National Institute of Statistics (ISTAT)

Unit SED (Index of production in construction - IPC)

Economic Statistics Directorate - Short Term Statistics Division

Via Tuscolana, 1788

00173 Rome

Monthly index of total production in construction.

Nace Rev. 2 Section F;  CC breakdown: CC1 (“Buildings”) and  CC2 (“Civil engineering”).

Sector: Nace Rev.2 Section F

     Size classes covered

     Hours worked: in principle, enterprises with no employees or more.

     Industrial turnover: units employing 20 employees or more.

Activities covered: NACE Rev. 2 Section F

Theoretical statistical population is represented by the total of the enterprises of the construction sector (more or less 500.000 enterprises).

 The area covered by the IPC is the whole national territory.

1995 is the starting point for the general index time series (coverage: Nace rev.2  section F).

From 2000 onwards also CC1 and CC2 time series are available (but not released at national level).

2010.

Index.

Month.

Data are sent only to Eurostat (not to other international organizations).

• Aggregated data on hours worked provided by BWWF are processed  and stored in a database, but not released.
• The two indices referred to item CC1 and CC2 are only transmitted confidentially to Eurostat.
• The same applies also to the Index of production for the first transmission to Eurostat (t+45), because it occurs before the national dissemination. After the national dissemination, data become free for Eurostat.

Once a year, in December, ISTAT publishes an official calendar providing the exact release dates for the coming year. The calendar is distributed to the press and is available free of charge to all the interested parties.

The release calendar is also disseminated on ISTAT's website (http://www.istat.it/en/information/journalists/press-calendar)

Monthly.

Every month data are released simultaneously to the all interested parties by issuing the monthly press release. IPC is calculated for  NACE Rev. 2 Section F, in the unadjusted, seasonally and working day adjusted form. The press release (Italian version and English short  version) is subsequently posted on the ISTAT Internet website http://www.istat.it/en/archive/constructions+production. 

Basic information on methodological aspects concerning the index  compilation are provided through a “Methodological note”, attached to the press release.

This index is not considered in other publications.

The Index of production series are made available in the ISTAT on-line database called I.stat (http://dati.istat.it/?lang=en) under the theme "Industry and Construction". Database is open and free of charge.

Not available.

Data are sent to Eurostat to be used in European aggregates (CC1 and CC2 indices, confidentially transmitted to Eurostat)  and released also as national level (only the total index).

Methodology and sources of information for the index calculation are described in "Nota metodologica-Produzione nelle costruzioni-Base 2010", 27th March 2013 attached to the press release (http://www.istat.it/it/archivio/86341).  Other methodological general information, even those referred to the old base 2005, are available in the document “A new index of production for the construction sector based on input data”, Contributi ISTAT n.2, 2003.
A description of methodology and sources used in compiling statistics is also contained in the Information System for Survey Documentation and Quality Control (Siqual) that is available on ISTAT's Internet website (http://siqual.istat.it/SIQual/lang.do?language=UK). This system manages qualitative information related to each single phase of a survey's production process (http://siqual.istat.it/SIQual/visualizza.do?id=7779917 and  http://siqual.istat.it/SIQual/visualizza.do?id=7779918).

More information about procedures applied for quality management and quality assessment are reported in the document "Quality Guidelines For Statistical Processes" - December 2012

http://www.istat.it/en/files/2011/11/QualityGuidelines_EngVers_1.11.pdf

Istat sticks to the quality general principles as they are stated in “ESS Handbook for Quality Report” (http://ec.europa.eu/eurostat/documents/3859598/6651706/KS-GQ-15-003-EN-N.pdf)

The construction production index is the result of an indirect estimate and an integration of different sources of data  (econometric model is used to estimate the relation between production and specific indicators on worked hours, intermediate consumption and capital stock). The overall quality assessment can be considered satisfactory as all the main quality requirements (e.g. completeness, user satisfaction, relevance, reliability) are met. 

The released estimates  meet  the relevance requirements (Eurostat and Istat National Accounts are the main institutional users). 

Other institutional or private users, as National Building Contractors' Association,  are in general satisfied with Istat output, even though they  sometimes ask for  specific elaborations, not scheduled in the monthly or annual Istat agenda.

Not available.

Eurostat’s requirements in terms of time series are fulfilled.

IPC is estimated using and indirect method, based on a production function. Hours worked by manual worker employed in the sector represent the first indicator used to estimate the production. Its contribution to the production estimate is 44%.  At the moment, data from a  purposive panel of  28 BWWF (instead of  90), covering most of the  national territory, are used to compile IPC, even though a census survey has been conducted since January 2016. With the next rebasing, all the data collected (Census survey on 90 BWWF) will be considered to estimate the hours worked index for the IPC.

The second variable used, explaining 45% of the output, is the turnover of industrial enterprises producing and selling building materials typically used in the construction sector (intermediate consumption). The trend and the rate of change of this two variables (hours worked  and industrial turnover) are always aligned, confirming the high reliability of the two indicators in describing the evolution of the sector output. 

Hours worked: as regards hours worked, only a general assessment is available at the moment. Although they represent administrative information, data on labour input used for IPC are obtained  through a statistical survey. The purposive panel used at the moment covers more or less 50 % of the total employment referred to the total 90 BWWF operating on the whole national territory.

For preliminary estimate (t+45), the response  rate is on average around 85% (25 out of 28) and rises almost to 100% for its first revision, one month later (t+75), when the responding units are generally 27-28.  For the preliminary estimate, 19-21 BWWF  data are used, representing  more than 70% of the responding units ( 19-21 out of 25-26). For the t+75 estimate, the number of the units used to estimate IPC rises to 24-26 (i.e. 90% of the 28 responding units).

In both cases, before being used for the IPC compilation, data are generally upward revised because they are not yet not complete at the time of the first estimate (non-responding enterprises, or enterprises transmitting data to BWWF after the Istat deadline).  Correction factors, estimated on the basis of the past data, are applied  to take into account the lacking information in the provisional hours worked data that enterprises transmit to BWWF. 

The correction factor is in the range 4.0-5.2%  for preliminary estimate of the hours worked index and falls within the range  1.6%-2.1% one month later, when preliminary data are revised for the first time.

After one year, when final and complete information are available, the  “adjusted”  data used for the provisional estimate are compared to them. The ratio between the total amount of hours worked revaluated (estimated) and used for the provisional index and the corresponding amount available at the end of the year (final data) is on average close to 98%  (e.g. in 2014, 97.9% on average).

Industrial turnover: please, see metadata on “Industrial turnover index”.

An assessment of the sampling error for IPC is not available, as IPC estimate is based on an indirect method.

For hours worked: see section 13.1 for a general assessment of the accuracy.

For industrial turnover: see metadata on “Industrial turnover index” (a “cut-off” sampling method is used).

Hours worked: the response rate is 90% on average for the provisional estimates (t+45) and near to 100% for the first revision estimates (t+75).

Industrial turnover: please, see metadata on “Industrial turnover index”.

A provisional monthly index is confidentially  sent to Eurostat   45 days after the reference period for total production and for production in “Civil engineering” and “Buildings”. Differently from CC1 and CC2 indices, the general  index is released through a national press release 48-50 days from the reference period. IPC is then revised the following month (first revision) and becomes final one year later (annual revision), when new National Account annual estimates of capital are taken into account and updated information on hours worked and industrial turnover can be used. 

European and national deadlines are always met. 

Data used for the index compilation cover a substantial  part of the national territory, as Welfare Funds involved in BWWF survey are spread throughout the main geographical regions. Industrial turnover data  used to estimate the intermediate consumption index refer to the whole national economy.

Data are comparable over time. No breaks in time series.

A comparison with data on building permits has been made in order to check cross domain coherence. This comparison has shown an overall coherent dynamic between the two indicators.

Italy quarterly National Accounts VA estimates are based on monthly IPC, therefore a good coherence is ensured between NA and STS figures on construction output figures.

IPC is always consistent with its sub-aggregates (items CC1 and CC2), as CC1 is obtained as a difference between CC and CC2. Internal coherence is also ensured for calendar adjusted data, as indirect adjustment is performed.

The cost estimate is compiled in terms of total number of hours worked (THW). It is equal to the product of three terms: (total) Annual Working Days (AWD), (total) Hours Worked per Day (HWD), Number of Persons involved in the Survey (NPS).

Statistical burden on BWWF that provide Istat administrative data on hours worked is not high (and no burden at all on enterprises that are required to transmit information to  BWWF for administrative reasons). Intermediate consumption and capital indices are based on Istat existing information.  

The general revision policy is predetermined,  reasonably stable from year to year and  adopted in order to incorporate new and improved information. In addition to the monthly current revisions, IPC is generally revised each year to take into account the updated data on hours worked, turnover industrial indices and the new NA estimates on annual capital stock used to estimate a monthly capital index.  Other IPC revisions can be performed when extra data are available or in case of correction due to errors in data input.

In particular, the revision policy is based on:

 - routine revisions: as data are provisional at the time of the press release, they are revised one month later for taking into account updated information and sometimes additional new information transmitted by BWWF; also monthly revisions of industrial turnover are considered;

 -annual revisions: once a year, a deeper revision is carried out  to take into account other possible updating of hours worked transmitted by BWWF, the annual revision of industrial turnover data and the new NA estimates on annual capital stock; on the occasion of the annual revision, also seasonal adjustment models are tested again.

 - major revisions performed every five years:  in the framework of the rebasing, when  revisions are  mainly due to the effects of the new weighting systems.

Users are always informed about the revisions and, at the time of each dissemination, they are also informed if data are preliminary or revised and the reasons why data have been revised are explained too (e.g. availability of additional new data, revisions/changes in data sources).

On the occasion of the annual current revision (or during the rebasing), also seasonal adjustment model can be modified, if necessary.

Coherence requirements and deadlines in the revision policy are met both for data nationally  released and transmitted to Eurostat.

As concerns non-scheduled revisions, if inaccuracies are found out, they are generally corrected with the first revision of the preliminary estimate. Otherwise, corrections are generally postponed and made when the annual revision is carried out.

Vintage database releases were stopped before the introduction of base year 2010=100  (March 2013) and they might be disseminated again in the future.

In the paragraph “Revisions”, attached to each press release,  methodological notes on revisions are reported (revision policy, source, periodicity, etc.). In terms of growth of rates, the maximum absolute difference between the second release index (t+75) and the provisional one (t+45) is in general lower than 0.6%. This difference  is due to the more updated and complete information concerning worked hours transmitted by Welfare Funds to Istat for the t+75 estimate and  to possible corrections on industrial turnover elementary data.

Monthly capital index used to estimate IPC is obtained from NA estimates of gross capital stock in a process benchmarked on annual basis against NA figures. For the annual revision, also more updated information on labour and intermediate inputs are taken into account. 

The IPC estimate is an example of integration between statistical and administrative information.    
For the monthly updating of the index, input data are taken from different sources: administrative data on hours worked from BWWF but collected through a monthly  survey; information on intermediate input from ISTAT current survey on industrial turnover, and monthly estimates of gross capital stock derived  from National Accounts data.

For the base year, a production function is estimated using SBS data.

Monthly.

As a first step, the coherence of the time series of worked hours data is checked, as a total amount and for each Welfare Fund providing data. Other general checks are made on the plausibility of hours worked growth rates in comparison with the trend of the number of employees (workers) and the number of enterprises they are referred to.

Outliers are detected and investigated and responding units (BWWF) are contacted again when necessary. To identify other outliers or other reasons of possible inconsistencies, also for industrial turnover data preliminary checks are scheduled (e.g. turnover figures checked against orders). IPC is estimated after checking the global alignment between hours worked and intermediate input indices (in terms of growth rates) or investigating possible economic/statistical reasons why they are not.

As regards the output (IPC), data  can be compared with other statistical domains information. 

Hours worked: when estimating hours worked index, “non-response” issue is not tackled, as the monthly growth rate for the current year is applied to the previous year  index and only the units responding in the two subsequent years are considered for this purpose (units responding both in the current and in the previous year). But, as construction sector enterprises might delay the data transmission to BWWF, information on hours worked are not yet complete when the provisional index has to be estimated. An upward correction is therefore necessary to take into account this lack of information (please, see 13.1).

Industrial turnover: see metadata on “Industrial turnover index”

IPC, CC1 and CC2 indices are calendar adjusted and corrected for seasonal effects by TRAMO-SEATS procedure (version 2010 for Linux). Adjustment models are completely reviewed once a year and parameters are re-estimated each month, when seasonally adjusted series are revised from January 1995.

A “Reg-Arima” approach is used for working days and seasonal adjustment. A multiplicative seasonal adjustment breakdown is adopted. Calendar adjustment is performed by using Italy national calendar which also considers national holidays. Also leap year is adjusted for IPC. Easter effect, preliminary tested, resulted not significant. One transitory change and three additive outliers (one being due to a particularly cold winter)  have been identified for IPC. The critical value for the outlier detection and the filter length are automatically chosen, whereas model/filter selection is manual.

For calendar adjusted data the consistency of different aggregation level is ensured by the indirect calendar adjustment method (please, see point 15.4 "Coherence - internal").

No constrains are applied on seasonally adjusted and raw data, but their annual average growth rates are generally very close.

None.