account should
be taken in the methods adopted of the seasonal factors in consumption
and prices (I, 10 (1)).
1
Similar problems are met with, in the case of vegetables and fruits
and clothing items.

COST-OF-LIVING STATISTICS

Provision should also be made for taking account of seasonal
factors in purchases of clothing and fuel (I, 10 (2)).
N E E D FOR N E W STUDIES OF CONSUMER PURCHASES

Changes in Consumption Patterns
Most of the consumer purchases studies from which the
weights for consumer price indices are derived were made in the
1920's and 1930's. Even in the normal course of events, economic change would have outmoded these old consumption
patterns in the intervening years. As a result of the war and
post-war economic upheavals and interruptions to trade production and supply, very great changes, though in some cases
of a transient character, have taken place in consumer purchases. Even today, and perhaps for the next few years to
come, the shortages of consumer goods and the economic
difficulties of certain countries make it unlikely that present
consumption habits will long remain typical.
The problem of obtaining a budget of commodities and
services representative of current consumption patterns has thus
assumed an acute form as a result of the major dislocations
during the recent war period and the great changes in consumption which have taken place. The extent of these dislocations varies from country to country. The specific problem
raised by such changes is due to their effect in impairing the
accuracy of the index numbers as a true measure of cost-ofliving changes.
This subject of changing consumption patterns was touched
upon in a general way by the Second International Conference
of Labour Statisticians in April 1925. The possibility that the
consumption weights used in the constitution of index numbers
might have become obsolete was covered in the following terms :
"When changes in consumption render the weights inappropriate a new series of index numbers based on the new weights
should be calculated." 1
The Conference recommended that new family budget
enquiries, which were described as the most satisfactory method
of determining the weights of the different items and groups a,
should be undertaken in those countries in which ho family
1

See International Standardisation of Labour Statistics, Studies and
Reports,
Series N, No. 25 (Montreal, 1943), p. 98 [236].
ä
Ibid., p. 99 [241].

WEIGHTING

DIAGRAMS

budget enquiries have been held since 1920-1921, "as soon
as economic conditions are sufficiently favourable". 1
In this connection, the Third International Conference of
Labour Statisticians in October 1926 adopted a general recommendation for surveys of consumption patterns a "at intervals
generally of not more than ten years".
The basic fact of changes in consumption patterns was
recognised in many countries by changes in the quantities
adopted for weights in the cost-of-living index, or by splicing
in new quantities in the old budget weights, or both.
In a number of countries recent family-living studies have
been utilised as a basis for new weights in the cost-of-living
index number. In the United States an extensive survey made
in 1934-1936 and in Canada a study in 1937 furnished the basis
for revised weights; and in both countries cost-of-living index
numbers were recalculated on the base 1935-1939. In Germany
in 1934 a new system of weights for the cost-of-living index
based on the family budget study of 1927-1928 was introduced. 3
A family budget enquiry was undertaken in 1937 and a preliminary report based on a small sample of the schedules was
published, but the results were apparently never utilised for
purposes of replacing weights in the cost-of-living index during
the war.
In Great Britain, a study of consumption in 1937-1938 was
made, but the results were not fully available at the outbreak
of the war. They became available, however, in 1941, but it
was decided that owing to wartime restrictions on consumption
and for other reasons the time was not then appropriate to
embark on a new basis for the index number. The old system
of weights based on consumption habits in 1904 was therefore
continued. A representative Committee is, however, now
considering the question of revising the basis ; and it is anticipated
that its recommendations will be available before August.
In New Zealand, a new wartime prices index was established
and the old index suspended. "Weights generally have been
revised to represent relative present-day consumption of the
various commodities and groups covered by the index." 4
1
2
3

Ibid., p. 99 [246].
Ibid., p. 116 [326].
See "Neuberechnung der Reichsindexzifïer für die Lebenshaltungskosten". Wirtschaft und Statistik, Vol. 14, No. 19, 1 Oct. 1934, pp. 626631.4
Official Year Book, 1943, p. 534.

COST-OF-LIVING STATISTICS

Revisions of cost-of-living index numbers have been undertaken or new indices have been set up in Mexico, Peru and other
countries.
A new and extensive study of family expenditures as well as
a revision of the cost-of-living index is being planned in Norway.
In Sweden, repeated surveys of consumption have been
undertaken. Changes in habits of consumption, especially of
foodstuffs, were found to be so great that it was decided to
make surveys of food consumption every quarter and to modify
the system of weights once a year to agree with the latest survey
of consumption habits. 1 In addition, the chain index method
was adopted in order to indicate trends over a period when the
index base is continually changing.8
In summary, various countries have thus indicated that, in
the judgment of the officials charged with the collection of costof-living data, the changes in quantities consumed during the
recent war period have been of sufficient importance to justify
new surveys to obtain up-to-date consumption data. Many of these countries have introduced these changes in the weighting
of their index numbers. One has inaugurated repeated surveys,
especially of food items, to keep its quantity weights in line
with national consumption patterns.
Criteria for Determining the Need for New Studies of Consumer
Purchases
In general there is agreement on the proposition that the
weights employed in cost-of-living index numbers should reflect
actual consumption of the particular group to which the index
numbers relate. As already suggested, however, there is
difference of opinion as to whether in calculating index numbers
of the cost of living over the recent period the best base to use is
the pre-war or post-war consumption, or an average of the two.
There is, however, general agreement that the weights used
should reflect the family expenditure of the group under consideration at some time within a fairly recent period. In principle,
the weighting diagram should correspond to actual consumption.
1
Finansdepartementet : Statens offentliga Utredninger 1943. Betänkande angáende Levnadskostnadsindex
Avgivet au särskilt tUlkallade
sakkunniga (Stockholm, 1943).
8
The chain index method was also adopted in Germany from t h e
beginning of the war. See Sodala Meddelanden, 1942, No. 3, p p . 197198.

W E I G H T I N G DIAGRAMS

The longer the period since the time to which the familyliving study relates, the more subject the cost-of-living index is to
bias. In general, the tendency is for the cost-of-living numbers
calculated in the usual way on a fixed base to depart more and
more as time passes from the true figure and for the index number thus calculated to overstate the true increase in the cost of
living. On this reasoning, it would appear advisable to make
new surveys of cost of living for purposes of determining the costof-living weights at intervals of not more than five or ten years.
The criterion in this proceeding is the departure of the actual
consumption from the consumption adopted for weighting
purposes; the greater the divergence, the more in error the
index is likely to be. It follows that the wider the variation
in consumption over a period of time, such as during the
war period and in the immediate post-war era, the greater
is the need for new budget studies to collect and bring up to
date the weighting diagrams used for purposes of calculating
the cost-of-living index numbers.
In assessing the need for new studies of consumer purchases,
an important point is the economic condition of the country
with special reference to its influence upon consumption
patterns. For example, in a country which has suffered heavy
damage from war the pattern of consumer expenditures during
the immediate post-war period might be materially affected
by the heavy expenses for needed reconstruction and repairs.
New studies of consumer purchases would show these factors
of consumption expenditure, but where changes in patterns are
rapid and continuous, the findings might need to be supplemented from time to time by new surveys.
Use of Small Sample Surveys
An important point in connection with this need for new
studies is the practicability of the use of small sample studies
of consumer purchases to indicate the extent and direction of
changes in consumption patterns. A small sample study can be
usefully made to throw light on the importance of these changes.
The results of such studies could be used either directly as a
basis for modifying the weighting diagram or for recommending
a new and comprehensive study of consumer purchases where
the changes indicated are sufficiently great to warrant a fullscale study.

COST-OF-LIVING

STATISTICS

In this connection mention may be made of the experience
of the Swedish studies on changes in food consumption during
the war period. These studies have been employed as a basis for
revising the weights for the food part of the index and for
utilising the results in the chain index procedure to take account
of changes in consumer purchases.
On these points the Office makes the following proposal:
The use of small sample studies of consumer purchases is recommended in order to provide the basis for estimating annual changes in
consumption patterns to indicate the need for changes in weighting
diagrams. Such studies do not obviate the need for more comprehensive benchmark surveys from time to time (I, 11).
Need for Full-Scale Studies of Consumption
The criterion for determining whether to make a full-scale
study of consumption is the extent of changes in consumption
since the previous survey.
In view of the sharp changes which have taken place in
various countries in the post-war period, and of the unsettled
conditions, it appears doubtful that full economic recovery
can be expected for some years after the war. In connection
with the general censuses of 1950 in population and agriculture,
comprehensive family-living studies or consumer purchases
studies might perhaps be undertaken around that period: the
results could then be linked up with population census data.
The Conference might therefore wish to consider a Resolution
to the effect that countries prepare plans for major consumer
purchases studies for the period around 1950.
Special Studies of Conditions of Family Living in Less WellDeveloped Areas
A special problem arises in less well-developed areas where
no comprehensive studies of family living have been undertaken, and where therefore the basic materials for weights for
cost-of-living indices are lacking. Furthermore, in areas such
as the Netherlands East Indies, for example, the taking of
family-living studies presents special technical difficulties.
On these points the Office proposes the following Resolution:
The Sixth International Conference of Labour Statisticians:
Recognising the need for new studies of family consumption in
order to furnish up-to-date weights for cost-of-living index numbers,
and in particular the special need for studies of family living in less
well-developed areas where such studies have not been made hitherto ;

WEIGHTING DIAGRAMS

Requests the Governing Body of the International Labour Office
to place on the agenda of an early session of the International Conference of Labour Statisticians the subject of the taking of family-living
studies, including the application of new sampling techniques to the
selection of families; and to consider the advisability of referring the
special problems of taking family-living studies in less well-developed
areas to a technical committee of specially qualified experts (V).

METHODS OF INTRODUCING N E W WEIGHTING DIAGRAMS

In general the method of introducing new weighting diagrams
is the method of linking. The objective of this method is to
ensure that the resulting index numbers should measure price
changes only, without any element of a change in quality or
type of article. The technique followed is, then, to compare,
within each pricing period, identical articles and identical
qualities, so far as this can be achieved. Changes in articles
or qualities may be introduced; but when this is done, the price
changes between the month when the change is first introduced
and the preceding month are calculated on the basis of the old
list of items and qualities, and between the month and the
succeeding month on the basis of the new list of items and
qualities, the two being connected by a link relative.
In practice, this is often effected by means of a change
introduced in the items in the base period; in place of the
original item in the base period, a new item is substituted in
such a way that the ratio of its cost at the time the item is
introduced to its cost in the base period is the same as that
obtained by the operation of the linking method just described.
With this method, the subsequent calculations can be prepared
more speedily than if the linking technique is performed each
month.
The ordinary revision of quantity weights, that is, the
substitution of one item for another, or of one quality for
another, as circumstances compel changes in the items included
in the index, is made in general on this plan. Where study of
consumption changes indicates that the weights applicable to a
particular item need to be changed, this technique of splicing
new items into the old series of weights is commonly followed;
or alternatively, items can be spliced out. For example, when
the shutting down of the manufacture of new automobiles in
the United States meant that new cars were not available in the
market, the item for "new cars" which appeared in the cost-of-

COST-OF-LIVING STATISTICS

living index with a relatively small weight was replaced by an
item for second-hand cars. At the point of change from new
to second-hand cars, the substitution of the latter for the
former is made in such a way that the value weight remains
unaltered.
In principle, the technique suffers from the disadvantage
that the weighting assigned to the second-hand car may no
longer correspond closely to its actual importance to the family
in the budget after the point of substitution. In short, the
drawback to this method of revision of quantity weights is
that it is only a temporary makeshift and is useful only for
piecemeal adjustments.
With a completely new weighting diagram, two possibilities
exist. In the first place, with the introduction of a new weighting diagram, all the index numbers can be recalculated on the
basis of the new weights and the old series be completely replaced. The second method is to provide for splicing, by joining
series using different weights in such a way that the indices are
equated at the point of juncture of the two series. This is the
familiar system of chain indexing. If a new set of weights is
available each year, the indices can be linked on a year-to-year
basis. Where new weights are available only every five or ten
years, when new budget enquiries furnish materials for a new
system of weights, these can also be spliced together in such
a way that, so far as possible, the weights applicable to each
year correspond as closely as possible with the actual consumption of that year.

CHAPTER IV
PROPOSED RESOLUTIONS
I
The Sixth International Conference of Labour Statisticians:
Having been convened by the Governing Body of the
International Labour Office, and
Having met at Montreal from 4 August 1947 to
,
and
Recognising the importance of an adequate statistical basis
for the proper weighting of cost-of-living index numbers,
and the need for refining methods and techniques so as
to ensure that the indices reflect correctly the movements
of actual market prices,
adopts, this
day of August 1947 the following
Resolution :
1. Once the purpose of a cost-of-living index has been
determined, every possible safeguard should be taken to ensure
that its preparation reflects actual market price. The public
should be informed both of the purpose of the index and of
these safeguards.
2. (1) The purposes which may be served by different
types of cost-of-living indices in the post-war period are the
measuring of changes over time in retail prices currently charged
consumers for:
(a) the pre-war consumption level of a given economic group
in specified types of communities;
(£>) the current consumption level of a given economic group
in specified types of communities;
(c) a consumption level meeting a given standard of adequacy
taking account of price-induced substitutions which may
be made without altering the level of satisfactions provided.
(2) The purpose of an index for measuring changes in retail
prices currently charged consumers for the current consumption
level of a given consumers' group in specified types of communities may be:
(a) to measure the variations in the cost of maintaining unchanged the pattern of consumption in a specific postwar year; or

COST-OF-LIVING

STATISTICS

(b) to measure price changes using the current pattern of
consumption as a weighting diagram adjusted as necessary
year by year to conform to changes in the consumption
pattern.
F?i (3) Where the purpose of the index is to measure changes
in retail prices charged for a post-war consumption pattern, the
weighting diagram should correspond to the current consumption pattern and should be adjusted as necessary year by year,
the index being constructed on the link method.
(4) Where circumstances are such as to render it desirable
to measure the variation in the cost of maintaining unchanged
the pattern of consumption in a specific year, the pattern of
consumption should be examined, and the weighting diagram
adjusted, if necessary, at intervals of five or ten years to correspond with the changes in the consumption pattern.
3. In countries where it is desirable in the immediate future
to introduce a new weighting diagram representing current
consumption :
(a) the index based on the new weights should be carried
as far back into 1946 or 1947 as the availability of representative
price quotations permits;
(b) consideration might be given to splicing the new
indices by groups of items to the indices by groups of items as
previously computed, and a new all-items index computed back
to YJ Day.
4. Separate price indices should be established to measure
changes over time in the cost of living for the most important
groups in the city and in the rural population, and for the total
population, using appropriate price series and weighting diagrams.
5. In view of the importance of the proper selection of
outlets for purposes of pricing, a statement of methods followed
in the preparation of the cost-of-living indices should be published in connection with such index numbers and should
include a brief review of the different types of outlets covered
by the pricing techniques, the weights assigned to each type,
and the sampling methods employed in their selection.
6. The most effective administration of the price collecting
programme requires the careful selection and training of price
collection personnel by the agency responsible for making the
index. The periodic use of "check pricing" or "purchase
checking", in which price quotations are verified by means of
duplicate prices obtained by different agents, or in actual purchases of the goods priced, is recommended.
7. In a period of price control:
(ö) illegal prices charged in outlets regularly and openly
patronised by the groups to which the index applies should be
regularly collected and included in the index;

PROPOSED

RESOLUTIONS

(b) black market prices charged for essential goods in
hideaway outlets should be included in so far as they can be
obtained with reasonable accuracy and are paid by the group
to which the index applies.
8. Studies of the correlation of price changes in different
communities should be made to determine the number and
identity of the communities required to be covered for a satisfactory national average index for a given group.
9. In establishing the weighting diagram for a cost-ofliving index for a particular group, individual items priced should
be assigned weights corresponding to the consumption expenditures not only on the individual items themselves but also on
other articles not priced, in accordance with the principle that
the weights for items not priced may be added to the weights
for those priced, when the price movements of the latter are
representative of the price movements of the former. In this
manner, all items purchased by the group covered can be
represented in the index, although not all are priced.
10. (1) Every effort should be made to include a suitable
representation of fresh, canned, dried and frozen fruits and vegetables in the list of goods priced, and whenever necessary and
feasible account should be taken in the methods adopted of the
seasonal factors in consumption and prices.
(2) Provision should also be made for taking account of
seasonal factors in purchases of clothing and fuel.
11. The use of small sample studies of consumer purchases
is recommended in order to provide the basis for estimating
annual changes in consumption patterns to indicate the need for
changes in weighting diagrams. Such studies do not obviate
the need for more comprehensive benchmark surveys from time
to time.
12. In order to promote understanding of the nature and
uses of indices of retail prices charged a particular group, the
term "cost-of-living index" should be replaced in respect of
such indices by the term "retail price index" or "consumer
prices index" specifying the particular group to which the
index applies.
13. Studies should be made of current consumption
patterns in relation to the extent to which the consumption
levels of different groups meet accepted standards of adequacy.
II
The Sixth International Conference of Labour Statisticians:
Noting the importance of developing accurate methods for
international, inter-regional and inter-city comparisons of
costs of living and recognising the theoretical and practical
difficulties in the way of obtaining satisfactory results;
Requests the Governing Body of the International Labour

COST-OF-LIVING

STATISTICS

Office to direct the Office to continue its studies of these problems, and to place the subject on the agenda of a future International Conference of Labour Statisticians when appropriate
methods and techniques have been further developed and
tested.
Ill
The Sixth International Conference of Labour Statisticians:
Recognising the importance of developing and appraising
different measures of consumption-goods price movements for
different purposes;
Requests the Governing Body of the International Labour
Office to direct the Office to undertake a study of a variety of
measures of consumption-goods price movements with a view
to determining the relative advantages of each for different
purposes, having regard both to the choice of weights and to
the prices employed in each type, and, in particular, for the
purposes of deflating the national income.
IV
The Sixth International Conference of Labour Statisticians:
Recognising that the measurement of price changes over
time or of place-to-place difference in prices involves the establishment of equivalence in the quality of goods priced at different
times, or in different places at the same time;
Requests the Governing Body of the Internationa] Labour
Office to direct the Office to make a study of the whole problem
of quality specifications in relation to the measurement of cost
of living.
V
The Sixth International Conference of Labour Statisticians:
Recognising the need for new studies of family consumption
in order to furnish up-to-date weights for cost-of-living index
numbers, and in particular the special need for studies of family
living in less well-developed areas where such studies have not
been made hitherto;
Requests the Governing Body of the International Labour
Office to place on the agenda of an early session of the International Conference of Labour Statisticians the subject of the
taking of family-living studies, including the application of
new sampling techniques to the selection of families; and to
consider the advisability of referring the special problems of
taking family-living studies in less well-developed areas to a
technical committee of specially qualified experts.

APPENDICES

APPENDIX I
RESOLUTIONS OF THE SECOND INTERNATIONAL CONFERENCE
OF LAROUR STATISTICIANS, GENEVA, APRIL 1925
COST-OF-LIVING INDEX NUMBERS

(intended to measure changes in the cost of living in a given country at
different periods)
(1) In each country statistics should be published showing changes
in the cost of living. They should cover the food, clothing, heating
and lighting, rent, and miscellaneous groups. Each group should
include the most important items of consumption, and especially
those articles the price changes of which may be taken as representative
of changes in the prices of articles not included.
Separate index numbers should be published for each of the groups
and general index numbers for all the groups combined. In calculating
these index numbers weights should be used proportionate to the
relative importance in consumption of the items in each group and
of the different groups. Changes in the cost of living in a given period
should be calculated by using weights which remain unchanged
throughout the period. When changes in consumption render the
weights inappropriate a new series of index numbers based on new
weights should be calculated.
It would be desirable that as far as possible the information
published should enable the differences which would result in the
index numbers from the inclusion or exclusion of direct taxes to be
determined.
(2) A series of national index numbers, calculated by using weights
based on the consumption of industrial workers generally, should be
published. In cases where desirable, index numbers may also be
calculated for other classes of workers.
In countries in which there are marked differences in economic
conditions in different districts or among different sections of the
working classes, it is desirable that separate index numbers should
be published for such areas and classes. In these cases the weights
used should, as far as possible, be based on working-class consumption
in the districts or among the sections of workers to which the index
numbers relate.
It may be convenient that the national index should be computed
by combining the separate indices, account being taken of the relative
populations in the different groups.

COST-OF-LIVING

STATISTICS

(3) In view of the existing inadequacy of statistics of consumption, the most satisfactory method of determining the weights of the
different items and groups appears to be that based on a standard
family budget obtained as the result of an enquiry into the average
expenditure of a number of families during a given period.
Where statistics of family budgets are not available, statistics of
aggregate consumption may be of some value in the determination
of the relative importance of certain articles of consumption in the
country as a whole, account being taken of the fact that certain
articles are used for purposes other than human consumption.
Theoretical budgets computed from all information available may
also be used.
(4) Care should be taken to ensure that the quality of each item.
for which prices are obtained at different dates and the methods of
collection and calculation remain unchanged during the period covered.
(5) Index numbers should be computed as often as may be
necessary having regard to the extent to which prices are changing.
In present conditions they may with advantage be computed at
monthly intervals, but under more stable conditions their publication
at less frequent intervals may be adequate.
(6) It is desirable that in those countries in which no family
budget enquiries have been held since 1920-1921 such enquiries should
be undertaken as soon as economic conditions are sufficiently favourable, and, if possible, not later than the year 1928.
(7) It is desirable that the Governments, in collaboration with
the International Labour Office, should agree to adopt the same year
as base for the calculation of a new series of index numbers. It is
suggested that the year 1930 would be suitable for this purpose.
(8) It is of great importance that the principles adopted in the
selection of articles and of the different qualities of such articles, in
the collection and computation of the price data and in the fixing
of the weights, should be indicated in full detail when such index
numbers are first published, also the precise data to which the index
numbers refer.

APPENDIX II

UNITED KINGDOM: COST OF LIVING ADVISORY COMMITTEE,
INTERIM REPORT1
The Committee was appointed 2 to advise the Minister of Labour
and National Service on the basis of the official cost-of-living index
figure, with immediate reference to the question whether any revision
of the basis of the figure was practicable or desirable in present conditions and, if so, the revision that might be made.
The Committee first of all addressed themselves to the question
whether any revision of the basis of the official cost-of-living index
figure is desirable under present conditions. On more than one
occasion during the inter-war years the question of revising the basis
of the index was considered until, finally, in 1936, it was decided to
institute an enquiry into the distribution of working-class expenditure
for the purpose of obtaining information on which to base a revision.
This enquiry was carried out in the years 1937 and 19383 and, but for
the war, some new index would already have been prepared. The
1914 basis and the 1937-38 budgets are compared in Appendix I to
the present report. The main change disclosed by the 1937-38
enquiry was a reduction in the proportion spent on food and a considerable increase in the proportion spent on "other items ", i.e., items
other than food, rent, clothing, fuel and light.
After careful consideration, the Committee came to the conclusion
that the official index was quite out of date, and they unanimously
recommend that the use of the present basis of the index should be
terminated.
In considering what should take its place, the Committee did not
feel precluded from considering the fundamental question of the purpose for which any revised index should be used. Discussion in the
Committee disclosed that the entire conception of an index purporting
to measure the changes in the cost of maintaining an unchanging
standard of living of a section of the community might be strongly
challenged, the description of an index as a "cost-of-living " index
being felt to be inappropriate and misleading.
Alternative plans for a new index considered by the Committee
included one which would measure the movement of retail prices
"weighted " on a basis which would be kept continuously up to date.
Many of the Committee felt that this might be the most reliable and
scientific basis on which to judge changes in retail prices for any
1
Interim Report of the Cost of Living Advisory Committee, Cmd. 7077.
Summary
printed in the Ministry of Labour Gazette, Mar. 1947, p. 81.
2
Idem, Jan. 1947, p. 16.
3
Idem, Dec. 1940, pp. 300-305; Jan. 1941, pp. 7-11; and Feb. 1941,
pp. 28-31.

COST-OF-LIVING

STATISTICS

section of the community; but it was considered that more time was
required in order to make a thorough study of the relative merits of different kinds of index numbers in order to arrive a t a recommendation
on the important question of a permanent index to replace the present
index. It was noted that the Statistical Commission of the United
Nations had expressed interest in this question and also that the International Conference of Labour Statisticians to be held in Montreal
in August next would be considering the principle involved in a currently weighted index. The Committee felt that it would be useful
in their further consideration of this matter to take into account the
result of the Montreal discussions. Accordingly, they propose to
continue their consideration of this matter and recommend that such
budget and other enquiries and tests be put in hand as will enable
them to express a considered view on the subject in due course.
The Committee then considered whether there was any interim
proposal which they could put forward to enable price changes to be
measured in the meantime without prejudice to the institution of
a new index of a more permanent character, but which would, nevertheless, serve a useful purpose for the time being, be easy to understand and be an improvement on the existing index. After careful
consideration the majority of the Committee felt that the best interim
arrangement of this kind would be an index making use of the information on the distribution of working-class expenditure obtained in the
enquiry of 1937-38. This enquiry had been thorough and the general
pattern of consumption which it presented could, in general, be taken
to be typical of the habits and customs of the immediate pre-war
period. The Committee, therefore, recommend that an interim
index of this nature should be started, the level of prices at the date
when the index was started being represented by 100 and future
monthly changes in price levels expressed as index numbers on this
basis. The Committee emphasise that the proposal should not be
regarded as providing a new cost-of-living index but as an interim
index to provide a measure, for the time being, of future changes
in retail prices of the things in which ordinary working-class households are interested.
The Committee further recommend that a Technical Committee
should be appointed immediately, to prepare a scheme showing how
the information from the 1937-38 enquiry could best be adapted for
the purpose of this interim index, with reference, for example, to the
selection of items for which prices are to be regularly collected, the most
suitable machinery for obtaining reliable information as to the prices
actually being paid, and the localities from which price quotations
should be obtained. The proposed interim index cannot be started
until the Technical Committee have completed this task. This is
unavoidable since the index will be based on prices still to be collected,
many of which will not be available for any past dates.
The Committee also recommend that the Technical Committee
should thereafter address themselves to the problems involved in
instituting a regular series of budget collections which would be needed,
amongst other things, to assist in reaching conclusions on the nature
of a permanent index to be introduced in due course..

PUBLICATIONS OF THE INTERNATIONAL LABOUR OFFICE
R e a d y shortly

EMPLOYMENT, UNEMPLOYMENT
AND LABOUR FORCE STATISTICS
Studies a n d Reports, New Series, N o . 7, P a r t 1
The h e a v y demands of t h e w a r have focused a t t e n t i o n on t h e
labour force as a m o s t i m p o r t a n t factor of economic a n d social policy.
In contrast with t h e period following t h e First W o r l d W a r , w h e n
u n e m p l o y m e n t was a major social problem, emphasis h a s now shifted
t o t h e positive aspects of employment. This development of t h e idea
of full employment has increased t h e need for a d e q u a t e statistics, a n d
a report on t h e subject was accordingly submitted t o t h e Sixth International Conference of L a b o u r Statisticians, which m e t in Montreal
in August 1947. T h e report is now being reprinted for general distribution in its original form with a few minor corrections.
CONTENTS
CHAPTER I.
Objectives.
CHAPTER I I .
Definitions.
The Employed — The Unemployed — Labour Force — Time Reference — Other
Definitions.
CHAPTER I I I . Reporting
Systems.
Employment — Unemployment.

CHAPTER IV.

Types of Employment

Series:

National

Practice.

Total Employment — Employment by Branches of Economic Activity — Employment
by Geographical Area — Employment by Sex, Age and Hours worked.

CHAPTER V.

Types

of Employment

Series:

General

Problems.

Agricultural Employment — Non-Agricultural Employment — Employment by
Industry — Total Employment — Labour Force — Employment by Geographic
Region — Employment by Age and Sex — Employment by Occupations and
Hours Worked — Partial Employment — SeasonaUy Adjusted Series.
CHAPTER V I . Employment:
Coverage.

Status — Condition — Age.
CHAPTER V I I . Employment:
Sample.
Establishment Sample Statistics — Labour Force Surveys.

CHAPTER V I I I .

Establishment

List.

National Lists — Non-Reporting — Comparisons based on Identical Firms.
CHAPTER I X .
Trend Bias.
National Experiences — Adjustment for Bias.
CHAPTER X .
Collection of Data.

CHAPTER X I .

Types of Unemployment

Series.

National Practice — General Problems.
CHAPTER X I I .
Unemployment:
Sample.
Registration at Employment Exchanges — Trade Union Statistics — Social Insurance
Statistics — Labour Force Surveys.

CHAPTER X I I I .

Unemployment:

Coverage.

Employment Exchange Statistics — Trade Union Statistics — Social Insurance
Systems — Labour Force Survey Statistics.
CHAPTER X I V .
Consistency.

CHAPTER X V .

International

Comparability

of Data.

Purposes — Industrial Classification — Occupational Classification.
CHAPTER X V I .
Publication.
Promptness of Publication — Presentation.

CHAPTER X V I I .
General Issues.
CHAPTER X V I I I . Proposed
Resolutions.
A P P E N D I X : Resolutions of the Second International

Statisticians,

Conference of Labour

Geneva, April 1925: II. Unemployment

Statistics.

PUBLICATIONS OF THE INTERNATIONAL LABOUR OFFICE

YEAR BOOK
OF LABOUR STATISTICS
9th Issue, 1945-46
This annual publication, now issued in a trilingual edition (English,
French and Spanish), presents in tabular form the most complete
data obtainable in the principal countries of the world relating to:
Gainfully Occupied Population
Employment and Unemployment
Hours of Work
Wages
Cost of Living and Retail Prices

Family Living Studies
Migration
Industrial Accidents
Industrial Disputes
Economic Statistics

The tables cover the last fifteen years. They are compiled from
the official statistics of more than 60 countries in all parts of the
world, supplemented in some cases by data obtained from private
sources. Recent enquiries of the I.L.O. on wages, hours of work
and cost of living in various countries are reproduced in full.
An introductory note to each chapter calls attention to the principal
questions of method to be borne in mind in interpreting the data,
especially when utilising them for international comparisons. The
statistics are presented in a systematic way with this object in view.
"Indispensable to statisticians of all countries . . . Illustrates the wealth and
complexity of the knowledge collected for experts to interpret . . . Through the
co-ordination and, so far as possible comparison of the official figures published by different
countries it is possible to gauge world economic trends in a way which was out of the
question twenty years ago."—The Times, London.
"The eighth issue of the Year Book is a noteworthy achievement, especially when viewed
in the light of wartime difficulties. The editor modestly credits the surprisingly small
wartime impairment of statistical data to the increased consciousness of the need for
adequate statistics in many fields important to the war and to post-war activities. A
reviewer may properly note, however, the evidence of diligence in obtaining and skill in
compiling the information, which is highly diversified alike in national origins, in subject
matter, and in degree of comparability."—The Journal of Political Economo, Chicago,
Illinois.
"The Year Book continues to be invaluable to those interested in international comparisons of labour conditions."—Economic Becord, Melbourne, Australia.
xv + 284 pages

Price: paper bound . . . .
cloth bound

$2.50; 10s.
$3.75; 15s.