69/2003 - 18 June 2003

May 2003

EU15 down to 1.8%

Euro-zone annual inflation fell from 2.1% in April to 1.9% in May¹ 2003, Eurostat – the Statistical Office of the European Communities in Luxembourg – reports today. A year earlier the rate was 2.0%.

EU15 annual inflation fell from 2.0% in April to 1.8% in May 2003. A year earlier the rate was 1.8%.

EEA annual inflation was 1.8% in May 2003.

In May 2003, the highest annual rates were recorded in Ireland (3.9%), Portugal (3.7%) and Greece (3.5%); the lowest rates were observed in Germany (0.6%), Belgium and Austria (both 0.9%).

Compared with April 2003, annual inflation fell in twelve Member States, rose in two and remained stable in one Member State. Compared with May 2002, the biggest relative falls were in Austria (1.7% to 0.9%), Germany (1.1% to 0.6%) and Finland (1.8% to 1.1%), and the biggest relative rises were in Luxembourg (1.3% to 2.3%), the United Kingdom (0.8% to 1.2%) and France (1.5% to 1.8%).

Lowest 12-month averages² up to May 2003 were in Germany (1.1%), Belgium (1.2%) and the United Kingdom (1.3%); highest were in Ireland (4.6%), Portugal (3.9%) and Greece (3.7%).

In the USA annual inflation fell from 2.2% in April 2003 to 2.1% in May, and in Switzerland it fell from 0.7% to 0.4% for the same period. These indices are not strictly comparable with EU harmonized indices.

Source: Eurostat r: revised p :provisional

Source: Eurostat * FoodAlcTob: Food, alcohol and tobacco r: revised p: provisional

The annual rate measures the price change between the current month and the same month of the previous year. This measure is responsive to recent changes in price levels but can be influenced by one-off effects in either month.

The 12-month average rate² overcomes this volatility by comparing average Harmonized Indices of Consumer Prices (HICPs) in the latest 12 months to the average of the previous 12 months. This measure is less sensitive to transient changes in prices.

The monthly rate compares price levels between the two latest months. Although up-to-date, it can be affected by seasonal and other effects.

HICPs are harmonized inflation figures required under Article 121 of the Treaty of Amsterdam (109j of the Treaty on European Union). They are designed for international comparison of consumer price inflation. The focus is on quality and comparability among the indices of different countries as well as on their relative movements.

Price changes as measured by the HICPs, the Monetary Union Index of Consumer Prices (MUICP), the European Index of Consumer Prices (EICP) and the European Economic Area Index of Consumer Prices (EEAICP) are used as measures of inflation in the Member States, in the euro-zone, in the European Union, and in the European Economic Area.

The MUICP is used by the European Central Bank (ECB) as the indicator for price stability in the euro-zone³.

The Member States’ HICPs are supplied by the National Statistical Institutes; the MUICP, EICP and EEAICP are compiled by Eurostat. The HICP is computed as an annual chain index allowing weights to be changed each year. HICP aggregates are calculated as weighted averages of the HICPs using the weights of the countries and sub-indices concerned. The weight of a country is its share of the household final monetary consumption expenditure in the euro-zone total. For the MUICP this is expressed in euro, for the EICP and the EEAICP it is expressed in purchasing power standards. The MUICP is treated as a single entity within the EICP and EEAICP starting from 1999.

More information⁴ on HICPs can be obtained from the monthly publication Statistics in focus, Economy and Finance, Harmonized indices of consumer prices. About 100 MUICP, EICP, EEAICP and HICP sub-indices with corresponding weights are available on Eurostat’s database NewCronos through Eurostat’s Data Shop network.

Provisional publication dates for the coming months are:

1. The MUICP flash estimate for May 2003, published on 2 June 2003, was 1.9%.
2. Measure used to determine price stability in Convergence reports of 1998 and 2000 by the Commission to the Council.
3. See ECB press release, 8 May 2003.
4. For technical notes on HICPs see: Eurostat News Release 21/97, 5 March 1997, Harmonizing the way EU measures inflation and Eurostat Memo 8/98, 4 May 1998, New monetary union index of consumer prices (MUICP), Eurostat Memo 02/00, 18 February 2000, Improved EU Harmonized Index of Consumer Prices: Extended coverage and earlier release dates for the HICP. Further details can be found in the Compendium of HICP reference documents – (2/2001/B/5). All are available on Eurostat's Internet server.

The most significant part of the total effect took place between December 2001 and January 2002, where the impact is estimated to fall within the range of 0.09% to 0.28% ().

The analysis presented here updates and extends Eurostat’s previous analyses on the euro changeover effects in the euro-zone HICP. It covers the period since the changeover using the most up-to-date HICP data available ().

There is no way to be sure what inflation would have been if the euro changeover had not taken place. Even at the lowest level of detail, there are many factors - regular, irregular, random and systematic ones - which overlap with potential changeover effects, and there are a multitude of potential changeover effects which may amplify each other or cancel out. Furthermore, as the time period is extended beyond the changeover, more price data becomes available for analysis but also there is more opportunity for other effects - not related to changeover - to influence the analysis.

The approach used for this analysis has again been to identify regular and special non-euro factors affecting the HICP in the period covered by the study. In more detail, the approach has been as follows ():

Eurostat has used statistical techniques to identify those HICP sub-indices showing a price change which is outside what the past data would predict, and then compared them with the price development in non-euro-zone countries. The predicted price changes were adjusted where the price development in non-euro-zone countries historically have been similar to the euro-zone price development.

Possible euro changeover effects have been estimated by forecasting, other things being equal, 2002 price developments on the basis of past price developments. This gives an "expected price development". The difference between the expected and the actual price development, measured by the euro-zone HICP, is regarded as an unusual price change. Unusual price changes which could not be plausibly explained by other regular or exceptional factors were attributed to the euro, expressed in terms of inflation rate points.

The estimated euro effect on the monthly rate of change for a particular sub-index is thus the difference between the actual and the expected rate of change multiplied by the corresponding expenditure weight.

The individual contributions of particular expenditure groups (sub-indices) are roughly additive so that they sum up to the overall euro changeover effect.

Following the approach outlined above, a large majority of expenditure groups, corresponding to around 54% of the total weight of the HICP, do not show any unusual price developments with a significant effect on the all-items index during 2002. That is, the price movements were either in line with the expected range or the deviations were so small that they have no effect on the all-items index.

Expenditure groups accounting for further 20% of the total weight show irregular price movements – but ones that can be explained by special non-euro factors. For example, no direct link to the euro changeover can be found for the energy sector where volatile prices result from external factors. The price movements for package holidays and travel by air seem to be the result of rising energy prices and additional security measures following September 11 rather than the result of the changeover to the euro. Tobacco prices were affected by significant tax changes in early 2002 (and again in early 2003).

The remaining expenditure groups, the "grey zone", account for about 26% of the total weight.

The expenditure groups within the "grey zone" show significant price movements beyond what past performance would indicate. However, some of these expenditure groups have an irregular pattern of past changes, show only smooth changes during 2001 and 2002, or show no unusual price change between December 2001 and January 2002. Although the price development for these expenditure groups do not necessarily match an expected euro changeover pattern, there are no other obvious non-euro factors that could explain the observed price behaviour. Therefore, they are considered as likely to have been affected by the changeover to the euro.

Many food items including bread, fruit, vegetables, and beer, as well as some services including maintenance and repair of cars and accommodation services dominate this group.

The expenditure groups with a likely effect include both items where an effect in early 2002 quickly disappeared and items where the initial effect was smaller than the estimated effect at the end of 2002. The impact in total was up to 0.19 rate points between December 2001 and January 2002. The persistent effect was slightly lower: the impact on the annual average rate of change between 2001 and 2002 was up to 0.17 rate points.

What remains is a number of expenditure groups where the changeover could most likely explain the price changes between December 2001 and January 2002 in most euro-zone countries.

These product groups can be primarily found within the services sector, in particular restaurants and cafes, and ‘small shop services’ such as hairdressers, various repair and cleaning services, and recreational, cultural and sporting services.

The impact from these expenditure groups was up to 0.09 rate points between December 2001 and January 2002. The persistent effect was slightly higher: the impact from these product groups is up to 0.12 rate points on the annual average rate of change between 2001 and 2002.

Adding together the "likely" and "most likely" effects gives an upper bound for the euro changeover effect of 0.29 rate points on the annual average rate of change between 2001 and 2002. The impact from the expenditure groups considered "most likely" to have been affected by the changeover gives a lower bound of 0.12 rate points. The impact of the euro changeover is thus likely to fall between the range of 0.12 to 0.29 rate points when comparing the average of 2002 with 2001.

Most of the impact seems to have taken place between December 2001 and January 2002 where the impact is likely to have fallen between 0.09 and 0.28 rate points.

Several services may be said to be the main source of euro induced price increases. They make up all of the expenditure groups considered "most likely" where their contribution is 0.12 rate points to the annual average rate of change. Services also contribute to the category "likely" adding 0.03 rate points to the upper bound.

The food sector is also considered a "likely" source of euro induced price increases. However, the picture is less clear because most of the unusual price changes, might also be explained by the outbreaks of BSE and foot and mouth disease, and by higher transport costs following from higher energy prices. Exceptionally bad weather played certainly a significant role for fruit and vegetables in early 2002. Nevertheless, as a result of a systematic comparison with non-euro-zone countries, some of these unusual price changes are now attributed to the "likely" category, particularly in the immediate period after the changeover. Their contribution to the upper bound was estimated as 0.13 rate points to the annual average rate.

Only few expenditure groups within the non-energy industrial goods show unusual price changes that may be linked to the euro changeover. Generally, these effects tend to be rather small. Particular attention was paid to clothing. Clothing prices seem to have risen unexpectedly outside the regular sales periods, but price reductions within sales periods seem to have been stronger than before. The total effect seems thus close to neutral. Non-energy industrial goods altogether contributed up to 0.01 rate points to the annual average rate.

In summary the effects are shown in table 1 by expenditure group:

Table 1: Estimates of the impact of the changeover by expenditure group ()
(Inflation rate points)

The following conclusions may be drawn from these estimates:

Table 2 below shows the annual average rates of change in 2002 for the euro-zone and some non euro-zone countries (DK, S, UK and CH), for some sub-indices selected among those which are likely or most likely to have been affected by the euro changeover.

While Table 2 shows that even outside the euro-zone some significant price changes also occurred for the sub-indices concerned, it does not show whether these rates of change were exceptional or not. It also demonstrates that it is only by looking at comprehensive price statistics that the full picture may be seen.

As has been shown above, the euro changeover had a marked effect on prices for some expenditure groups, mainly in early 2002. It seems that the price development of late 2002 has continued at the expected rate in the first part of 2003 for most expenditure groups, and it is not possible to identify any unusual price changes that can be directly related to the euro-change over.

Furthermore, given the passage of time and the length of the period analysed it becomes even more difficult to evaluate whether a change in the pattern of price movements should be attributed to the euro changeover or to other factors.

The euro changeover may also have influenced prices and inflation before notes and coins came into circulation on 1 January 2002. For example, large producers could, in anticipation of the euro, have adjusted prices upward in some countries and downward in other countries in order to remove traditional price differentials. Also, when retailers changed their prices during 2001 some will have taken the opportunity to move at that time to more "psychologically attractive" euro prices.

Eurostat has analysed the HICP data for 2001 using the same approach it applied to the data for 2002. Although there is clear evidence of a large number of unusual prices changes beyond those which past development would have predicted, it is difficult to link them directly to the expected pre-euro effects in the relatively aggregate data available to Eurostat. A large part of the inflationary developments seen in the HICP data during 2001 could be explained by normal inflation patterns combined with some important effects which seem most likely to be unrelated to the euro changeover.

Prices, in particular in the food sector, rose significantly in the first half of 2001. Most of these price changes seem to be related to non-euro factors and similar price movements can be seen in some non-euro countries. BSE, foot and mouth disease, a difficult wheat harvest in northern Europe, and some other factors all affected prices. The unusual price changes in the food sector added as much as 0.5 rate points to inflation in 2001. Eurostat’s analysis of possible pre-euro effects in 2001 gives no conclusive answers. It is only on the basis of detailed micro-data, which is not available to Eurostat, that parts of this question could be answered

Several euro-zone national statistical institutes have conducted their own special studies of the possible impact of the euro changeover on inflation rates in 2001, some based on analyses of detailed micro-data. Studies for some countries indicate that there probably was a changeover impact in the latter part of 2001, though generally it seemed to have been smaller than in 2002. Some references for these studies are given at the end of this note.

For many years there has been a close and straightforward statistical relationship between the balance of consumer perceptions of inflation in the euro-zone () and the all-items HICP. The close relationship between the two clearly broke down in 2002, with the introduction of euro notes and coins.

Some, among many possible factors () which might explain this divergence, are analysed below.

Consumers are well informed about actual price changes from their own experience. As well as taking their own personal experience into account, an event as important as the changeover will have meant that consumers were also very well aware of price increases reported in the official statistics and in the media. Every reported increase in the official inflation rate can thus have a cumulative or snowball effect on consumer perceptions. In practice, Eurostat’s analysis shows that it is possible to establish a very close relationship between the all-items HICP and consumer perceptions during 2002, and for the earlier years, by estimating the change in perceptions each month as a function of the previous level of the index plus the change in the all items HICP in the current month. This can be seen as giving some support to the hypothesis of a snowball effect.

Complementary to the hypothesis of a snowball effect, it could be expected that consumers form their own views of inflation by paying the most attention to their most frequent purchases (e.g. daily or weekly food purchases).

Eurostat has constructed an index including a number of frequently purchased items such as food products, petrol, transport by train and bus, newspapers and magazines, restaurants and cafes, and hairdressers. Prices for frequently purchase items have risen faster than the all-items index since mid 1999.

Higher than average price increases for frequent purchases may explain up to a half of the drift between the all-items HICP and the index of consumer perceptions. They do not therefore explain in full the discrepancy. If there are price increases for regular low cost purchases, but similar price increases are not seen for some more expensive goods, then the index of consumer perceptions could be expected to rise somewhat faster than the all-items HICP.

Another way to look at the role of regular purchases is to ask which types of expenditure do consumers seem to be have paid most attention to in different time periods. Eurostat has analysed this by estimating from HICP data which types of purchases, with which weights, would best fit the index of consumer perceptions.

Eurostat’s analyses on this question show that the data give some support to the hypothesis that consumers may have paid more attention to the prices of regular purchases, and paid less attention to the prices for less frequent, more expensive, purchases. The divergence between the index of consumer perceptions and the all-items HICP since early 2002 can be viewed as a consequence of euro changeover induced price increases and the significant prices increases for alcohol and tobacco.