Reconciling regional growth data: what do the latest data tell us?


Reconciling regional growth data: what do the latest data tell us?

By Stuart McIntyre

UK real GDP growth estimates were released by the ONS last week. These showed growth of 0.0% in Q4 2019 relative to the quarter before. Growth for 2019 as a whole was +1.1% relative to the same quarter the year before; and +1.4% on a rolling 4 quarter basis. It is the latter measure that our model uses.

In order to shed light on how this growth is distributed across the regions and nations of the UK, we have once again run our regional nowcasting model (KMMP).

For more details on the model used see the paper here.

Our headline growth estimates are contained in the table below.

Regional growth in the year to 2019 Q4 (i.e. on a rolling 4 quarter basis)

  2019 Q4


East of England 2.1%
North West 2.0%
South East 1.8%
Scotland 1.6%
UK 1.4%
Yorkshire and The Humber 1.1%
North East 1.1%
Northern Ireland 1.1%
West Midlands 0.8%
Wales 0.8%
South West 0.6%
East Midlands 0.1%

In addition to producing new estimates for 2019 Q4 for each region, our model has also produced revised historical GVA growth estimates. Importantly, these now condition on the annual GVA(B) data for regional growth for 2018 (and previous years) released by the ONS in December 2019.

In addition to these updated annual data, the ONS have been producing new quarterly Regional Short Term Indicators (RSTIs) since September of last year. These are ‘experimental statistics’ and use VAT turnover data (which is also used to produce UK GDP measures) to produce a timelier and higher-frequency measure of economic activity in the English regions and Wales. There are also separate short term growth estimates for Northern Ireland and Scotland produced by the devolved administrations.

To see how these new RSTI data compare historically with the ONS’s annual GVA(B) data and our model-based estimates we contrast the three sets of output growth estimates for 2018. Aware of data revisions, in fact we consider two RSTI/devolved administration data vintages (for October 2019 and February 2020). These now four sets of regional growth estimates are contrasted in the chart below. The KMMP (model-based) estimates shown (in pink) are those produced prior to the model conditioning on the ONS’s estimates for annual regional growth in 2018, as published in December 2019. By construction, the latest KMMP estimates of growth in 2018 equal those from the ONS, shown by the black squares for each region. The grey bars indicate the growth estimates from the October 2019 RSTI publication (and the devolved administration releases around the same time); and the red bars show the latest RSTI estimates as well as the latest estimates from the devolved administrations.

Chart 1: Comparing estimates of regional growth in 2018

This chart therefore shows that, in broad terms, our KMMP model-based estimates from last year proved to be consistent with the pattern of growth across the UK regions as subsequently estimated by the ONS and the devolved administrations in their various publications. However, if we focus on the ONS’s GVA(B) data for 2018 (as published in December 2019) as the target outcome, we do see some differences.

The most obvious of these is for Northern Ireland. Our KMMP estimate for 2018 was +1.0%. In October 2019, the Northern Ireland Statistics and Research Agency (NISRA), who produce the short term output measure for Northern Ireland, similarly put growth in 2018 at +1.1%. In contrast, ONS in December 2019, in their ‘National Statistics’ data release, put growth in 2018 at -0.9%. In January, after the release of ONS GVA(B) data, NISRA estimated growth in 2018 to be 1.2% (red bar in the chart above).

Similarly, in Scotland, the Scottish Government produce a short term output measure (of GDP growth). Their latest estimate for 2018 is that the Scottish economy grew by 1.4% (this is basically unchanged across their October 2019 estimate and the latest data released last month). Our last KMMP estimate put growth in Scotland in 2018 at 1.3%. This is close to the estimate from the Scottish Government, but higher than the ONS’s December 2019 estimate of +0.9%.

It is important to note that the devolved administrations take a different methodological approach to calculating economic activity than the ONS, which may help explain the difference between these estimates. The KMMP model-based estimates are also, of course, produced ahead of these ‘official’ estimates from the ONS.

Aside from the estimates for Scotland and Wales, with the advantage of hindsight our model appears to have been too optimistic about growth in the South West of England (+2.2% v. +0.8%) and East Midlands (+2.1% v. +1.1%), and too pessimistic about growth in the North West (+0.5% v. +1.3%) and Wales (+0.4% v. +1.4%). In most other cases, our estimates are broadly similar to those subsequently published by the ONS.

Having a number of different estimates which tell a slightly different story is not ideal of course. However, it is important to note that the RSTI data are new, and thus very much still in development. Our own estimates are just that – model-based estimates of the likely allocation of UK growth based on the historic pattern of regional growth, national growth and macroeconomic indicators. Like any estimate, especially one produced ahead of ‘official’ data, they are estimated with uncertainty; so we caution users from reading too much into small differences between the point estimates from KMMP and ‘official’ estimates of growth subsequently published by the ONS and the devolved administrations.

ONS continue to develop their RSTIs. Meanwhile, our work in the coming months is to incorporate these RSTIs into our model. This will help produce more accurate model-based estimates of regional growth that condition on and are consistent with the various ‘official’ estimates of regional output growth produced by the ONS and the devolved administrations.

You can access our full UK Regional GVA Quarterly Estimates spreadsheet from here.

ESCoE blogs are published to further debate.  Any views expressed are solely those of the author(s) and so cannot be taken to represent those of the ESCoE, its partner institutions or the Office for National Statistics.

About the authors

Research Projects

Related publications