New base for the Icelandic consumer price index

I.  Introduction
  

The Icelandic legislation on the Consumer Price Index (currently Act No.12/1995) stipulates that Statistics Iceland shall make a comprehensive survey of household expenditures, a household budget survey, at an interval of no less than five years Accordingly a household budget survey (HBS) was conducted in 1995 following similar surveys in 1990 and 1985. The processing of the HBS of 1995 was completed in early 1997 resulting in a new base for the CPI. The compilation of the CPI according to the new base was carried out for the first time in April 1997.

The CPI is applied to a variety of purposes and it is therefore vitally important that the underlying work for its construction be performed with as much care as possible. The CPI is based on a collection of data on all kinds of household expenditure, not only data on everyday purchases of food, toilet articles, clothing, etc. but also the cost of electricity, space heating, telephone, operation of private car, transport and recreation. The base thus reflects more or less all kinds of expenditure relating to a household and the everyday life of people in general. There is no attempt whatsoever to evaluate whether or not this expenditure is considered necessary. The main purpose of the HBSs is to extract a base for calculating the CPI, i.e. to assess the weights of the different categories of goods and services constituting the CPI base. Consumer patterns are constantly changing and for this reason it is necessary to revise the composition of the CPI regularly by household expenditure surveys. Furthermore, the information thus gathered is an important source for examining changes in consumer habits and for investigating household expenditure and its composition with regard to various social and economic factors such as location of the household within the country, type of household, employment and income of household members.

The CPI is used for a number of purposes, chief among which are the following:

• Measurement of general price changes.
• Indexation of financial obligations.
• Computation of the price adjustment factor in financial statements of enterprises.
• Computation of purchasing power.
• Deflation of GDP and various other economic aggregates.
• Assessment of the price impact of economic policy.
• International comparison.
   

II . Household budget survey

Sample and sampling method

The 1995 HBS was conducted along similar lines as the one of 1990 by using a random sample drawn from the National Register of Persons without regard to domicile, household size, income or occupation. A HBS attempts to measure household expenditure and therefore the sampling unit is the household. A sample that would produce the same sampling probability for all Icelandic households would indeed be desirable. This, however, is not feasible since it is impossible to distinguish separate households in multi-dwelling houses in the Register of Persons. The Register does, however, identify so-called nuclear families, i.e. households consisting of couples (married or cohabiting) or lone parents with children under the age of 16. Each nuclear family has its family number and the same applies to single people. On reaching the age of 16, people are assigned their own family number and no longer counted as a part of a nuclear family.

The sample for the 1995 HBS was drawn from the family numbers of people aged 18–74 with domicile in Iceland at 1 December 1994. Sixteen and seventeen-year old teenagers were assumed to be still living with their parents. The upper age limit was set at 74 years in view of the fact that the HBS demands considerable effort on part of the participants and experience has shown that older people usually feel less confident of being able to perform the work required or less capable to do so. However, individuals 75 years or over were not excluded from the survey if they were members of sampled households. The probability of households being selected was not the same for all types of households. Thus, there was a greater probability of selection for households composed of several individuals 18 years and older. In households of selected individuals every household member automatically became a participant. The sample was not supposed to include institutionalised people staying in hospitals or residential homes for the elderly.

Response rates

The original sample selected included 2,910 family numbers, of which 203 did not qualify according to the sampling rules. The majority of these were individuals residing abroad although domiciled in Iceland. The final sample, therefore, totalled 2,707 households. Out of these, 1,649 households agreed to take part in the survey, 1,004 declined and 54 households could not be contacted. Data from 1,375 households turned out to be satisfactory, 233 households dropped out during the survey period and data from 41 households proved unacceptable. The net response rate in the 1995 HBS was therefore 50.8%. For comparison, 790 households took part in the 1990 HBS and 376 in the one conducted in 1985/86. The 1,375 households taking full part in the survey were composed of 4,519 individuals or 1.7% of the resident population according to the National Register at 1 December 1995. The willingness to participate in the survey varied between different types of households. Households consisting of married or cohabiting couples with children proved most cooperative whereas single individuals were the least willing to participate. The response rate did not vary much between regions, however. Of the participating households, 857 were in the capital region, 351 in other urban areas and 167 in rural areas.

Execution

The survey was conducted in two stages. On one hand, participants were supposed keep a household diary over a period of two weeks and record every item of expenditure of every household member. The purchase of every item of goods or services had to be recorded separately. In order to ease this burden of recording, participants were permitted to return itemized bar-code cash receipts and record only the total amount. This phase of the survey lasted a whole year with a new group of households keeping household diaries for every two-week period. As the household diary period was only two weeks there was a risk of losing information on some vital but irregular expenditure items. The participants were therefore requested to provide information on certain large expenditure items relating to a three-month period. Such items involved e.g. operating and maintenance costs for housing and car, the purchase of durable goods such as household appliances and furniture, travelling costs, etc. This part of the survey was conducted by means of special account books sent to participants living outside the capital region that they were asked to fill in. In the capital region, on the other hand, participants were interviewed face-to-face and their expenses recorded on the spot. The 1995 HBS was the first survey of its kind conducted by Statistics Iceland in which interviewers used laptop computers.

Processing

The final results were obtained by combining data from the quarterly accounts/interviews and the household diaries. Data on purchases of food, beverages, clothing, and sanitary and toilet articles was gathered exclusively from the diaries while the quarterly accounts provided data on the operation of housing and car, expenditure on education, child day care and transport services. Expenditures were calculated on an annual basis.

Different sampling probability and response rates

Because of the unequal probability of households being selected and the varying response rates of different types of households, the results of the survey had to be adjusted for these factors. For this purpose, information on households refusing to take part in the survey had to be gathered specifically and a weight was calculated for each responding household in the sample taking into account the sampling probability of that household and the response rate of the household type in question. The household expenditures were then adjusted by applying these weights. These adjustments make the survey results more reliable than would otherwise be the case.

Mean household size

The mean size of household in the 1995 HBS was 2.82 persons as compared with 3.63 in the survey of 1990. However, these figures are not comparable because of the adjustments made in the 1995 survey for different probabilities of selection and varying response rates. In the 1990 HBS no such adjustments were made. In the 1995 survey, household size varied considerably between regions; it averaged 2.65 persons in the capital region, 3.00 persons in other urban areas and 3.40 persons in rural areas.

Type of household

Households were divided into five categories in the HBS according to the family relationship of household members. Of the 1,375 responding households, there were 185 single-person households, 209 households of married or cohabiting couples without children, 686 households of couples with children, 116 households of lone parents and their children, and 179 households of some other composition. In defining the category married or cohabiting couple with children, the upper age limit of the children was set at 24 years. If children in a household were over 24 the household was classified to the category "other type of household".

Housing

Just over 81% of the households lived in owner-occupied dwellings. Furthermore, 34% of the households lived in detached houses, 28% in blocks of flats, 22% in 2-5-dwelling houses, 15% in terraced or semi-detached houses and 1% in separate rooms. Housing categories varied greatly between regions. In the capital region, 62 % of the participating households lived in blocks of flats or 2-5-dwelling houses while elsewhere in the country 73% lived in detached or terraced houses. The average size of dwellings was 114.5 m² and there was an average of 4 rooms per dwelling. Dwellings outside the capital region were somewhat larger than those in the capital region.

III. New CPI base

Harmonization within the European Economic Area

During the past several years, a great deal of effort has been devoted to the harmonization of CPIs within member states of the European Economic Area (EEA). The chief objective of this effort has been the coordination of methodologies applied by the states in their compilation of the CPI, collection of price data, classification of expenses and construction of a new base. The member states have taken great pains to ensure the adoption of the best available methods. This work has not been completed as yet although harmonized CPIs based on the work done so far are now published each month for all the EEA member states.

Statistics Iceland has taken an active part in this cooperation. Although already apparent in the initial stages that the compilation of the CPI in Iceland stood comparison with the best procedures in other EEA states, the coordination nevertheless led to several changes in the Icelandic CPI base. The most important of these is the adoption of a new classification of expenditures as well as a redefinition of certain items. The compilation of the CPI has also been made subject to EEA regulations that on the whole are in line with procedures already applied at Statistics Iceland. A new provision in the regulations stipulates that the CPI will in the future be subject to quality adjustments based on the results developed under the auspices of Eurostat.

A new international classification

The new international classification COICOP (Classification of Individual Consumption by Purpose) has been adopted for classifying commodities and services in the HBS and the new CPI base. It comprises twelve main expenditure divisions instead of nine in the previous system. The new classification has three new divisions; education, communications and hotels, cafés and restaurants.

In addition, the new classification involves changes in the grouping of certain commodities and service items. Beverages, for instance, are now classified to food and cover tea, coffee and fruit juices while alcohol and tobacco have been made into a separate division that includes non-alcoholic and ordinary beer. Package holidays become a separate subgroup belonging under recreation and culture. Similarly, the group social protection is classified under the division miscellaneous goods and services, a division also comprising all types of insurance, including car insurance that formerly used to be classified under operation of personal car. Several other changes have been made that need not be enumerated in this context.

From now on the CPI will be released with more detailed breakdown than before. The classification has twelve divisions and 39 groups. The following example shows the structure of the system:
 

• 01 Food and non-alcoholic beverages (Division)
• 011 Food (Group)
• 0111 Bread and cereals (Class)
• 01111 Rice (Basic heading)
• 01111aa Rice in shop (Item in a shop)
    

Cars

The item personal car has been redefined. Expenditures involving the purchase of a new car are now calculated net, i.e. as the difference between the purchase price and the price obtained for a sold car, whereas earlier this expenditure was based on the gross annual renewal price in relation to the average car life. This change is the result of the European harmonization referred to above. The personal car item in the CPI will be revised every year and this revision will apply both to car makes and the relative weight of car purchases in the total household expenditure. Prices for different makes of personal cars have been revised annually for a number of years whereas the CPI weight of cars has not. The CPI base of March 1997 relates to car imports in 1996.

Housing

According to the household budget survey, 81 per cent of households lived in their own housing as compared with 89% in the 1990 survey. In the labour force survey conducted by Statistics Iceland in November 1996, questions were included on owner occupancy as opposed to rented housing. The results were similar to those of the HBS, thereby confirming the HBS findings that a relatively larger portion of the population now lives in rented housing than before. From now on, Statistics Iceland will assess rentals for housing, partly by means of special surveys.

The assessment of the value of owner-occupancy in terms of money is not an easy one. The present CPI relies on the same user-cost method as the previous one in calculating the housing component as imputed rents for all participants living in their own housing. The basis for this calculation is the real estate value of dwellings.

Reassessment of HBS items

The survey revealed that average household expenditures on lotteries, TV and newspapers were presumably somewhat underestimated because regular credit card payments of such expenditures were left out in the household diaries. These items were therefore reassessed based on information from the enterprises concerned on operating revenue of the lotteries, media subscription figures and figures on printed copies of newspapers.

IV.  Calculation of the CPI

Selection of items

The breakdown of expenditure in the CPI between different classes of commodities and services is based on the results of the HBS. As soon as these results are available the next step is to determine what goods and services should constitute each class of the CPI base. The HBS data is made up of a much larger variety of goods and services than can possibly be used in monthly calculations of the CPI or are needed for that purpose. For this reason each class of expenditures is created by means of a selection of goods and services that are considered either important or typical. The price changes of these particular goods and services are then monitored and their weighted average is subsequently used as a measure of price changes for that class.

The 1995 household budget survey provided more detailed information on the commodities involved than had been available before. In order to ease the response burden of the survey participants, they were permitted to return bar-code cash receipts from stores whenever these were properly itemized. This experiment turned out to be more successful than anticipated and 53% of all entries for food and beverage items were obtained in this manner and almost 41% of the total number of entries in the survey. The cash receipts usually show information regarding type of commodity, brand, quantity, unit price and place of purchase. This information was utilized to select outlets and commodities for the new CPI base.

The information collected refers to cash prices for goods and services and takes into account cash discounts available to everyone as well as special offers. So far sales prices have not been observed although there are plans to do so in the near future. Information on all items of goods and services is collected and recorded in the first two days of each month. The data covers some 730 basic headings for the CPI and approximately 15,000 prices for 2,100 items.

The data is collected in a number of ways. Statistics Iceland staff members collect data on food, beverages and clothing in selected outlets. Data on other goods and service items is collected by enquiries through fax, telephone, letters, e-mail or other electronic form directly from sellers. Price data is collected for those goods and services that are most widely purchased and no attempt is made to find comparable goods or services offered at a lower price. The most important concern is to deal with exactly the same goods and services from one month to the next. Information on certain seasonal goods and services, for instance summer or winter tours, school fees, fishing licences etc., is recorded as soon as such prices are determined, i.e. once or twice a year as appropriate. In between, the prices remain unchanged in the CPI. No adjustments are made for seasonal prices of such items as vegetables, rented cars etc.

The Consumer Price Index Act (of March 1995) provides that the index must reflect prices throughout country as far as practicable. For this purpose Statistics Iceland has collected prices in selected outlets in four different places of the country since the middle of 1995. As of March 1997, such data is fully incorporated in the CPI.

The method of calculating mean prices has been changed adopting the geometric mean instead of the arithmetic mean. This method has various technical advantages and has recently been adopted in many European countries. The calculation of mean prices for the whole country is based on four consumer price indices weighted by the estimated market share of individual shops or chain stores. These estimates are based on information on turnover from tax authorities and from the household budget survey.

For practical reasons it was decided to maintain the base of May 1988 = 100 as reference base for the new CPI. The CPI base compiled in this manner will be released for each of the 12 divisions of goods and services while the base of March 1997 = 100 will be used for reference as regards subdivisions (groups and classes). The reason for this anomaly is the change in classification already described.

V.  Main conclusions

Table I shows the breakdown of expenditures in the CPI in March 1997 according to the old and the new bases. On one hand, there has been a reduction in expenditures in the divisions food and beverages; alcoholic beverages and tobacco; clothing and footwear; housing, water, electricity etc., and recreation and culture as percentage of the total household expenditure. On the other hand, there has been an increase in the weight of divisions such as furnishing, household equipment etc.; health; transport; communications; education; hotels, cafés and restaurants, and miscellaneous goods and services. The changes involved are mostly minor ones and some of them are the result of changes in methodology, definitions and classification rather than actual changes in the composition of household expenditures.

The proportionate contraction in food expenditure is partly a result of changed consumption patterns involving an increase in the consumption of fast foods, which are classified to restaurants, at the cost of food purchases. Within food purchases the consumption of meat and fish has decreased while the opposite applies to pasta. There has also been an increase in the consumption of confectionery, soft drinks and fruit juices.

The decrease in expenditure on clothing may not be actual because of changes in definitions and survey methods. It is quite possible that the 1990 survey produced an overestimate under this heading, which was due to difficulties in breaking down and classifying presents.

As mentioned earlier the housing component of the new CPI weighs less than it did in the former CPI base. This is primarily due to a change in the definition of house maintenance as a result of the new classification of expenditures and international harmonization of definitions. According to these changes major maintenance and renovation of housing is defined as investment whereas only minor maintenance and repairs are counted as part of household expenditure. In the previous CPI base, however, both types of maintenance were considered part of private consumption and therefore included in the housing component of the CPI.

The transport component shows an increase from the last survey. This is to a considerable extent caused by increased purchases of cars. Car purchases in the new base refer to the year 1996. Counteracting the increases in car purchases is a contraction in the operating cost of cars. Among other transport items measured in the HBS, an increase has occurred in expenditure on air transport, part of which may be caused by different methods of breakdown between package holidays and scheduled flights.

Finally, the restaurant component has increased considerably, a fact that can be ascribed to an increase in purchases of fast food and increased consumer expenditure at restaurant and cafés.

For comparing the old and the new CPI's, the CPI according to the new base has been calculated backwards for the approximately two years proceding its adoption in March 1997. A comparision of the overall measured price change according to the new and the old and the new CPI's for that period only shows a marginal difference. Thus from, the 1995 average to March 1997 the total price change is 0,2% smaller than as measured by the old one.

Table II shows average household expenditure per year by regions. In this connection it must be kept in mind that the compostion of household types differs quite considerably between densly populated and sparsely populated aread. Thus, single-person households are much more common in the capital region than elsewhere in the country. In rural areas, households of the type married couple with children and other types of households constitute as much as 76% of all households while in the capital area this percentage is 58%. These two types of households have the greatest number of household members, which means that rural households are larger than urban ones on average. Therefore, the difference in the composition of household expenditure cannot be ascribed to residence alone but also to type and size of household. The largest categories of household expenditures are housing, food and transport, making up approximately half of all household expenditures. Food expenditures are relatively lowest in the capital region and highest in rural areas. The larger size of rural households explains their higher food expenditures. Housing costs, on the other hand, are highest in the capital region. Within the housing component, however, the cost of electricity and space heating is lower in the capital region than elsewhere in the country while housing rents are higher in the capital region. Imputed rents is calculated on the basis of the real estate value of dwellings, which is lower outside the capital region than within it. Transport costs are relatively lowest in the capital region, partly because refunding of transport expenses is more common there than elsewhere in the country.

Table III shows average expenditures per year by type of household. It is somewhat difficult to compare expenditures by type of household because of their uneven size. Expenses may vary in different proportions with varying size of household. To discover differences in consumer patterns the easiest way is to compare relative shares. Food expenditures are relatively highest in the category couple with children and other type of household, the types that constitute the largest households and include the largest number of grown-up individuals. Food expenditures of single-person households are similarly relatively lower than those of other households whereas housing costs are relatively higher. Single persons live in rented housing to a greater extent than other. Also, their expenses on alcoholic beverages and tobacco are higher those of other households. Naturally, households with children spend more on education and childcare than other types of households.