Brad DeLong described the “pig” economist as someone who dives into the details of the economy and wallows around in it until he gets some insight about what is happening. Meanwhile, Nick Rowe is wondering whether we are moving closer or father from being an “apple” economy. This post is a pig’s response.
Upside Down V
This first thing to note is the “V” shape in the capital-output ratio, from the mid-60s until the present day, with the peak occurring around 1982. The following Fred graph measures the tangible assets of non-farm non-financial businesses as a proportion of GDP.
Tangible assets consist of equipment and software together with real estate, with real estate the more volatile component (primarily due to valuation swings):
Whereas the trend in equipment and software, which are not subject to the same valuation swings, is more clear, particularly when we normalize by potential GDP:
Meanwhile, the gap between gross and net output is widening. I.e. a greater share of gross output consists of capital consumption:
A different story emerges if we look at the financial assets of non-financial businesses, normalized by GDP:
Note that this data comes from Z.1, and presumably includes goodwill, purchased patents and copyrights, etc, but I couldn’t find confirmation for this digging around the Fed’s website (and for some reason they can’t put the handbook online!). If anyone can verify, please let me know.
As an alternate measure, we can use the BEA Fixed Asset Stock tables. I used the quantity index of private equipment and software, and compared this to potential real GDP. Note that we are using two different indexes with different bases, but we can still compare the growth rate of one with the growth rate of the other. This series ends in 2009 and is only available annually.
Also note that computing quantity indexes is difficult for software and technology components: just because MS office has 10 times more features does not mean that 10 times the quantity has been supplied over the previous version, or that firms purchasing MS Office for their workers are making 10 times the capital investment that they were making when they purchased the previous version. You can imagine a pareto distribution of feature-use, with the most obscure features used very rarely by a handful of customers, etc. Similarly, if processors are 1o times faster than 5 years ago, this does not mean that 10 times as many processors were supplied, nor does it mean that this component of the capital stock has increased ten-fold. IMO this is why the BLS has some whacky price deflators for semiconductors and software in comparison to more plausible price deflators for things like capital equipment.