Bringing energy back in

For a general discussion of topics relating broadly to power and political economy (e.g., capital-as-power, Marxism, neo-classical economics, institutionalism).

Moderator: sanha926

Bringing energy back in

Postby blairfix » Mon Jan 20, 2014 4:53 pm

For a while now I've been labouring to apply the Capital as Power framework to my other research, which is mostly biophysical. It's been difficult, but here is one preliminary result.

Energy & Corporate Employment.png
Energy & Corporate Employment.png (161.43 KiB) Viewed 3461 times


This figure shows the correlation between corporate concentration (measured by % employment of the top 10 firms) in relationship to per capita energy consumption. Energy and labour force data are from the World Bank, all corporate data is from Compustat.

What are we to make of this correlation?

Energy per capita is a very good proxy for the level of industrial "development". It would seem that the concentration of corporate power and said development go hand in hand.
blairfix
 
Posts: 88
Joined: Fri Dec 20, 2013 3:45 pm

Re: Bringing energy back in

Postby sanha926 » Fri Feb 21, 2014 3:48 am

Hi Blair,

I have nothing new to say about this interesting chart, other than to agree with you that there seems to be a clear correlation between industrial "development" and corporate concentration.

My question has to do with the data source: how did you compile your list of the top 10 corporations and their share of total employment for such an impressive sample of countries?

Thanks,

Sandy
User avatar
sanha926
 
Posts: 103
Joined: Mon Nov 24, 2008 6:20 pm

Re: Bringing energy back in

Postby blairfix » Fri Feb 21, 2014 10:59 am

Hi Sandy,

All corporate data is from Compustat Global - Fundamentals Annual. I just filtered out the top 10 companies (by employment) for each country in the chosen year. I guess a caveat is that this is the Compustat top 10 ... I have no idea how complete their list is, so some companies might be missing.

Total employment data is from the World Bank. Dividing the two series gives the results shown.

This was inspired by a paper I recently read arguing that higher concentration of employment by large corporations "causes" a more equal society. I think the paper is bunk, but the empirical technique is interesting.

http://webuser.bus.umich.edu/gfdavis/Papers/davis_cobb_10_ROB.pdf
blairfix
 
Posts: 88
Joined: Fri Dec 20, 2013 3:45 pm

Re: Bringing energy back in

Postby blairfix » Sun Mar 16, 2014 1:49 pm

There seems to be a clear connection between breadth (growth in employment by dominant capital) and growth in energy consumption. Here are some US time series that make the case. The first shows employment concentration in the construction and manufacturing sector and this sector's energy use per labour hour. The second chart shows the same calculations for the service sector.

The most fascinating finding is the transition that occurred in the early 1970s, both in energy and corporate breadth, from a growth to stagnation regime. This correlates nicely with both the peaking of US domestic oil production and the transition from exponential to linear growth in the world oil supply.

Energy & Breadth 1.png
Energy & Breadth 1.png (175.4 KiB) Viewed 3294 times

Energy & Breadth 2.png
Energy & Breadth 2.png (181.18 KiB) Viewed 3294 times


I can find no explanation in existing political economy as to why breadth and energy growth aught to go together. This might follow if companies were pursuing only greenfield investment, spurring the growth in consumption. But if the majority of breadth occurs by way of mergers and acquisitions, why should this be related to energy?

I have no answers.
blairfix
 
Posts: 88
Joined: Fri Dec 20, 2013 3:45 pm

Re: Bringing energy back in

Postby DT Cochrane » Sun Mar 23, 2014 8:27 am

Sorry I can't give an answer either, but I do want to commend you on the research, Blair.

It is very interesting. After a question has been asked, it often feels obvious. But, we should recall that perhaps the most important part of good research is just coming up with the questions.
User avatar
DT Cochrane
 
Posts: 66
Joined: Thu Nov 21, 2013 6:49 am

Re: Bringing energy back in

Postby Mladen Ostojic » Mon Jul 28, 2014 1:12 am

Blair, thank you for sharing your research with us.

I'd like to comment on the last two graphs you posted, which seem to suggest a connection between breadth and growth in energy consumption. It occurs to me that some part of this connection might actually be between the employment and labour hours built into your data.

If I'm following, your measure of breadth involves dividing by a sector's employment and your measure of growth in energy consumption involves dividing by that same sector's labour hours. Since employment and labour hours should be correlated, they might account for at least some of the connection between your data series.
Mladen Ostojic
 
Posts: 5
Joined: Mon Dec 09, 2013 11:04 pm

Re: Bringing energy back in

Postby blairfix » Mon Jul 28, 2014 2:54 pm

Good point, Mladen.

I checked, and auto-correlation is not a problem. If we remove the denominator from each series we still get very high correlation. Here is the total employment of the top 200 compustat corporations (left, in thousands) and the total energy consumption of the US (right, in Exajoules):

Energy.jpg
Energy.jpg (16.55 KiB) Viewed 3019 times




Actually, I've recently become aware of a problem with least squares regressions. Any smooth, monotonic curve will be highly correlated with any other curve with the same properties.

This puts a great deal of econometrics in jeopardy, including much of economic growth theory. Real GDP growth is almost perfectly exponential in the long-run, meaning any other smoothly growing curve will give extremely high correlation. This is why neoclassical production functions work. We could just as easily assert that GDP is an exponential function of time: Y= a*exp(bt). Both will yield high correlation, both are non-nonsensical.
blairfix
 
Posts: 88
Joined: Fri Dec 20, 2013 3:45 pm

Re: Bringing energy back in

Postby DT Cochrane » Mon Jul 28, 2014 3:55 pm

Blair,
Econometrics is well aware of the correlation issue. One solution is to calculate the correlation of first differences. Make each series the simple change from one year to the next and then check the correlation. Many seeming correlations will disappear.
User avatar
DT Cochrane
 
Posts: 66
Joined: Thu Nov 21, 2013 6:49 am

Re: Bringing energy back in

Postby blairfix » Wed Nov 26, 2014 7:28 pm

Here's a little more data relating corporate employment concentration and energy consumption. This time I look at the global level. Data is only available for a short time period, from 1990 onward.

If we include China in the analysis, the results are not great (top graph), but if we exclude China the results are as expected (bottom graph). Given the results posted earlier in this forum, we expect corporate employment concentration to track with energy consumption. As the data shows, without China, both series
have remained approximately constant over the last 20 years.

Global 500 and Energy.png
Global 500 and Energy.png (188.51 KiB) Viewed 2795 times
blairfix
 
Posts: 88
Joined: Fri Dec 20, 2013 3:45 pm


Return to Political Economy

Who is online

Users browsing this forum: No registered users and 1 guest

cron