Consumer Price Index
Technical Manual

This technical manual was prepared by the National Statistics Office of Georgia in accordance with internationally recognised methods and practices, and is based on the following handbooks:

• Consumer Price Index Manual, Concepts and Methods — International Monetary Fund and others, Washington 2020.

  • Responsible organisations: IMF, ILO, OECD, Eurostat, UN, EBRD, World Bank
  • https://www.ilo.org/public/english/bureau/stat/guides/cpi/CPI_Manual.html
• Practical Guide to Producing Consumer Price Indices — United Nations, New York and Geneva, 2009.

  • Responsible organisations: UNECE, ILO, IMF, OECD, Eurostat, UN, World Bank, UK Office for National Statistics
  • https://www.unece.org/.../Practical_Guide_to_Producing_CPI.pdf
• Consumer Price Indices — Ralph Turvey, International Labor Organization, 1993.
• Harmonised Index of Consumer Prices (HICP), Methodological Manual — Eurostat, Luxembourg, 2018.

  • Responsible organisation: Statistics Office of the European Union (Eurostat)
  • https://ec.europa.eu/eurostat/documents/...

Goods and services acquired by consumers are represented by the fixed consumer basket, which reflects the structure of consumption expenditures for an average consumer. Items in the basket are selected based on data from the National Accounts and the Integrated Household Survey, and are grouped according to the Classification of Individual Consumption by Purpose (COICOP).

The Consumer Price Index is used for the following purposes:

• The CPI is the only index used to measure the inflation rate in Georgia, and is a key factor in inflation targeting.
• The CPI plays an important role in the indexation of incomes, social assistance payments, contract amounts, and similar obligations.
• The CPI is used as a deflator to eliminate the influence of inflation when calculating various economic indicators.

In addition to the CPI, Georgia also produces the Harmonised Index of Consumer Prices (HICP). It is used primarily by EU member states and is an important indicator of price stability. The calculation and dissemination of this indicator therefore has great significance in terms of Georgia's approximation to EU standards.

The HICP is based on a unified, harmonised methodology and regulations developed by the EU statistical office — Eurostat. Strict observance of these requirements is mandatory to ensure the international comparability of the index. For the purposes of this index, the European version of the Classification of Individual Consumption by Purpose — ECOICOP — is used.

The Consumer Price Index reflects changes in expenditures on goods and services acquired for private consumption by residents, both within and outside the territory of the country. In contrast, the coverage of the HICP includes expenditures on goods and services acquired for private consumption by both residents and non-residents, but only within the economic territory of the country.

The index does not cover expenditures of government or institutional households (hospitals, prisons, shelters, etc.).

Information about active business entities is used as the data source for outlet sampling. Sampling is conducted for each city, based on the commodity groups represented in the basket (in some cases for individual goods and services), using the sequential probability-proportional-to-size (PPS) sampling method, with the following formula:

After calculating Q_i ranking variables for each outlet, they are sorted in ascending order and the first n units are included in the sample. The distribution of outlets across city districts is also taken into account. A selected outlet that is not suitable is replaced by the next one on the list. The same method applies to outlets that close during the survey period. The outlet sample is updated annually.

After outlet sampling, a market survey is conducted to identify the most commonly sold representative items for each basket category — the brands most in demand. Price changes of these representative items reflect the dynamics of consumer prices and tariffs.

Detailed specifications are defined for each representative item alongside its selection. Strict adherence to specifications is essential, so that recorded price differences reflect only genuine price changes, not changes in the item or its characteristics. In some cases, broader specifications may be used to capture rapidly shifting market demand — for example, clothing, where fashion and detailed characteristics change frequently.

A standard unit of measurement is also defined for each item. Standard unit prices are used for data validation and logical control by both price collectors and the central office.

The number of price observations depends on the price variability of each good or service and its weight in the consumer basket. For most goods and services, the target is at least 6 prices per city; for electronic goods (e.g. computers, mobile phones), 20–25 prices are collected to ensure sufficient comparable observations given their frequently changing characteristics. Where a city does not have enough suitable outlets, fewer than 6 observations may be accepted. For utility tariffs, a single observation is permitted.

Special cases:

Prices of used cars are registered on the last weekend of the collection period, as those markets operate most intensively on Saturdays and Sundays. Prices are therefore registered on the same day of the week rather than the same date each month.

Where an outlet regularly applies discounts on a fixed day of the week, prices are registered on the discount day nearest to the previous month's registration date.

Prices of perishable goods (fruits and vegetables) are collected between 11 a.m. and 4 p.m., to avoid the significant differences between morning and evening prices.

If an outlet closes temporarily (for stock-taking, repairs, etc.), the price collector returns on the next day or on the predefined date of the following month, depending on the reason. In exceptional circumstances (public holidays, etc.), collection may take place up to two days before or after the scheduled date.

There are two main methods of price collection: local price collection and central price collection.

3.2.1 Local price collection is used for most goods and services in the consumer basket. It can be conducted in the following ways:

• Price collectors visit outlets in person and register prices for predetermined goods or services;
• Price collectors register prices via the outlet's website;
• Price collectors obtain prices by telephone. This applies to:
  • Transport services (train tariffs, air flight fees, public road transport tariffs);
  • Communications (monthly telephone fees, intercity call tariffs, mobile phone tariffs, internet service tariffs);
  • Cable television fee;
  • Public utility fees (natural gas, water, electricity, waste collection);
  • Medical services (therapist and dentist services, surgical and maternity services, blood tests);
  • Tourist trips abroad, etc.

Prices are collected from the following types of retail outlet:

• Retail markets and market-type outlets;
• Stores;
• Supermarkets;
• Specialised stores and retail chains;
• Newsstands;
• Household service establishments;
• Recreation and entertainment facilities;
• Health and education institutions;
• Catering establishments, etc.

At market-type outlets, price collectors behave as ordinary shoppers. After asking 3–5 different sellers, they record the most prevalent — the dominant price.

At shops and supermarkets, label prices are recorded or verified with a sales assistant. If an item is unavailable for three consecutive months (12 months for seasonal items; see 3.3), the price collector substitutes it with a comparable item and notes the replacement item's characteristics in the questionnaire. The replacement item's price is registered in the third month since the original item became unavailable.

Prices registered for index calculation include any service charge added to the item (e.g. a restaurant meal, coffee in a café). Price collectors therefore register prices inclusive of any service surcharge.

Special cases:

A discount price is recorded only if it applies to an item of the same quality as in the previous period and is available to all consumers. Discounts restricted to particular groups, or associated with the clearance of damaged or deteriorated stock, are disregarded.

Temporary bonuses, extras and gifts:

• When additional quantity is offered (e.g. "buy 2 get 1 free", or "extra 20% free"), the extra quantity is not included in the price;
• An item's price is not registered if it can only be identified as part of a bundle with other items and cannot be priced separately.

If special offers become permanent, price collectors inform the GeoStat central office so the change can be taken into account. Loyalty-card rebates and credit-card payment discounts are excluded.

3.2.2 Central price collection is conducted by GeoStat headquarters staff. It covers centrally fixed prices obtained from the head offices of retail chains. Checks are made to confirm that the relevant goods and services are actually available in the specified outlets in each city.

3.2.3 Collection of centrally fixed prices covers goods and services that have the same price nationwide, or that are sold through national retail chains. Such prices may be registered by field collectors or the central office, and the recorded price is then extended to all cities. Items in this category include:

• Pharmaceutical products;
• New and used cars;
• Fuel (petrol, diesel);
• Train tariffs;
• Air flight fees;
• Mobile phone tariffs;
• Banking service fees;
• Intercity telephone call tariffs, etc.

3.2.4 Web scraping — automated download of data from trade and service outlets' websites. This relatively new method, carried out under agreement with the outlets, allows large volumes of detailed data to be downloaded quickly for subsequent analysis and inflation calculation. It currently covers selected basket categories, including electrical appliances and used cars.

3.2.5 Scanner data forms part of so-called "Big Data". It consists of transaction-level information on products sold — identified by barcode — supplied by retailers. Compared with traditional methods, scanner data is significantly more accurate, timely, and comprehensive, and its use substantially improves the precision of the inflation estimate. Scanner data has monthly periodicity, covering sales from the 1st to the 10th of each month, by retail chain and city.

If a representative item is unavailable for three consecutive months (12 months for seasonal items), it is replaced by a similar item. The replacement must be qualitatively and quantitatively as close as possible to the original.

The replacement is selected in the same outlet where the original became unavailable. If no suitable replacement is found there, the collector looks in a nearby outlet of the same type that is included in the sample.

An item is considered seasonal if:

• it is not available on the market during certain periods of the year; or
• it is available year-round but its price fluctuates sharply across seasons.

Seasonal items include fruits and vegetables, clothing and footwear, and leisure and entertainment services. Prices for seasonal items are registered during the relevant season; outside that season, the index is estimated using the imputation technique (see Chapter 4).

Comparability criteria ensure that quality changes are not recorded as price changes. A comparable item must satisfy the pre-defined specifications of the consumer basket. When comparing representative items across periods, price collectors follow these rules:

• Two items from different outlets are not considered comparable;
• Two items produced in different countries are not considered comparable;
• Two items with approximately the same quality and characteristics may be treated as comparable (e.g. clothing items that differ only in colour);
• Where a replacement item differs materially in size or weight, it is added as a new representative item; otherwise it is treated as comparable to the replaced item.

When a representative item is unavailable, its price must be imputed. The imputed price is calculated by multiplying the item's price in the previous period by the elementary aggregate index for the current period (see 8.1) — that is, the geometric mean of comparable item price ratios. If no comparable prices exist for the previous period, an upper-level group index is used instead.

For example, if no price is available in March for Brand D, the price is imputed as follows (Table 2):

When the item reappears on the market, the previous month's imputed price serves as the comparable baseline for the current period's price.

When a missing item is replaced by one that differs qualitatively or quantitatively, a quality adjustment method must be applied to make prices comparable.

When a price collector knows in advance that an item will be missing or an outlet will close in the following month, the overlap method is used: prices are collected for both the replaced and the replacement item in the same period. The price difference between the two items is treated as the full value of the quality difference. The replaced item's price is used in the current period's index calculation; the replacement item's price is used from the following period onwards.

For example, in March a collector is informed that Brand C will be unavailable in April. The collector registers the price of the replacement item (Brand D) in March, which is then used in the April index calculation (Table 3).

If an item is missing for three consecutive months (12 months for seasonal items), the replacement procedure described in 3.3 is followed. Table 4 illustrates index calculation in this case.

The elementary aggregate indices for March, April and May are calculated solely from the geometric means of comparable item price ratios, which are also used to impute Brand C's price in those months. In May, the replacement item is priced and used in index calculation from June onwards. The price difference between Brands C and D recorded in May is treated entirely as a quality differential.

Data validation is carried out in two stages:

Stage 1 takes place simultaneously with fieldwork. If a recorded price falls outside a pre-set acceptable range for that item, the software alerts the price collector to verify the price. The collector confirms accuracy by adding a comment. Data are sent to the central office on the same day as collection, where the analyst responsible for that city reviews the information and requests re-verification if necessary.

Stage 2 is carried out by the central office after calculating indices and average prices, and involves:

• Comparing detailed and group indices across all six cities, and verifying significant deviations from the average;
• Analysing the dynamics of average prices and detailed indices.

An elementary aggregate index is considered reliable once it has passed both validation stages, regardless of the number of comparable prices registered in that city.

The weights for basket groups and individual goods and services are based on the consumption structure derived from the National Accounts, reflecting the most recent expenditure patterns across the country. The Integrated Household Survey provides supplementary data for weight calculation. The final weights represent each item's share of total monetary consumer expenditure.

City weights are calculated as each region's share of total national monetary consumer expenditure, also based on National Accounts data and the Integrated Household Survey.

For HICP weights, price updating is additionally applied: the t–2 expenditure data are adjusted for price changes between period t–2 and period t. This serves two purposes:

• To provide a comparable, harmonised calculation process across countries — enabling aggregation and analysis of the HICP across countries when the same weight reference period is used;
• To align the weight and price reference periods — since expenditure data become available with a lag, the weight reference period (t–2) precedes the price reference period (December t–1); price updating brings these two base periods into alignment.

It should be noted that price updating of weights is not an alternative to the regular annual weights update, as it does not change the quantities consumed.

Each good or service in the consumer basket is represented by a set of representative items (brands) at the city level. Diagram 1 shows the structure of the consumer basket for a city, where the elementary aggregate index is represented by elementary aggregates such as rice, buckwheat, etc. The short-term elementary aggregate index is calculated as the ratio of the geometric means of comparable item prices in periods t and t–1:

The long-term elementary aggregate index relative to the price reference period is obtained by multiplying short-term indices. The national index is calculated by weighting city-level indices with the corresponding city weights.

Diagram 1. Structure of the consumer basket for a city

The long-term CPI relative to the price reference period is calculated using a Laspeyres-type formula:

The same formula applies to higher-level aggregate indices at city level. A higher-level aggregate index is calculated by weighting the long-term elementary aggregate indices within the aggregate relative to the price reference period, where the sum of weights equals 1.

Short-term indices for separate aggregates or the whole basket, compared to the previous month, are obtained as ratios of long-term indices in the reporting and reference periods.

The national-level index is a weighted arithmetic mean of elementary or higher-level aggregate indices calculated for individual cities, where weights represent each city's share of national consumer expenditure. Short-term national indices compared to the previous month are calculated as ratios of long-term indices in the current and previous months, as at the city level.

During the annual update of the basket and specifications in December, prices are collected for both old and new consumer goods and services simultaneously. This enables two successive baskets to be chain-linked, allowing higher-level aggregate indices to be calculated relative to any base period regardless of changes in weights or basket components.

For example, before December 2011 all-items indices with a reference period of December 2009 were calculated using w_i weights; from 2012 onwards, indices with a reference period of December 2011 are calculated using k_i weights (see Table 5). X₁ is the chain index for January 2012 with base period December 2009:

The same result is obtained by chain-linking: I^{12.2011/12.2009} × I^{01.2012/12.2011} = 106 × 102/100 ≈ 108.

Core inflation is an indicator of the underlying trend in inflation, obtained by excluding the seasonal component and the most highly volatile prices. Its purpose is to represent the main dynamic of inflation, making it a significant analytical tool for assessing inflationary processes in the country. Core inflation is calculated using the same weight structure and formula type as the overall CPI.

In Georgia, core inflation is calculated by excluding the following groups from the consumer basket:

• Food and non-alcoholic beverages;
• Tobacco (in certain cases);
• Energy (petrol, diesel, gas fuel, liquefied gas, firewood);
• Administered prices (water, waste collection, electricity and natural gas);
• Transport (suburban and long-distance rail, suburban bus, intercity road transport, intercity taxi and air travel).

A CPI press release is published on the GeoStat website every month, covering monthly and annual inflation rates, group contributions to index formation, and time-series graphs.

Group contributions to the percentage change in the all-items index provide a powerful analytical tool. The contribution of a component is defined as the percentage change in the overall CPI caused by the change in that component's index alone, all else equal. At the city level, the contribution of component i is:

Where weights change between the two periods, the contribution of component i is:

Example: calculation of group contributions in the case of a weight change

Using the formula above, the contribution of food and non-alcoholic beverages to the overall CPI for October 2012, given weights of 0.35 (2011) and 0.28 (2012), is calculated as follows (Table 6):

Thus, the contribution of food and non-alcoholic beverages to the all-items CPI in October 2012 amounted to 0.8 percentage points.

The following CPI time series are published monthly alongside the press release:

• CPI relative to the previous month;
• CPI relative to a long-term base period;
• CPI relative to the same month of the previous year;
• CPI — 12-month average compared to the previous 12-month average.

The following HICP time series are also published monthly:

• Harmonised Indices of Consumer Prices compared to the previous month;
• Harmonised Indices of Consumer Prices compared to the corresponding month of the previous year.

Published indices are rounded to four decimal places and represent final data. Time series data are published with corresponding graphs on the website.

Data are also available via PC-Axis — a data dissemination system developed by Statistics Sweden — which allows users to obtain statistical information from the GeoStat website (www.geostat.ge) in various formats (text, tables, graphs, etc.). CPI data are also disseminated through an Android mobile application.

Diagram 2 shows the stages of CPI calculation and their periodicity.

Diagram 2. Stages of CPI Calculation