Can Intangible Investment Explain the UK Productivity Puzzle?

We have a new paper on this. An outline is below and the paper is available from here.  A comment on the interesting post by Hugh Small is below.

Here's the puzzle.

Between 2007 and 2009 UK market sector value added fell by 5.8%. Hours worked fell by 1.9%% and hence productivity fell by 3.9%.

In 2009, hours started to grow again, but output has grown very slowly. Between 2011 and 2012Q3, the latest period for which market sector data are available, hours have grown by 2.3% but market sector value added by 1.3%. Hence productivity has fallen by 1%. Why?

The explanation for the initial fall in productivity is labour hoarding. This seems reasonable.  Firms cut output but keep labour in reserve for the recovery. Productivity, output per worker, falls at first, but then recovers as the firm uses the reserve inputs.

This explanation seems to carry less and less weight for the post 2008 years, for it seems very unlikely that firms are still carrying underutilised workers four years on.  So it must be something else.
 
In this paper we thus examine the role of intangibles. Our starting point is the observation that whilst investment in tangibles, plant/vehicles/buildings has fallen and stayed low, a point perhaps not noticed is that investment in intangibles, specifically R&D and software has risen since the recession (software fell and has then been rising, R&D was flat and then rose). Consider then a firm who has reduced production but maintained investment in intangibles. Its skill level rises, since intangible investment typically requires high qualified workers. Its measured output falls, since the output of e.g. R&D projects might not manifest itself for a few years. Thus labour productivity falls, in a pattern that looks just like labour hoarding.

We also investigate some other mechanisms on TFP, see the paper.  But what do we find on labour productivity?

Our main findings are:

1. Because intangible investment has been growing but is not counted as value added, measured value added is understated.
2. In fact, market sector real value added growth since the start of 2011, at 1.3%, is understated by 1.1% (about 0.5%pa);
3. In terms of the labour productivity puzzle then, true value added is growing faster than measured, 2.4% rather then 1.3%, and since hours growth has been 2.3% over this period, productivity has not been -1% but +0.1%. 
4. Thus we believe that unmeasured intangibles are part of the explanation, but not all of it. 

For convenience, here is the key diagram on the different behaviour of intangibles and tangibles over the recession


Source: figure 3 of  Goodridge, Haskel and Wallis, 2013, paper available from here.

Hugh Small in an interesting post of Feb 1st 2013, has essentially advanced the same argument.  He correctly points out that the intangible investment is high fraction of GDP and that it is not measured.  He shows some data from the World Bank, on the fraction of GDP accounted for by intangible investment.  I am more familiar with our European project data that can be downloaded for free from www.intan-invest.net with a paper describing it here Corrado, Carol; Jonathan Haskel, Cecilia Jona-Lasinio and Massimiliano Iommi, (2012), "Intangible Capital and Growth in Advanced Economies: Measurement Methods and Comparative Results" Working Paper, June, available at http://www.intan-invest.net.  Figure 5 of that paper shows the fraction of GDP accounted for by the intangible investment we measure (software, R&D, design, investment in artistic originals, branding, training and business process engineering):



There are two points to note.  First, some of this investment is already counted in GDP, notably software with R&D to follow soon.  Second, whilst this is a potential effect on GDP levels, the effect on GDP growth is different. As set out in our paper, the addition to the growth of value added is the growth in real intangible investment (not in GDP) over and above real value added growth, times the share of that investment in overall GDP. That share is about 10%, real intangibles are growing faster than GDP by about 5% and hence the undermeasurement of real GDP is 0.5%.  (0.10 times 0.05). 

So I hope that this piece advances the points made by Small and is a contribution to the understanding of the puzzle.  My conclusion: it's not going to be one big thing, but a string of little things that will add up: Zombie firms for example are likely part of the puzzle too.

Zombie firms and low UK Productivty

The collapse in UK productivity since the recession is ever more mysterious.  Might it be the fault of Zombie firms: low productivity firms who are kept in business by the forbearance of banks and who would otherwise go out of business and raise productivity  via the beneficial averaging effect of their exit?

What do we know?

1. Disney, Haskel and Heden document that 50% of productivity growth in UK manufacturing over a decade is driven by the batting average effect of entry and exit of high and low productivity firms respectively.

2. The FT have a recent feature, with some interesting work  cited from the Bank

"The Bank of England recently lent the theory some weight, pointing out that about 30 per cent of companies were lossmaking in 2010, a bigger proportion than in the 1990s recession, yet corporate insolvency rates during this downturn have been much lower than in previous ones."

3. The ONS have an important new paper looking at changes in firm productivity up to 2009, using their very comprehensive data.  Since they have access to the company register, this is a very large sample of firms.

Their data suggest there may be something to the Zombie view. The figures below shows , for market services , the average productivity of firms in the lowest and highest quartiles of the labour productivity.  There is a fall in 2009 and a hint that the lower tail of firms are lower in those years than before.  Note that before the recession the upper tail had been widening.  So the recession seems to have cut off the very highest performers and lengthened lowest tail, widening the productivity distribution. 








One thing that is very puzzling however, is that small firms in market services are much more productive than large ones as their Figure 3 shows  (size class 4 are above 250 and class 1 below 20).

 

 And the averaging effect is shown here:





This shows that over the 2000s most employment growth was in the low productivity firms (Quartile 1). the left hand bars, which should have retarded productivity.  So what's going on?  My guess is this: large service sector firms employ likely a lot of part-timers (e.g. large retailers).   So per labour hour they are more productive, but per employee as shown here, less so.  Thus the interpretation of Figure 42 is that the sorting effect has beeen working in the 2000s to raise true productivity since its been working in favour of the large firms.  So what's been happening in the recession?  In 2008 and 2009  it looks like the smaller firms, that is the right hand lines, are gaining employment relatively more than the larger ones.  That is, the left blue line is lower relative to the right hand lines.  This agains suggets the sorting effect is working less well.


More puzzles but credit the ONS for using their considerable data resources to try to figure out this problem.