National Accounts - Quarterlies December 2020

Gross domestic product: December 2020 quarter – changes and revisions Introduction Gross domestic product (GDP) sources and methods are well documented (see: Quarterly gross domestic product: Sources and methods (second edition), with a further update in Quarterly gross domestic product: Sources and methods (fifth edition) ). The quarterly indicators used to compile our estimate of GDP are the best that we have available at that point in time. When compiling our quarterly estimates, we usually face trade-offs between timeliness and the quality, coverage, length of historical series, and/or conceptual alignment of available data sources to the component it is meant to represent. As a result, the methodologies that we use vary depending on the data available and are underpinned by certain assumptions. In normal periods of activity, these assumptions provide us with reasonable estimates of activity. In periods of abrupt disruption (such as the alert level 4 lockdown) resulting in changes to activity and temporary shifts in the structure of the economy, some of these relationships weakened or no longer held. Changes made due to COVID-19 There are fewer series that required alternative indicators in the December 2020 quarter. We’ve continued to review and refine approaches applied in the March, June, and September 2020 quarters by confronting our usual quarterly indicators against other data sources to identify instances where our usual quarterly indicators were unable to represent the actual level of activity through the quarter. If appropriate, we used alternative data sources in place of the usual quarterly indicator. See December 2020 quarter gross domestic product and COVID-19, for details of the alternative data sources and changes in methods applied. Seasonal adjustment treatment Sharp changes in activity, such as those caused by the lockdown, pose challenges for our usual seasonal adjustment processes. We have reviewed all series and have applied outliering where required. This has the effect of subduing the impact of unusual data points on the seasonal adjustment process. Revisions to previously published statistics Revisions are usually driven by new or improved data that affect historical results, or changes to the underlying measure. In both cases we aim to better represent the real-life situation of a changing society, economy, and environment. The most significant revisions introduced this quarter are described below. Data errors in previous quarters During input data processing for the December 2020 quarter release, we spotted that some input data for previous quarters had either missed being loaded or had been loaded incorrectly. This impacted on: . changes intended to be introduced during annual incorporation for retail trade and accommodation were unintentionally omitted when an updated program was not moved from test to production. The impact of the error is an upward revision of approximately $148m to Retail in the March 2019 year and $189m to the March 2020 year, with the biggest quarterly impacts being an upward revision to Retail of $50m. Revisions apply from June 2018 quarter to Sept 2020 quarter. . dairy prices for inventories (affected change in inventories for the March, June, and September 2020 quarters) . data relating to package holidays within household consumption expenditure (HCE) for the September 2020 quarter. These issues have been rectified and additional processes put in place to lessen the likelihood of these issues being repeated. Incorporating annual agricultural production statistics We incorporated data from Agricultural production statistics: June 2020 (provisional). The data replaced earlier published data and modelled stock levels. Incorporating the data led to revisions to agricultural activity in GDP, and to agricultural inventories in the expenditure measure of GDP. These changes caused revisions from the March 2019 quarter onwards. Methodological changes Updated financial data from the business financial data collection (BFD) has led to some revisions where we have been using the BFD data collection as alternative indicators from our normal methodology. This contributed to revisions in: · mining . manufacturing . retail trade and accommodation . transport, postal and warehousing · local government administration · healthcare and social assistance · arts and recreation · household consumption expenditure . gross fixed capital formation. We also reviewed our data sources and methods to identify methods that would be unable to capture the change in activity caused by COVID-19 and efforts to contain it. During this process we identified inconsistencies in how interpolation methods were being applied. We have standardised the approach used, which resulted in revisions from the June 2019 quarter onwards to the following series: . metal ore, non-metallic mineral mining, and quarrying in mining . own account capital formation for electricity generation and on-selling, and electricity transmission and distribution in electricity, gas, water, and waste services . water transport, other transport, postal and courier services, and warehousing and storage in transport, postal, and warehousing services . library and other information services in information, media, and telecommunications . non-residential property operators in rental, hiring, and real estate services . public order, safety and regulator services, and local government nonmarket in central government administration, defence, and public safety . consumption of fixed capital in local government expenditure. Revisions due to updated indicator data We received updated respondent information and other source data, causing revisions to: · forestry and logging · fishing · mining . manufacturing · electricity, gas, water, and waste services . construction . wholesale trade · information, media, and telecommunications . financial and insurance services · rental, hiring, and real estate services · local government administration · household consumption expenditure . central government final consumption expenditure . local government final consumption expenditure . gross fixed capital formation. . exports of goods and services . imports of goods and services. Revisions to inventories Revisions were also made during the process of reconciling change in inventories with other movements in production and expenditure series. This process is designed to minimise inconsistencies in timing and valuation. Revisions table The following table shows previously published and revised quarterly movements for the June 2018–September 2020 quarters’ GDP and expenditure on GDP (GDE). Previously published and revised quarterly movements Gross domestic product Gross domestic expenditure Percentage change from previous quarter Quarter Previously published Revised 18 Mar 2020 Previously published Revised 18 Mar 2020 June 2018 1.1 1.2 1.4 1.4 September 2018 0.2 0.2 0.6 0.5 December 2018 1.1 1.1 1 1.1 March 2019 0.4 0.5 0.9 0.9 June 2019 0.4 0.5 0.3 0.4 September 2019 0.7 0.7 0.7 0.6 December 2019 0.1 0.1 0.2 0.4 March 2020 -1.2 -1.2 -0.9 -1.1 June 2020 -11 -11 -9.5 -9.1 September 2020 14 13.9 15.6 14.3 Source: Stats NZ Magnitude of revisions for March, June and September 2020 quarters Revisions to GDE were larger than revisions to GDP. Methods and data sources used in GDE are generally weaker and less stable compared to GDP. There is a higher level of uncertainty associated with measuring the significant changes in economic activity resulting from the pandemic. We have used extra data and careful analysis to minimise this uncertainty. As more detailed and complete data becomes available, including through the 2021 and 2022 annual national accounts cycles, we expect revisions to our initial quarterly estimates. It is likely that these revisions will be of a greater magnitude compared to those observed in previous quarters and revision cycles. In real terms (i.e. constant price actuals) GDE was revised upward by around $200m in the June 2020 quarter and downward $480m in the September 2020 quarter. The three most significant revisions and their causes were: Household consumption expenditure Household consumption expenditure was revised upward by around $100m in June and downward $120m in September due to revisions in HCE Services and expenditure on non-durable goods. HCE services were revised due to updated BFD data and revisions to some deflators. HCE non-durables revised in the June 2020 quarter largely due to updated travel credit ratios that are used to apportion travel credits to HCE classes. Change in inventories Change in inventories was revised upward by around $130m in June 2020 quarter and downward $290m in September 2020 quarter due to revised data from the following sources: . Agricultural inventories incorporated data from the latest agricultural production survey . Manufacturing and distribution inventories were revised following revised inventory data published by the quarterly manufacturing survey and the retail trade survey. The regular process of reconciling our change in inventories with movements in other production and expenditure series also contributed to revisions in manufacturing and distribution inventories in the June and September 2020 quarters. Gross fixed capital formation Gross fixed capital formation was revised downward by around $60m and $5m in the March and June 2020 quarters, and upward by around $80m in the September 2020 quarter. Revisions to gross fixed capital formation in the March to September 2020 quarters were largely due to revised quarterly building activity survey data data (impacting on residential buildings and non-residential buildings), revised BFD data (impacting on other construction), revised plant, machinery and equipment input data from the quarterly manufacturing survey, and updated employment monthly survey data impacting on intangibles (research and development). en-NZ