A change to the GDP model Nicole Hodges nhodges@arlingclose.com

As I’m sure you already know, the Office of National Statistics (ONS) releases quarterly publications of GDP data that estimate the economic activity in the UK. What you may not appreciate is the underlying publication model the ONS has been using for quite some time. How this briefly works is a ‘Preliminary estimate of GDP’ is released 25 days after the end of the quarter to which it refers. This is a pretty fast turnaround, beating the USA by 5 days to lay claim to the title of fastest-preliminary-estimate-of-GDP-in-the-G7! However, this only includes data on the value of the output of goods and services (‘gross value added’ or GVA) and no data on the income earned from them and expenditure spent on them (together the three approaches to measuring GDP).

As you can guess, the data content used in the preliminary estimate is therefore fairly low at around 45% and also only includes forecasts for the final month of the quarter in question, simply because the data is not yet available. 30 days after the Preliminary estimate, so 55 days after quarter-end, a Second estimate is published which includes output, income and expenditure data so by this time the data content has increased to around 65%. 30 days later, 85 days since quarter-end, the Third estimate is published and includes an average data content of 90%. This model has been used for a fairly long time but following a public consultation in 2017, the ONS has now (as of July 2018) introduced a new model with the aim to strike an improved balance between the timeliness of GDP estimates to aid policy makers and the quality of the available data. 

The new – and now current - model involves a change in the quarterly releases and the introduction of a monthly estimate of GDP. 

Instead of three GDP releases being published per quarter, there will now only be two. The first will be released 40 days after quarter-end, so just over two weeks later than the preliminary estimate would have been released in the previous model, but this time it will include data on income and expenditure as well as output. Despite sacrificing some timeliness, this means that the data content within the first estimate will be much greater, providing a more balanced picture of economic growth. The new system also means data on expenditure and income will actually come earlier than in the past, as these were previously only available in the old Second estimate. The most comprehensive estimate of GDP – the new Second estimate - will continue to be released 85 days after quarter-end.  

The newly introduced monthly GDP release will be published on a ‘short-term economic indicators theme day’, alongside the publication of Index of Services (moved forward by two weeks), Index of Production, Index of Construction and UK trade data. The monthly GDP estimate will only include information on the value of the output of goods and services (the GVA measure) as it is the timeliest of the three GDP measures and is the only one available on a monthly basis. The release will contain single-month estimates but will focus on rolling three-month growth rates, as data concentrating on a month on its own is likely to be more volatile and so not provide a good picture of underlying GDP growth. The rolling three-month data is calculated by comparing growth in a three-month period with growth in the previous three-month period, for example growth in March to May compared with the previous December to February. So we will now receive a monthly GDP update with a fairly high data content within one release. 

The first monthly GDP release was published on 10th July 2018 and showed the headline measure of three-month growth to May 2018 at 0.2%, up from 0.0% in the three months to April. The individual monthly growth rates were 0.0% for March, 0.2% for April and 0.3% for May.

The key point to understand about this transformation to a new model is the trading of timeliness in exchange for accuracy with regard to the quarterly GDP estimates. The old system released information earlier with a lower data content while the new model releases data two weeks later, but incorporating a much larger data content. The ONS hopes this will mean their estimates are more reliable and helpful, and ultimately lead to greater confidence in the GDP data. The introduction of monthly GDP estimates also means that extra data is being provided more quickly. This should be beneficial for economic analysis and help give policymakers the vital information they need.