Interesting paper from 2005 published in Structural Change and Economic Dynamics: http://www.fraw.org.uk/files/economics/ayres_2005.pdf
This paper starts by noting that under normal assumptions, i.e. a constant returns to scale production function, changing factor endowments cannot explain the change in economic output over the 20th century. We need to postulate an additional factor: usually `technological progress'. This conclusion is qualitatively unaltered even if we expand the factors we consider from labour and capital to also include energy inputs.
However, if raw energy inputs are converted into `useful work' using estimated efficiency factors, then the change in labour, capital and useful work, at least over the 1900 - 1970 period, does seem to explain the change in economic output.
Is this an anodyne statement? Is every technological improvement basically just an improvement in the conversion of energy resources into useful work? Should we be surprised by this result? In particular, are we comfortable with idea that the productivity of labour has stayed constant and it's just that each unit of labour has more capital and joules to play with?
The paper can be criticised for introducing non-standard production functions that just serve to confuse the issue - they may fit the data better, but if the driver is just the use of useful work as a factor of production then they should keep it simple by explaining the issue purely in these terms. The authors claim that their results hold even if we just use a Cobb-Douglas production function.
The other interesting issue brought out by this paper is the divergence of their economic output as predicted from labour, capital and useful work, with actual economic output, post 1970. The authors postulate that this could be due to labour and capital using useful work more efficiently as prices increased in the 70's oil shock (this explanation is akin to Hassler, Krusell & Olovsson) or perhaps to to the rise of information technology. My only thoughts here were that it's interesting that this is also the point in time at which there is the divergence (at least in the US) of GDP per capita and median income, and between the GDP deflator and the CPI index (so that the economic statistics are "either overstating inflation (and hence understating income gains) or overstating economic growth").