Why doesn't UK growth reflect red hot business surveys?

While several surveys are roaring higher, the strength implied by these indicators isn't apparent in official data.

Deutsche Bank's chief UK economist, George Buckley, has looked into the problem ahead of the latest GDP figures, due on the 28 Janaury.

Analysts expect fourth quarter GDP to come in slightly lower than the previous two quarters (both of which saw growth of 0.8 per cent).

Buckley looks at three key measures for more clues.

Composite PMI

The composite purchasing managers' index should "track the rate of overall economic growth well", as long as you exclude retail sales, agriculture, the non-manufacturing portion of industrial production and government services - all of which are not included.

Quarterly GDP-ex growth has a 72 per cent correlation with the composite PMI since 1988, slightly lower than the 69 per cent correlation with overall GDP.

With composite PMI averaging modestly above 60 in both the third and fourth quarters of 2013, we would expect much higher growth than has been seen - in the region of 1.3 per cent a quarter.

CBI Industrial Trends Survey

A regression suggests quarterly manufacturing growth of around 0.5 per cent - lower than the second and third quarter average of 0.9 per cent.

But that might be in line with what might be expected for the fourth quarter of 2013 if manufacturing output growth in December was flat.

BCC Survey

The correlation between these two - weighted BCC and quarterly GDP growth - is just over 70 per cent.

Buckley notes that there's a similar story here as with the CBI data - "the [BCC] survey published early in the quarter is correlated with the official GDP print from the previous quarter which is reasonable given that the survey would have been conducted in the final month of the previous quarter."

A regression here implies growth of one per cent over the past two quarters - stronger than the 0.8 per cent growth rates seen in practice.

Summing up

Using all the surveys available (including some not mentioned above) Deutsche Bank estimates GDP growth at one per cent in the third quarter of 2013.

Buckley lists the following possible reasons for the discrepancy:

  • First, GDP and the surveys do not measure the same thing. The former measures the depth of output growth across the economy whereas the latter typically measure its breadth (the PMI, for example, simply asks respondents to indicate whether output has risen, fallen or stayed the same relative to the previous month – thus if all firms surveyed indicated that output had risen but that it had risen very modestly, the PMI would be far stronger than actual GDP growth).
  • The business surveys and GDP will include different firms within their respective samples.
  • The official data uses deflators to obtain a volume series whereas the surveys normally ask firms directly if their volume of output has gone up, down or remained the same.
  • When corporate insolvencies rise this will depress GDP as output will be lower, but it might support the business surveys as the average growth of the remaining firms will improve (i.e. survivor bias). Of course there have not been as many bankruptcies this time round as in the early 1990s recession, but they have nonetheless risen relative to pre-crisis. A more general point is that the surveys may not be good at capturing structural shifts in output growth.
  • At levels of over 60 on average in the final two quarters of this year, the PMIs were outside of the range we have seen in the previous history of this survey. As a result, we do not know if there are non-linearities in the relationship between the PMIs and GDP at these levels, raising questions about the usefulness of standard linear regression techniques.
  • Finally, the ONS has typically revised its forecasts for economic growth upwards in the past, something that the Bank of England captures through its ‘backcasts’. The final estimates of GDP published many years from now may show far stronger rates of growth, more consistent with the PMIs, than is currently the case. Moreover, upward revisions are typically associated with periods of stronger growth in the first place – such as now.