Croatian Bureau of Statistics

Construction Statistics Department

Ilica 3

10000 Zagreb

Croatia

Index of Production in Construction (IPC) is designed to be an indicator of the economic cycle. In Croatia Index is based on the number of actually worked hours on construction sites. The IPC is broken down into two sub-indices: building construction and civil engineering. Data are collected by using the reporting method on the Monthly and the Quarterly Report on Construction (GRAĐ-21/M and GRAĐ-21/3M forms).

NACE Rev.2. is used. The breakdown of the indicator into building and civil engineering works is based on the Classification of Types of Constructions (CC).

Construction covers economic activities listed in section F of NACE Rev. 2. – Construction (divisions 41, 42 and 43). The breakdown into building and civil engineering works required for certain indicator is based on the Classification of Types of Constructions (CC).

The coverage based on the activity (NACE) classification is a cut off survey on the basis of the number of employees. Included are business entities (legal entities and tradesmen) and parts thereof employing 20 or more persons.

The exclusion of small business entities does not cause a significant bias. If we compare year to year growth rate of hours worked on sites produced by the Annual survey (business entities employing 5 or more persons) and index produced by the Monthly survey (business entities employing 20 or more persons), trend is similar (for example, in 2021 growth rate was 9.4 in the Monthly survey and 12.7 in the Annual survey).

The objective of the production index is to measure changes in the volume of output at close and regular intervals. It provides a measure of the volume trend in value added over a given reference period.

The volume index of construction is compiled from the number of actually worked hours on construction sites, and covers all types of construction works (new constructions and reconstruction, repair work and maintenance) and all types of constructions (broken down to residential buildings; non-residential buildings; transport infrastructure; pipelines, communication and electricity lines; complex construction of industrial sites; other civil engineering).

Included are hours of work done by employees directly involved in the production process and overtime. Excluded are hours of work in business administration and any paid non-working period (holidays, sick leave, etc.).

Productivity factors are applied to adjust data series of hours worked. The productivity factor is estimated on the annual basis by dividing the deflated value of works done with hours worked, separately for buildings and for civil engineering works. The value of works done are deflated with construction value added from National Accounts. The fluctuation of individual values of the productivity factor is alleviated by using the exponential smoothing of the time series. After the calculation of indices for buildings and for civil engineering works, the next phase includes the calculation of the index for the total construction indirectly by aggregating these two semi-indices applying the Laspeyres’ formula. Weights are defined according to the share of the value added from section F – Construction taking into account the production factors costs from structural business statistics in a base year.

Since the productivity factor is estimated at the annual level on the basis of the data from the previous year, data for the whole reference year is issued as provisional results. When the calculation of the actual productivity for the reference year is done, final results are issued.

The breakdown of production in construction into buildings and civil engineering works is based on the Classification of Types of Constructions (CC).

Detailed definitions of the indicators are described in the Regulation (EU) 2019/2152 of the European Parliament and of the Council, and in the Commission Implementing Regulation (EU) 2020/1197.

During the transitional period (2021 -2023), detailed definitions of the indicators described in the Commission Regulation 1503/2006 are still relevant.

Data are collected monthly and include hours worked on construction sites during the reporting month.

Units of statistical observation for which data are provided are business entities (legal entities and tradesmen) with main activity in divisions 41, 42 and 43. They coincide with the reporting units. For business engaged in construction but with main activity outside divisions 41, 42 and 43, statistical unit is kind of activity unit.

Section F of NACE broken down into building construction and civil engineering works according to the Classification of Types of Constructions (CC).

This survey is based on the non-probability (cut-off) sampling, which means that for business entities with 20 and more employees very exhaustive survey is carried out. For business entities less than 20 employees’ data are not collected and sampling is not applied. 

The target population are all active business entities in the Statistical Business Register which are classified in division 41, 42 and 43, and employing 20 or more persons.

The whole territory of the Republic of Croatia - no regions are excluded. The regional coverage has not changed in any way.

Construction production index has been produced using the current method (based on hours worked on construction sites) since 2000 for the main aggregates, for the total, building construction and civil engineering works. Time series are not available before 2000.

Base year is 2015

Reference year is 2015

The base year is the same as reference year and this is the year when the index equals 100 (monthly average of the base year = 100).

The same base and reference periods are applied to data released nationally and to those transmitted to Eurostat.  

Indices, percentage changes (%).

Month

The legal basis for the Monthly/Quarterly Statistical Survey on Construction (GRAĐ-21/M and GRAĐ-21/3M forms) are:

The Official Statistics Act (OG, No 25/20)

Zakon o službenoj statistici (nn.hr)

Programme of Statistical Surveys of the Republic of Croatia 2021 – 2027 (OG, No 29/22)

Program statističkih aktivnosti Republike Hrvatske 2021. – 2027. (nn.hr)

The Annual Implementation Plans of Statistical Activities of the Republic of Croatia

REGULATION (EU) 2019/2152 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 27 November 2019 on European business statistics, repealing ten legal acts in the field of business statistics

COMMISSION IMPLEMENTING REGULATION (EU) 2020/1197 of 30 July 2020 laying down technical specifications and arrangements pursuant to Regulation (EU) 2019/2152 of the European Parliament and of the Council on European business statistics repealing ten legal acts in the field of business statistics

Still relevant during transitional period (2021 – 2023):

The Council Regulation No 1165/98 of 19 May 1998 concerning short-term statistics, amended by the Regulation (EC) No 1158/2005 of 6 July 2005 of the European Parliament and of the Council concerning short-term statistics

The Commission Regulation (EC) No 1893/2006 of 20 December 2006 establishing the statistical classification of economic activities NACE Revision 2

The definitions of short-term statistics variables are laid down in Commission Regulation No 1503/2006 of 28 September 2006

Response is obligatory. The obligation on units to submit the report is based on the Article 45 of the Official Statistics Act (OG, No 25/20). In case of the refusal to provide data, the provision of incomplete or incorrect data or the provision of data out of the time schedule, penalty provisions defined in Article 76 of the Act shall apply.

Zakon o službenoj statistici (nn.hr)

Data are regularly transmitted to Eurostat and to other international institutions (e.g. International Monetary Fund).

National legislation:

Terms and conditions, under which official statistics are produced, including confidentiality of individual responses, are governed by the Official Statistic Act (OG, No 25/20). Individual data are strictly confidential and must not be published or reported.

Zakon o službenoj statistici (nn.hr)

Confidentiality and reference to the Official Statistics Act are available on the data collection questionnaires.

European legislation:

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

Confidential data are treated according to policy rules (see point 7.1).

Data are published only on the level at which they are not confidential.

A release schedule for the upcoming year, specifying dates, reference period and level of statistical publication, is published each October in the Calendar of Statistical Data Issues and each December in the CBS Publishing Programme. These documents are available in both Croatian and English.

Web | Državni zavod za statistiku (dzs.hr)

Statistical publications are released simultaneously to all interested parties at the same time (11:00) by the issuance of the press release, first release - Construction Work Indices.

No users have prior access to the data.

There is no internal government pre-release access to the statistics for construction production indices.

Data on construction production index - production (total - F), production of building construction (F_CC1), and production of civil engineering (F_CC2) are transmitted to Eurostat monthly, beginning with January of 2000 with the required level of detail and timeliness, on a same day when data are published in the national publication (average 56 days after the end of the reference period).

Monthly - both to Eurostat and nationally.

Data are published monthly according to release calendar in the First release "Construction work indices", on the CBS website: 

Construction Works | Državni zavod za statistiku (dzs.hr)

News releases are issued at 11 a.m on CBS website. First release contains indices for total construction and two subindices, for buildings and civil engineering works.

Indices of construction works are presented as gross (non-adjusted) indices, seasonally adjusted indices, working-day adjusted indices, trend indices and as various types of growth rates in tabular form. Seasonally adjusted indices are also presented in graphic form.

Not available.

Data on production indices are available on the STS Database (the database of short-term statistics) at Reports (dzs.hr) 

The conditions under which certain users can have access to microdata are regulated by The Ordinance on the conditions and manner of use of statistical data for scientific purposes.

Monthly indicators are transmitted to Eurostat in form which is acceptable in the transitional period (2021 - 2023) - total production, buildings and civil engineering, on the same day when the data are published in the national publication (average 56 days after the end of the reference period).

European business statistics compilers’ manual for short-term business statistics — 2021 edition.

European business statistics methodological manual for compiling the monthly index of production in construction — 2021 edition.

Short methodological explanations are available in the First release "Construction work indices", available on the CBS website (Construction Works | Državni zavod za statistiku (dzs.hr)

More detailed notes have been published as methodological guidelines “Monthly and Quarterly Report on Construction (Revision)” (Handbook, No 75/17 of the Croatian Bureau of Statistics), and are available on the CBS website Microsoft Word - MU-75.doc (dzs.hr).

All results are displayed exclusively in the form of indices. The monthly index of production in construction (2015=100) is published for total construction and for buildings and civil engineering works.

Indices of construction works are presented as gross (non-adjusted) indices, seasonally adjusted indices, working-day adjusted indices, trend indices and as various types of growth rates in tabular form. Seasonally adjusted indices are also presented in graphic form.

Following the example of the European Statistical System, the Croatian Bureau of Statistics has developed a model of the Total Quality Management based on the Code of Practice of European Statistics. It is the model that is fully adjusted to the statistical purposes.

The documentation is available at CBS website Državni zavod za statistiku - Quality (gov.hr).

Quality checks and validation of data are done throughout the whole process. To ensure the quality of results, CBS develops, produces and disseminates data based on the same standards and harmonized methods. In this respect, CBS applies the criteria of relevance, accuracy, timeliness, accessibility and clarity, comparability according to the Official Statistic Act (OG, No 25/20).

The quality of the data corresponds with requirements of relevance, accuracy, timeliness and punctuality and comparability. The index of production in construction is an important indicator used both nationally and by the EU. All users’ needs are fulfilled. Coverage of the indices is the whole Croatian territory. Indices are produced at 1-digit level of CC (for buildings, civil engineering works and total). Indices are based on reliable data. Publications of result are almost punctual. Before results are presented to users, additional quality control of statistical data is performed and data are compared with previous and corresponding period and checks with different source are made. All data have been available since January 2000 and are comparable and free of any break in series.

Following the example of the European Statistical System, the Croatian Bureau of Statistics has developed a model of the Total Quality Management based on the Code of Practice of European Statistics. It is the model that is fully adjusted to the statistical purposes.

The user’s needs are fulfilled. The indices are used by different users, both national and EU ones. The index of production in construction has an essential role in the calculation of GDP. Users of statistical data on construction production indices can be found in all areas of economic and social life, beginning with individual citizens, business entities, media, economic analysts, researches and students, to state government bodies, professional business organisations (the Croatian Chamber of Economy), non-profit organisations (the Croatian Chamber of Trades and Crafts, the Croatian Employers' Association), financial institutions (the Croatian National Bank, Raiffeisen bank, etc.), as well as Eurostat and internal users in CBS (national accounts).

Not available.

The international and national requirements of time series are fulfilled.

Indices at 1-digit level of CC (for buildings, civil engineering works and total) are available on a monthly basis from January 2000.

It is in accordance with the requirements of Regulation No 2019/2152  and Regulation No 2020/1197, as well as Regulation No 1165/98 (still relevant during transitional period 2021 – 2023).

The comprehensive survey is performed on a cut-off census (included are business entities and parts thereof employing 20 or more persons). Data are not collected, nor the sampling procedure is done for business entities with less than 20 employees.

The exclusion of small business entities does not cause a significant bias. If we compare year to year growth rate of hours worked on sites produced by the Annual survey (business entities employing 5 or more persons) and index produced by the Monthly survey (business entities employing 20 or more persons), trend is similar (for example, in 2021 growth rate was 9.4 in the Monthly survey and 12.7 in the Annual survey).

There was no significant deviation betwen the provisional and the final results in 2022.

Final results for 2022 are published together with the provisional results for January 2023 (3 months after the end of the reporting year).

Not relevant. Quantifying sampling errors is not possible because the sample is not based on random sampling. The comprehensive survey is performed on a cut-off census (included are business entities and parts thereof employing 20 or more persons). For business entities less than 20 employees' data are not collected and sampling procedure is not applied.

It is not possible to evaluate the accuracy of the change in the occurrence in the selected model due to the non-existence of a secondary data source for the whole set. We can evaluate only the accuracy of the trend of the change in the occurrence at the monthly and annual level. The evaluation for the non-sampled part of the population (business entities employing less than 20 persons) is not carried out due to the specific character of the construction activity (the major number of small business entities executes all works irrespective of the registered activity). Therefore, it is impossible to distinguish, on the basis of the registered activity, how many working hours these business entities did on buildings and how many on civil engineering works. Rates calculated according to data obtained from the Fina's Annual Report for 2021 show that business entities employing less than 20 persons, participate in total turnover with 30.3%.

The reasons why sampling isn't applied are large number of small units which should be included in the sample, huge respondent burden and huge cost of data gathering for the CBS.

The exclusion of small business entities does not cause a significant bias. If we compare year to year growth rate of hours worked on sites produced by the Annual survey (business entities employing 5 or more persons) and index produced by the Monthly survey (business entities employing 20 or more persons), trend is similar (for example, in 2021 growth rate was 9.4 in the Monthly survey and 12.7 in the Annual survey).

Unit non response

The survey is compulsory. Response rate is 98.4%. All data are available according to deadlines. For late or missing data, units are contacted by phone or mail and asked to provide the missing data. If units are not available, credible estimates are used to compensate for the late or missing data (approx. 1.6% of all observed units).

Weighted unit response rate is 99.1%. Weighting variable is value of works done with own workers.

Item non-response rates

No case of item non-response is involved in the survey, since during data editing phase, units which did not fully complete the questionnaire are contacted by telephone and asked to provide the missing data. Every month  item response rate of 100% is achieved.

The undercoverage is mainly caused by misclassification of the business entity in the Statistical Business Register – SPR (e.g., registering an activity different from construction although the business entity is engaged in construction works). There is no register on the number of non-covered reporting units due to the misclassification. Since the comparison of the SPR data with another data source (it does not exist) was not possible at the time when the contingent of reporting units was formed, estimates for them cannot be made because their significance is unknown. Consequently, the target population in this survey is limited to the one covered by the frame, while the influence of non-covered ones is neglected. Nevertheless, great efforts are done in the Croatian Bureau of Statistics to minimise this error. Specifically, while collecting, controlling reports and contacting reporting units for incorrect and doubtful reports and perceiving that the activity of an reporting unit included in their coverage was misclassified, experts in various Units submit information on that to the Statistical Business Register Unit. This Unit regularly updates its data base and therefore it may be expected that such misclassifications will be gradually cut down in the Statistical Business Register.

Overcoverage: As it is the case with the undercoverage, this error is mainly caused by misclassification of an activity in the SPR (a business entity has been classified in the construction activity, while in fact it is engaged in an activity other than construction). This kind of error is regularly disclosed during the process of collecting and controlling the data as well as while contacting reporting units. There were 6 reporting units in 2022 included in the coverage that were in fact engaged in an activity other than construction, which made 0.5% of all units included in the coverage. The major part of those units was engaged in the industrial or trade activity. The obtained information was submitted to the Statistical Business Register Unit for data updating.

Double classification: A risk/probability for this error is zero, since the coverage is formed on the basis of SPR. To date, multiple coverage of reporting units has not been recorded.

Measurement errors: Although an alternative data source (that is, data bases/registers to compare data obtained on the Monthly/Quarterly Report on Construction in order to define measurement errors) does not exist, the Construction Statistics Unit compares the trend in volume indices of construction works (which is deduced from hours actually worked) with data on the trend in the number of persons in employment in construction compiled in the Labour Force Statistics Unit. Still, at this comparison one should take into account the difference in methodologies of data collection. According to the methodology applied in the Labour Force Statistics Unit, persons are considered employed as long as their employment record card is kept by the business entity, irrespective of whether they actually work and get paid or not, while the volume index in construction is based on hours actually worked on sites. This is why, in case of economic turmoil, the total number of persons in employment in the construction activity tends to fall at a slower pace than the construction production indices.

Causes of errors in measurements when collecting data that occur due to ignorance, negligence, conscious/unconscious presentation of inaccurate data by respondents due to unclear questions in the questionnaire, measurement units or instructions for filling in the forms are minor due to the fact that the availability of data was taken into consideration while defining their contents. In addition, reporting units are welcome to contact Branch Units of the Croatian Bureau of Statistics (who helps reporting units in filling in reports and controls the completeness and accuracy of data before they are submitted to the Croatian Bureau of Statistics) or directly the Croatian Bureau of Statistics in case they need information or help in resolving ambiguities.

These measurement errors do not represent a major issue, since they are revealed during the visual and calculation control of data before they are entered in the computer as well as by the default computerised logical and calculation control during data processing. When such illogicalities or errors are spotted, reporting units are contacted for the check-up of inaccurate and incomplete data and for the correction of inaccurate data.

The provisional monthly indicators are transmitted to Eurostat in form which is acceptable in the transitional period (2021 - 2023) - total production, buildings and civil engineering, on the same day when the data are published in the national publication (average 56 days after the end of the reference period).

Since the productivity factor is estimated at the annual level on the basis of data from the previous year, data for the whole reference year are issued as provisional results. When the calculation of the actual productivity for the reference year is done, final results will be issued. Final results for 2022 are published together with the provisional results for January 2023 (3 months after the end of the reporting year).

Postal questionnaires are sent to the reporting units once (for the whole year) at the end of December. The deadline for replies is the 20th day in a month at the latest, for the previous month. The data collection is completed approximately 15 days after the deadline for replies.

All new releases are published as announced in the release calendar: 

Web | Državni zavod za statistiku (dzs.hr)

There is no time lag between the delivery dates announced in the calendar and the time the target data are actually delivered.

Comparability of the methodology used in observing the activity, observation period, dates and modes of data publication for EU Member States are provided in the REGULATION (EU) 2019/2152 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 27 November 2019 on European business statistics, repealing ten legal acts in the field of business statistics and in the COMMISSION IMPLEMENTING REGULATION (EU) 2020/1197 of 30 July 2020 laying down technical specifications and arrangements pursuant to Regulation (EU) 2019/2152 of the European Parliament and of the Council on European business statistics repealing ten legal acts in the field of business statistics, as well as in the Regulations No 1165/1998 and No 1503/2006 (still relevant during transitional period 2021 - 2023).

This ensures a good comparability of national data.

All data have been available since 2000 and are comparable and free of any break in series.

Comparisons are made on “ad hoc” basis and are not published.

Confrontation is made with other available data within CBS (value added from National accounts – section F (construction), number of persons employed in construction from Employment), as well as with monthly/quarterly data published in Eurostat's publications.

Results are:

a) the correlation coefficient of the quarterly growth rates of the IPC and of the value added of the Section F from QNA was computed for the period 2001Q1-2022Q4. Multiple R for this period is 0.82 for calendar-adjusted series and 0.72 for seasonally adjusted series.

b) despite the differences in the methodology (see point 13.3 – measurement errors), when comparing annually data between hours worked (GRAĐ-21/M and GRAĐ-21/3M forms) and total number of persons employed in construction (data from Employment statistics unit), we can conclude that there is a strong correlation between these two variables. Multiple R for the period from 2001 to 2022 is 0.84.

Data are internal coherent.

In 2022:

Burden of respondents: 4370 hours annually for all enterprises.

Total costs of NSI: 8 850 hours annually

The users of statistical data are informed about revisions (preliminary, final data) on the CBS website.

Correction of errors, major changes in the classifications, methodology, base year changes and changes of seasonally adjusted data is possible. Notice about minor changes is given at the time the change is introduced.

The same revision policy is applied to data released nationally and transmitted to Eurostat.

The data publishing revision policy in the Croatian Bureau of Statistics includes the following:

1.    Updating of the Calendar of Statistical Data Issues

2.    Publishing of corrigenda of the published data

1) Updating of the Calendar of Statistical Data Issues

The data publishing revision policy of the Croatian Bureau of Statistics, which is carried out according to the defined deadlines, is carried out by updating the Calendar of Statistical Data Issues. According to the Official Statistics Act, changes in the publishing deadlines for publications/data are announced in advance. The Croatian Bureau of Statistics is obliged to announce each change in the publishing deadlines (the Calendar update) at least three days prior to the date originally announced in the Calendar at the web site of the Croatian Bureau of Statistics.

The update of the publishing deadlines for data or publications (the update of the Calendar) is announced without delay on the Croatian Bureau of Statistics’ website by updating the date in the Calendar and putting the notice beside the Calendar, which says: updated: (the accurate new date).

Therefore, the Calendar that contains the original publishing deadline for a particular publication/statistical information is replaced with the new Calendar which includes the updated publishing deadline. The original Calendar is kept in the Publishing Department which registers every change in publishing deadlines for publications in the Calendar Update Base.

2) Publishing of the corrigenda of the published data

In cases when an error is observed in the published data, the Croatian Bureau of Statistics announces the correction as soon as possible and as simple as possible in order to help the users to clearly see what was changed in comparison to the originally published data. The correction is carried out by marking the incorrect data with an asterisk and creating the corrigendum.

There is an internal vintage database where data on non-adjusted data are stocked. But, a vintage database with all published data on all revisions of calendar adjusted indices, seasonally adjusted indices and trend indices as well as an audit trail of all revisions going back since 2000, is not available.

No special revision calendar exists.

Benchmarking (confrontation) with other statistics is carried out but not published.

Regular revision analyses for the national data sent to Eurostat are not carried out.

Major changes in the methodology, base year changes, are announced in the First release at the time the change is introduced.

The general practice is that all changes in methodology are explained when data based on the new methodology are published for the first time.

Since the productivity factor is estimated at the annual level on the basis of data from the previous year, data for the whole reference year are issued as provisional results. When the calculation of the actual productivity for 2022 is done, final results are published together with a change of base and reference year.

Quality indicators for the period 2020M1-2022M12:

Gross (non-adjusted) growth rates

1)            MAR = 0.085

2)            MR = -0.018

Working-day adjusted growth rates

1)            MAR = 0.771 

2)            MR = -0.087

Routine revisions, related to replacing estimated data with the subsequently reported data, are not carried out, because all data are available according to the deadlines.

If error occurs, corrected data are published in the next issue of First release. Data changed in relation to the first released are disseminated with a comment.

Gross indices and data revised due to the process of Seasonally or Working day adjustment are published each month and are available for the period from January 2000 in the STS Database (the database of short-term statistics) at Reports (dzs.hr).

Size of these revisions has not been calculated yet.

Use of benchmarking and domain (indicator) specific characteristics of revisions are not available.

Statistical survey

The Address Book for the monthly survey maintained by the CBS Construction Statistic Unit, derived from the Statistical Business Register.

This survey is based on the non-probability (cut-off) sampling, which means that for business entities with 20 and more employees very exhaustive survey is carried out. For business entities with less than 20 employees’ data are not collected and sampling is not applied.

Data are collected monthly.

Data are collected by using the reporting method. Two types of questionnaires are used: The Monthly and the Quarterly Report on construction (GRAĐ-21/M and GRAĐ-21/3M forms). The Monthly Report on Construction is collected for each month, except for March, June, September and December. The data for these months are collected on the Quarterly Report on Construction.

Electronic and postal questionnaires are used for gathering data from business in the survey: the Monthly and  the Quarterly Report on construction (GRAĐ-21/M i GRAĐ-21/3M forms).

Postal questionnaires for all 12 months of a year are sent out at the beginning of each year.

Actions taken to speed up or increase the rate of response: written remainders (postal and e-mail) and telephone contacts.

Electronic questionnaires are automatically checked during the data entry by reporting units with a set of warnings and errors, and in the CBS the data entry is again checked by automated error-checking procedures.

Reports that fail to meet the quality standard are subject to verification and are corrected as required.

Postal questionnaires are firstly checked manually and visually.

All questionnaires (electronic and postal) are then verified by automated error-checking procedures in the branch statistics of the CBS, on micro and macro level.

Treatment on micro-data level

Collected data are first checked and verified by staff from the Construction Statistics Unit and then during the automated process of data entering. Quality checking is carried out by using a set of strict manual and electronic logical/calculation controls. Respondents’, coding and data entry errors and inconsistencies are detected and corrected using manual and electronic logical/calculation quality control while editing. Reports that fail to meet the quality standards are subject to verification and are corrected as required. We organised correction procedures by priority: severe errors are generally corrected manually and more effort is spent on these, while minor errors require less attention and are mostly corrected automatically by computer.

Treatment on macro-data level

The macro data tables are verified by experts in construction unit. Data from the actual reporting period is compared with the previous and the corresponding reporting period of the previous year. Furthermore the indices are crosschecked with the number of person employed of the same NACE Rev.2 category. If inconsistencies are detected the data must be verified at micro level again by using raw data of the questionnaries and possibly changed.

Apart from previously mentioned checks done for the purpose of national statistics, neither additional checks are done nor additional tools are used before data are transmitted to Eurostat.

Data are collected on the base of cut-off survey (census for business entities employing 20 and more persons), without a random sample of business entities with less than 20 persons employed and without an estimate for smaller units.

There are no unit non-response rates and item non-response rates, because reporting units that did not send filled-in questionnaires or did not fully complete the questionnaire are contacted by phone and asked to send the filled-in questionnaire for the current month or to provide the missing data.

Index of Production in Construction (IPC) is based on the number of hours worked on sites adjusted with a productivity factor. The productivity factor is estimated on the annual basis by dividing the deflated value of works done with hours worked, separately for buildings and for civil engineering works. The value of works done is deflated with construction value added from National Accounts. The fluctuation of individual values of the productivity factor is alleviated by using the exponential smoothing of the time series. After the calculation of indices for buildings and for civil engineering works, the next phase includes the calculation of the index for the total construction indirectly by aggregating these two semi-indices applying the Laspeyres’ formula. Weights are defined according to the share of the value added at factor costs from section F – Construction, from structural business statistics in the base year.

Since the productivity factor is estimated at the annual level on the basis of data from the previous year, data for the whole reference year are issued as provisional results. When the calculation of the actual productivity for the reference year is done, final results will be issued. Final results for 2022 are published together with the provisional results for January 2023 (3 months after the end of the reporting year).

IPC are compiled as unadjusted, WDA, SA and trend-cycle indices.

Seasonal and working-day adjustment has been done by applying the programme interface JDemetra+ (version 2.2.2) by using the X13 ARIMA method on monthly index series (Ø2015 = 100), which starts with January 2000 index. The effect of a random component in the seasonal adjustment process of time series of indices has been excluded.

The calculated series of volume indices of construction works are comparable to those published by the Eurostat.

The average index value of the base year (reference year) is not 100 for the calendar adjusted series and for the calendar and seasonally adjusted series due to the fact that IPC is seasonally adjusted directly without imposing any constraints on seasonally adjusted series.

The model/filter selection - manual.

Parameters re-estimation - ref. periods April, August and December (in other months application of forecasted factors). Annual revision of models, filters, calendar regressors for ref. period m12.

Seasonally adjusted time series are revised in April, August and December (all values backwards).

Model used: (0,1,1)(0,1,1)

Critical value for outlier detection – 4.

Seasonal filters are fixed during the year (revised at the end of a year if necessary):

-          Total construction (F) - 3x3 seasonal filter

-          Buildings (CC_1) - 3x5 seasonal filter

-          Civil engineering (CC_2) - 3x5 seasonal filter

The date of seasonal breaks in the series - 31/12/2005

Direct adjustment (via components) is used.

Checking residual seasonality - qs test and f-test on residual seasonality are checked for every series.

There is no consistency amongst the different levels of breakdown.

Not available