Statistics Belgium

(A division of the Federal Public Service Economy, SMEs, Self-employed and Energy)

 Economic Statistics - Price Statistics Unit

Statistics Belgium

North Gate - Boulevard du Roi Albert II, 16 - 1000 Brussels

The harmonised index of consumer prices (HICP) is a consumer price index (CPI) that is calculated according to a harmonised approach. It measures the change over time of the prices of consumer goods and services acquired by households (inflation).

Due to the common methodology, the HICPs of the countries and European aggregates can be directly compared.

European classification of individual consumption according to purpose (ECOICOP)

The HICP covers the final monetary consumption expenditure of the household sector.

The main statistical variables are price indices.

The basic unit of statistical observation are prices for consumer products.

3.6.1. Statistical target population

The target statistical universe is the 'household final monetary consumption expenditure' (HFMCE) on the economic territory of the country by both resident and non-resident households. The household sector to which the definition refers, includes all individuals or groups of individuals irrespective of, in particular, the type of area in which they live, their position in the income distribution and their nationality or residence status. These definitions follow the national accounts concepts in the European System of Accounts.

3.6.2. Coverage error population

Not available

3.7.1. Geographical coverage

The HICP refers to the economic territory of a country as referred to in paragraph 2.05 of Annex A to ESA 2010, with the exception that the extraterritorial enclaves situated within the boundaries of a Member State or a country are included and the territorial enclaves situated in the rest of the world are excluded.

3.7.2. Coverage error regions

Not applicable

3.8.1. Start of time series

The HICP series started in January 1997.

3.8.2. Start of time series - national specifics

The Belgian HICP has been produced since the start of the HICP in January 1997 and HICP-CT since December 2002.

See the HICP database.

2015=100

The following units are used:

• Index point
• Percentage change on the same period of the previous year (rates);
• Percentage change on the previous period (rates);
• Percentage share of the total (weights).

HICP is a monthly statistics.

Harmonised Indices of Consumer Prices (HICPs) are harmonised inflation figures required under the Treaty on the Functioning of the European Union. Regulation (EU) 2016/792 of the European Parliament and the Council of 11 May 2016 (OJ L 135) sets the legal basis for establishing a harmonised methodology for the compilation of the HICP and the HICP-CT.

This regulation is implemented by Commission Regulation (EU) 2020/1148 of 31 July 2020.

Further documentation, can be found in Eurostat’s website - HICP dedicated section, namely recommendations on specific topics, under the methodology page, and guidelines, under the quality page.

Not applicable.

All Belgian HICP indices at an ECOICOP level are non-confidential. Everything below this level is confidential and not available to the general public. This is in line with the Belgian CPI which is also only published on this level. The aim is to calculate reliable and provide reliable indices from the published level. 

Regulation (EC) No 223/2009 of the European Parliament and of the Council, of 11 March 2009, on the transmission of data subject to statistical confidentiality to the Statistical Office of the European Communities.

Belgian Statistical Law of 22 March 2006.

All Belgian HICP indices at an ECOICOP level are non-confidential, indices at lower level are confidential and therefore not published.

In line with the Community legal framework and the European Statistics Code of Practice Eurostat disseminates European statistics on Eurostat's website (see point 10 - 'Accessibility and clarity') respecting professional independence and in an objective, professional and transparent manner in which all users are treated equitably. The detailed arrangements are governed by the Protocol on impartial access to Eurostat data for users.

The HICP is released according to Eurostat’s Release calendar.

The calendar is publically available and published at the end of the year for the full following year.

The release calendar can be found in the Statistics Belgium website.

On the day of the release of the Belgian HICP this is made clear on the home page of the website of Statistics Belgium and a press release is published and send out to a list of subscribers. The Belgian Observatory of Prices, the Federal Planning Bureau and the National Bank of Belgium have pre-access.

Monthly

The HICP is disseminated on the website of Statistics Belgium. Data can be downloaded in various formats.

The press release of the HICP contains an analysis of the inflation rate. It contains a link to where the data can be obtained

Monthly publication of the HICP and of the sub-indices on the website of Statistics Belgium. The HICP is used in the quarterly report of the Belgian Observatory of Prices. The HICP flash estimate is published in the press release of the CPI.

HICP database

Only for the Observatory of Prices, the National Bank of Belgium and the Federal Planning Bureau.

Not available.

The HICP Methodological Manual provides the reference methodology for the production of HICP. (https://ec.europa.eu/eurostat/web/products-manuals-and-guidelines/-/KS-GQ-17-015)

10.6.1. Documentation on methodology - national specifics

Scanner data

Web scraping

Multilateral methods

Inflation measurement during COVID-19

See Eurostat's Compliance Monitoring Report of 2014 and follow-up report of 2018.

Statistics Belgium tries to apply the European Statistics Code of Practice in all its processes. 

11.1.1. Quality management - Compliance Monitoring

Compliance Monitoring

11.1.2. Quality assurance - national specifics

Statistics Belgium has an own developed quality management system using an independent validation service which validates data before publication or transfer to Eurostat. Another aspect is the use of Quality Indicators for all statistics. Every year the quality of statistics produced by Statistics Belgium are assessed using quality indicators. The criteria used are relevance, accuracy, timeliness, coherence and comparability, cost and burden, process quality and accessibility.

Whenever developing a methodology for sub-indices HICP guidelines are always taken into account, if these deviate from practices for the CPI then indices are compiled using two methods (see section 15.1 for the differences between the HICP and national CPI). Indices which are used in the Laspeyeres aggregation are flagged in the IT system if strong deviations arise compared to the expected index. The production system also provides other warnings such as when not all prices from the price collectors have been received, when there are problems in the aggregation (weights don't correctly sum up), etc.

11.2.1. Compliance monitoring - last report and main results

The last available compliance or follow-up report can be found in the dedicated HICP section ‘Compliance Monitoring’ of Eurostat’s website, under ‘Methodology’.

11.2.2. Quality assessment - national specifics

Every year the quality of statistics produced by Statistics Belgium are assessed using quality indicators. The criteria used are relevance, accuracy, timeliness, coherence and comparability, cost and burden, process quality and accessibility. 

In addition to being a general measure of inflation, the HICP is also used in the areas of:

• wages, social benefit and contract indexation;
• economic forecasting and analysis;
• measuring specific price trends;
• accounting purposes and deflating other series;
• inflation targeting by central banks;
• cross-country economic comparisons.

The euro area (evolving composition) index is used by the European Central Bank (ECB) as the main indicator for monetary policy management. The ECB and the European Commission's Directorate-General for Economic and Financial Affairs (DG ECFIN) use the HICP for assessing price stability and price convergence required for entry into European Monetary Union.

Other users include: National Central Banks, financial institutions, economic analysts, the media and the public at large.

12.1.1. User Needs - national specifics

The three most important users of the HICP are the Observatory of Prices, National Bank of Belgium and the Federal Planning Bureau.

The Observatory of Prices uses the HICP as the basis for their quarterly reports in which they analyse the price evolution in Belgium compared to its neighbouring countries. The Belgian Central Bank uses the HICP also for price and economic analysis on a monthly basis.

Both of the most important users receive specific statistical output as well as customised analysis.

The most important user for the national CPI is the Index Commission which consists of trade unions, employers' associations and academics. The CPI is used for indexation purposes such as wages, social benefits and rents.

A user satisfaction survey is carried out by Statistics Belgium. The survey and the results can be found in Statistics Belgium website, in French and Dutch.

For the CPI there is an Index Commission (=CPI board) which consists of the main stakeholders (trade unions, employers associations, academics, etc). This commission validates the methodology of the CPI carried out by Statistics Belgium. The standard methodology for the CPI is the same as the one used in the HICP except for cases where the CPI board determines otherwise.

Indices are compiled for all HICP product groups that need to be covered (=above the expenditure threshold). For those indices all taxes that need to be taken into account are covered in the HICP-CT.

The accuracy is guaranteed by observing prices of a great number of diverse products. The accuracy cannot be quantified however due to the use of non-probability sampling and complexity of price index structures. Accuracy on the overall HICP level is higher than on a disaggregated level. The main sources of error are: 

• Errors in the sample of data suppliers,
• Errors in the choice of products in the sample and the estimate of quality adjustment in case of replacements,
• Errors during price collection,
• Errors in determining the HICP weights,
• Non response.

Non-probability sampling is used. Probability sampling is not possible due to a lack of data and due to the use product specifications to allow for quality control.

Not available

The full set of HICPs is published each month according to Eurostat’s Release calendar, usually between 15 and 18 days after the end of the reference month.

Each year, the January release is published at the end of February to allow for the annual update of the weights, both of individual product groups and the relative country weights of Members States in the country group aggregates.

The euro area flash estimate is published on the last working day of the reference month or shortly after that.

Since the March 1997, launch of the HICP release, the HICP for the country groups aggregates has always been published on the dates announced in Eurostat’s Release calendar.

HICPs across Member States aim to be comparable. Any differences at all levels of detail should only reflect differences in price changes or expenditure patterns.

To this end, concepts and methods have been harmonised by means of legislation. HICPs that deviate from these concepts and methods are deemed comparable if they result in an index that is estimated to differ systematically by less than or equal to 0.1 percentage points on average over one year against the previous year (Article 4 of Council and Parliament Regulation (EU) 2016/792).

Belgian HICP indices are almost fully comparable over time. There have been several improvements in methodology since the HICP was introduced with the aim of improving the reliability and comparability of the HICP. Breaks in the series exist due to the inclusion of sales periods in 2001 and ending seasonal adjustment for package holidays in 2011.

Differences between the HICP and national CPI

•  The CPI uses the household budget survey for weights while the HICP uses the national accounts.
•  For scanner data: 3 weeks are used in the HICP and 1 or 2 weeks in the CPI (due to the early publication date of the CPI).
•  Some segments are included the CPI (such as road taxes) and not in the HICP. While others are only in the HICP (concerts), but not in the CPI.
•  Consumption expenditure by institutional households and non-residents in the economic territory is included in the HICP but excluded from national CPI.
•  Following consumption expenditure is included in national CPI but excluded from HICP: expenditure of national residents abroad.
•  Seasonal adjustment is not applied in the HICP, but is applied in the CPI for package holidays and stays in holiday villages.
•  Sales periods are not included in the CPI, but are included in the HICP.
•  A weighted 12-month average is applied in the CPI for domestic heating oil.

The HICPs are internally coherent. Higher level aggregations are derived from detailed indices according to well-defined procedures.

Not available

The HICP series, including back data, is revisable under the terms set in Articles 17-20 of Commission Implementing Regulation (EU) 2020/1148.

17.1.1. Data revision - policy - national specifics

The national CPI is not revisable due to the use of that index in wage and contract indexations. The HICP may be revised for corrections as specified in  Articles 17-20 of Commission Implementing Regulation (EU) 2020/1148. The general revision policy of Statistics Belgium website.

The HICP was revised due to the introduction of the seasonal adjustment regulation. Due to sharing a production system with the CPI up until 2014, package holidays were seasonally adjusted. They were revised to non-seasonal adjusted indices from 2011. The overall effect on average yearly inflation rate was below 0.1%. A revision also occurred when providing historical data with introduction of the ECOICOP 5, but this had no effect on the overall index

18.1.1. Weights

As of 2010, the major source for the HICP weights are the national accounts. National accounts are used up to ECOICOP level 5 since the introduction of ESA 2010. Before the introduction of ESA 2010, national accounts were only available at COICOP level 4. Therefore, the most recent household budget survey was used for the ECOICOP. Below the ECOICOP level the weights are based on various other sources (turnover data, retail trade statistics, market research information, scanner data, etc). Depending on the source they are updated yearly or less regularly.

18.1.1.1. Compilation at elementary aggregate level

A variety of sources is used to determine the weights below ECOICOP level: 

• For the groups where scanner data are used, weights are based on scanner data,
• For stratification expenditure shares from household budget survey, turnover data and market research information,
• Data from household budget survey is used,
• Data from regulators is used: i.e. natural  gas  and electricity,
• Data from service providers is used: i.e. telecommunication services,
• Administrative data is used: i.e. health care and cars,
• Market research information is used: i.e. newspapers and consumer electronics,
• Regional weighting information is obtained from the household budget survey.

The level of which weights are used depends on the product group. For some product groups there might be 3 or 4 more levels below ECOICOP, while for others this might just be 2.

A top-down structure is used to compile the weights. The main source is national account, which is used to ECOICOP 5. Below the ECOICOP level various sources are used, this depends on the product group (see above). Depending on the source they are updated yearly (i.e. scanner data groups or telecommunication) or less regularly (i.e. the HBS is every two years).

18.1.1.2. Compilation of sub-index weights

The expenditure shares derived from the national accounts t-2 at a COICOP 5-digit level (ECOICOP) were made representative for t-1 using expenditure information obtained from the quarterly accounts 2022 (thus t-1). Meaning the sub-index weights from t-2 are multiplied with the rate of change derived from the quarterly accounts t-1. Information from Q1 to Q4 2022 was used.

National (annual) accounts t-2 and quarterly accounts t-1 are the main sources of weights for all of the sub-index weights, since these are available at a ECOICOP level. The only adjustments made were to bring them inline with the HFMCE concept used in the HICP.

Preliminary information from the quarterly accounts was used for the 4th quarter. This quarter is therefore treated in a similar way compared to the other quarters.

18.1.1.3. Compilation of sub-index weights

All weights are based on the national accounts of t-1.

18.1.1.4. Weights – plausibility checking

The plausibility check of the annual weights update are done by comparing the relative weight from the current year with the previous year. Large shifts are verified with the people responsible for estimating the national accounts if the results are not self-explanatory. An extra plausibility check is also carried out by comparing the weights with the national CPI (which mostly uses the HBS instead of national accounts).

The lower level periodic review is verified in a similar manner. For segments that are available since the last review a comparison can be made between the current and the previous weight. In case of no self-explanatory reasons an extra verification is carried out with the data sources used for the estimation of these weights. The lower level weights are also always verified by multiple price statisticians.

18.1.1.5. Price updating

Since the expenditure shares were already representative for t-1, no price update between t-2 and t-1 was carried out.

The expenditure shares for t-1 were price updated using the standard procedure, thus multiplied by the difference between average of 2022 and December 2022 for each corresponding index. The price update is carried out starting from the elementary aggregate level.

18.1.1.6. Compilation of total household final monetary consumption expenditure

Quarterly accounts information 2022 was used to determine the HFMCE. The adjustments were carried out by the national accounts unit using preliminary information that was available to them. The HICP weights and the HFMCE are consistent, since they were derived using the same data sources.

The same preliminary Q4 information used to compile the sub-index weights was used to compile the total HFMCE

18.1.2. Prices

Price data is based on a mix of multiple data sources or on data from other statistical activities.

18.1.2.1. Data Source - overview  

18.1.2.2. Scanner data - general information

As of 2016 scanner data of the 4 largest supermarket chains are used for the following ECOICOP groups: 01, 02, 05.6.1, 09.3.4.2, 09.5.4 and 12.3.1. The total market shares of these 4 retailers amounts to around 85%.

As of 2022 scanner data is also used for consumer electronics (laptops, tablets, smartphones and televisions) and household appliances (washing machines, dryers, refrigerators, freezers, cooking hobs and dishwashers)

18.1.2.3. Web scraping - general information

Bulk web scraping is used for the following segments:

• Clothing and footwear (ECOICOP 03)
• One discounter chain. This concerns the following ECOICOPs: 01 (food an non alcoholic beverages), 02 (alcoholic beverages and tobacco), 05.6.1 (non-durable household goods), 09.3.4.2 (products for pets) and 12.1.3 (other appliances, articles and products for personal care)
• DVDs, Blu-rays & video games (ECOICOP 09.1.4.1 and 09.3.1.1)
• Hire of second residence (student homes, ECOICOP 04.1.2.1)
• Second hand cars (ECOICOP 07.1.1.2)
• International train travel (part of ECOICOP 07.3.1.1)
• Package domestic holidays (ECOICOP 09.6.0.1)  and hotels (ECOICOP 11.2.0.1) 
• Vitamins (part of ECOICOP 06.1.1)
• Sun glasses (part of ECOICOP 12.3.2.9)
• Car accessories (part of ECOICOP 07.2.1.3)

Web scraped price evolution is used as a proxy for price evolution of physical shops therefor they assume 100% of the weight of that outlet. Foreign websites are included if the (parent) company is in some way listed in the business register and declares VAT. This is a necessary condition to apply retailer level weights, these are usually derived from VAT/Turnover statistics.

Also a robot tool is used to monitor prices for a range of products (e.g. printing of digital photos). However, here a manual price collector has to go to the website first and mark which price needs to be tracked automatically (we don't consider this to be bulk web scraping and is thus ignored in this answer).

18.1.3. Sampling

18.1.3.1. Sampling design: locations for survey

Local price collection is carried out in a sample of around 50 representative economic areas covering the whole territory of Belgium.

18.1.3.2. Sampling design: outlets

Outlets are for the most part chosen by the price collectors under guidance of the central office. For certain product groups, outlets are also determined by the central office. They must represent the existing trade and services network in the area and they are chosen usually on three main criteria: 

• Popularity with consumers,
• Significant turnover from consumer sales and
• Availability of goods and services included in the basket.

Wholesalers, thrift shops and factory outlets are excluded. For central price collection usually turnover information is used to sample outlets.

18.1.3.3. Sampling design: newly significant goods and services

Every two years, the most recent Household Budget Survey is used to determine which product groups have an increase in expenditure below the ECOICOP level.  Market analysis and feedback of price collectors are also used.

If a suitable method is immediately available which allows to include these new goods and services then they are immediately added to the basket. New goods that have entered the index the last years are smoke detectors, orthopaedic soles, aerosol devices, sacharometer, hire of apnoea machine, PC maintenance, orchid, streaming services (audio & video), tutoring and pedicure.

Price data is collected every month.

18.3.1. Price collection surveys

Price are collected in the field during the first three weeks of the month using tablets. Centrally collected prices are collected around the 15th of the month. For prices known to vary during the month (like airfares, package holidays, hotels, international train tickets) prices are collected for multiple departure dates.

18.3.2. Timing of price collection

Price collection in the field is carried out during the first three weeks of the month (starting from the 1^st of the month).

The data are checked for internal consistency and completeness. Output files show the annual rates of change and monthly rates of change for the elementary aggregates, the contribution to the index (in points).

18.4.1. Data validation - price data

Test and verification checks are in place in the tablet application which is used to collect prices and also in the application used by the central office. Automatic listings with deviations of +/-20% are systematically verified. Listings can also be obtained for smaller variations, namely +/-5% or +/-10%. Obviously, data entry errors can be corrected by the central office.  Other price observations of similar products are used to reject or adjust the reported prices (as well as by verifying on the website of the retailer when possible). The ability to contact the price collector is also foreseen in the application (feedback messages can be directly send via the application).

18.5.1. Elementary price index formulae

The Belgian HICP is a Laspeyres-type index that is annually chained in December. At the elementary level, most products and services are calculated with the Jevons formula. This index was chosen because it takes into account substitution effects and it is indifferent to the price level of the observations.

• Price observations: for most goods and services 2 decimals, when prices are expressed with more decimals (such as fuels) then up to 4 decimals are used.
• Weights: 8 decimals are used.
• Calculation of index figures and rates of change: 12 decimals for the index and 16 for the rates of change.
• Publication of index figures and rates of change: 2 decimals are used for the index level and 1 decimal is used for the rate of change.

In all instances 'round half away from zero' is used when rounding numbers.

18.5.2. Aggregation of different data sources

Standard Laspeyres aggregation is used to obtain higher and top level indices.

18.5.3. Chaining, linking and splicing methods

The indices for the different ECOICOP groups are chained in December. All elementary indexes are therefore calculated with December of the previous year as the reference period for the prices.

Splicing is used for all groups for which supermarket scanner data is used, as well as for clothing and footwear. A 25 month window with HASP is applied.

Splicing is also applied for second hand cars and sunglasses, using the fixed base moving window method. The window length is 13 months and December acts as the fixed base month. The index itself is estimated using a time dummy hedonic method.

18.5.4. Quality adjustment – Detailed information

Explicit quality adjustment tends to be avoided. This means that quality adjustments are mostly implicit and done using the bridged overlap method where there is not risk of downward bias due to products having high entry prices and low exit prices on an item level. Option pricing is used for cars, while direct comparison is normally used for products such as video games, CDs, DVDs, books and clothing and footwear, given that these segments tend to have high entry prices and low exit prices on an item level. The number of quality adjustments carried out is not kept in a global database, so no figures can be given. They need to be separately compiled for most segments.

Hedonics is used for second-hand cars sunglasses. A time dummy hedonic method is used. Hedonics are also used for consumer electronics (laptops, tablets, smartphones and televisions) and household appliances (washing machines, dryers, refrigerators, freezers, cooking hobs and dishwashers). The method there is the ITGEKS method.

In the context of the recommendations on bridged overlap, which quality adjustment methods are used in the following situations:

• The last price of the replaced (old) product-offer is a reduced price. —> direct comparison (with quantity adjustment) with most suitable replacement in the case no hedonics is used.
• The first price of the new product-offer is a reduced price. —> direct comparison (with quantity adjustment) with old item in the case no hedonics is used.
• The first price of the new product-offer is unusually high —> direct comparison (with quantity adjustment) with old item in the case no hedonics is used.
• The matched sample of product-offers includes reduced or atypical prices, or shows a downward price trend during the product life cycle. —> for such product groups no bridged overlap is used

18.5.5. Seasonal items

Fresh fruit, fresh vegetables and fresh fish are considered to be strongly seasonal products. The indices are compiled using scanner data with the same method as for the other supermarket scanner data product groups.

Not applicable.