Statistical Implications of Inflation Targeting
Chapter

8 Eurostat’s HICP and the European Central Bank’s Definition of Price Stability

Author(s):
Carol Carson, Claudia Dziobek, and Charles Enoch
Published Date:
September 2002
Share
  • ShareShare
Show Summary Details
Author(s)
Vítor gaspar

The European Central Bank (ECB) announced its stability-oriented 1 monetary policy strategy on October 13, 1998 (ECB, 1998). The strategy includes three main elements: first and foremost, a precise definition of price stability. Price stability has been defined as an annual increase in the harmonized index of consumer prices (HICP)1 of below 2 percent for the euro area as a whole. Price stability is to be maintained over the medium term. The definition clearly signals the ECB’s commitment to maintaining price stability. Second, the stability-oriented strategy includes analyses assigning a prominent role to money. The third element of the strategy is represented by analyses based on a wide range of other models and indicators.2

It may seem surprising to find the definition of price stability as one of the main elements of the ECB’s monetary policy strategy. Price stability is the primary goal of the ECB as prescribed in the European Union Treaty (EUT, or simply “the Treaty”) and in the Statute of the European System of Central Banks (ESCB) and of the ECB (“the Statute”).3 Therefore, the price stability mandate is given to the ECB through a legal text of a constitutional nature, ratified by all member states of the European Union. The Treaty and the Statute also provide the ECB with the means to pursue its mandate, namely, the authority to define and implement the monetary policy of the euro area.

The setting defined by the Treaty reflects the belief that macroeconomic stability—including price stability—is a key component of an economic regime fostering sustainable economic growth and other economic goals specified in the EUT. These goals include a high level of employment and a rising standard of living and quality of life.

The arguments above make it clear that the institutional commitment to price stability precedes the choice of the monetary policy strategy by the ECB’s Governing Council. The Treaty and the Statute do not, however, provide a precise, numerical definition of price stability based on a specific statistical indicator. It is the precise definition of price stability that constitutes a main element of the ECB’s monetary policy strategy.

The announcement of a precise, quantitative definition of price stability on the basis of a specific indicator served mainly three purposes. First, the definition aimed at making the primary objective easier to understand, thereby rendering monetary policy more transparent and predictable. The strategy induces a systematic pattern of policy responses compatible with the maintenance of price stability over the medium term (rule-like behavior). Second, the definition made ex post monitoring by an independent third party possible. Thus it contributed to making monetary policy accountable. Third, the definition—together with the medium-term orientation—helped to anchor long-term inflation expectations, thereby making the policy more credible.

Most central bankers agree that price stability means low and stable inflation. The ECB’s definition of price stability is fully in line with this consensus view. Moreover, when they were deciding on the definition of price stability, the members of the Governing Council of the ECB were not operating in a historical vacuum. This chapter will provide some historical background relevant to understanding some of the aspects of the ECB’s definition of price stability. Three main developments are worth noting: the long process of disinflation made necessary by the “great inflation” of the 1970s and the early 1980s; the convergence process in the European Union and the formulation of the price stability convergence criterion in the Treaty; and the process of statistical harmonization stemming from the need to evaluate convergence on the basis of comparable data. All three are important to understanding the origins of the definition of price stability. The definition of price stability and its background are also useful when approaching issues relating to the conceptual framework for price measurement for monetary policy purposes. This is particularly so when determining the scope of the relevant index. In other words, what definition of the domain of prices should be included in (and, implicitly, excluded from) the calculation?

The chapter is structured as follows. The next section presents some background on the HICP and on the definition of price stability in the European Union. It will make clear why the wording chosen by the ECB’s Governing Council can be seen as a natural one. The chapter will then discuss some conceptual issues related to the choice of price index to be used in a definition of price stability. These include questions ranging from the choice of the domain of prices to be covered to practices aimed at ensuring the representativity of the prices used to compute the index. Some possible criteria will be presented and discussed. Subsequently, the main characteristics of the HICP are described. The concluding section identifies a number of areas for further work.

A Brief History of the HICP and of the Definition of Price Stability in the European Union

The wording for the definition of price stability by the Governing Council of the ECB was chosen in a specific historical context. Once the main aspects of this context are recognized, it becomes easier to understand some of the details of the formulation that was chosen. There are three aspects to consider: first, the long process of disinflation; second, the formulation of the price stability convergence criterion in the EUT; and, third, the process of statistical harmonization required to calculate consumer price inflation in a way that would make comparisons possible.

First, the 1990s were a decade of disinflation in the European Union. After decades of inflation fighting, a satisfactory degree of price stability was finally achieved (see Figure 8.1). This trend has been shared by most of the world economy. In any case, the end of disinflation is recent. It was only during 1996 that the monthly, year-on-year inflation rate fell below 2 percent for the aggregate of the 11 countries that would constitute the euro area at the beginning of stage III of the Economic and Monetary Union (EMU), in January 1999. The annual average fell below 2 percent in 1997 only. Going back to 1990, all these 11 countries had inflation rates above 2 percent. The same was true for Greece, which joined the euro area later. In fact, inflation at the time in Greece and also in Portugal was still at double-digit levels (see Figure 8.1). It was in the middle of this process of disinflation that the Council of Ministers wrote, in its Broad Economic Policy Guidelines:

Further progress towards price stability must be made … Those Member States that are currently expected to experience rates of inflation between 2 percent and 3 percent should maintain a policy aimed at preventing any resurgence of inflation and at progressing towards or going below 2 percent. Other countries need to increase their efforts, in some cases substantially if they are to meet the guidelines.4

Figure 8.1.Euro-Area Countries: Inflation Rates in the 1990s

Source: Eurostat.

The use of an upper bound to limit the rate of inflation is a natural choice when drafting in a context of disinflation.

Upon commencement, the ECB inherited a situation of price stability from the national central banks of the euro area. Price stability was apparent in terms of both current price developments and the credibility of price stability (as reflected in, for example, long-term bond yields and indicators of inflation expectations). This was a valuable inheritance made possible by a long and steadfast struggle against inflation. It was therefore key to signal continuity with past practices by the national central bank. The value of 2 percent had traditionally been used by those European central banks with the best records in terms of price stability, including the Deutsche Bundesbank.5

Second, progress toward economic and monetary union required meeting some convergence conditions. The EUT specifies a number of requirements that member states have to meet before participating in the euro area. These requirements include the convergence criteria. There are four such criteria in the Treaty (Article 121): on price stability, on the government budgetary position, on participation in the exchange rate mechanism, and on the convergence of interest rates. A Protocol annexed to the EUT deals specifically with issues related to the convergence criteria. Article 1 of the Protocol states that, for the purpose of evaluating the price stability criterion: “Inflation shall be measured by means of the consumer price index on a comparable basis, taking into account differences in national definitions.” The formulation in the Protocol clearly shows that the drafters of the EUT intended consumer price inflation to be the yardstick to assess price stability.

Third, harmonization efforts led to the production and publication of HICPs for the member states and for the euro area. The second and third developments are very closely related. The Protocol on the convergence criteria established a clear link between the notion of price stability in the EUT and price measurement on the basis of consumer price indices (CPIs). It constrained the domain of prices to consider when assessing the success of the ECB in delivering on its price stability mandate. Beyond doubt, the motivation for the production of the HICP is closely linked to the process leading to economic and monetary union in Europe. According to Article 1 of the 1995 HICP framework regulation,6 the aim was to establish bases “for arriving at the calculation of comparable indices of consumer prices at Community level” (italics added).

Nevertheless, the need for a monetary union indicator was foreseen from the start. The need is explicitly reflected in a “whereas” clause in the 1995 framework regulation.7 In any case the need to compute a monetary union indicator, labeled monetary union index of consumer prices (MUICP) in Article 2 (c) of the regulation, was not the dominant concern. Most Council and Eurostat regulations would have been enacted even without the specific requirements deriving from the need to calculate an index for the euro area. The main exception relates to the definition of geographical coverage. While clearly the diversity of conceptual approaches across member states called for harmonization, the option in favor of the “domestic concept” was justified by the need to aggregate national HICPs to obtain the MUICP. Specifically, the option relates to the reliance on the “domestic concept.” In fact, the HICP includes prices incurred “on the economic territory of Member States … by households irrespective of nationality or residence status.”8 Reliance on the “domestic concept” ensured that the statistical framework for the MUICP was able to deal with the initial unknown concerning the set of member states participating in monetary union. It also ensures the ability to encompass an increasing number of countries.

As required by the framework regulation, HICPs have been produced and published for each member state since March 1997.9 These indices were immediately accepted by the Commission and the European Monetary Institute as satisfactory for assessing the achievement of convergence. The official publication of the HICP for the euro area as a whole started only in January 1999, the starting date of the single monetary policy.10 As important users, the European Monetary Institute and, subsequently, the ECB have been closely involved in all the conceptual work underlying the harmonization of national CPIs. This work has been led throughout by the European Commission (EC; Eurostat) in cooperation with the national statistical institutes. The process of harmonization has been based on several EC regulations agreed to, as the case may be, by the EU Council or the EU Commission.11

It is worthwhile to summarize the information that is now (at the beginning of 2002) being made available by Eurostat on a regular basis. Each month Eurostat releases the HICP for the euro area as a whole. It starts with a flash estimate, which is made public on about the last working day of the corresponding month. Then, between day 16 and day 19 of each month, the HICPs corresponding to the previous month are released.12 The information made available includes disaggregated data on developments in almost 100 different price items. At the same time HICPs for each of the 15 member states of the European Union are also made available. The same disaggregated information is available for each member state. The HICP for the euro area is a weighted average of the HICPs of the 12 member states of the euro area. The weights are given by the share of each country in the household final monetary consumption expenditure of the euro area as a whole. The country weights and the product weights are generally available (see Table 8.1).

Table 8.1.Euro Area: HICP Weights in 2001(Two-digit aggregation level)
AustriaBelgiumGermanySpainFinlandFranceGreeceIrelandItalyLuxembourgNetherlandsPortugalEuro Area
Food and nonalcoholic beverages4.355.6242.4022.452.7434.435.112.2732.200.288.174.63164.57
Alcoholic beverages, tobacco1.041.0514.353.381.158.1812.61.065.210.292.620.6840.25
Clothing and footwear2.312.3022.4210.550.8311.793.010.5819.920.183.381.4678.68
Housing, water, electricity, gas, and other fuels4.7453465.7411.912.5430.082.561.0220.070.2610.432.00156.98
Furnishings, household equipment, and routine maintenance of the house2.752.5822.956.550.8014.021.960.5420.550.254.771.6579.37
Health1.371.3312.203.040.778.521.420.267.020.032.231.2439.43
Transport4.894.6647.7516.212.5737.253.211.4327.720.477.054.42157.67
Communication0.950.796.042.630.385.630.580.175.870.030.980.4024.44
Recreation and culture3.734.5634.546.971.8219.221.041.3514.570.275.960.8294.93
Education0.000.000.000.000.000.000.000.000.000.000.000.000.00
Restaurants and hotels4.402.7915.5115.891.4917.472.782.3719.420.233.742.7688.75
Miscellaneous goods and services2.172.2825.174.860.8018.891.340.6814.460.163.190.8874.92
Country total weight32.7033.50309.08104.4415.90205.4624.2811.72187.002.4652.5220.941,000.00
Source: Eurostat. HICP, Harmonized index of consumer prices.
Source: Eurostat. HICP, Harmonized index of consumer prices.

The achievements through the process of statistical harmonization are remarkable. It has produced national HICPs that are comparable and a new statistic, the euro-area HICP, which is used in the ECB’s definition of price stability.

On October 13, 1998, the Governing Council of the ECB reached an agreement on the main features of the stability-oriented monetary policy strategy. As part of the strategy, the Governing Council stated that price stability meant an annual increase in the HICP of below 2 percent for the euro area as a whole. The definition as such is timeless. However, when announcing the definition of price stability, the ECB’s Governing Council also clarified that price stability is to be maintained over the medium term. The focus on the medium term reflects, in part, the time lags associated with the effects of monetary policy. Unanticipated price shocks can be offset only with delay. This delay means that price developments cannot be controlled on a quarterly or even annual basis, making some short-term volatility in inflation unavoidable. More important, a medium-term orientation is compatible with the role of monetary policy in the overall context of stability-oriented macroeconomic policies. The idea is that a credible ex ante commitment to a medium-term orientation allows for the maintenance of price stability while, at the same time, avoiding unnecessary volatility in output, employment, and interest rates.

The announcement of a precise, quantitative definition of price stability on the basis of a specific indicator served mainly three purposes. First, the definition is a main element of the stability-oriented monetary policy strategy, which induces a systematic pattern of policy responses compatible with the maintenance of price stability over the medium term. In other words, the strategy induces rule-like behavior. Gaspar, Masuch, and Pill (2001) argued that the announcement of the strategy aimed at reducing strategic uncertainty. The definition aimed at making the primary objective easier to understand, thereby rendering monetary policy more transparent and predictable. Second, the definition made ex post monitoring by an independent third party possible. Thus it contributed to make monetary policy accountable. Third, the definition—together with the medium-term orientation—helped to anchor longer-term inflation expectations, thereby making the policy more credible.

There are a number of features in this definition. First, the definition focuses on a specific, well-defined statistical indicator: the HICP for the euro area as a whole. No special attention will be given to sectoral, regional, or national price developments. They will be considered only insofar as they affect current and prospective price developments in the area as a whole. This focus reflects the recognition that monetary policy cannot offset idiosyncratic price movements. Second, the definition excludes both inflation and deflation. Significant inflation is excluded by imposing an upper ceiling of 2 percent. Deflation is excluded by the use of the expression “annual increase.” Third, the definition acknowledges the possibility of imperfections in the measurement of prices. One of the factors that may be used to justify an upper bound of 2 percent is the potential existence of a positive measurement bias.13

A number of elements in the ECB’s definition of price stability follow directly from the historical developments summarized in this section. First, the reference to annual price increases of below 2 percent follows the tradition established by the European central banks with the best track record in terms of price stability (including the Bundesbank). It also reflects the widespread understanding of the notion, as is clear from the wording of the 1995 Broad Economic Policy Guidelines. As made clear above, the formulation adopted is meant to be symmetric, as it excludes references to both inflation and deflation. Second, the focus on CPIs follows from the definition of the price stability convergence criterion in the Treaty. Third, the euro area HICP was the natural indicator to choose. Its development followed from the need to evaluate the price stability convergence criterion. But perhaps more important, its development through the process of harmonization reflected the vision of statisticians at Eurostat and the national statistical institutes who foresaw the need for an area-wide statistic at an early stage. Their efforts provided the ECB with a price index ready to use to define price stability for monetary policy purposes.

Conceptual Issues in Price Measurement

This section will review a number of issues relevant to the choice of a price index as the basis for defining price stability. Similar issues have been discussed in Bloem, Armknecht, and Zieschang (this volume) from the viewpoint of an inflation-targeting monetary policy strategy. Following Diewert (forthcoming), two main questions need to be considered:

  • the choice of the appropriate scope for the index—that is, the definition of the domain of transactions to include and therefore (implicitly) the domain of transactions to exclude from the index; and

  • the choice of practices that will ensure that the prices reflected in the index are representative of the universe of transactions to be covered.

Diewert (forthcoming) quotes Fisher (1911) as arguing that, ideally, a price index aimed at measuring inflation should cover all monetary transactions occurring in the economy. However, Fisher himself regarded this concept as impractical. He noted: “It is, of course, utterly impossible to secure data for all exchanges, nor would this be advisable. Only articles which are standardized, and only those the use of which remains for many years, are available and important enough to include … thus leaving practically nothing but wholesale prices of commodities to be included in the list of goods.” Given the predominance of financial transactions in present-day economies, it is clear that the focus on an index attempting to cover all monetary transactions would be misplaced.

The issue then is one of identifying the relevant subset of transactions to be covered. To do this, it is necessary to identify both the relevant set of traders and the kinds of transactions to cover. Bloem, Armknecht, and Zieschang (Chapter 10 in this volume) and Diewert (forthcoming) approach this question by focusing on alternative sets of traders following the classification of economic agents and institutional units in the system of national accounts. Both papers also consider alternative domains, including flow transactions and stock transactions.

According to the System of National Accounts 1993 (1993 SNA), the economic agents resident in an economy can be divided into five institutional sectors: nonfinancial corporations, financial corporations, general government, households, and nonprofit institutions serving households.

The full set of possible issues is too broad to be discussed here. Therefore, only a few will be covered. Concerning whether to include transactions in equities, it will be sufficient to note the broad consensus on excluding asset prices from the definition of price stability for monetary policy purposes. Asset prices should be taken into account only insofar as they provide information, in general, on the current situation and prospects for the economy and, in particular, on future price developments. Asset prices should be used, loosely speaking, as information variables and not as targeting variables.14

Bloem, Armknecht, and Zieschang remark that, if asset prices are included in a monetary policy reaction function, they can be included separately. This allows for the weight on asset prices to be set much lower than on goods and services prices.

Bloem, Armknecht, and Zieschang (this volume), Hill (1996), and Woolford (1999) argue that focusing on households’ consumption expenditure would lead to an inflation concept that is too narrow for monetary policy purposes. The preferred measure for Woolford (1999) is the domestic final purchases (DFP) price index. DFPs take net exports out of GDP and therefore include consumption, investment, and general government’s final demand.15 The measure favored by Bloem and others is what they call the adjusted domestic supply price index (ADSPI). The ADSPI would reflect the implicit output price index (IOPI)—which includes intermediate goods—the import price index, and the price index relative to sales of nonfinancial assets.16

In the context of the 1993 SNA, Diewert (forthcoming) argues that a broad concept of inflation could be based on consumption, C, consumption plus government final expenditure, C + G, consumption plus government expenditure plus investment, C + I + G, or, finally, consumption plus government final expenditure plus investment plus exports, C + I + G + X. The author argues strongly against the use of the GDP deflator (based on C + G + I + X − M). When doing so, he remarks that in the literature on Taylor rules, the GDP deflator is usually chosen as the measure of inflation (see, for example, Taylor, 1993; Clarida, Galí, and Gertler, 1998; or Judd and Rudebusch, 1998). Diewert’s view on the use of the GDP deflator as a measure of inflation is sound. The GDP deflator does not provide a solid foundation for measuring inflation, simply because it is not an index of product prices: when computing the GDP deflator as a weighted average of product prices, some weights (namely imports) are negative.

Diewert (forthcoming) concludes: “Within the components of final demand, our preferred domain of definition for the target inflation index is just consumption expenditures, C, but cases could be made for C + G, C + G + I and C + G + I + X as possible domains of definition.”

The preference for using consumption expenditure as the basis for the definition of the main price aggregate for monetary policy purposes is based on a number of arguments. First, focusing on the components of final demand, one can argue for the exclusion of exports, investment, and, finally, government final expenditures. Diewert (forthcoming) does exactly this. He argues for the exclusion of exports given that they correspond to expenditures made by the rest of the world. He argues for the exclusion of investment because the price of investment goods is not relevant for deflating current period household expenditures on goods and services. Finally, he argues for the exclusion of government expenditure on practical grounds. In practice it is very difficult to obtain reasonable deflators for government final demand. Diewert (forthcoming) concludes that the CPI is perhaps the best indicator of inflation in the economy.

Second, and more important, human beings are the ultimate justification for all activities in human societies. The evolution of consumption over time is what society cares about from a welfare point of view. Households are the owners of financial and nonfinancial corporations. Households are directly or indirectly—through firms—taxpayers. Measures of inflation based on consumer prices seem to be well understood by the public and a focal point for what the public cares about. Last, there is clear evidence that when the question of which measure of purchasing power of money to use as a standard is raised at an institutional level, the choice falls on CPIs. All inflation-targeting central banks focus on some measure of consumer price inflation. When central banks provide a precise, quantitative definition of price stability, they do so on the basis of some CPI. As seen above, in the European Union the reference to consumer price inflation as the standard for price stability was included in the EUT itself.

As stated above, a further important conceptual issue is representativity. The concern with representativity arises in the context of the fixed basket approach. In the context of the preparatory work leading to the HICP, there was never an extensive discussion of the general index formula. The time pressures involved and practices already in use in the member states were clearly factors. When harmonization efforts started in the early 1990s, almost all member states were applying a Laspeyres-type (base-weighted) index. As the harmonization efforts were based, as much as possible and appropriate, on methodologies already in use in the member states, it was agreed that the HICP would also be a Laspeyres-type index. It is worthwhile to highlight that Diewert (forthcoming) reports an amazing result. He looks at “best” index number formulas from different viewpoints: the fixed basket approach (with the aim of preserving representativity), the axiomatic approach, the stochastic approach, and the test approach. From all these approaches, three index formulas appear to be “best”: the Walsh, the Fisher, and the Tornqvist-Theil. Moreover, the three alternatives are likely to lead to very close results. In fact, Diewert (1978) has shown that the three formulas will approximate each other to the second order.

The issue of representativity is particularly important when it is recognized that price indices have to account for “new” goods and quality change. This point is clearly acknowledged by Eurostat and the national statistical institutes involved in methodological work for the HICP. Indexed “new goods” and “quality change” have been identified as areas in which the Laspeyres concept does not lead to clear guidance for practice. Work in these particular areas is currently in progress.

The Main Characteristics of the HICP

The HICP17 is conceptually linked to the concept of households’ final monetary consumption expenditure of the European system of accounts (ESA). The latest version of ESA is ESA95. The link with the system of national accounts is one of the strengths of the HICP. The HICP is a Laspeyres-type index.18 In order to ensure that the base period basket is kept representative, the weights must be updated frequently and regularly. At this point, it has been agreed that the weights must refer to a period of consumption no more than seven years before. Weight updating intervals vary between one and five years across member states. The change in the index literally measures the change in expenditure necessary to ensure the feasibility to the base period’s consumption pattern (under unchanged population composition). Thus the HICP follows a fixed basket approach. The fixed basket approach is frequently presented as an alternative to the cost of living index (COLI) approach.19 These points are summarized very clearly in Eurostat’s 2000 report to the European Union Council (Commission of the European Communities, 2000, p. 13):

The HICP may thus be described as a Laspeyres-type “consumer inflation” or “pure price index” measuring average price change on the basis of the changed expenditure of maintaining the consumption pattern of households and the composition of the consumer population in the base or reference period. “Pure” means that, strictly speaking, it is only the changes in prices that are reflected in the measure between the current and the base reference period. The HICP is not a cost of living index. That is, it is not a measure of the change in the minimum cost for achieving the same “standard of living” (i.e., constant utility) from two different consumption patterns realized in the two periods compared and where factors other than pure price changes may enter the index [underlining in the original].

The HICP is based on actual monetary market transactions. This fact explains the decision made to exclude the possibility of including owner-occupied housing on the basis of either the rental cost equivalence approach or the user cost approach (see Astin, 1999). Astin (1999) argues that as the cost of borrowed money, interest payments are neither a good nor a service. However, the argument by Woolford (1999) for excluding interest rates from a price index is more convincing. The argument emphasizes that interest rates are not contemporaneous prices. An interest rate necessarily refers to a period elapsing between two points in time. Interest rates, therefore, can be seen as intertemporal prices, which should not be considered when comparing prices between two periods (see also Diewert, forthcoming).20

Given the concept of households’ monetary expenditures, prices are measured including net taxes (that is, adding indirect taxes and subtracting subsidies). The classification of goods and services used in the HICP follows the classification adopted for the system of national accounts. The national accounts classification of individual expenditures by purpose (COICOP) was slightly changed to obtain the so-called COICOP/HICP. The main changes involve the omission of certain categories of expenditures (for example, narcotics, prostitution) and the combination of some others, in order to minimize the risk of having zero or insignificant weights for any country’s subindex.

The HICP for the euro area is obtained by aggregation of the HICPs of the member states participating in the euro area. Aggregation is achieved through the calculation of a weighted average. The weights are given by the share of each country’s households’ final monetary consumption expenditure in the corresponding total for the euro area as a whole. These weights are allowed to change every year.21 As explained above, the computation of the MUICP was not the only goal of building an HICP. Had it been, the statisticians might have considered designing a representative sample for the entire euro area.22 As it was, the result has been a larger and better sample than what would have been required for the MUICP alone. The challenge was the harmonization of methodologies and practices so that the results could be aggregated to obtain the MUICP. The need for harmonization is intimately related to the need to ensure the comparability referred to above.

Most of the regulations produced for the HICP would have been produced even without the special requirements stemming from the MUICP. In particular, given that for the foreseeable future member states will want to be able to compute their own national CPIs, national baskets were seen as desirable both for the HICPs and for the MUICP. The most relevant exception relates to the definition of geographic coverage. In this regard, the need to have an MUICP led to the adoption of a domestic concept. According to the domestic concept, the scope of expenditure to be considered is defined as expenditure incurred “on the economic territory of the member state … by households irrespective of nationality or residence status.”23 The use of the domestic concept was the only approach possible to avoid gaps or double counting given that, ex ante, the membership of the euro area was not known. Moreover, it was always clear that the membership of the euro area would expand over time.

Following the analysis presented in the previous section, a further issue to be covered is representativity. The concern with representativity arises in the context of the fixed basket approach. For the HICP, this concern led to rules concerning the updating of the weights used in the computation of the index. Quoting again from Eurostat’s 2000 report to the European Union Council (Commission of the European Communities, 2000, p. 14):

The weights used in the compilation of the HICPs may relate to a reference period up to seven years prior to the current year. However, adjustments must be made each year for especially large changes in the expenditure pattern. This minimizes any disparities arising from different update frequencies [across different member states].

The harmonization process, leading to comparability of the national HICPs, has been completed successfully. From a monetary policy perspective, the HICP both provides the domain a price index should cover and guarantees the representativity of the price index. However, comparability of national indices and compliance with those issues discussed in the previous section would not be enough to justify the use of the euro area HICP in the ECB’s definition of price stability. In their contributions to the Center for Economic Policy Research (CEPR)/ECB Workshop on Issues in the Measurement of Price Indices (held on November 16–17, 2001), Eugenio Domingo-Solans (2001) and Otmar Issing (2001) made it clear that, in 1998, the HICP was the only serious contender for the measurement of inflation in the euro area. The main reasons are the following:

  • The HICP is obtained through aggregation of national indices in a consistent manner without gaps or overlaps, ensuring a broad coverage of household final monetary consumption expenditure in the euro area.

  • The HICP is available monthly and released in a timely manner.

  • The HICP is subject to only minor revisions.

  • The HICP is based on actual monetary transactions and records prices as faced by households (including indirect taxes and subtracting subsidies).

  • The HICP does not directly reflect interest payments.

On the same occasion, Otmar Issing emphasized that, in 1998, the Governing Council of the ECB took the view that the HICP could already be used as the statistical basis for the ECB’s definition of price stability. This reflects the outstanding work carried out by European statisticians from Eurostat and the national statistical institutes.

Conclusions: Some Areas for Future Improvement

Despite the achievements described in the previous sections, which have led to the production of a high-quality statistical product, the HICP can and will benefit from further improvements.

Four areas can be briefly mentioned. First, a more systematic documentation of methodologies and practices should be provided. Second, the methods used to treat quality change and the introduction of new goods should be spelled out. Third, the treatment of owner-occupied housing is a problem area and it may be useful to present alternative approaches in supplementary tables. And, finally, it would be useful to take another look at the issue of the type of index number formula used in an effort to obtain a formula that uses weights that are representative of both periods being compared. At the same time, doing so may lead to an index that would complement the existing index but would be published on a somewhat delayed basis. The U.S. Bureau of Labor Statistics has recently started to publish such an index using a Tornqvist formula.

The achievements from the harmonization process have been remarkable. National HICPs are comparable and, since March 1997, an HICP covering the entire euro area has been produced and released. However, as stressed by Wynne and Rodriguez-Palenzuela (forthcoming), the production of the HICP for the euro area is highly decentralized. The task of producing the national HICPs is the responsibility of national statistical institutes. The national HICPs are then aggregated to obtain the HICP for the euro area. The two reports of Eurostat to the Council of Ministers of the European Union (Commission of the European Communities, 1998 and 2000), and Eurostat’s comprehensive compendium of HICP reference documents (Eurostat, 2001) are valuable help for interested professionals. Moreover, Eurostat’s plan to make available a handbook on the elaboration of the HICP would improve matters still further. However, systematic comparable information on the methodologies and practices of the various national statistics institutes does not seem to be readily available.

A second issue that deserves to be singled out relates to quality change in goods and services.24 The importance of quality change is closely related to the dynamism of modern economies. In fact, this issue is not very relevant when standard goods dominate output. This was arguably the case in most industrialized economies in the first half of the twentieth century. The situation has substantially changed in at least two ways. First, services now have the largest share of output and consumption. In services, heterogeneity and product differentiation are prevalent. Second, the pace of innovation leading to the introduction of new goods has substantially increased. This faster pace leads to shorter product cycles entailing rapid degradation of product samples, in particular for consumer durable goods.25 Quality change ranks high on Eurostat’s priority list. The approach followed is a pragmatic one, as illustrated by the definition of quality change by the Commission of the European Communities (2000):

“Quality Change” occurs whenever a change in specification has resulted in a significant difference in utility to the consumer between a new variety or a model of a good or service and a good or service previously selected for pricing in the HICP for which it is substituted.

Quality change, particularly in the form of product innovation, leads to rapid degradation of base period samples and is a good example of the importance of adopting methodologies and practices aimed at preserving the representativity of the index.

A third issue has to do with the treatment of owner-occupied homes. It is an important issue for a number of reasons. First, shelter is an important necessity for households. Housing services are an important component of consumption. Therefore, it seems appropriate to adequately cover this item when assessing consumer price inflation. This links to the representativity of the index. Second, the importance of rental homes versus owner-occupied homes varies widely across member states. For example, in Germany about 60 percent of all homes are rented, while in Spain the proportion drops to 15 percent. This difference raises the question of comparability. Both representativity and comparability help to explain why the issue of owner-occupied homes ranks high on Eurostat’s priorities. In fact, for a number of member states, a parallel index of housing costs, based on the net acquisition cost approach, is going to be computed. Eventually decisions will be made concerning the extension of this methodology to all member states and the inclusion of the results in the overall HICP.26

Finally there is the issue of the index type. Apparently, there has never been any extensive discussion of this issue in the context of the HICPs, mainly because of two factors: the time constraints on the harmonization process, and the reliance on methods and practices used in member states. In fact, almost all national statistical institutes were constructing indices that could be regarded as Laspeyres-type (base-weighted) indices.

This chapter mentioned that Diewert (forthcoming) shows that three index formulas appear to be best from different viewpoints. One of the viewpoints under study was the issue of preserving representativity in the fixed basket approach. The index formulas ranked as best were the Walsh, Fisher, and Tornqvist-Theil indices. Diewert (1978) has also shown that the three formulas will approximate each other to the second order.

would like to thank Ignazio Angeloni, John Astin, Peter Bull, Gonzalo Camba-Mendez, Erwin Diewert, Kirstin Hubrich, Geoff Kenny, Gerard Korteweg, Bettina Landau, Diego Rodriguez-Palenzuela, Wolfgang Schill, Michel Stubbe, Oreste Tristani, Juan Luis Vega, and Mark Wynne for helpful comments. It is only fair to single out John Astin, from whom I learned the history leading to the production of the HICP; Erwin Diewert, with whom I had the privilege of discussing price measurement issues; and Gonzalo Camba-Mendez, with whom 1 have been working on related issues. I would also like to thank Andres Manzanares for producing the tables and charts and Patricia Kearns-Endres for administrative assistance. The opinions expressed are the author’s own and do not necessarily reflect those of the ECB or the Eurosystem. The remaining errors are mine.

The HICP is released monthly by Eurostat, the Statistical Office of the European Union. See Astin (1999).

See ECB (1999, 2000, 2001); Caspar, Masuch, and Pill (2001); and Issing and others (2001) for a detailed account of the ECB’s monetary policy strategy.

Article 105, paragraph 1, of the EUT, reads: “The primary objective of the ESCB shall be to maintain price stability. Without prejudice to the objective of price stability, the ESCB shall support the general economic policies in the Community with a view to contributing to the achievement of the objectives of the Community as laid down in Article 2. The ESCB shall act in accordance with the principle of an open market economy with free competition, favouring an efficient allocation of resources, and in compliance with the principles set out in article 4.” Article 105 (1) is repeated as Article 2 of the Statute.

Council Recommendation of July 10, 1995, on the broad guidelines of the economic policies of the member states and of the Community.

The Bundesbank referred to a normative rate of inflation of 1.5 percent to 2 percent when deriving the monetary growth target. For example, Issing (1997, p, 70) states: “The Bundesbank’s monetary targets include a normative, medium-term rate of inflation of not more than 2 percent. Such a rate, on the one hand, allows for some inaccuracy in the measurement of inflation, and on the other, is low enough that any residual inflation should not lead to allocative distortions or impede economic activity.”

Council Regulation (EC) No. 2494/95 of October 23, 1995, concerning harmonized indices of consumer prices.

At its meeting in December 1991, the Statistical Programme Committee, composed of the presidents or directors general of the national statistical institutes, agreed to establish a new working party to discuss the harmonization of national CPIs. The working party established several task forces. The draft framework Council Regulation was finally adopted in October 1995 (see footnote 6).

Council Regulation (EC) No. 1688/98 of July 20, 1998, Article 1 (1).

The interim indices of consumer prices (IICPs) were released in February 1996 and reflected the progress in harmonization achieved up to then. The Convergence Report noted: “Though not yet fully comparable, IICPs are considered to constitute a much better basis for the assessment of inflation convergence among member states than the national CPIs.”

See footnote 22.

See Eurostat (2001) for a complete list of relevant references.

The only exception concerns the publication of the index for January, because of the need to proceed with the revision of the weights.

See Issing and others (2001) and ECB (2001) for further discussion concerning the content of the ECB’s definition of price stability.

The statement in the text is clearly a simplification. For a discussion of relevant conceptual issues, see Bernanke and Gertler (1999) and Cecchetti and others (2000). For the use of asset prices as information variables, see Shiratsuka (1999) and Good hart and Hof-mann (2000). The latter authors and Diewert (forthcoming) discuss the particularly important issue of housing and real estate.

This measure is, also favored in Peter Hill (1996).

Hill (1996) favors using either C + G + I or C + I + G + X.

See Astin (1999), Berglund (1999), Commission of the European Communities (1998 and 2000), Diewert (forthcoming), and Wynne and Rodriguez-Palenzuela (forthcoming) for more systematic and complete presentations on the characteristics of the HICP.

According to the Commission of the European Communities (2000), the HICP is” … a Laspeyres-type index that is based on the prices of goods and services available for purchase in the economic territory of the Member States for the purpose of directly satisfying consumer needs” (p. 13, underlining in original).

The COLI approach aims at measuring the change in the minimum expenditure necessary to ensure the same standard of living (utility) in the two periods being compared.

For a formulation of the household’s intertemporal budget constraint making explicit the role of interest rates, see Pollak (1989).

For example, the weights for 2000 are derived from the national accounts data for 1998, updated according to the price changes observed up to December 1999.

There is, however, a problem associated with the fact that membership in the euro area is a subset of membership in the European Union. The composition of the euro area was unknown in 1995, at the time of publication of the framework regulation. The publication of the MUICP started in January 1999, the official starting date for the third stage of economic and monetary union. As soon as the initial membership of the euro area was settled, in May 1998, Eurostat started the publication of a preliminary version of the MUICP. In any case, participation in the euro area is open—subject to the fulfillment of the convergence requirements—to all member states of the European Union. Thus it is necessary that the statistical framework for the MUICP be designed in such a way as to be able to accommodate an increasing number of countries. It may have been the case that the subtle implications of these points were not fully grasped at the stage of launching the harmonization work in the early 1990s.

Council Regulation (EC) No. 1688/98 of July 20, 1998, Article 1 (1).

Issues relevant for quality adjustment are covered in Commission Regulation (EC) No. 1749/96. These are further commented upon in the two reports from the Commission to the Council (Commission to the European Communities, 1998 and 2000).

Silver and Heravi (2001) found, using scanner data, that 50 percent of the washing machines sold at the end of 1998 were not available in the beginning of the year. Diewert, quoting the Bureau of Labor Statistics (1984), argues that this rate of sample degradation is not atypical.

Diewert (forthcoming) systematically compares the net acquisition cost approach with the user cost approach and the rental equivalence approach.

    Other Resources Citing This Publication