Implications of the differences
between the Consumer Prices Index
and Retail Prices Index
Introduction
The Office for National Statistics (ONS) produces two main measures of consumer price inflation –
the Consumer Prices Index (CPI) and the Retail Prices Index (RPI). The statistics are produced in
accordance with the Code of Practice for Official Statistics1 and consequently ONS has a
responsibility to users to prepare and disseminate sufficient supporting information and analysis to
aid the interpretation and use of the statistics.
This article is in response to a requirement of the UK Statistics Authority’s assessment of ONS’s
consumer price indices in 2010, and explains the implications that the differences in scope and
coverage of the CPI and RPI have for their uses both as macroeconomic indicators of inflation and
as measures of compensation.
It is important to note that ONS is responsible for the production and dissemination of a wide range
of official statistics, including the CPI and RPI, to enable informed decision making by government,
public services, business, researchers and the public. ONS has a role to play in explaining the
implications of using the statistics for particular purposes. However, it is for other decision-makers
outside ONS to determine which statistics they use for their purposes. The use of the CPI, RPI and
their derivative indices for government policy-making purposes, in particular, has been and
remains a political decision, made by the government of the day.
This article looks at the implications of the differences between the CPI and RPI in relation to their
main uses and in particular assesses their qualities when used as macroeconomic indicators of
inflation and for compensation purposes. These objectives are explored by first summarizing the
history of the CPI and RPI and looking at how their evolution has led to both similarities and
differences between their designs and their main uses. The article then looks more broadly at the
statistical considerations for index design and highlights how these considerations help inform how
the price indices should be used. Given these statistical considerations. and the current design
constraints of the CPI and RPI, an indicative assessment of their relative merits when used as
macroeconomic indicators of inflation and for compensation purposes follows. This information is
presented in the following series of annexes:
Office for National Statistics
1
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Annex 1. A summary of the history of the CPI and RPI
Annex 2. An indication of the main similarities and differences between the CPI and RPI
Annex 3. A summary of the main uses to which price indices are put
Annex 4. The decision making process relating to index use
Annex 5. An indication of the statistical considerations for index design
Annex 6. An indicative assessment of the CPI and RPI, highlighting the implications of their
differences when used as macroeconomic indicators of inflation
Annex 7. An indicative assessment of the CPI and RPI, highlighting the implications of their
differences when used for compensation purposes
Annex 8. Technical information on the different price aggregation techniques used in the CPI
and RPI
Annex 9. A brief summary of current international use of price indices for macroeconomic and
compensation purposes
Context
During 2010, the UK Statistics Authority assessed the Office for National Statistics’ Consumer
Price Indices against the Code of Practice for Official Statistics and considered the communication
of consumer price indices more generally. The conclusions were published as Assessment Report
792 and Monitoring Brief 7/20103. The UK Statistics Authority concluded that the statistics should
continue to be designated as National Statistics subject to ONS implementing five enhancements
or requirements. The third of these required ONS to:
Publish information about the history and the reasons for the differences in scope and
methods between the Consumer Prices Index (CPI) and the Retail Prices Index (RPI); and
explain the implications that these differences have for the uses to which these statistics
are put.
This article addresses the latter part of that requirement and follows on from an article by Phil
Gooding (History of and differences between the CPI and RPI July 2011)4, which primarily
addressed the first part of the requirement. The article also marks the beginning of the analysis
work to address suggestion 2(a) of the UK Statistics Authority Monitoring Brief to:
Carry out more analysis of the strengths and weaknesses of the CPI and RPI as
macroeconomic measures of inflation and as compensation indices.
ONS will continue to engage with users on how best to take forward further analysis work.
Feedback on this article will be welcomed and the establishment of a Prices User Group in the
near future will aid this user engagement process.
Office for National Statistics
2
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Summary
The CPI and RPI have many strengths and have met users’ needs over a long period of time and
continue to be widely accepted and used. No one price index can meet all needs perfectly and how
well an index fulfils users’ needs is dependent on what particular factors matter to particular users.
The main differences between the CPI and RPI relate to:
• population base – the RPI excludes very high and low income households and hence the
CPI has a wider population coverage than the RPI
• commodity coverage - the CPI excludes owner occupiers’ housing costs and hence the
RPI has wider commodity coverage than the CPI
• formulae used to combine prices at the first stage of aggregation - the CPI uses a
combination of geometric means and arithmetic means, whereas the RPI only uses
arithmetic means
The implications that these differences have in relation to the main uses of the CPI and RPI are
summarised below:
Use of the CPI and RPI for macroeconomic purposes – (that is, as a measure of inflation
within the economy)
The CPI’s design means it has greater consistency with National Accounts’ principles and wider
population coverage than the RPI. This gives the CPI a number of advantages over the RPI as a
measure of macroeconomic inflation and also for the deflation of the National Accounts. However a
major disadvantage of the CPI for many users is that it does not currently include a measure of
owner occupiers’ housing costs (OOH). ONS is pursuing the introduction of OOH into the UK CPI
as a matter of priority, with implementation expected by 2013.
Use of the CPI and RPI for compensation purposes – (that is, to uprate incomes and
benefits to provide compensation for the adverse impact of inflation)
Both the CPI and the RPI have desirable properties in relation to their use for compensation
purposes. They both provide robust measures of price change but their distinct differences in
population and commodity coverage are likely to be the deciding factors that influence users on the
choice of index for their particular purposes. For example, the inclusion/exclusion of owner
occupiers’ housing costs and the coverage or otherwise of very high and low income households is
likely to feature in this decision.
However there are two other issues that continue to be widely debated but as yet, have no
definitive resolution:
1. Target population – this refers to the particular sub-population that is the target of the
compensation measure. Should an index used for compensation purposes reflect the
movements of prices for the whole population or for the specific demographic groups to
which the measure is targeted?
Office for National Statistics
3
Implications of the differences between the Consumer Prices Index and Retail Prices Index
2. Consumer substitution behaviour - this refers to the concept that as prices rise,
consumers may substitute away from relatively more expensive products and choose
different brands/varieties of a similar product that are relatively cheaper. For example,
choose a different type of loaf within ‘white sliced bread’. Even if it is accepted that
consumer substitution behaviour exists, adopting it for all compensation purposes has been
questioned; do poorer population sub-groups have the means to benefit from flexible
shopping to achieve the best prices?
The key issue of consumer substitution in relation to the CPI and RPI is that they embody different
assumptions about the degree of consumer substitution behaviour between different varieties or
brands of products and these assumptions lie behind the "formula" effect (that is, the differences
caused by using the geometric and arithmetic mean at the first stage of price aggregation)5.
The actual level of consumer substitution in the market place is not clear. If consumer substitution
behaviour is prevalent and considered to be appropriate in the inclusion of an index used for
compensation purposes then the CPI provides the closer match to this ideal. If consumer
substitution behaviour is limited or is not considered to be appropriate for inclusion in an index
used for compensation purposes then the RPI provides a closer match to this ideal.
Public debate continues to present challenges to the use of price indices for macroeconomic and
compensation purposes. To respond to these and other challenges, a robust and efficient
procedure to implement improvements to the CPI and RPI is in place. The UK Statistics Authority
has established the Consumer Prices Advisory Committee (CPAC) to enable clear and efficient
decision making processes, greater transparency and access to external expertise on inflation. In
addition ONS is always keen to hear from users with suggestions on how best to face the
challenges that lie ahead and ONS intends to further improve the mechanisms to do this by
establishing a Prices User Group to complement the existing panel of advisors to help plan the way
forward.
In parallel to this, ONS has already started a research project to investigate whether data exists to
allow better analysis and modelling of consumer substitution behaviour. The project will consider
what information is available to understand consumer behaviour when faced with changes in
relative prices. ONS will aim to analyse the extent to which consumers switch between goods and
services as prices change. This evidence will be used to consider the appropriateness of using
either a geometric or arithmetic mean for combining prices at the first stage of price aggregation.
The tentative plan is to produce an initial report on this work later in 2011/12.
This particular project is part of a wider development plan by ONS to improve the consumer price
indices it produces, details of which can be found in the CPAC Annual Report6.
Office for National Statistics
4
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Annex 1: History of the CPI and RPI
History of the RPI
The RPI was initially developed as a compensation index, derived from an index designed as an
aid to protect ordinary workers from price increases associated with the First World War. The RPI
provides estimates of inflation from 1947 onwards with the first official Retail Prices Index being
produced in January 1956.
The RPI has evolved in response to users’ needs and over the years has been utilised for a wide
variety of purposes. Government uses of the RPI have included uprating of benefits, taxation
allowances and thresholds, indexation of index-linked gilts and the price regulation of privatised
utilities. It has also been used as the main macroeconomic indicator of inflation by economic
analysts and policy makers. This multi-purpose role has helped to shape its development over
time.
A range of indices based on the RPI have also been developed over the years in response to the
widening range of user needs. For example, the RPI excluding mortgage interest payments,
known as RPIX was introduced and became the UK’s official inflation target in October 1992 and
the RPIY - the RPI excluding mortgage interest payments and indirect taxes - was designed to
give a measure of price change largely unaffected by Government-driven changes.
The RPI covers, in principle, only those transactions incurred by private households (overseas
visitors and persons living in institutions are excluded) and excludes very high and low-earning
households (the top 4 per cent of households by income and pensioner households where 75 per
cent of their income is derived from state pensions and benefits are excluded from the RPI). The
RPI is currently used for indexing private contracts, tax allowances and thresholds, and
government gilts. Its objective is to measure the average change in prices of goods and services
bought by the vast majority of households in the UK for consumption purposes (that is, used for
the direct satisfaction of individual needs or desires). The aim of the RPI is to measure changes in
prices on a specified day within a calendar month.
History of the CPI
The CPI has a much shorter history than the RPI and was launched in 1996, first known as the
Harmonised Index of Consumer Prices (HICP) and designed from a macro-economic perspective.
This means the index was designed to be used as an indicator of the rate of inflation within an
economy. The HICP was developed as a comparable measure of inflation across European Union
(EU) Member States and is defined in a series of legally binding regulations drafted by Eurostat in
conjunction with Member States. In December 2003, the National Statistician announced that the
UK version of the HICP would be renamed the CPI, at the same time as the Chancellor
announced that the UK’s inflation target would in future be based on the new CPI, replacing the
RPIX. This is a role the CPI continues to fulfil.
Office for National Statistics
5
Implications of the differences between the Consumer Prices Index and Retail Prices Index
The CPI has also evolved over time in response to users’ needs and new European regulations
with a series of developments to extend its coverage. A selection of indices based on the CPI has
also been designed to meet the specific needs of users. For example, the CPIY was introduced in
2006, designed to give a measure of price changes excluding the effect of indirect taxes, along
with another index - CPI at constant tax rates (CPI-CT). The differences between CPI and CPI-CT
were designed to show the contribution of tax changes to the overall CPI inflation figures.
The CPI covers, in principle, expenditure made by all households within the UK, all residents of
institutional households and foreign visitors to the UK, on goods or services that are used for
consumption purposes (that is for the direct satisfaction of individual needs or desires). The CPI
currently excludes costs associated with owner occupied housing. The aim of the CPI is to
measure price changes incurred within a calendar month, that is, not on a specific day. However,
this concept is not yet achieved in practice.
A fuller history of the RPI and CPI is included in an article by Phil Gooding (History of and
differences between the CPI and RPI July 2011).
Office for National Statistics
6
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Annex 2: Similarities and differences between the CPI and RPI
Similarities between the CPI and RPI
Price collection
The CPI and RPI both measure the average change in price of a fixed basket of goods and
services over time. Both indices are based on a comprehensive price collection that combines the
price movements of around 180,000 price quotes collected each month, for a range of over 650
representative goods and services, from a selection of over 20,000 retail outlets in over 140
locations within the UK, as well as including a range of prices obtained from the internet.
Fixed basket approach
The basket of goods and services priced each month is fixed within each year and is selected to be
representative of the items bought by UK households. Fixing the basket of goods and services
means that as prices change within each year, the relative quantities of each type of product
purchased remain constant. Thus within year movements, for example the monthly inflation rates,
reflect only changes in prices.
Annual update of weighting framework and basket of goods and services
The contents of the CPI and RPI baskets and the expenditure weights associated with them are
updated annually and the resulting within year indices chained together to form a single price index
spanning many years. This regular annual updating helps to ensure that the indices remain
representative of consumer spending over time.
Differences between the CPI and RPI
Table 2.1 – Differences between the CPI and RPI
Population base
CPI
RPI
Includes spending by foreign
visitors to the UK but spending
by UK households abroad is
excluded.
Excludes spending by foreign
visitors but includes spending by
UK households when abroad.
Expenditure data are sourced
from the Household Final
Monetary Consumption
Expenditure component of the
National Accounts.
Based on all spending by all
private and institutional
households.
Expenditure data sourced from
ONS’s household budget survey Living Costs and Food Survey.
Excludes the top 4 per cent of
households by income and
pensioner households with three
quarters of their income coming
from state pensions and benefits.
Office for National Statistics
7
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Commodity coverage
CPI
RPI
Excludes the following items
that are included in the RPI:
Excludes the following items that
are included in the CPI:
•
mortgage interest
payments
•
university accommodation
fees
•
house depreciation
•
•
foreign students’ university
tuition fees
buildings insurance
and ground rent
•
unit trust and stockbrokers
charges
•
estate agents’ fees,
home buyers’ survey
costs and
conveyancing charges
•
council tax
•
television and road
fund licences
•
Commodity measurement
of:
1. New car prices
2. Insurance
expenditure
trades union
subscriptions
1. New car prices are
based on list prices
and quality adjusted
using an option
costing1 technique.
2. Amount paid out in
claims is distributed
among other spending
categories according to
the nature of the claim,
for example a motor
insurance claim would
be allocated to
‘motoring’. The residual
(service charge) is then
allocated to the
relevant insurance
heading.
1. New car prices are imputed
on the basis of movements
of second-hand car prices.
2. All expenditure on
insurance is allocated to
the relevant insurance
heading (for example,
motor insurance
premiums).
1
Option costing is a quality adjustment technique that can be used when old and new items differ by easily
identifiable characteristics which can be separately priced (as options).
Office for National Statistics
8
Implications of the differences between the Consumer Prices Index and Retail Prices Index
CPI
RPI
Index construction
formulae2 at the first stage
of aggregation
The CPI predominantly uses
the geometric mean,
supplemented by a particular
form of the arithmetic mean
known as the ratio of average
prices.
The RPI uses two different types
of arithmetic mean:
Product coding
(the classification system
used for aggregation and
publication of results)
Based on National Accounts’
principles – an internationally
recognised system for
classifying household
expenditure known as
Classification of Individual
Consumption According to
Purpose (COICOP).
Rounding
CPI monthly and twelve-month
rates are calculating using
unpublished3, unrounded
indices.
RPI monthly and twelve-month
rates are calculated from
published rounded indices.
Governance
Governed by a series of legally
binding European Regulations.
Governed by the Statistics and
Registration Services Act 2007.
The UK can only make certain
changes to the CPI. More
fundamental changes would
require ONS to introduce a
national version of the CPI
distinct from the HICP
produced to meet international
requirements.
Any methodological changes to
the RPI require the approval of the
UK Statistics Authority before
being referred to the Bank of
England. If the change is
considered materially detrimental
to the interests of the holders of
index-linked gilt edged securities,
the consent of the Chancellor of
the Exchequer is also required.
2
3
1. Ratio of average prices
(RA)
2. Average of price relatives
(AR)
Based on a long standing
classification system specified and
developed by earlier Cost-of-Living
and RPI Advisory Committees and
unique to the UK.
See Annex 8 for an explanation of aggregation methods and their implication.
Price indices to 3 decimal places are available from ONS on request.
Office for National Statistics
9
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Annex 3: Index purpose
Consumer price indices are put to a number of different uses. Broadly speaking, these uses split
into two categories:
1. Macroeconomic indicators of inflation – this refers to the use of the indices as indicators of the
current status of inflation within the economy.
2. Compensation – this refers to the use of the indices in the public and private sector for income
adjustment and price adjustment, for example uprating state benefits and determining pay
awards.
This broad categorisation can be split further into five more detailed categories:
1.1 For measuring macroeconomic inflation and to inform policy-making decisions.
1.2 For deflation – that is when price indices are used in national accounts to determine growth
in the economy without the effect of price change.
and:
2.1. For indexation of wages, benefits and pensions.
2.2. For indexing gilts – the redemption values of certain gilt-edged securities can be linked to
price indices.
2.3. For indexing rates in private contracts.
It is unlikely that any single statistical index could be constructed to exactly match what is required
for all of these different uses. Neither is there a unique match between index design and purpose several designs can be relevant to a single overall purpose. Hence a consensus approach is
difficult to reach even amongst users with the same or very similar needs. For example, a price
index to adjust the basic state pension could be based on the general change in prices for the
whole population, or the change in prices relevant to the population group who are receiving the
state pension.
Regulations underpinning the production of the European HICP, the equivalent of the UK CPI,
have reached broad agreement within Europe on minimum standards required for the production of
a cross-country comparable macroeconomic indicator of inflation. If cross-country comparability is
of paramount importance, other compromises may be necessary and these need to be
disseminated to users. For example, owner occupiers’ housing costs cannot be measured
consistently across Europe so are currently excluded from the European HICP.
There continues to be a lot of international debate in relation to price indices used for
compensation purposes. A price index used for this purpose is often considered to be
compensating for changes in the ‘cost of living’. The cost-of-living is difficult to define in practice.
Some use it to mean a measure of the cost of buying sufficient quantities of various items to
maintain some minimal standard of living. Defining this standard is subjective however and if it
rises over time, this would influence such an index. Another definition is of an index calculated in
the same way as the CPI and RPI but restricted to basic essentials. However, the definition of a
Office for National Statistics
10
Implications of the differences between the Consumer Prices Index and Retail Prices Index
basic essential is also subjective and tends to change over time with former luxuries such as
telephones now usually considered essential.
In economic terms, a cost-of-living index can be defined as the minimum expenditure at this
month’s prices required to achieve the same level of utility as in some earlier period, relative to the
expenditure in that earlier period. There is no assumption that the relative quantities of goods and
services purchased in the two periods is the same, so the concept is quite different to fixed basket
indices such as the CPI and RPI. A consequence of the word ‘minimum’ in the definition is that a
cost-of-living index usually gives a lower rate of inflation than the CPI or RPI. In other words when
prices rise, provided consumers have a choice, they can always achieve a given standard of living
at lower cost by varying the relative quantities of the goods and services they purchase compared
with the increase in overall spending on a fixed bundle of goods and services. Consumers basically
substitute purchases of relatively expensive items for similar goods that have become relatively
cheaper. This helps limit the rise in the cost of the shopping basket when there is a general
increase in the cost of goods and services overall. It should be noted that neither the CPI nor the
RPI is a cost-of-living index.
International debate focuses on the level of consumer substitution behaviour and to what extent
this should be reflected in a price index. There is evidence to indicate that consumers do substitute
some goods and services in response to relative price changes, though this is believed to vary
between demographic groups. Another widely debated topic is whether in particular, an index used
for compensation purposes should reflect the movements of prices for the whole population or for
the specific demographic groups to which the measure is targeted. These issues have been widely
debated and as yet, there is no definitive resolution.
Office for National Statistics
11
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Annex 4: Decisions on the use of the CPI and RPI
Until the introduction of the HICP in the mid 1990s, the RPI or one of its derivatives such as the
RPIX, were the only measures of UK consumer price inflation available to users. Consequently the
same price index was often used for multiple purposes, for example, as both an indicator of
inflation for macroeconomic purposes and for compensation purposes.
However, with the introduction of the European HICP in 1996, there became a statutory need in
the UK for not one, but two main consumer price indices and this provided users with additional
choices. ONS was and still is responsible for the production and dissemination of these price
indices to meet the requirements of informed decision-making by government, business and
researchers. However, it is for policy makers within government and other decision makers to
determine which statistics they use for their purpose. The use of the CPI, RPI and their derivative
indices in the public domain has been, and remains, a political decision made by the government
of the day. For example:
• the Rossi index – the all-items RPI excluding rent, mortgage interest payments, council tax
and depreciation costs was until recently the index used to uprate state income-related
benefits. The index was named after Hugh Rossi, the social security minister responsible
for its introduction and definition
• in December 2003, the Chancellor of the Exchequer made the decision to use the CPI as
the UK government’s measure of inflation (replacing the RPIX) while at the same time,
confirming that pensions, benefits and index-linked gilts would continue to be calculated on
the same basis as previously, that is, with reference to the all items RPI or its derivatives
• in June 2010 the Chancellor announced that the government would adopt the CPI as the
basis for the indexation of benefits, tax credits and public service pensions from April 2011.
This replaced the previous use of the RPI and Rossi index
• the Budget 2011 announced that from April 2012 the default indexation assumption for
direct taxes will switch from the RPI to the CPI while the employer National Insurance
Contributions threshold, and the age-related allowance and other thresholds for older
people will increase by the equivalent of the RPI. The personal allowance will increase
from 2013-14 by at least the equivalent of the RPI, until the Government’s goal of
increasing the personal allowance to £10,000 is achieved
• the default indexation assumption for indirect taxes and index-linked gilts is currently the
RPI. The Government expects to undertake a formal consultation on the issuance of CPIlinked gilts in 2011-2012. The Government will also review the use of the CPI for indirect
taxes
Office for National Statistics
12
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Annex 5: Statistical considerations
When designing a price index from scratch, there are statistical considerations that need to be
addressed. To put this in context, these considerations have been aligned with the key differences
between the CPI and RPI and are described in the table below:
Table 5.1 – Statistical considerations
Design attribute
Considerations
Stated purpose
For what purpose is the price index mainly used? Is the index
primarily for compensation or macroeconomic purposes? How
can that purpose be best represented in the design of the
index?
Scope (population base)
Which particular consumers or households is the index
designed to represent and how can the different spending
patterns of those particular consumers be best represented?
Coverage (of commodities)
What range of goods and services should be included?
Aggregation formulae
Is the arithmetic mean or geometric mean more appropriate in
the first stage of index aggregation? (a key consideration is
the degree of substitution behaviour assumed in the
population). Should substitution behaviour be included in the
index design at all? Is it valid for all demographic sub groups?
Do all consumers have the means to shop around to gain
efficient substitutions? To what extent should the index take
account of consumer substitution behaviour?
Weighting data
Which is more suitable - expenditure data consistent with the
National Accounts’ classification system (COICOP) or
expenditure data from a household budget survey? Are
sufficient expenditure data available?
Price recording and
treatment of new goods
How and when should new goods be included in the price
index? How should the prices of these goods be measured
and recorded?
Special commodities
For example, what types of housing costs are relevant and
how should they be measured?
Even if agreement could be reached on the main purpose of the price index and assuming all the
statistical considerations detailed above are resolved, there are often practical constraints of timely
index production and data availability that compromise these ideals. No single index could meet
the requirements of all users and compromises are often necessary.
Office for National Statistics
13
Annex 6: Implications of the differences between the CPI and RPI when used as macroeconomic
indicators of inflation
Table 6.1 – Implications of the differences between the CPI and RPI when used as macroeconomic indicators of inflation
Design attribute
Purpose
Primary considerations
Macroeconomic indicator of
inflation.
Population base Coverage of household
expenditure.
(including
expenditure
weights)
UK CPI
RPI
Specifically designed for the purpose of
measuring inflation at macroeconomic
level. Hence consistent with most National
Accounts concepts and consistent with the
European HICP.
Originally designed as a compensation index but
has evolved over time in response to users’
needs. Also used as the main domestic measure
of inflation although not specifically designed as
such.
Advantage over the RPI.
Not designed to be consistent with National
Accounts’ concepts.
Unique to the UK.
All individuals in the domestic territory,
including private households, institutional
households (such as nursing and
residential homes) and foreign visitors.
Expenditures of high and low income households
are excluded from the RPI. The exclusion of such
households is not common in other countries and
is against the principles of the HICP, which
focuses on European comparability.
Source of spending data is household final
monetary consumption expenditure taken
from the National Accounts. Compliant with
European legislation and this ensures
greater coherence with other economic
data.
Wider coverage than the RPI.
Advantage over the RPI.
Office for National Statistics
14
Based on a household budget survey (Living
Costs and Food Survey).
Exclusion of some pensioners and high earners
mean the RPI has a narrower population
coverage than the CPI. However it could be
argued that the exclusion of high and low income
households mean the population represents a
more typical household.
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Design attribute
Primary considerations
UK CPI
RPI
Commodity
coverage
Scope and range of coverage.
The CPI currently excludes owneroccupiers’ housing costs and hence has
narrower coverage than the RPI.
Item
measurement cars and
insurance
Robust price measurement
processes.
Cars
Cars
New car prices are based on list prices and New car prices are imputed on the basis of
movements of second-hand car prices.
quality adjusted using an option costing4
technique. A better method than is used for
the RPI.
Insurance
Insurance
The amount paid out in claims is distributed All expenditure on insurance is considered to
belong to the relevant insurance heading (for
among other spending categories
example housing or motor insurance premiums).
according to the nature of the claim with
only the service charge allocated to the
relevant insurance heading
Advantage over the RPI.
Should reflect consumer
behaviour and changing
markets.
Annual update of goods and associated
weights.
(these are the two
commodities that
have detailed
differences in the
way in which they
are measured and
included in the CPI
and RPI)
Aggregation
Includes detailed treatment of owner-occupiers’
housing (OOH) costs.
OOH costs are a significant proportion of
expenditure; hence the RPI has an advantage
over the CPI by better reflecting household
expenditure.
Annual update of goods and associated weights.
Use of the arithmetic mean (average of price
relatives) can lead to an upward bias across the
chain link period.
If it is decided that it is right to include
substitution then the CPI more closely
resembles consumer behaviour by using
geometric means for products where
significant substitution takes place and
arithmetic means where there is little or no
substitution.
Effectively assumes little or no substitution
between products meaning quantities purchased
remain the same regardless of their changes in
price relative to other products. This is unlikely to
be a realistic reflection of consumer behaviour.
Advantage over the RPI.
4
Option costing is a quality adjustment technique that can be used when old and new items differ by easily identifiable characteristics which can be
separately priced (as options).
Office for National Statistics
15
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Design attribute
Primary considerations
Product coding
(classification
system)
International comparability
Rounding
Maximum precision
UK CPI
RPI
Uses COICOP which aids international
comparability.
The RPI employs its own unique classification
system. It is not internationally comparable but
the classification structure it uses is more detailed
than COICOP.
Advantage dependent on users’ needs.
Suitability for UK purposes
Also suitable for UK purposes.
Advantage dependent on users’ needs.
Transparency
Indices are calculated using full precision
as are the monthly and twelve-month rates.
The full precision of the CPI is a benefit for
macroeconomic purposes as an inflation
target.
Advantage over the RPI for precision.
Public
acceptability
Well recognised and accepted
as the most appropriate
measure
Widely recognised as the UK’s main
inflation measure for macroeconomic
purposes although the CPI may still not be
as familiar as the RPI in the public domain.
Some public scepticism arising from the
formula effect.
Public opinion will vary and is
challenging to measure.
Office for National Statistics
16
Indices are calculated using full precision but the
monthly and twelve-month rates are calculated
from published rounded indices.
The calculation of monthly and twelve-month
rates from rounded indices makes inflation rate
calculations transparent.
Advantage over the CPI for transparency.
Familiarity and credibility built up over time may
give the RPI and its derivatives an advantage in
the public eye, though the CPI is increasingly
becoming the UK’s primary measure of price
change.
Public opinion will vary and is challenging to
measure.
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Annex 7: Implications of the differences between the CPI and RPI when used for compensation
purposes
Table 7.1 - Implications of the differences between the CPI and RPI when used for compensation purposes
Design attribute
Purpose
Primary
considerations
Compensation Index
UK CPI
CPI explicitly developed as a macroeconomic
measure of inflation.
Advantage dependent on precise use for
compensation.
Population base
Target population
All individuals in the domestic territory,
including private households, institutional
households and foreign visitors.
If it is more appropriate to consider domestic
spending in total, regardless of consumer,
then the CPI has an advantage.
Weighting
Expenditure data
RPI
The RPI began life as a compensation index and
historically served that purpose well. It has
evolved over time and later, it came to be used as
the main domestic measure of inflation, which
represented a change to its original intended
purpose.
Advantage dependent on precise use for
compensation.
Excludes some parts of the population –
extremely high and low-earning households.
If it is more appropriate to restrict spending to a
subset of domestic households but regardless of
whether spending is in the UK or overseas than
the RPI has an advantage.
Wide population coverage is a strength in that
the desired sub-population is likely to be
included, but a weakness is the inclusion of
everyone else, including foreign visitors.
If it is appropriate to include all UK
households including those in institutional
households for the purposes of
compensation then the CPI has an
advantage.
By excluding very low income households the RPI
may exclude portions of the sub- population that
are being targeted for compensation.
Based on the final household consumption
expenditure component of the National
Accounts.
Based on an ONS household budget survey
Office for National Statistics
If it is appropriate to exclude very high and low
income households and those in institutional
households for the purposes of compensation
then the RPI has an advantage.
Survey data potentially allowing analysis of
17
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Design attribute
Primary
considerations
UK CPI
RPI
spending by demographic group or sub group (for
Challenging to adapt national expenditure data targeted compensation purposes).
to a sub-population (for targeted compensation
purposes).
Advantage dependent on precise use for
Advantage dependent on precise use for
compensation
compensation.
Commodity
coverage
Target population
Item measurement - Robust price
cars and insurance measurement
(these are the two
commodities that have
detailed differences in
the way in which they
are measured and
included in the CPI and
RPI)
Excludes owner occupiers’ housing costs,
council tax, television and road fund licenses
and trades union subscriptions.
Includes owner occupiers’ housing costs, council
tax, television and road fund licenses and trades
union subscriptions
The exclusion of owner-occupiers’ housing
costs and other commodities is a weakness if,
these costs are faced by the target subpopulation and if it is considered desirable to
compensate for them; it is a strength
otherwise.
Advantage dependent on how well the
commodity coverage fits the target
population.
Inclusion of owner occupiers’ housing costs and
other commodities is an advantage if people face
these costs. In addition these costs can always be
excluded if necessary.
Cars
New car prices are based on list prices and
quality adjusted using an option costing5
technique. A better method than is used for
the RPI.
Insurance
The amount paid out in claims is distributed
among other spending categories according to
the nature of the claim with only the service
charge allocated to the relevant insurance
heading
Advantage over the RPI.
Cars
New car prices are imputed on the basis of
movements of second-hand car prices.
Advantage dependent on how well the
commodity coverage fits the target population.
Insurance
All expenditure on insurance is considered to
belong to the relevant insurance heading (for
example housing or motor insurance premiums)
5
Option costing is a quality adjustment technique that can be used when old and new items differ by easily identifiable characteristics which can be
separately priced (as options).
Office for National Statistics
18
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Design attribute
Aggregation
Primary
considerations
Consumer substitution
UK CPI
The use of the geometric mean in the CPI at
the first stage of aggregation mimics the
practice of consumer substitution between
different brands/varieties of a similar item in
response to relative price changes.
If it is decided that it is right to include
substitution then the CPI more closely
resembles consumer behaviour by using
geometric means for products where
significant substitution takes place and
arithmetic means where there is little or no
substitution.
RPI
Use of arithmetic mean in the RPI at the first stage
of aggregation assumes little or no substitution
between different varieties or brands of products
in response to price change. and hence assumes
that consumers continue to buy the same
quantities of these goods and services
irrespective of their price change relative to other
products.
If consumer substitution is limited or it is decided
to exclude consumer substitution behaviour from
the price index, the RPI reflects this behaviour
better than the CPI.
Advantage dependent on the prevalence of Advantage dependent on the prevalence of
consumer substitution in the target
consumer substitution in the target
population.
population.
Product coding
(classification
system)
Suitability for UK
purposes
Uses COICOP, an internationally recognised The RPI employs its own unique classification
classification system and which is also suitable system, suitable for UK purposes and more
for UK purposes.
detailed than COICOP.
Advantage for more detailed classification
Advantage for international comparisons. requirements within the UK.
Rounding
Maximum precision
Indices are calculated using full precision as
are the monthly and twelve-month rates.
Transparency
Advantage over the RPI for precision.
Office for National Statistics
Indices are calculated using full precision but the
monthly and twelve-month rates are calculated
from published rounded indices.
The calculation of monthly and twelve-month rates
from rounded indices makes inflation rate
calculations transparent.
Advantage over the CPI for transparency.
19
Implications of the differences between the Consumer Prices Index and Retail Prices Index
Design attribute
Primary
considerations
Public acceptability Well recognised and
accepted as the most
appropriate measure.
UK CPI
RPI
The role of the CPI for compensation purposes
is a new one and hence public debate
continues on its use for this particular purpose.
Public opinion will vary and is challenging
to measure.
Office for National Statistics
Familiarity and credibility built up over time give
the RPI and its derivatives an advantage in the
public eye.
Public opinion will vary and is challenging to
measure.
20
Annex 8: Aggregation methods
The basket of goods and services priced for the CPI and RPI is based on a detailed and
comprehensive weighting framework which is updated annually. Internationally there tends to be a
lack of reliable expenditure data for products at the lowest level of price collection, meaning it is not
always possible or meaningful to derive weights in either quantity or value terms for individual
products within these tightly defined commodity groupings (for example, for a large white sliced
loaf purchased in an independent store in Birmingham). In these circumstances it is internationally
accepted practice to assign an equal weight to each price observation. Price indices at the lowest
level (where prices enter the index) are predominantly calculated using a simple arithmetic or
geometric mean. The three averaging techniques used in the first stage of price aggregation of the
CPI and RPI, together with their basic assumptions, are listed below:
y ratio of average prices (RA) – used in the RPI for tightly defined items, for example food,
alcohol and tobacco, and in the CPI for items where the opportunity for consumer
substitution is limited, for example petrol and diesel products. This method implicitly gives
greatest importance to the highest priced products in estimating overall price change and
assumes the quantities purchased remain the same regardless of their change in price
relative to other products.
y average of price relatives (AR) – used in the RPI where wider variations in price, resulting
from broader item descriptions limits the application of RA, for example, clothing and
furniture. Its use is prohibited in the CPI because it can be shown in certain circumstances
that the use of AR combined with chain linking of in-year indices introduces an upward bias
known as ’chain drift’6. AR is calculated as the arithmetic average of price relatives7 and
assumes equal expenditure on each product in the reference period.
y geometric mean of price relatives (GM) – used for the first stage of aggregation of around
70 per cent of the CPI. GM assumes value shares of each item are the same each period.
As a product becomes more expensive relative to others, the quantity purchased declines,
in such a way that the expenditure on that product remains the same.
In summary:
For the CPI the choice of aggregation formulae is between GM and RA. The GM is used for the
majority of items, unless the opportunity for consumer substitution between similar products is
likely to be limited as in the purchase of petrol and diesel products.
For the RPI, the choice of aggregation formulae is between RA and AR. For items that are tightly
defined as is the case for food, alcohol and tobacco then RA is used. For items where there are
likely to be large variations in prices, for example with furniture, AR is used.
6
Price bounce is a bias in the AR aggregation method as a result of negatively correlated prices across the chain link period which can result in what is known as ‘chain drift’. 7
A price relative is the ratio of the price of a specific product in one period to the price of the same product in some other period. Office for National Statistics
21
Implication of the differences between the Consumer Prices Index and Retail Prices Index
Annex 9: International practice
A quick review of international practice indicates that other countries do not generally use a single
price index for both macroeconomic and compensation purposes although in some instances, for
example in France, the differences between the two indices can be minimal.
Table 9.1 - Practice in other countries, including a brief explanation of the major differences
between the indices used for macroeconomic and compensation purposes
Country
Australia
Index used for
macroeconomic
purposes
National CPI
Denmark
HICP
Finland
HICP
France
HICP
Index used for
compensation
purposes
Specialist index Pensioner and
Beneficiary Living
Costs Index (PBLCI) is
used to index social
security pensions
when the PBLCI is
greater than the CPI.
Average wages used
for the majority of
compensation
purposes.
National CPI used for
regulating child
benefits.
National CPI used for
compensation together
with wage and salary
earnings.
National CPI used for
compensation
purposes.
How the indices
differ
PBLCI includes:
• interest charges
and excludes:
• house purchase
PBLCI collects only
concessional prices;
food is given a higher
weight than in the CPI
and education a much
lower weight.
National CPI includes:
• owner occupied
housing (OOH)
National CPI is a costof-living index and
includes:
• OOH
• games of chance
• vehicle tax
• interest rates for
consumer loans
Minor differences
relating to health
insurance (National
CPI is net of insurance
reimbursement).
Office for National Statistics
22
Implication of the differences between the Consumer Prices Index and Retail Prices Index
Country
Germany
Netherlands
Norway
Index used for
macroeconomic
purposes
HICP – for EU
comparisons
National CPI or
sometimes the
national retail trade
prices index.
HICP
Version of the national
CPI adjusted for
changes in taxes and
excluding energy (CPIATE) is used by the
Norwegian central
bank as a target for
monetary policy.
Index used for
compensation
purposes
National CPI used for
compensation in
private contracts (e.g.
occupational
pensions) together
with the national retail
trade prices index.
Specialist index for the
adjustment of social
benefits (Hartz IV)
National CPI or CPICT (constant taxes)
generally used for
compensation
National CPI used for
compensation
purposes.
How the indices
differ
Hartz IV index
includes:
• consumption
habits of lower
income consumers
and excludes:
• alcohol, tobacco,
cars and package
holidays
National CPI:
• uses a national
concept - all
expenditure by
Dutch residents
included
and includes:
• dog tax and motor
vehicle tax
• OOH
• gross insurance
premiums
There are also a few
minor classification
differences.
National CPI:
• uses a domestic
approach and its
stated concept is a
cost-of-living index
• Includes OOH
Office for National Statistics
23
Implication of the differences between the Consumer Prices Index and Retail Prices Index
Country
USA
Index used for
macroeconomic
purposes
National CPI
Index used for
compensation
purposes
CPI-W (CPI for wage
earners and clerical
workers) is used to
determine annual
social security and
federal retirement
cost-of-living
adjustments and
collective wage
bargaining.
How the indices
differ
CPI-W excludes:
• professional,
managerial and
technical workers
• self-employed
• unemployed and
others not in the
work force (e.g.
retired workers)
Acknowledgements
The author would like to thank Jeff Ralph for his contribution to this paper.
Further Information
For further information please contact Tegwen Green 01633 455789
Alternatively email: [email protected]
© Crown Copyright 2011
References
1
Code of Practice for Official Statistics available at: http://www.statisticsauthority.gov.uk/assessment/codeof-practice/index.html
2
UK Statistics Authority Assessment Report 79: Consumer Price Indices available at:
http://www.statisticsauthority.gov.uk/assessment/assessment/assessment-reports/index.html
3
UK Statistics Authority Monitoring Brief – Communicating Inflation, available at:
http://www.statisticsauthority.gov.uk/assessment/monitoring/monitoring-briefs/monitoring-brief-7-2010--communicating-inflation.pdf
4
Gooding P (July 2011) ‘History of and differences between the Consumer Prices Index and Retail Prices
Index’, available at: http://www.statistics.gov.uk/downloads/theme_economy/history-diff-cpi.pdf
5
For more information on the formula effect and differences between the arithmetic and geometric mean
see: ONS (2003) ‘The New Inflation Target: the Statistical Perspective’, pp 23-25, available at:
http://www.statistics.gov.uk/downloads/theme_economy/New_inflation_target_031210.pdf
6
‘Consumer Prices Advisory Committee – 2010 Annual Report to the UK Statistics Authority’, available at:
http://www.statistics.gov.uk/downloads/theme_economy/cpac-annual-report2010.pdf
Office for National Statistics
24