Current methodology - Value of Building Work Put in Place from December 2014 Quarter
Current methodology - Value of Building Work Put in Place from December 2014 Quarter
The value of building work put in place measures activity in the construction sector, and complements building consents issued information (which represents the intention to build).
These quarterly releases provide estimates of the value and volume of work put in place on construction jobs in New Zealand. The value of building work includes residential building work and non-residential building work, which are summed to give all building work. Non-building construction work, such as roads and bridges, is excluded.
The population frame for value of building work put in place is a list of building consents issued by territorial authorities (TAs) throughout New Zealand. Each month, TAs provide us with new building consents they have issued. These details include the consent value, job description, contact details of owner and builder, and the building site address. Building consents valued under $5000 are excluded, as are consents for non-building construction.
Methodology from December 2014 quarter
We introduced a new methodology for estimating the value of building work put in place in the March 2015 quarter, and applied it from the December 2014 quarter. From a methodology based on a sample survey, it is now based on modelled administrative data supplemented by a sample survey.
Between 1965 and 2014, the value of building work put in place was estimated from a sample survey. During this time the sample design did not change significantly, although we kept the sample optimal.
See Methodology and classification changes to value of building work put in place statistics for more details about the changes.
Related information: Implementing classification and other changes to building consent statistics
COVID-19 from the March 2020 quarter onwards
The behaviour of modelled building projects changed due to COVID-19, so we have surveyed an additional sample of medium-value building projects since the beginning of the pandemic. We calculate model adjustments by comparing the results of this survey (with outliers removed) against our model’s predictions.
Model adjustments account for the impacts of COVID-19 on building activity. These adjustments are revised from time to time as improved information becomes available for previously non-responding building projects. Downward adjustments reflect building activity that was delayed or cancelled (for example due to alert level 4 lockdowns or supply constraints). Upward adjustments reflect a catch-up on delayed construction in later quarters.
The current adjustments are:
|March 2020||Down 8.2 percent (5 working days)||Final|
|June 2020||Down 21.7 percent||Final|
|September 2020 to June 2021||Up 3.5 percent||Final|
|September 2021||Down 15.8 percent in Auckland (no adjustment elsewhere in New Zealand)||Provisional|
|December 2021||Down 5.4 percent||Provisional|
The methodology aims to reduce the number of questionnaires sent to respondents by estimating the quarterly value of building work, based on the age and value of building consents. Large consented projects (new and ongoing) will continue to be surveyed each quarter for the duration of the building project.
Modelled and modelled low value
Building consents are modelled when the value is under the postal value cut-offs in the table below. Modelled low value consents are under the low value threshold.
Modelling has two main aspects:
- The building consent value is rated up to reflect that the value of work put in place typically exceeds the building consent value.
- The rated-up value of building work is apportioned across several quarters to reflect typical time taken to complete building projects.
The modelling equation is: quarterly value of work put in place = building consent value * rate-up factor * apportionment ratio
We apply different rate-up factors and apportionment ratios depending on the building type.
Modelled low-value consents are rated up but completed in one quarter, not apportioned across several quarters.
Estimates for the value of building work put in place are modelled for a portion of the population. Modelling errors measure the variability that occurs when we apply a statistical model to produce estimates, which quantifies the effect of model 'imperfections'. The relative modelling error is expressed as a percentage of the estimate at the 95 percent confidence interval limit.
Modelled rate is the percentage of the estimated work put in place that we modelled for a building category.
Quarterly Building Activity Survey (QBAS)
The following table shows the building consent cut-off values for the survey components.
|Value of building work - building consent value cut-offs for survey|
|March 2015||$800,000 +||$1,800,000 +|
|June 2016||$900,000 +||$1,900,000 +|
|December 2017||$1,130,000 +||$1,900,000 +|
|December 2018||$1,180,000 +||$1,900,000 +|
|December 2019||$1,220,000 +||$1,900,000 +|
|December 2020||$1,250,000 +||$1,900,000 +|
|December 2021||$1,400,000 +||$1,900,000 +|
While the majority of projects are modelled, projects with the highest consent value are all directly surveyed. All staged building consents are also surveyed.
Changes in the December 2017 to 2021 quarters
We increased the residential building consent value cut-off for postal survey to from $900,000 to $1,130,000 in the December 2017 quarter, to $1,180,000 in the December 2018 quarter, to $1,220,000 in the December 2019 quarter, to $1,250,000 in the December 2020 quarter, and to $1,400,000 in the December 2021 quarter. We calculated new modelling parameters from completed projects to use for consents valued between $900,000 and the upper boundary. These higher value residential projects have a higher rate-up factor than most residential consents, and the apportionment ratio reflects that less of the building work is completed in the first three quarters than for lower value residential projects. The modelling parameters for residential consents valued under $900,000 were not changed.
There were no changes for non-residential buildings – the consent value cut-off remained $1,900,000 and the modelling parameters were unchanged.
We introduced the higher residential value modelled strata to reduce respondent burden, as the number of residential projects in the postal survey increased markedly in recent quarters. The higher survey count was due to both the large volume of construction activity being undertaken and higher building costs reflected in building consent values.
For building projects where no survey response is received, we impute values for work put in place, based on responses for comparable projects.
Imputation rate is the percentage of the estimated quarterly work put in place that we imputed for a building category.
Quarterly quality measures
See Infoshare for building activity quality measures. (Industry sectors/ Building Activity Survey - BAS/ Building activity quality (Qrtly - Mar/Jun/Sep/Dec))
- Modelling error
- Modelled rate
- Imputed rate
- Sampling error (for data up to September 2014 quarter)
Note, rates tend to be lower for building categories and regions with the greatest number of units.
Interpreting the data
Volumes (constant price series)
Current values include both a quantity and price component, whereas constant price series (volumes) have had the effect of price changes removed. Removal of price change (deflation) leaves just the volume (or quantity) component, enabling comparisons across different time periods without the distortion caused by price inflation (or deflation).
Quarterly values for residential building work and non-residential building work are separately deflated by the residential buildings and non-residential buildings sub-indexes from the capital goods price index (Included in Business Price Indexes.) The deflated quarterly values are expressed at a constant pricing level, using September 1999 quarter prices. Deflated values for all building activity are calculated as the sum of the deflated values for residential and non-residential building activity.
We deflate prices before seasonal adjustment and estimation of trend values.
Seasonally adjusted series
Seasonal adjustment removes the estimated impact of regular seasonal events, such as summer holidays and pre-Christmas purchasing, from statistical series. The makes figures for adjacent periods more comparable.
We recalculate the seasonally adjusted series quarterly when each new quarter's data becomes available. Figures are therefore subject to revision, with the largest changes normally occurring in the latest quarters.
We use the X-13ARIMA-SEATS seasonal adjustment program, developed in the U.S, Census Bureau, to produce the seasonally adjusted and trend estimates.
See Seasonal adjustment in Statistics New Zealand for more information.
Trend calculation removes the estimated impact of regular seasonal events and irregular short-term variation from statistical series. This reveals turning points and the underlying direction of movement over time.
We recalculate trend series quarterly when each new quarter's data becomes available. Figures are therefore subject to revision, with the largest changes normally occurring in the latest quarters. Revisions can be large if values are initially treated as outliers but are later found to be part of the underlying trend.
We use X-13ARIMA-SEATS seasonal adjustment program to produce the seasonally adjusted and trend estimates. Irregular short-term variation is removed by smoothing the seasonally adjusted series using optimal weighted moving averages.
Comparison with building consent statistics
Building consent statistics provide an indication of upcoming building activity, but comparisons may be affected by variable timing and valuation differences, particularly following the Canterbury earthquakes.