Professional Documents
Culture Documents
Measurement Issues
in European Consumer
Price Indices and the
Conceptual Framework
of the HICP
Publisher
European Central Bank,
Frankfurt am Main
Centre forEconomic Policy Research,
London
Editors
Gonzalo Camba-Mendez
Vítor Gaspar
Mark Wynne
Design
Dirk Stähling
Print
Kern & Birner
June 2002
ISBN 92-9181-306-0
16th – 17th November 2001
Measurement Issues in
European Consumer Price Indices
and the Conceptual Framework
of the HICP:
Summary and Conclusions of the
CEPR/ECB Workshop on Issues
in the Measurement of Price Indices
The views expressed in this paper are those of the authors and do not necessarily reflect those
of the European Central Bank, the Eurosystem, the Federal Reserve Bank of Dallas or the
Federal Reserve System.The authors would like to thank Henning Ahnert, John Astin, Ernst
Berndt, Peter Bull, Erwin Diewert, David Fenwick, Johannes Hoffmann, Omar Licandro,
Jacques Mairesse, Pedro Neves, Diego Rodriguez-Palenzuela,Wolfgang Schill, Mick Silver and
Gian Maria Tomat for helpful comments and suggestions.
A Workshop on Issues in the
Measurement of Price Indices was
hosted by the European Central Bank
(ECB) on the 16th and 17th of
November, 2001.The main objective of
the workshop was to provide a forum
for central bankers, statisticians and
academics to discuss price measurement
issues, particularly in relation to the
Harmonized Index of Consumer Prices
(HICP). No comprehensive research on
measurement issues related to the HICP
was previously available. The Workshop
was a first step towards filling this gap,
and advanced our knowledge in two
major areas: i) Partial estimates of
measurement biases in European
consumer price indices; ii) The
conceptual framework of the HICP.The
Workshop was jointly organized with
the Center for Economic Policy Research
(CEPR). The composition of the
Scientific Committee responsible for
the programme was as follows: Ernst
R. Berndt (Massachusetts Institute of
Technology), Gonzalo Camba-Mendez
(ECB), Vítor Gaspar (ECB) and
Dietmar Harhoff (Universität
München and CEPR). Participants in
the Workshop included staff of the ECB,
several National Central Banks of the
Eurosystem, Eurostat, several European
National Statistical Institutes, as well as
leading academics from North America
and Europe.
1. Introduction 8
3. Main Conclusions
of the Workshop 15
1
Stigler, G. J. (1961):‘The Price Statistics of the Federal Government: Report to the Office of Statistical Standards.’
Bureau of the Budget. NewYork: National Bureau of Economic Research.
2
Boskin, M. J.; Dulberger, E. R.; Gordon, R. J; Griliches, Z. and Jorgenson, D.W. (1996):‘Towards a More Accurate Measure of the Cost of
Living.’ Final Report to the Senate Finance Committee from the Advisory Commission to the Study of the Consumer Price Index. 9
3
National Research Council (2001).AtWhat Price? Conceptualizing and Measuring Cost-of-Living and Price Indexes. Panel on Conceptual,
Measurement, and Other Statistical Issues in Developing Cost-of-Living Indexes, Charles L. Schultze and Christopher Mackie, Editors.
Committee on National Statistics, Division of Behavioral and Social Sciences and Education.Washington, DC: National Academy Press.
contained a series of recommendations price changes. The National Statistical
for the BLS to further improve the US Institutes (NSIs) of the member states
CPI (some of which have already been and the European Commission (Eurostat)
adopted). took on the task of developing the
In Europe, the prospect of Economic “Harmonized Index of Consumer
and Monetary Union (EMU) brought a Prices” (HICP). The coverage of the
new dimension to the problem of initial HICP was mainly that which was
accurately measuring consumer price common to the CPIs of member
inflation. The Maastricht Treaty set the countries.The coverage extends now to
convergence criteria to guide decisions almost all items of consumers’
on the passage to stage three of EMU, expenditure.The HICP has been adopted
i.e. how to decide which The HICP has been by the European Central Bank
countries would qualify for adopted by the (ECB) in its quantitative
European Central
adoption of the single currency. Bank (ECB) in its definition of price stability. It is
One of the criteria was (and quantitative therefore very important from a
definition of price
still is for countries that might stability. It is European perspective to identify
join the euro area in the therefore very weaknesses, if any, in its
important from
future) that inflation does not a European conceptual framework and
exceed by more than 1.5 perspective to construction, as well as ways in
identify
percentage points the inflation weaknesses, if any, which both can be improved.
rate of (at most) the three in its conceptual
best-performing member states framework and
construction, as
The Workshop on Measurement
in terms of price stability. In well as ways in Issues in Price Indices aimed
which both can be
October 1995 the Council of improved. at providing a forum where
Ministers of the European economists and statisticians
Union adopted the framework regulation could discuss areas of common interest.
for the establishment of a harmonized While the bulk of the papers presented
methodology for the construction of at the workshop focused on problems
consumer price indices in EU member associated with the measurement of
states for the purposes of assessing inflation at the consumer level, the
convergence4.The conceptual frameworks importance of measurement issues
of the existing consumer price indices extends well beyond the domain of
of the member countries of the consumer spending. The rapid pace of
European Union were different, as were technological innovation in the
their coverage of items of consumption information technology (IT) sector in
expenditure and the measurement of recent years has created major challenges
10 4
Council Regulation (EC) No 2494/95 (OJ L 257, 27.10.1995, p.1).
for the construction of accurate deflators discussed during the workshop are
for high-tech goods. The US has used presented in section 6. Section 7 discusses
hedonic prices rather more than potential areas for future research
European countries for the deflation of work.
investment spending in the national
accounts, and some have argued that this
accounts for part of the better measured
performance of the US economy in the
latter half of the 1990s. Understatement
of the true rate of price decline for
important categories of investment goods
may lead to understatement of Understatement of
GDP growth. For the purposes the true rate of
of monetary policy, accurate price decline for
important
measures of real GDP and categories of
productivity growth are investmentmay lead to
goods
14 6
Domingo Solans, E. (2001):‘Issues Concerning the Use and Measurement of the HICP’. Speech delivered at the CEPR/ECB Workshop.
Downloadable from http://www.ecb.int
3.
Main Conclusions
of the Workshop
This workshop contributed to our also for the euro area aggregate HICP
knowledge on two major areas: (the MUICP) which is a weighted
average of the results of all euro area
1. Partial estimates of measurement countries. The workshop did, however,
biases in European consumer put together a collection of studies
price indices. conducted in Europe, and was therefore
The Workshop confirmed the view that a step in the right direction in filling this
the potential bias in the HICP should gap. The conclusions from these studies
not be regarded as constant over time, validate the views of those who strongly
but should rather be understood as warned against i) extrapolating the
changing as a function of the economic results obtained in the US to Europe,
environment. Substantial evidence was and ii) transferring the results for certain
presented that suggested that Substantial index sub-components (such as
the measurement bias for evidence was IT goods) to other components
presented that
certain items is downward suggested that the in the index. As an illustration
rather than upward, contrary measurement bias in this regard, some evidence
to the widely held belief that for certain items is
downward rather
was presented of a slight
measurement error always than upward, understatement of rent inflation
causes inflation statistics to contrary to the in the rent component of the
widely held belief
overstate the true rate of that measurement consumer price index for West
inflation. In recent years, many error always Germany. Evidence was also
causes inflation
analysts have noted the absence statistics to presented of measurement
of a comprehensive report on overstate the true problems that may result from
rate of inflation.
the potential for measurement the methods commonly used
bias in European price indices. No by NSIs to control for quality changes,
broad studies of the accuracy of i.e. the matched model approach. This
European consumer prices indices suggested that the matched models
comparable to those conducted in the approach is not always well suited to
US by the Boskin Commission in 1996 today’s dynamic markets. In this
or by the National Academies in 2001 is regard, it was shown that both the size
available in the European context, and and sign of any resulting biases depend
therefore no overall assessment of the crucially on the structure of the markets
sign or size on any bias in the HICP is concerned.
possible. This is true for each of the
national HICPs in the European Union
and euro area, and, as a consequence,
16
2. The conceptual framework of
the HICP.
There was general agreement on the
appropriateness of building the HICP
on the foundations of economic theory
(consumer theory or inflation theory).
However, the precise theoretical concept
was not regarded as major issue by most
participants in view of the actual
practice of statistical agencies. For
example, there was consensus that one
source of bias which is often cited in the
discussion about “cost of living indices”
(COLI) – namely the substitution bias –
may not be a major issue in practice in
Europe, because of the use of
There was general
frequently updated weights in agreement on the
price indices. In addition the appropriateness
of building the
practice of the NSIs to HICP on the
carefully ensure that the foundations of
economic theory.
sample basket of goods is
representative of the consumption
patterns, may also serve to minimize
any substitution bias. On the basis of
this broad consensus, discussions held
suggested that, rather than focusing on
theoretical debates, the main issue was
to focus on improvements for price
statistics which lead to better practices
for the treatment of new goods,
housing, quality adjustment or the use
of new data sources (e.g. scanner data).
17
4. Partial Estimates of
Biases in European Consumer
Price Indices
4.1. Results with Scanner Data estimates. David Fenwick, acting as
Mick Silver presented results from a discussant for this paper, pointed that
7
hedonic and matched model experiment there is indeed the potential for biases as
using scanner data. Results in the paper large as that found in this work for
showed that a form of sampling bias, not certain items of the UK RPI. The use
previously identified, may exist as a made of scanner data in the UK Office
by-product of the use of the matched for National Statistics points in this
model method to control for quality direction for some cases. For example,
changes, because the matching excludes some of the most popular models of
unmatched new and old models from dishwasher representing 70 per cent of
the sample (i.e. there is sample sales in scanner data figures, represent
depletion). This is not the case with the only 2 per cent of the quotes in the
hedonic method. The failure of the construction of the RPI. Nonetheless,
matched model method is therefore more work is needed before techniques
more acute in very dynamic markets such as those used by Mick Silver can be
where new models replace old The failure of the implemented by NSIs. The
ones at a fast pace. The results matched model reasons for this are not only
method is more
presented were computed using acute in very related to the robustness of
scanner data on monthly prices dynamic markets hedonic regressions, but just as
where new models
for washing machines in the replace old ones at importantly, the quality of
UK for 1998. Sample depletion a fast pace. scanner data and the lack of
was severe over the course of the year: complete correspondence between
for example, only 53 per cent of the what is provided in scanner data and the
January 1998 basket of varieties were definitions used in the construction of
used for the December/January index, the RPI. This is not to say that scanner
and those accounted for just 48.2 per data is of no use in the computation of
cent of the transactions in December. consumer price indices. Scanner data
The results showed that quality adjusted can play a very important role even now
prices fell faster for the matched sample, in checking and controlling the
than for the unmatched samples. The representivity of the base sample in
exclusion of unmatched data seriously assisting with an effective replacement
overstated these price falls. In any case strategy as well as providing a mechanism
all the estimates presented, whether for adjusting for quality changes. He
based on hedonics or matched models, emphasised the importance of the sort
gave estimates of price changes that were of quality assurance work that had been
vastly below the corresponding official conducted in the UK and that further
7
Heravi, S. and Silver, M. (2001):‘Why the CPI Matched Models Method May Fail Us: Results from an Hedonic and Matched Experiment 19
Using Scanner Data’. Paper downloadable from http://www.ecb.int
work needed to be done to identify and early part of Gordon’s sample.This means
test solutions. In this respect it was that the prices found in catalogues do not
important that countries learned from reflect changes in the price of textiles,
one another and found common and this might alter the results presented.
solutions.
4.3. Price Indices for Housing
4.2. Apparel Prices in the US There are difficulties in measuring price
8
Robert J. Gordon presented a paper changes for goods and services whose
on apparel prices in the US using Sears characteristics are changing rapidly. As
Catalog data for the period 1914 to highlighted in the previous two papers,
1993. Results in the paper showed also the matched model method, though
that the matched model indices may be widely used in statistical offices, might
flawed because price increases that not always be the solution. Hedonic
occur at the same time as model methods have been introduced to deal
changes are excluded from the matched with such problems, but there are also
model index. Gordon showed There are difficulties associated with the
that a hedonic price index for difficulties in use of hedonic techniques. One
measuring price
women’s dresses always changes for goods major difficulty is the fact
increases faster than the and services whose that omitted unobserved
corresponding matched model characteristics are
changing rapidly.
characteristics correlated with
index, and suggested that his those included in the hedonic
findings raise the possibility that price regression can bias the hedonic estimates.
indices for goods subject to frequent Olympia Bover9 presented a hedonic
fashion or taste changes (such as clothing) house price index for Spain that
may be subject to significant downward controls for characteristics which are
bias. John Muellbauer suggested unobservable in their sample such as
some caution before extrapolating this location, transport, traffic and so on.
estimated bias to the figures reported in This was possible by relying on data that
the CPI for apparel prices. There had provided information on a site identifier
probably been significant methodological for flats, and also information on the
changes over the period under study, number of dwellings built on that site.
and it might be that some of the Data employed in this study is regularly
divergences in prices found in the study collected by the Spanish Ministerio de
could be attributed to this. Ernst Fomento. The study reports a difference
Berndt pointed to the fact that many with respect to the Spanish residential
clothes were made at home during the construction deflator used in the National
20 8
Gordon, R. J. (2001):‘Apparel Prices and the Hulten-Bruegel Paradox’. Paper downloadable from http://www.ecb.int
9
Bover, O. and Velilla, P. (2001):‘Hedonic House Prices without Characteristics:The Case of New Multi-unit Housing’. ECB Working Paper
No17. Paper downloadable from http://www.ecb.int
Accounts of around -3.5 percent per slight understatement of rent inflation
year. The authors suggest that the of around 0.5 percent in the official rent
disparity may not all be due to quality index for West Germany. Most of the
adjustment but also to the fact the divergence from the official rent index
Spanish deflator is an index based on the occurred in the years 1991 to 1994,
cost of the construction inputs, while when the housing market was under
this was not the case for the hedonic stress. The data employed was taken
index reported. Spain has one of the form the German Socio-Economic
highest owner-occupancy rates in Europe Panel.
(85 percent), and new dwellings account
for most of the residential construction, Stephen Cecchetti acted as discussant
which in turn represents a substantial for these two papers on housing prices.
part of Gross Fixed Capital Formation. He started his discussion by pointing to
The impact of biases in the house price three major uses of house price indices.
index can therefore have a significant First, they are needed to measure real
impact on overall GDP. The relevance GDP. Second, housing costs are relevant
of mismeasurement of house prices in the computation of a Cost of Living
could also be of major concern if Index. Third, house price inflation is
owner occupied housing is included in important for monetary policy purposes.
the HICP and measured by the price of The type of house price index that we
new dwellings. may prefer will depend on the use we
want to make of it. For a deflator for
Johannes Hoffman and Claudia that part of Gross Fixed Capital
Kurz10 presented a study on rent indices Formation that corresponds to residential
for housing in West Germany, covering houses, the price index for new houses
the years 1985-1998. Contrary to Spain, as computed in the paper presented by
nearly 60 percent of households in Olympia Bover is the right one, though
Germany live in rented dwellings. This it may not be for a cost of living index.
explains why the weight of rents in the This is because house owners are hedged
German HICP is much higher than for against changes in these prices. But this
other European countries, 8.6 percent approach could work with a percentage
in 2001, compared with around 1.5 per correction to account for new house
cent in the Spanish HICP. A hedonic owners only. Cecchetti expressed the
housing rent price index computed in view that the net acquisitions price
this study concluded that for the period approach is the right approach for
under review there is evidence of a monetary policy purposes. This is so
10
Hoffmann, J. and Kurz, C. (2001):‘Rent Indices for Housing in West Germany 1985-1998’. ECB Working Paper No 116. 21
Paper downloadable from http://www.ecb.int
because it is important to have a measure of housing) can lead to huge cross
of asset prices. Following this reasoning, country discrepancies. This applies also
it would be very important to incorporate to the disparities across countries
housing in the HICP. concerning the depreciation rate for
housing. Ernst Berndt highlighted the
Robert Gordon pointed out that the fact that homeowners in the US pay
price of property includes the price of taxes in accordance with the value of
land and this is problematic as it should their properties, which generates a large
be separated from the price of housing. business for appraisers, and noted that
This may not be very important for the the use of hedonic techniques for this
case of Germany because the housing purpose is widespread.
stock is much more homogeneous, but
this entails more problems in the case of 4.4. Price Indices for Cars
the US. John Astin noted the interest A collection of papers on the potential
of the Eurostat in the analysis presented for quality bias in car price indices for
by Olympia Bover. Eurostat is interes- Spain, Italy and the Netherlands was
ted in incorporating owner-occupied presented at the workshop. Omar
housing on a net acquisitions basis in the Licandro 11 presented results of a
HICP. Erwin Diewert commented on quality-adjusted index for new cars in
the need to disentangle the price of the Spain.The method adopted to adjust for
land from the price of housing, but once quality was that of hedonic regressions.
more this depends on the purpose of the The data used are collected by the
index. In a price index for the National ‘Instituto de Estudios de Automocion’
Accounts the land should play no role, in Spain. This dataset covers the whole
and the imputation should be exclusively population for new cars sold in Spain,
for the structure. If on the other hand, together with information on 28
one wants to compare owner occupied characteristics. Information on prices
homes with rental homes, in order to refers to catalogue (or list) prices rather
compare like with like, the price of land than transaction prices.The Spanish NSI
should be included. The rent has the uses only a sample of this population in
price of land included and it is therefore computing its car price index. To avoid
necessary to incorporate the opportunity multicollinearity, the authors adopt the
cost of land in the home ownership strategy of grouping the indicators in a
category as well. Differences in the quality index by making use of principal
practices of NSIs (in terms of including components.The quality-adjusted index
or excluding the price of land in the cost computed is then compared to an index
22 11
Izquierdo, M., Licandro, O. and Maydeu,A. (2001):‘Car Quality Improvements and Price Indices in Spain’.
Paper downloadable from http://www.ecb.int
built in line with the Spanish NSI improve the measurement of the cost
strategy, namely the matched model of living by focusing on prices of the
method. Results show an overstatement fundamental services rather than prices
of the price of new cars by as much as of inputs. It attempts to measure the
3.1 percent per year over the period cost of attaining a given level of utility.
1997-2000 in the official index. Gian Two separate indices were built, a fixed
Maria Tomat 12 presented results that cost index that included total
suggested that the official Italian depreciation, insurance premium and
automobile CPI is also characterized by road tax, and an operating costs index
a positive quality bias of around 2.2 which includes maintenance and repairs
percent over the period 1988-98. The and fuel consumption.The data used for
data employed in the analysis is the the analysis is built from three sources.
Quattroruote Price and Characteristics Data- First the Royal Dutch Tourist Office
base. The paper presented discussed (ANWB) collects detailed information
measurement problems The problem of on prices and technical
associated with the hedonic unobserved characteristics of cars. The RAI
characteristics
regression method, in particular introduces Data Center (RDC), responsible
problems associated with not nuisance for car registrations in the
parameters in the
being able to define in a hedonic regression Netherlands, provides the
quantifiable manner all relevant function, which required information on sales
in turn can affect
characteristics.The problem of the precision of and a limited set of
unobserved characteristics this method. characteristics. The data on
introduces nuisance parameters in the user costs come from the ANWB. The
hedonic regression function, which in annual growth of total user costs for the
turn can affect the precision of this CPI based index was 1.45 percent over
method. The analysis presented in the the period 1992-1999, while the car
paper suggests that the use of a chain model based user-cost index increases
index system should help to reduce at an annual average growth of 0.67
measurement problems. Ben Bode 13 percent. This difference is mostly
presented a paper that examined the explained by large disparities in the
difference between car model based indices for operating costs.
user-cost indices and CPI based indices
in the Netherlands. The user-cost Jack Triplett acted as discussant of this
approach presents an alternative to session and pointed that the correlation
hedonic methods for quality adjustment. among the characteristics is a problem
The user-cost approach attempts to not on its own but if it comes together
12
Tomat, G. (2001):‘Durable Goods, Price Indexes and Quality Change:An Application to Automobile Prices in Italy 1988-98’. 23
ECB Working Paper No 118. Paper downloadable from http://www.ecb.int
13
Bode, B. and van Dalen, J. (2001):‘The Cost of Private Transportation in the Netherlands, 1992-1999’. Paper downloadable from
http://www.ecb.int
with large errors in the data. He warned with that in countries like the UK and
that employing principal components secondly because the study presented
brings the problem of interpreting the focused exclusively on food items,
coefficients. He warned that the user while these big retail outlets sell many
cost approach tends to over-adjust as other goods. The final impact of outlet
some of the savings in user costs might substitution bias on the Portuguese CPI
already be incorporated in the prices. might thus be much larger.
Triplett also noted that the findings in
the papers contrasted with what was
typically found in US studies of quality
adjustment of auto prices using hedonics,
specifically that the hedonically adjusted
prices tend to rise faster than the official
indices.
31
6.
Some Other Issues
The workshop made clear that research people to be in doubt about changes in
by economists is providing valid tools prices. Stephen Cecchetti pointed that
aimed at circumventing measurement the implicit levels of inflation computed
issues in price indices. It is worth from survey data are usually wrong, but
mentioning that the mandate of the there is valid information in this survey
committee of experts of the NAS about changes in inflation. He pointed
mentioned above was to provide new out that it is also important to look not
methods and recommendations to only at first and second moments, but
improve measurement of consumer the third and fourth moments as well.
prices. This was a de facto The third moment of inflation is
acknowledgement that a useful input important to know whether there are
from economists is expected in this area. downside price rigidities. Mick Silver
Papers at the workshop covered a range asked whether the form in which the
of techniques to deal with measurement question on price changes is put to the
problems. company managers in the
The workshop
made clear that French Business Survey includes
Jacques Mairesse presented research by
20
quality adjustment.
economists is
an analysis of the consistency providing valid
of the French producer price tools aimed at Ian Crawford 21 presented a
circumventing
index with data on price measurement revealed preference method
changes published in the French issues in price for calculating a virtual price
indices.
Business Survey. They found for valuing new goods. Rather
that the dispersion between these two than a virtual price, the nonparametric
types of measures was extremely large method proposed computes a lower
and attempted to develop a model of bound for that virtual price. This
friction that could explain this technique allows the computation of a
inconsistency. Peter Hill acting as cost of living index when the number of
discussant of this paper pointed that the goods available changes over two
two French enquiries are not treated periods. He applied this technique to
symmetrically, and therefore one does the case of the UK National Lottery, and
not know whether to expect consistency showed that the failure to include the
or not. As for the model of friction National Lottery in a superlative measure
developed, he pointed that the thresholds of consumer price inflation in the UK
(for a no-change in prices reply in the caused an upward bias of 0.156
French Business Survey) appear to be percentage points. Pedro Neves pointed
too large to be those expected from that the revealed preference approach
20
Desplatz, R. and Mairesse, J. (2001):‘Assessing the Consistency of Price Changes at the Firm Level in Two Widely Different Data Sources: 33
The French Producer Price Surveys and the French Business Surveys’. Paper downloadable from http://www.ecb.int
21
Blow, L. and Crawford, I. (2001):‘A Non-Parametric Method for Valuing New Goods’. Paper downloadable from http://www.ecb.int
could be used to deal with other types of Nicole Jonker22 compared the hedonic
biases such as quality bias and substitution method for quality adjustment with a
bias. But the success of this technique method based on the use of discrete
depended very much on having enough choice models. She pointed that while
variation in relative prices. He also the hedonic method was based on both
noticed that, despite the theoretical consumer and producer theory, the
merits of the revealed preference discrete choice model concentrates on
approach, its implementation raises some consumer utility maximising behaviour.
difficulties for statisticians, given the very A drawback of the discrete choice-model
strong data requirements. Gabriel is that its computation requires
Fagan questioned the use of these information on individual consumers
techniques by NSIs. He pointed that income, information that is not typically
looking at the practicalities of the HICP, available to working statisticians on a
NSIs take time before they incorporate real time basis.
new goods, and therefore information on
previous prices is available at the time this Irwin Collier23 computed comparable
good is incorporated. David Fenwick cost of living indices for East and West
pointed out that the lottery is not in the Germany over the period 1989-1998.
UK RPI and that the latter is not a COL The official price statistics available are
index. He questioned therefore the only able to observe the parallel
appropriateness of using the revealed development of consumer prices in East
preference approach outside the and West Germany, but in order to
framework of a COL index. In addition calculate the purchasing power of the
he also cast doubt on the realism DM in East and West Germany detailed
underpinning the restrictive conditions price comparisons are needed. On the
under which revealed preference theory basis of a study on comparisons of
can be applied. Mick Silver mentioned consumer prices for 1993 Collier gauged
that for a COL index it is essential to do the difference in the costs of living.
something about new goods, but given Axel Weber pointed to the fact that price
the difficulties of the method proposed by deregulation needs to be accounted for in
Ian Crawford, he pointed that it might be the computation of the index. Price
possible to use an alternative ‘dirty changes related to changes in regulation
method’ like for example to conduct a are particularly drastic in the energy
survey on consumers asking if they would and house rental markets.
be willing to buy a certain good if the
price went down by a certain amounts.
34 22
Jonker, N. (2001):‘Construction Quality Adjusted Price Indexes:A Comparison of Hedonic and Discrete Choice Methods’.
Paper downloadable from http://www.ecb.int
23
Collier, I. (2001):‘The DM and the Ossi Consumer: Price Indices During Transition’. Paper downloadable from http://www.ecb.int
24
Cummins, J. G. and Violante, G. (2001):‘Equipment-Embodied Technical Change in the US (1947-1999): Measurement and
Gianluca Violante 24 presented a paper when technical change is embodied in
in which by extrapolating the quality-bias new capital goods and that the
in the official price indices measured in contribution of embodied technical
Gordon (1990)25, they constructed change to per capita GDP growth in the
quality adjusted price indices by type of US is as large as 69 per cent.
equipment from 1947-2000 in the US
and used them to measure embodied
technical change at the aggregate level.
The results presented showed a great
variation in the rates of embodied
technical progress across industries.
Embodied technical change, and not
total factor productivity, was shown to
be the main driving force of growth
outside durable goods manufacturing.
Results also showed that the aggregate
rate of equipment embodied technical
change accelerated to more than 6
percent a year during the 1990s (from a
postwar average of 4 percent), and that
the technological gap between the
frontier (or state of the art) technology
and the average machine increased
substantially in all industries since the
1970s.