31995D0309

COMMISSION DECISION of 18 July 1995 specifying the principles for estimating dwelling services for the purpose of implementing Article 1 of Council Directive 89/130/EEC, Euratom on the harmonization of the compilation of gross national product at market prices (Text with EEA relevance) (95/309/EC, Euratom)

THE COMMISSION OF THE EUROPEAN COMMUNITIES,

Having regard to the Treaty establishing the European Community,

Having regard to the Treaty establishing the European Atomic Energy Community,

Having regard to Council Directive 89/130/EEC, Euratom of 13 February 1989 on the harmonization of the compilation of gross national product at market prices (1) and in particular Article 1 thereof,

Whereas for the purpose of the definition of gross national product at market prices (GNPmp) pursuant to Article 1 of Directive 89/130/EEC, Euratom, it is necessary to clarify the principles for the estimation of dwelling services as used for the purpose of the European System of Integrated Economic Accounts;

Whereas the measures provided for in this Decision are in accordance with the opinion of the Committee established by Article 6 of Directive 89/130/EEC, Euratom,

HAS ADOPTED THIS DECISION:

Article 1

For the purpose of the implementation of Article 1 of Directive 89/130/EEC, Euratom, the following principles for the estimation of dwelling services shall be applied:

(a) to compile the output of dwelling services Member States shall apply the stratification method based on actual rents;

(b) Member States shall use tabular analyses or statistical techniques to derive significant stratification criteria;

(c) the actual rent shall be understood as the rent due for the right to use an unfurnished dwelling;

(d) to compile imputed rents, actual rents from all contracts shall be exploited relating to privately-owned dwellings;

(e) in countries where the privately rented sector is small, the corresponding level of rent in the private sector may be obtained either using increased (public) rents or, in justified exceptional cases, employing other objective methods like the user-cost method;

(f) rents for furnished dwellings may similarly be used to enlarge the basis for imputed rents if scaled down to exclude the payment for the use of the furniture;

(g) Member States should extrapolate a given base year figure using appropriate quantity, price and quality indicators.

The specification of the principles for the estimation of dwelling services is contained in the attached Annex.

Article 2

For the years 1988 onwards, the estimation of dwelling services, in accordance with the principles set out in the Annex, are to be submitted to the Commission (Eurostat) no later than 30 September 1996.

Article 3

This Decision is addressed to the Member States.

Done at Brussels, 18 July 1995.

For the Commission

Yves-Thibault DE SILGUY

Member of the Commission

(1) OJ No L 49, 21. 2. 1989, p. 26.

ANNEX

For the purpose of the implementation of Article 1 of Directive 89/130/EEC, Euratom, the following points aim to clarify the principles for the estimation of output, intermediate consumption and transactions with the rest of the world, regarding dwelling services.

1. OUTPUT OF DWELLING SERVICES

1.1. Basic Method

In national accounts, by convention, the output of housing services comprises not only the services produced by rented dwellings but also those provided by owner-occupied dwellings. With regard to the valuation of output of housing services, the European System of Integrated Economic Accounts (ESA) stipulates in paragraph 315 i, that 'services produced by the ownership of dwellings are measured by the value of the rents if rented, or by the value of the rents of similar dwellings if owner-occupied`. The major problem in applying this rule relates to the interpretation of 'similar` in the case of owner-occupied dwellings. In practice, basically two different methods are used to make this rule operational (1). On one hand, the majority of countries apply the stratification method, which combines information from the the total housing stock, broken down by various strata, with information on actual rents paid in each stratum. On the other hand, some countries use the self-assessment method for owner-occupied dwellings, asking owner-occupiers to estimate a potential rent for their property.

This difference in procedure may to a certain extent be explained by the fact that the percentage of owner-occupied dwellings in different Member States varies between 42 % and 77 %. In the case of a high percentage of owner-occupied dwellings it may be difficult under certain circumstances to obtain an actual rent for similar dwellings as required by the stratification method. In contrast to this, it should be stressed that some countries having a very high percentage of owner-occupiers do in fact apply the stratification method. On the other hand, the major problem of the self-assessment method consists of the largely subjective influence on the estimate. This leads to substantial uncertainties because of over- of underestimates (depending on the precise circumstances) and will increase the error margin of the GNP estimate in parallel with the growing importance of owner-occupied dwellings. Further, the error margin over time does not seem likely to remain stable with changing circumstances.

Given that the self-assessment method is largely subjective, it would seem reasonable from a statistical viewpoint to recommend the stratification method. In general terms, the stratification method uses information about actual rents from rented dwellings to obtain an estimate of the rental value of the total stock of dwellings. This may be interpreted as a grossing-up procedure based on a quantity-price approach. A stratification of the housing stock is required to obtain a more reliable estimate and to include relative price differences properly. Since rents will vary with the characteristics of a dwelling, it would seem indispensable to take these into account. Subsequently, the average rent per stratum is applied to the number of dwellings in that particular stratum. If the available information is derived from sample of dwellings in that particular stratum. If the available information is derived from sample surveys, the grossing-up relates to both a part of the rented and all owner-occupied dwellings. The detailed procedure to determine a rent per stratum is usually carried out for a base year and this result is then extrapolated to the current years.

The difficulties with this method, where for certain strata of owner-occupied dwellings an actual rent is missing, can probably be overcome in most cases by applying more sophisticated statistical methods, like regression techniques, or an extended version of the Dutch points system. Obviously this does not solve the problem in the extreme case of all dwellings being owner-occupied. As an objective assessment for such a case the user-cost method could be applied. Briefly, this consists of adding relevant cost items, like intermediate and capital consumption as well as some allowance for the net operating surplus, including the interest on mortgages. Details of this method should be established by the GNP Management Committee as soon as it is required.

Principle 1:

To compile the output of dwelling services Member States shall apply the stratification method based on actual rents, either by direct extrapolation or by means of econometric regression. In respect of owner-occupied dwellings, this implies the use of actual rents for similar rented dwellings. In the justified and exceptional case where actual rents are missing or statistically unreliable for certain strata, other objective methods, like the user-cost method, may be employed.

1.2. Stratification of the housing stock

1.2.1. Factors affecting the rent level

A first set of variables determining the level of actual rents concern the characteristics of the dwelling and the building. To start with, the size of a dwelling will be important, both in terms to the area and the number of rooms. The bigger the dwelling, the higher will be the actual rent. However, the rent per square metre tends to fall with the size of a dwelling. Another important factor will concern the amenities of a dwelling. These may cover variables like the existence of a bathroom, a balcony/terrace, special flooring or wall covering, an open fire, central heating, air-conditioning, special glazing and other noise or thermal insulation measures; the layout of the dwelling is also relevant. With regard to the building, certain facilities may have an influence like a garage, a lift, a swimming-pool, a (roof-) garden or even the position of a dwelling within the building. In addition, the type of the building (detached, semi-detached house, flat), the architecture, the age, or the number of dwellings in a building may also affect the rent.

A second set of variables relate to environmental characteristics. A well-known factor is the rent difference between a comparable dwelling in a city and a remote location. The distance to an economic centre or the form of the landscape (flat land, mountainous) may not be negligible factors. In addition, neighbourhood factors like the view, surrounding green areas, transport facilities and access, shops and schools or the reputation and security of a district tend to have an influence on the actual rent.

Another set of variables may be summarized as socio-economic factors. For instance, in most countries rents are influenced by general government regulations like rent restrictions or subsidies. Further, the age of the tenancy agreement, the type of contract (temporary, permanent), the number of inhabitants per dwelling (flat-sharing community), the type of owner (public, housing association, private, employer) or the rent policy of the landlord may also affect the rents.

Obviously, several additional variables may have an influence on rents. But to collect all the abovementioned factors could lead to overburdened questionnaires. Therefore the use of capital values for stratification purposes may be considered. The rationale behind the use of the capital value of a dwelling is that it reflects all its important characteristics. The capital value can thus be considered as an implicit stratification factor. Using the ratio of the capital value to actual rent may be regarded as a feasible approach, especially in those countries where the rented dwellings represent a minor part of the housing stock. Provided the ratio is stable, such a method would enable the rental value to be determined for those dwellings which only feature in the owner-occupied sector. In addition, capital values do not exclude the use of 'physical` stratification criteria; they may be combined. In this situation, capital values are assumed to reflect the missing 'physical` stratification criteria. In any case, capital values to be used in the rents calculation have to be based on an objective assessment established for an up-to-date reference year.

In practice the stratification differs from one Member State to another in respect of both the number of strata and the precise criteria used to define them. Although at first sight this may cause some concern, it should be stressed that certain basic criteria, like the size and the (geographical) location of a dwelling are used almost everywhere. Moreover, it should in general be accepted that the aptness of other features will vary between countries and that Member States themselves are in the best position to determine significant criteria. This in turn requires that a basic analysis is carried out thoroughly on the actual rents used to calculate the value for a stratum.

Principle 2:

For stratification purposes, Member States shall use important features of the dwellings. These may relate to the characteristics of the dwelling and the building, to environmental characteristics of the dwelling or to socio-economic factors. In addition, the use of up-to-date capital values is acceptable for stratification purposes, if these are based on an objective assessment.

1.2.2. Selection of stratification criteria

Given the various characteristics affecting the rent of a dwelling, the first task is to investigate those variables having a significant impact. One way of detecting, significant variables is to produce a tabular analysis of the statistical information available. This method probably forms the basis of present procedures in most Member States. To obtain an objective measure of assessment, the variance of the actual rents within a stratum would seem to be useful. This would create an incentive for possible improvements to the stratification by choosing strata in order to minimize the within-stratum variance. It is therefore recommended that the variance per stratum is calculated at least in those cases where the stratification affects the level of both actual and imputed rents.

A more sophisticated approach is offered by advanced statistical techniques like (multiple) regression analysis. Such a technique allows the influence of individual variables to be assessed, so that the variation in rents can be assigned to certain characteristics. Briefly, the explanatory power of a variable can be quantified (via the correlation coefficient). As a by-product, it enables the characteristics to be ranked according to their importance. This helps to determine where to stratify in greater detail. Combining the most important variables, using multiple regression techniques, shows their overall explanatory power. The use of advanced statistical techniques to select important variables is considered to be an efficient way of stratifying the housing stock. In addition, such techniques would seem to be a useful tool for estimating the average rent in cases where there are no corresponding observations in the rented sector (empty strata).

A further advantage of selecting stratification criteria on the basis of an advanced statistical technique is that it avoids the need to prescribe uniform criteria for all countries. To obtain a comparable result, it is sufficient to establish a ranking of the most important criteria in each country and to stipulate the required overall level of explanatory power. Obviously, such a regression analysis depends largely on the available statistical information. However, even in a situation of restricted statistical information, this could be an incentive for future improvements.

Given that the information about the different variables affecting rents mainly depends on the development of basic statistics, the possibility of using advanced statistical techniques may be restricted at present. Therefore a standard method is recommended, i.e. Member States shall apply all significant criteria as derived from tabular analyses. As a minimum, the size, the location and at least one other important feature of a dwelling have to be used to stratify the housing stock; this stratification should produce a minimum of 30 cells. The breakdown of the housing stock has to be meaningful and representative of the total stock of dwellings. An advanced statistical technique may be used to determine the important explanatory variable(s) for selecting the strata.

In practice, however, a Member State may prefer to use fewer variables, or variables other than those prescribed by the standard method. This is acceptable, as long as a (multiple) regression analysis has previously been carried out, showing that an acceptable level of explanatory power has been obtained. To guarantee comparable results, a correlation coefficient of at least 70 % is recommended as a threshold. This threshold value would be acceptable in the context of a large sample, with zero and cheap rents as well as outliers having been removed.

Principle 3:

Member States shall use tabular analyses or statistical techniques to derive significant stratification criteria. As a minimum, the size, the location and at least one other important feature of a dwelling have to be used. A minimum of 30 cells are to be produced and at least three size classes and two types of location shall be distinguished. The use of fewer or other variables is acceptable, if it has been proved previously that the (multiple) correlation coefficient reaches 70 %.

1.2.3. Actual and imputed rents

In general, the rent is defined as the price due for the right to use an unfurnished dwelling. Charges for heating, water, electricity etc. should therefore be excluded in principle, although it may sometimes be difficult to separate them out in practice. To be in line with the valuation rules of the ESA, the output of housing services should exclude invoiced value added tax.

With regard to actual rents, certain public supports are probably more important. For instance, a specific household as a consumer is entitled to a general government transfer (e.g. housing benefit) but which, for administrative reasons, is paid directly to the landlord. Depending on the information source, the actual rent figure may differ. If the information source is the tenant, it may therefore be necessary to correct the actual rent observed by adding back any specific rent allowance.

In addition, the use of actual rents for imputation purposes requires the clarification of several fundamental questions having an impact on the harmonization of the data. The first point relates to the question of whether use should be made of all actual rents or only those from new contracts, for the imputation procedure. Depending on the purpose, different theoretical arguments can be put forward supporting actual rents paid according to new contracts, contracts signed in the construction year, or 'average` contracts. Applying the general rule, i.e. to use rents of similar dwellings, it does not seem acceptable to limit the basis of imputation to rents from new contracts. Given that 'average` rents are used for the rented sector, the same should apply for owner-occupied dwellings. In addition, a different solution would probably pose great difficulties in many countries when applying the stratification method. Briefly, the conclusion is that the average actual rents from all contracts should be used to calculate imputed rents.

The second question concerns the problem as to whether rents of publicly-owned dwellings can be used for imputation purposes. Given that owner-occupied dwellings are mostly privately owned, in principle, only actual rents from the private sector should be used for imputation purposes. However, if not enough observations of actual rents of privately-owned dwellings are available to constitute a sufficient basis for the imputation, exceptionally, rents of public-owned dwellings may be used, provided they are increased for any subsidies paid only to public and not to private housing.

A further question relates to the use of rents from furnished dwellings to enlarge the basis for imputed rents. In principle, the basis for imputing a rental value for owner-occupied dwellings is unfurnished rents. Therefore, rents for furnished dwellings cannot be used directly. In order to avoid imputing the wrong level of rent, they should be scaled down to exclude the payment for the use of the furniture.

Principle 4:

The actual rent shall be understood as the rent due for the right to use an unfurnished dwelling. If the information source is the tenant, it may therefore be necessary to correct the actual rent observed by adding back any specific rent allowance, which for administrative reasons is paid directly to the landlord. To compile imputed rents, actual rents from all contracts shall be exploited relating to privately-owned dwellings. If necessary for statistical reasons, rents of publicly-owned dwellings may, exceptionally, be used, provided they are increased for any subsidies paid only to public and not to private owners. Similarly, rents from furnished dwellings may be included in the imputation basis after deduction of the rent differential between furnished and unfurnished dwellings.

1.3. Sources for the base year estimate and methods of extrapolation

1.3.1. Housing stock

An essential element of the calculation according to the stratification method is the information on the stock of dwellings. This information serves as a reference universe for extrapolation procedures. In general terms, the housing stock consists of all buildings or parts thereof which are used as dwellings. More details are given in the section on special problems. The major sources used throughout the Community to establish such a housing stock are building censuses, administrative building registers or population censuses. Mostly, the base year figure is updated to arrive at the current year estimate.

With regard to the base year housing stock, building censuses seem to be the least problematic and the most complete, especially when carried out together with a population census. Administrative building registers depend largely on legal procedures, which may cause uncertainties, for instance, about whether extensions, improvements, conversions and demolitions of dwellings are recorded properly. Using the information provided by households in a population census as the basis of the housing stock may cause problems because the results tend to underestimate second homes which are not occupied at the census date.

Principle 5:

For the compilation of the base year housing stock, Member States shall exploit either a building census or a population census or an administrative building register as an initial basis. Since a building census usually provides the highest degree of completeness, the use of administrative building registers and population censuses requires intensive and thorough checks to obtain exhaustiveness.

1.3.2. Actual rents

The second fundamental element for the calculation of the output of dwelling services, according to the stratification method, concerns the actual rents paid in the rented sector. Information on actual rents in the base year is either derived from a census (e.g. population census) or from a sample survey like family budget surveys. In the first case, actual rents are probably covered in total and the calculations only have an impact on the level of imputed rents. In the case of family budget surveys, the calculations affect the level of both actual and imputed rents. Obviously a census gives a broad basis for reliable information. But family budget surveys are normally also considered as fairly reliable, especially with regard to essential goods. However, differential non-response is known to be a general problem with this type of survey. If housing is considered as more of a luxury than as an essential item, this problem would have an undesirable impact on the results of the rents calculation, and should be offset. Another problem of a family budget survey, at least in some countries, concerns its small size which may restrict the possibility of stratifying the rents. In any case, available supplementary sources should also be exploited as far as possible. This may, for instance, be the case in countries where a high proportion of housing is under public control and housing agencies have to present accounts. Further, as an on-going task to improve the results, alternative sources, like specialized rent surveys, should be investigated.

Principle 6:

Member States shall exploit the broadest and most reliable sources to derive actual rents per stratum, for example a population census or a household survey. Alternative sources should be assessed to improve the reliability and exhaustiveness, and particularly the stratification.

1.3.3. Extrapolation of base year results

Only few Member States have the annual information needed to carry out the calculation of the output of owner-occupied dwellings each year afresh. In most countries the results for a given year are taken as the benchmark and subsequently updated to estimate the current year figure by means of indicators. In practice, differences seem to exist in so far as some countries proceed by updating the (total) base year output using a combined indicator whereas others separately extrapolate the housing stock and the rent per stratum. Although similar results could be expected in general, structural shifts, e.g. in the relationship between rented and owner-occupied dwellings, may cause differences. In addition, a separate calculation would permit plausibility checks.

With regard to the indicators used, the quantity index is mostly derived from the production of the construction industry. The price indicator on the other hand, is often based on the price index of rents paid from the consumer price index. This may cause distortions in those cases where the assumption that imputed rents follow the movement of the total is not justified, for instance, due to public rent controls. For the extrapolation of imputed rents, it therefore seems preferable to use, as in the base year, a price index reflecting the movement of privately rented dwellings. Further, attention has to be drawn to the fact that price indices will normally exclude price increases due to quality changes. The price indices therefore have to be supplemented by a quality indicator reflecting improvements.

Finally it seems useful to minimize the impact of structural changes on the results by restricting the extrapolation period. Taking account in this respect of the periodicity of the relevant basic statistics it seems appropriate to carry out a benchmarking of the housing stock every 10 years, i.e., the normal interval for population censuses. In addition, the benchmarking of the price element (rent per stratum) should be carried out at least every five years, i.e., the usual periodicity for family budget surveys.

Principle 7:

If it is not possible to carry out complete re-estimation of the output of dwelling services annually, Member States may extrapolate a given base year figure using appropriate quantity, price and quality indicators. The extrapolation of the housing stock and the average rent shall be carried out separately for each stratum. The extrapolation procedure shall distinguish between the calculation for actual and for imputed rents. If necessary, the number of strata used for extrapolation may be less than that used for the base year calculation. To extrapolate the imputed rent for owner-occupied dwellings, in general, a price index reflecting private rents shall be applied. In any case, the benchmarking of the housing stock should not exceed 10 years and that of the price element should not exceed five years.

1.4. Special problems

1.4.1. Rent-free and cheap dwellings

When collecting data on actual rents, sometimes zero or very low values will be observed. In the case of rent-free dwellings this leads to the strange situation that the dwelling service is actually provided but without a (visible) payment. It seems appropriate in such cases to adopt the solution that the observed actual zero rent be corrected. An analogous solution would appear logical for cheap dwellings.

Apart from interventions by general government, there are other reasons why rent-free or cheap dwellings may be observed. One example is an employee occupying an employer-owned dwelling at a reduced or zero rent. This may concern all kinds of employees, including housekeepers or guardians. In this case the actual rent has to be corrected and the difference between the actual and comparable rent will be treated as remuneration in kind (see ESA 1979 paragraph 408j). Another possibility is that dwellings are let at a zero or very low rent to relatives or friends. In this case the correction may be obtained by simply reclassifying those dwellings from the rented to the owner-occupied sector. Further, a similar correction seems appropriate in the case of lump sum payments by tenants, i.e., where the tenant makes a pre-payment of the rent for a longer than normal period.

Principle 8:

The actual rent observed in the case of rent-free and cheap dwellings shall be corrected to include the full dwelling service. Neither zero rents nor cheap rents shall be used to calculate imputed rents, on an uncorrected basis.

1.4.2. Holiday homes

Holiday homes cover all kinds of leisure-time dwellings like the nearby weekend-house which is used for short periods many times a year or the more distant resort-home which is used for longer periods but only few times a year. At first sight the case of rented holiday homes seems no problem, since the actual rent paid is taken as a measure for output. However, if the actual rents collected are on a monthly basis, the extrapolation to the yearly total may lead to over-estimates if no supplementary information on the average occupation time is included.

To calculate an imputed rent for owner-occupied holiday homes, the most logical approach is to stratify these properties and to apply the appropriate average annual rent for actually rented similar accommodation. The annual rent implicitly reflects the average occupation time. In the case of difficulties, a substitute method may be applied, i.e., to collect information on holiday homes in one stratum and to apply the average annual rent for actually rented holiday homes to homes in one stratum and to apply the average annual rent for actually rented holiday homes to the owner-occupied ones. Thirdly, it is acceptable to use the full annual rent for ordinary dwellings, in the same stratum of location, where holiday homes account for a very minor part of the housing stock or where they cannot be separated from other dwellings. Even in the case of resort homes these procedures seem reasonable when taking into account that they are always available to the owner and will also be used free of charge by his friends or relatives.

With regard to the activity classification of holiday homes, i.e., on the question of whether to allocate them to the dwellings branch or the hotels/restaurants sector, there are two opinions. On the one hand, it seems that the NACE classification would include activities related to holiday homers under hotels and restaurants. On the other hand, it could be useful to group all imputations together. In any case, if the third solution above is applied, a separation between holiday homes and other dwellings is difficult. Since from the point of view of the GNP level this question appears not to be of first priority, it could be decided upon by other expert groups.

Principle 9:

Holiday homes cover all kinds of leisure-time dwellings like the nearby weekend-house or the more distant resort-home. To estimate the output of holiday homes the annual average rents of similar facilities shall preferably be used. The annual rent implicitly reflects the average occupation time. Although stratification would seem desirable, holiday homes may be grouped in one stratum. If holiday homes account for a very minor part of the housing stock, the full annual rent of ordinary dwellings, in the same stratum of location, may be used. In the justified exceptional case of missing or statistically unreliable actual rents for certain strata, other objective methods, like the user-cost method, may be employed.

1.4.3. Time-share

In order to facilitate a common solution for the treatment of time-share properties it seems useful to recall the basic features of this new type of accommodation. In the case of time-share, a real estate agent sells the right to stay for a fixed period each year in a certain dwelling located in a tourist area and takes care of the administration of this property. The right is guaranteed by a certificate, which is issued after the initial payment. This certificate may be traded at the current price. Periodic payments are further due to cover administrative costs.

From this description it follows that the initial payment should be treated as an investment since the certificate issued is similar to a share. This is supported by the fact that at least in one Member State's law the purchaser acquires a real right. Thus it would seem useful to include the initial payment under intangible assets in the national accounts. Further, it would seem logical to consider the rent-free accommodation service as a dividend in kind paid by the real estate agent.

The fundamental problem is that a service is actually provided by the time-share accommodation which is not included in the output of the economy. Logically this requires a correction. To start with, the proposal to accept the periodic payment as a proxy implicitly means that no correction is made for the accommodation service, since the periodic payment covers a different service, namely management costs. Another theoretical possibility would be to consider the initial payment as a prepayment of the service provided and to distribute it over the relevant periods of occupation. Apart from the statistical problems of putting this model into practice, there seems to be a contradiction in legal terms since the implicit interpretation is a purchase of a service and not the acquisition of an asset.

A further possibility consists of deriving a proxi from annual actual rents for similar (self-catering) accommodation facilities. This solution is supported by the fact that time-share accommodations are located in tourist areas and coexist with actually rented holiday apartments. In the case of difficulties, the two other methods proposed for holiday homes are also acceptable for time-share properties. The imputed rent should be on a net basis, to avoid double-counting charges covered by the periodic payment. In any case, the imputation of a rent for time-share properties requires at the same time an adjustment to the income and expenditure approach.

With regard to the industry classification, time-share accommodation can either be treated as a second dwelling (housing industry) or as a hotel-type accommodation. It seems that time-share accommodations have characteristics of both the housing and the hotel-restaurant industry. Since from the point of view of the GNP level this question appears not to be of first priority, it could be decided upon by other expert groups.

Principle 10:

With regard to time-share accommodation, the same procedures shall be applied as for holiday homes.

1.4.4. Spare room lodgers

In most countries a large number of students are accommodated in spare rooms. Often this extends to other younger people or those who are employed in a job which involves being away from home. If the room is a part of a rented dwelling, i.e., is sublet, no major problems seem to exist. The spare room rent can be considered as a contribution to the actual main rent, i.e., a transfer between households. However, if the room is a part of an owner-occupied dwelling it would be double counting to include both the rent that the lodger pays as well as the imputed rent in its entirety. Probably the correct solution would be to take the actual rent paid by the lodger for the percentage of the dwelling he occupies and to impute a rent for the rest. However, this may not be practicable to implement. Instead one could consider the rent as a transfer involving the sharing of the expenses of the dwelling. This would be similar to the first case in so far as the actual spare room rent is considered as a contribution to the imputed main rent. As a consequence of this treatment, a correction will be due if the household sector is broken down by groups.

A further point is how to deal with subletting various rooms. In this case, it is suggested that the term spare room lodgers only applies when the owner or the main tenant himself also continues to occupy the dwelling. Otherwise, the subletting should be considered as a separate economic activity (housing service or pension).

Principle 11:

Rents paid for spare rooms within a dwelling shall be considered as a contribution to the main rent as long as the owner or the main tenant continues to occupy the dwelling.

1.4.5. Empty dwellings

First, a rented dwelling is always considered as occupied even if the tenant chooses to live elsewhere. In accordance with paragraph 315 i of ESA 1979, the value of the rent is considered to be the output. Secondly, in line with the general solution agreed for holiday homes and timeshare accommodation, the annual rent implicitly reflects the average time of occupancy. The problem of empty dwellings is therefore restricted to non-rented dwellings which are not used by the owner i.e., which are available to be sold or rented. In such cases no dwelling service is provided, so a zero rent should be inserted.

The necessary information to determine whether a non-rented property is empty or not can be based on the declaration of the owner or neighbours. In the absence of such information, the existence of furniture may be used as an indication that the property is occupied. In contrast, unfurnished dwellings can be considered as empty, since it is difficult to imagine that a housing service is being provided. Empty dwellings should also include dwellings which are repossessed following defaults of payments, or which are empty for a short period because a housing agency does not immediately find a new tenant. A borderline case is an empty dwelling which is fully furnished and can be used by the owner immediately. Here one might argue that no housing service is provided as long as it is not actually occupied by the owner. But since it is comparable to the case of a rented but empty dwelling, it seems appropriate to insert a rent. Therefore, furnished owner-occupied dwellings are generally regarded as occupied.

Finally, it should be noted that an empty dwelling may still incur costs, like current expenditure on maintenance, electricity, insurance premiums, taxes etc. These should be included under intermediate consumption of the housing industry. As in the case of an enterprise not producing any services, this may lead to a negative value added.

Principle 12:

For a non-rented dwelling, which is available to be sold or rented, a zero rent shall be inserted. A furnished owner-occupied dwelling in general shall be treated as an occupied dwelling.

1.4.6. Garages

Since some countries argued that the treatment of garages is not totally clear in the ESA 1979 (nor the SNA 1968), a proposal to clarify the situation seems useful. It may be recalled that the Member States responding to an earlier questionnaire were unanimously in favour of including garages in the output of dwelling services, if they are normally associated with the dwelling. This should include parking places since they probably have the same function.

An ESA condition for classifying durable goods as gross fixed capital formation is their use in the process of production. Durable goods not used for production purposes, e.g. consumer durables, are to be recorded as final consumption expenditure. Conversely this means that items included in gross fixed capital formation in general lead to a subsequent output. Since garages are a part of gross fixed capital formation, it would seem appropriate not only to include the service of the rented ones in the output of the economy but also to calculate an imputed output for owner-occupied garages. In both cases the garage represents an element of comfort of the dwelling like any other facility.

However, footnote 1 to paragraph 315j of ESA 1979 stipulates that 'the value of the imputed rents on non-residential buildings used directly by their owners is not accounted for separately.` Consequently for owner-occupied garages a rent has to be imputed only if they are considered as residential buildings. In practice, several cases may be distinguished. First, a garage is an integral part of a residential building. In such a case where a physical separation is not possible, the whole building will be classified as residential and an imputed rent should be calculated for those owner-occupied garages. In the second case of a separate owner-occupied garage, one may distinguish between those used in connection with the dwelling and those used for other purposes (e.g. parking near the workplace). Both will probably be classified as a non-residential building. In accordance with the abovementioned footnote, no rent has to be imputed for the use of separately located, owner-occupied garages.

In any case it should be stressed that usually there are more owner-occupied dwellings than rented dwellings with a garage. To include this structural difference properly, the best way seems to use the existence of a garage as a stratification criterion.

Principle 13:

Garages and parking places provide services to be included in dwelling services if structurally integrated in the dwelling.

2. INTERMEDIATE CONSUMPTION

Intermediate consumption of dwelling services is dealt with on a more precise level in the ESA input-output methodology. From a practical point of view problems seem to arise concerning the treatment of certain (communal) charges as well as repair and maintenance.

With regard to certain charges, the ESA 1979 input-output methodology (2) requires that charges for heating, water, electricity, common areas, security, lifts, etc. are excluded from intermediate consumption (and hence output). In practice, however, many countries include these on the basis that they are considered to be part of the rental service and, quite often, cannot be separated out. It is recommended that the current ESA rule be adhered to in order to obtain a parallel treatment of rented and owner-occupied dwellings and to avoid double-counting for owner-occupiers. On the other hand the GNP level should not be affected, if a gross treatment is carried out consistently for intermediate consumption and output.

With respect to repair and maintenance, the ESA 1979 is clear in so far that major repairs, defined as those improving, prolonging the life of, or reconstructing the property in question, should be allocated to capital formation. In connection with routine repairs, the input-output methodology of the ESA states that these 'should be included in the intermediate consumption whenever such expenses are borne by the owner, irrespective of whether the accommodation is let or owner-occupied. However, where these expenses are borne by tenants to whom the accommodation has been let, they should be recorded directly in the final consumption of households`. This implicitly means that intermediate consumption of an owner-occupied dwelling should be a bit higher than that of a rented dwelling, since it includes that expenditure regarded as a tenant's repair in the latter case. It seems that some countries depart from this rule and include, after having established output at market prices, repairs borne by tenants in the output and intermediate consumption of dwelling services, and then channel it through output to final consumption. This may be accepted if it has no impact on the level of GNP.

In addition to the point raised under empty dwellings, it is stressed that double-counting of intermediate consumption in the case of employer-owned dwellings should be avoided.

Principle 14:

Intermediate consumption of dwelling services shall be established in line with the definition of output. In general both items should exclude charges for heating, water, electricity etc. If for practical reasons a different treatment is preferred, this is acceptable as long as the levels of GDP and GNP are not affected.

3. TRANSACTIONS WITH THE REST OF THE WORLD

According to ESA 1979 rules (paragraphs 211 and 214) non-resident units are considered as notionally resident units in their capacity as owners of land or existing buildings on the economic territory of the country, but only in respect of transactions affecting such land or buildings. This simply means that the service provided by a dwelling which is owned by a non-resident is included in the output of the economy where the dwelling is located. However, in theory this is corrected during the transition from GDP to GNP under the heading property and entrepreneurial income (R 40). In paragraph 430 b of ESA 1979 it is confirmed that this item includes the net rent received by residents as owners of land and buildings in the rest of the world or vice versa. It represents the interest on the claim of the resident unit against himself in his capacity as a notional resident unit of the rest of the world.

In general few problems arise if the property owned by a non-resident is actually rented to a resident, since a monetary flow will be observed and included in the balance of payments. The situation, however, may be different in the case of owner-occupiers, since a similar monetary flow will probably not be observed. One possible solution would be to consider that when a resident owns and occupies a dwelling abroad he is in fact considered to be a resident of the country where the dwelling is located. Possibly this solution would not be favoured by countries where a lot of dwellings like holiday homes are occupied by non-residents. The second possible solution would be to introduce a correction to the item transactions in net rents with the rest of the world, as far as owner-occupied dwellings are concerned. This requires that a cross-table of dwellings owned and occupied by foreign residents is compiled, and reflects the fact that the nationality of a person is not sufficient to separate a resident from a non-resident. In addition, it would seem useful to obtain an agreement about the cross-table between the concerned Member States. However, a general problem is caused by statistical lacunae in this field. Although holiday homes owned by foreign residents matter most in this respect Member States explained unanimously that at present such information is not available.

A special issue in this regard relates to timeshare properties. Since for the same accounting period such a property may be occupied by residents of different countries, a direct allocation to the country of origin seems almost impossible. But comparable results may be obtained using a more feasible approach. First, the (imputed) value added generated by time-share accommodation is allocated to the country of origin of the owning company. Subsequently the owning company may be asked to provide a distribution by country of origin of the time-share owners, which may serve as a distribution key.

Principle 15:

According to ESA 1979, all dwellings on the economic territory of a Member State contribute to its GDP. The net rent received by non-residents as owners of land and buildings in that country are to be recorded as property income to the rest of the world and therefore to be deducted from GDP during the transition to GNP (and vice versa). The net rent shall be understood as the net operating surplus from actual and imputed renting of dwellings. To take account of owner-occupied dwellings owned by non-residents, an exchange of information between the Member States is required. The coordination of this information would be taken on by Eurostat, especially in connection with the compilation of a cross-table showing these dwellings by foreign residents.

(1) Until recently, one Member State had at its disposal an objective assessment of the imputed rent for owner-occupied dwellings, i.e., the rateable value which was a potential rent assessed by tax authorities.

(2) See Statistical Office of the European Communities, 'Community methodology for input-output tables 1965` (only existing in French and German).