Instituto Nacional de Estatística, Statistics Portugal

National Accounts Department/ Price Statistics Unit 

Av. António José de Almeida
1000-043 Lisboa
Portugal

The House Price Index (HPI) measures the evolution of the acquisition prices of the dwellings purchased by households in the residential market in Portugal.

The indexes are based on transaction prices and are compiled using fiscal administrative data provided by the Tax and Customs Authority on the Municipal Property Transfer Tax (IMT) and the Municipal Property Tax (IMI). The value of land is included in the prices and weights of the HPI. An overall indicator and two sub-indexes are provided in this domain. The sub-indexes cover the following aggregates:

• Purchases of newly built dwellings;
• Purchases of existing dwellings.

Indicators on the number and value of transactions are also provided together with the release of the HPI.

It is important to note that the detail of the information used to calculate the index differs throughout the available series. Before the 2^nd quarter of 2019, the administrative data used did not allow to know the institutional sector of those involved in the transactions, which means that it wasn´t possible to identify purchases made exclusively by households. The availability of new and more detailed data allowed some changes and for the period after the 1^st quarter of 2019, the HPI is fully in line with its basic and framework regulations (see section 6.1), i.e., being calculated exclusively based on purchases by Households.

The HPI is compiled using a classification system that separates newly built from existing dwellings.

The index reflects the price change associated with the purchase of residential properties made by the households. See section 3.1 for additional details.

The data source supporting the compilation of the price indexes provides information on transaction prices and covers the population of residential sales in Portugal (all dwelling transactions are covered). The transaction enters in the index in the quarter in which the IMT is paid, which typically is done some days before or at the moment of the property transfer. As mentioned above (section 3.3), the index reflects the price change associated with the purchase of residential properties made by the households. In total, three indexes are compiled on a quarterly basis: i) an overall price index and HPIs covering the (ii) existing and (iii) newly built dwelling market segments.

Residential dwelling unit.

The target population of the HPI comprises the residential structures in Portugal purchased by households (until the 1^st quarter of 2019 includes all sectors purchasing – see section 3.1).

The indicator covers the whole country.

The dissemination of the HPI started in July 2014. Data are available since the first quarter of 2008 for all index series. Since IMT data are only available from the beginning of 2009 onwards, 2008 indexes were estimated using 2008/2009 year-on-year rates of change derived from bank appraisals data (taken from a survey on commercial banks providing 90% of all mortgage credit in Portugal). The series of the number (and value) of dwelling transactions has its start in the first quarter of 2009.

The base 100 of the index is 2015 (2015=100).

The following units are used:

• index number, with 2015 as its reference year;
• Percentage change on a previous period (rates);
• no, number of transactions that occurred in the reference quarter.
• euros, value of transactions that occurred in the reference quarter.
• Percentage share of the total and in national currency (weights).

Quarter.

European:

- Regulation (EU) 2016/792 of the European Parliament and of the Council of 11 May 2016 on harmonised indices of consumer prices and the house price index;
- Commission Implementing Regulation (EU) 2023/1470 of 17 July 2023 laying down the methodological and technical specifications in accordance with Regulation (EU) 2016/792 of the European Parliament and of the Council as regards the house price index and the owner-occupied housing price index, and amending Commission Regulation (EU) 2020/1148.

National:

- There are no national legal acts related to the HPI.
- The production of the HPI is included in the annual work programme of Statistics Portugal approved by the National Statistical Council.

Not applicable.

The HPI is based on administrative data (provided by the Portuguese Tax Authority, based on a cooperation protocol between the two institutions).

Statistical confidentiality is one of the basic principles of the Law of the National Statistical System (Law No 22/2008 of 13 May). According to its article 6, all individual statistical data collected by statistical authorities is to be treated as confidential. This Law and other relevant national legislation on this topic (e.g. Law No 67/98 of 26 October on the protection of individual data) can be obtained on the website of Statistics Portugal (documents available only in Portuguese). As any statistic produced by Statistics Portugal, the HPI is subject to these laws.

Access to administrative microdata is limited to the staff in charge of the compilation of the HPI. No information is available to identify or locate the dwelling or the agents involved in the transaction.

The release dates of the HPI are available in advance on the website of Statistics Portugal.

An advance release calendar is available on the website of Statistics Portugal (see 8.1.).

HPI (and House Sales) data and dissemination formats are made available at the same time to all users through the institutional website of Statistics Portugal.

The dissemination policy of Statistics Portugal lays down the fundamental principles governing the dissemination of official statistics, directly or indirectly produced under its responsibility. It should have as main reference the applicable principles of the National Statistical System: technical independence, statistical confidentiality, impartiality and accessibility.

Quarterly.

The HPI is disseminated through an on-line press release (Destaque), which is made available on the website of Statistics Portugal (www.ine.pt).

Annual publications Construction and Housing Statistics and Statistical Yearbook.

A file, with HPI (and House Sales) data, is made available at the time the quarterly news release is disseminated. Information is also available in an online database.

Individual data are to be treated as confidential (cf. 7.1.). Researchers have access to the HPI micro-data base if they comply with the rules defined by Statistics Portugal.

Not available.

The methodology of the HPI (in Portuguese) is available on the website of Statistics Portugal.

Not available.

The general quality assurance conceptual framework is provided by Regulations no. 2016/792, 2023/1470 and by the RPPI international handbook. The quality framework that is used for the evaluation of the HPI is completed by the set of HICP reference documents, which provide additional guidance against which the quality of produced indicators are routinely checked.

Ensuring the quality of procedures involves the following aspects:

- Validation of the information received on a monthly basis. This is done analysing the match quality between databases, but also by assessing the size of the received flows, as well as validating the information received through descriptive analysis by variable.

- Definition of rules to include or exclude transactions.

- Processing of the location variables of the properties by the geoinformation department.

- Comparison of the results obtained with different methodologies (with and without quality adjustment) and alternative data sources

- Regular evaluation/revision of hedonic model specifications.

- Compilation and regular analysis of the HPI results with associated indicators of the national real estate sector.

No explicit quality assessment has been performed.

The HPI responds to the needs of those users interested in making comparable comparisons of housing price developments among EU member states and is an important indicator for the early detection of macroeconomic imbalances in EU member states. Finally, the HPI responds to the needs of all users interested to know how prices evolve in the Portuguese housing market.

Not available.

The index covers the whole country and all dwelling market transactions. It covers the expenditure on residential units purchased by households. See sections 3.4 and 3.6 for more details.

The overall accuracy of the index is good. The price corresponds to the transaction prices declared for IMT purposes. The degree of complexity and cross-checking of information by the Tax and Customs Authority means that the prices declared are very close or similar to the real transaction prices. The consistency of the results obtained using different methodologies is indicative of the quality of the procedures in place.

None. The HPI is based on an administrative data source, which covers all market transactions carried out in Portugal.

Data sent by the Tax and Customs Authority is routinely checked through the help of routines and procedures that help identifying coding and editing errors. A review of the ongoing automation procedures is carried out on a regular basis.

 The HPI is produced and transmitted in accordance with data transmission requirements established in article 7 of the regulation (EU) 2016/792 of the European Parliament and of the Council of 11 May 2016. House price index numbers (and House Sales indicators) are sent up to 85 days (on average 83 days) after the end of the reference quarter and weights are sent before 15 June of the year following the year to which the weights relate. The HPI is published approximately two months and a half after the end of the reference quarter. The HPI is released nationally and sent to Eurostat on the same day.

The HPI is produced and provided on time and always according to the previously defined schedule. Data files are sent in line with the transmission calendar agreed with Eurostat.

HPI data is comparable with the HPIs produced in other EU countries. The Portuguese HPI does not have a regional breakdown.

Data are comparable over time. Take note of the considerations made in sections 3.1 (Data description) and 3.6 (Statistical population).

The coherence of produced HPIs is regularly checked against other sources of information on the housing market (e.g., bank appraisals data).

All figures are checked for internal consistency. Higher-level aggregations are derived from elementary aggregates according to index chaining practices.

One senior statistician is in charge of the compilation of the indicator on a full-time basis, together with the OOHPI and House sales. In addition, the production of the HPI is supported by an IT expert and an expert in geo information, both on a part-time basis. Assistants and/or administrative staff are employed whenever it is necessary. The senior statistician participating in the production of the HPI has more than 10 years of experience in the compilation of real estate statistics and participates in workshops and technical meetings on real estate price statistics.

The HPI and the House Sales indicator are not subject to routine revisions.

Not applicable (see 17.1.).

See below.

The data used in the compilation of the HPI (both indexes) is derived from the combination of two different fiscal administrative data sources.

HPI weights are based on dwelling transaction values (IMT data). The partition of the total HPI weight into newly built and existing dwellings is done using information available in the IMI data. That is, the weights correspond, in essence, to the indicator of the value of purchases of dwellings by Households.

 The indicators for the number and value of dwelling transactions are calculated from the same information flows used in the compilation of the Housing Price Index, that is, the IMT and IMI (see 18.1.1.1).

The IMT and IMI data, which is used in the compilation of the HPI, is transmitted to Statistics Portugal on a monthly basis. The data received in month t refers to all IMT and IMI fiscal processes that have been completed during month t-1 (data received in month t affects transactions in months t-1, t-2 and t-3). 

The fiscal administrative data (IMT and IMI) is provided in txt files using a safe data transmission channel, which has been defined by Tax and Customs Authority and Statistics Portugal. The information is received on the 11th of each month.

The overall quality of received data is considered to be good, with variables not showing many cases with missing information, abnormal values or typing mistakes.

A first validation is done when the information is received. Monthly, through automatic procedures, it is ensured that IMT and IMI flows have the defined characteristics. Then, the data is subject to monthly checks, which include the monthly validation of received raw data through the visual inspection of key variables (e.g., transaction prices, floor area), the use of descriptive statistics and exploratory data analysis.

Every quarter, robust regression (median regression) is employed to see regression results are influenced by the presence of outliers. For each primary stratum (existing apartments, new apartments, existing houses and new houses) transaction price thresholds were defined.

See below.

See below.

HPI is a Laspeyres-type price index.

The aggregation of the index is performed as follows:

1)      In a first step, the price changes for the elementary aggregates of the index are estimated by ordinary least squares using the adjacent time dummy method. Thus, for each pair of contiguous quarters (Q-1 and Q), four regressions, covering the following strata/elementary aggregates, are run: i) Existing apartments, ii) Existing houses, iii) New apartments, and, iv) New houses.

The value of the price index in quarter Q, in comparison with quarter Q-1, is obtained directly through the parameter estimate of the time dummy variable included as a covariate in the hedonic regression model (which assumes the value 1 when a transaction was carried out in quarter Q and 0 otherwise). The rate of change between Q and Q-1 of the dwelling prices for stratum j, is obtained by exponentiation of the estimated parameter for the time variable. The index of each of these (4) strata in Q is obtained by multiplying the index in Q-1 and the rate of change between Q and Q-1.

 2)      The next phase of the compilation of the HPI involves the aggregation of the indices of the four elementary aggregates into the newly built and existing HPIs.  The HPI sub-indices for newly built and existing, result from the calculation of a weighted arithmetical average, of the indices of new apartments and new houses, in the case of the newly built HPI and of existing apartments and existing houses, for the existing HPI.  Each of the two sub-indices, are obtained by the sum of two parts (for apartments and houses) calculated by dividing each quarterly value of year Q (computed in step 1) by the 4^th quarterly value of year Q-1 and multiplying by its weight.

 3)      Finally, in the third step it is compiled the HPI based on the indices obtained in step 2. The HPI is computed as a weighted arithmetical average, of the indices for newly built and existing dwellings. The same procedures described in step 2 are applied here, that is, each quarterly value of the (two) sub-indices of year Q are divided by the value of the respective indices for the 4th quarter of year Q-1 and then multiplied by the respective weights. The value of the HPI index in quarter i of year Q results from the sum of the two previous parts.

The elementary aggregate weights are based on data on household expenditure on transacted dwellings.

The reference period for weight is year t-1 (the weights that are used in year t, reflect the dwelling transaction prices that occurred in year t-1).

Weights are updated every year.

All elementary aggregate weights are adjusted with a price change between the year t-1 and the fourth quarter of the year t-1.

The HPI is a Laspeyres-type index that is chained on a yearly basis using the fourth quarter of the previous year as the linking period. 

Both indices, newly built dwellings and existing dwellings, are based on hedonic methods combined with stratification. The stratification variables, as described in 18.5.1.2, are property type (apartment vs. house) and dwelling category (new vs. existing).

 The hedonic model used is based on the following characteristics:

-  The hedonic method used is the rolling time-dummy, with two periods (adjacent periods). The value of the price index in quarter Q, in comparison with quarter Q-1, is obtained directly through the parameter estimate of the time dummy variable included as a covariate in the hedonic regression model (which assumes the value 1 when a transaction was carried out in quarter Q and 0 otherwise). The rate of change between Q and Q-1 of the dwelling prices for each stratum, is obtained by exponentiation of the estimated parameter for the time variable.

 -  Price changes for the index are estimated by ordinary least squares.

 -  The functional form of model is semi-log.

 -  The dependent variable is the natural logarithm of transaction price.

-   Independent variables used, at least, in one of the four regressions performed (see annex).

The SPAR method is not used.

As mentioned in 18.5.1.6.1 the stratification method is combined with hedonic methods.

See below.

Prices are recorded when the transaction tax is paid by the buyer, which is done on the same day or some days before the signature of the transaction contract and when the property effectively changes hands.

In Portugal, the VAT is not applicable to property transactions.

Taxes, other than VAT, are not included in the price of dwellings. In Portugal, transaction tax, stamp duty and other taxes/commissions are paid separately from the price paid for the acquisition of a property.

Not applicable. Residential property purchases are not subsidized by the Government.

The price of land is included in both prices and weights. House Sales indicators also include the price of land.

Not applicable / non-existent in our real estate market. This type of housing tenure, for which transactions refer to the transmission of a share of the right to occupy and use a certain housing unit in a cooperative, is inexistent in Portugal. (There are cooperatives for the construction of new dwellings; but the cooperative usually ends after the dwelling is finished).

In view of the above, they are not considered in the HPI and in the indicators of number and value of transactions.

Transactions with clearly low (below 20,000 Euros) or high (above 3,000,000 Euros) values are excluded through the definition of thresholds. It should also be noted that some transactions considered as "non-market" are excluded a priori from the scope of the index as they are classified with a specific legal taxable event and not considered in the indicator in question.

Transactions with more than one property are only considered if it is possible to identify the transaction value for each property. Situations where this is not the case are residual. In the equally infrequent situations in which an identical value was assigned to each property transacted, the hedonic model has variables capable of correcting and adjusting such situations.

The scope of the index and of the number and value of transactions indicators only covers properties that have been traded in their totality. This means that properties are considered to be sold as a whole property, or in any other fraction where the sum of these fractions makes up the unit, e.g. two halves, three thirds, or one quarter together with three quarters.

The identification of outliers is carried out in two ways. On the one hand, by defining thresholds for the value of transactions, minimum and maximum, which allow identifying "non-market" transactions, but also outliers. A second way to identify outliers is through the analysis of some measures obtained within the hedonic regression model. In particular, the observations that have large residuals (i.e. studentized residuals that lie outside the [- 3,3] interval) and that have a cook’s distance higher that 4 are dropped out from final regressions.

It should also be noted that transactions of properties with a gross floor area of less than 20 m2 are considered outliers.

No extrapolation techniques are foreseen in case of insufficient coverage of data sources for housing sales indicators. The protocol signed between Statistics Portugal and the Tax and Customs Authority, foresees only the availability of the universe of dwelling transactions and not partial availability of databases.

No adjustments other than quality adjustment are applied to the HPI data.

None.