# How to estimate Robot density in a country?

I'm writing a thesis where a key element is the degree of automation (for Switzerland) through time.

I was counting on the IFR to get those data but I need to pay 2000k to get them... Does anyone have an idea of how to estimate that? maybe a proxy? I cannot think of anything...

Daron Acemoglu has written a fair bit about the impacts of automation on labour markets. Bearing in mind that mechanization and "robotization" are different in important ways (with that importance increasing as AI and machine learning become more sophisticated and ubiquitous) you could use his work as a starting point. Certainly, you'd be remiss not to cite it if you're writing a paper.

In https://www.nber.org/papers/w24196, Acemoglu outlines the challenges to the question you're asking and offers a framework of approach. In https://www.aeaweb.org/articles?id=10.1257/jep.33.2.3, the framework is applied, and there are links to the data which you can look at for insight into how to measure this stuff.

Friend,

You can construct and capital(specifically equipment) intensity (capital[equipment]/labor[hours} ratio) measure which can be used as a proxy for automation.

It would be useful to calculate as by dividing capital[equipment] input index to the number of total hours worked for a giving sector in the economy. This ratio would indicates that capital services (input) are increasing relative to labor (hours worked) and/or that capital[equipment] is becoming more important in the production process relative to labor.

This statistic can be an indicator of the presence of automation as firms achieve labor productivity gains through the substitution of capital to labor.

• How would this separate the automation component of capital? As you say this would only “indicate that capital services are increasing relative to labor” and have no bearing on the concentration of automation – Brennan Feb 6 '20 at 15:36
• In this case we don't have a measure of robots. So a good proxy would be to use capital but more specifically equipment as robots would most likely fit into this category. Sorry about not being more specific. the more use of equipment instead of human labor (hours) would represent automation. – Mike J Feb 6 '20 at 17:58
• It's not clear what level this work is being performed at (e.g., Masters' thesis or just motivation for a toy model), but in general it would be extremely difficult to argue that labour-augmenting technology improvements are a good instrument for the robot population. – heh Feb 6 '20 at 20:57
• True its not clear of the level of detail. However, if there is no data available one need to find some sort of proxy. – Mike J Feb 6 '20 at 21:20
• Im not convinced that this would work as a valid proxy, although, I can’t come up with a better one – Brennan Feb 6 '20 at 21:35