Recently, I've heard the claim that as the business cycle slowly turns toward recovery in the developed world, investment may not provide the impetus it has in previous upswings. The theory is that the economy is less capital-intensive now because services account for more economic activity, and so there is less need for investment.

Sounds reasonable. But I thought I would roll up my sleeves and see what the data say. There are various ways to measure capital intensity of production, but perhaps the best metric in this case is the capital-to-output ratio.

The red line in the first graph below shows the capital-to-output ratio for the US since 1997, excluding certain sectors. I've tried to keep the focus on business investment, so I have excluded government and housing related sectors. I have also excluded mining because much discussion at home treats mining separately. We see that capital intensity has actually gone up in the US, so capital is more important.

Nonetheless, it is still worth asking how much of this change is accounted for by changing industry mix. We can do this by a 'shift-share' analysis. Basically, we ask: what would the aggregate capital-to-output ratio be if we kept that ratio constant at the 1997 level for each of the 17 industries included in the graph, and let the industry mix of the economy evolve as it did historically (eg. if the manufacturing share fell from 20% to 10%, that would affect the ratio, but anything that happened to the capital to output ratio within manufacturing itself would not). This result is the blue line, which has basically tracked sideways, meaning the industry mix has been an unimportant influence in the US.

I've done the same exercise for Australia. The result is below.

Since 1990, both lines have moved down, but the red has moved down much more than the blue. Moreover, the blue line has basically moved sideways since the early 2000s. So, as in the US, industry mix has played a secondary role.

Since 1990 the largest contributor to the fall in the blue line is the decline of manufacturing. But that decline must be close to running its course. In 1990, manufacturing accounted for 15% of output. It now accounts for 7%. How much further could it fall? It's mathematically impossible for it to fall as much as it has in the last 17 years (8%), so my guess would be that this blue line will keep tracking sideways.

What I do I take from all of this? It looks like industry mix is an unimportant influence on the recovery in investment. I do think investment will be a less important driver of activity than in the past, but not for reasons related to the structure of the economy. Fundamental determinants of investment include multi-factor productivity growth, depreciation rates, and labour force growth. These have generally fallen over the past decade or so, and should have had an appreciable effect on investment demand. That may be the subject of another post.