News in Charts: Ignore the retail sales data – a contraction in UK output in Q3 is highly likely

Last week’s retail sales data tell us next to nothing about the health of the UK economy after the Brexit vote. We have put together our own indicator of underlying economic activity based purely on economic survey data. This (known as Fathom’s UK Economic Sentiment Indicator) fell sharply in July, to -0.4%. That was the lowest reading since early 2009. We will get a better picture of actual GDP growth in Q3 as we move through the quarter. Nevertheless, at this stage a contraction in output appears highly likely.

Refresh the chart in your browser | Edit chart in Datastream

Last week’s UK retail sales data were much stronger than expected. Total sales volumes rose by 1.4% in July – above even the highest estimate submitted to Thomson Reuters in their regular survey of economists. What does this tell us about underlying activity in the UK economy in the wake of the EU referendum? Probably next to nothing. Monthly retail sales data are notoriously erratic, and in the short term, depend more on the vagaries of the weather than on economic fundamentals.

Refresh the chart in your browser | Edit chart in Datastream

Retail sales and claimant count unemployment aside, we presently have very little in the way of hard economic data that relate to a period of time entirely after the Brexit vote. What we do have is a wide range of surveys. Some were moderately upbeat, and some were very weak. How should we weight all of these sometimes contradictory surveys together? Principal component analysis (PCA) offers one approach. Effectively, it is a method of distilling a wide range of economic data into just a few summary statistics.

UK Economic Sentiment Indicator

We apply PCA to the responses to a total of eleven questions from three closely-watched economic surveys. We consider five questions from the Bank of England Agents’ survey, four from the Markit PMIs, and two from the GfK consumer confidence survey. All of these surveys extend to July 2016. We find that the first principal component by itself is able to account for close to 60% of the variation in the underlying data. Moreover, each of the survey responses contributes positively to the first principal component. Because a more positive balance is seen as indicative of stronger economic activity in each case, this is a desirable property.

Refresh the chart in your browser | Edit chart in Datastream

We have transformed the first principal component so that it has the same mean, and the same variance, as quarterly GDP growth. The resulting monthly series, which we label the Fathom UK Economic Sentiment Indicator (UKESI), is shown alongside quarterly GDP growth in the first chart. The UKESI fell by 0.6 percentage points between June and July, from 0.2% to -0.4%. That was the biggest one-month drop in the near 20-year history of our indicator, while the July reading by itself was the weakest since early 2009.

How should we interpret the July reading? The UKESI is more persistent than GDP growth. By construction, it has the same mean and variance, but it displays less short-term volatility. In that sense, we might interpret it as a measure of underlying economic activity, rather than the best possible prediction of actual GDP growth. Actual GDP growth is likely to be more volatile in any given quarter than our survey-based UKESI simply because the ONS will be able to identify ‘lumpy’ economic transactions that, because of their design, the surveys cannot hope to capture. Over time, as we receive more official statistics – such as the Index of Production, Output in the Construction Industry, and the Index of Services – we hope to build a more reliable estimate of actual GDP growth in Q3. Nevertheless, for the time being we conclude that survey readings for July, if sustained, appear consistent with a contraction in output of around 0.4% in the third quarter.

Refresh the chart in your browser | Edit chart in Datastream

Refresh the chart in your browser | Edit chart in Datastream

Looking further ahead, we continue to believe that the UK, although flirting with technical recession through the second half of this year, will avoid a severe downturn. A reduction in the level of output in Q3 is very likely, not least because Q2 appeared erratically strong. But we expect to see a gradual recovery in economic activity, and in the survey balances, as we move towards the year end.