The team is joined by Guest Kats Rosie Burbidge, Stephen Jones, Mathilde Parvis, and Eibhlin Vardy, and by InternKats Verónica Rodríguez Arguijo, Hayleigh Bosher, Tian Lu and Cecilia Sbrolli.

Monday, 1 February 2010

Do more equal societies produce more patents? Some very preliminary data

Do more equal societies produce more patents? Well - leaving aside whether patent applications can measure "creativity" or "use of human resources" - first, what is a "more equal society"? Let's assume a society with a more equal income distribution, as measured by the Gini index (also called the Gini coefficient). You can find a list of countries ranked according to their Gini index (or other measures) here.
WIPO publishes as statistic on the number of patent [corr.] applications per resident; the latest (more or less complete) figures are from 2005.
To check for a relationship without using sophisticated statistical analysis, we can display the data using a scatter plot; plotting the Gini index on the y-axis and the number of patent applications per million residents on the x-axis. The result looks like this (click for larger version):

This graph is pretty useless. Because the variability of the number of patent applications is so much larger than the variability of the Gini index, most countries are concentrated in a small area on the left hand side. A logarithmic transformation of the data yields this:There does not seem to be much of a (linear) relationship between number of patent applications per resident and the Gini index as measurement of income equality. But the professor (allegedly) said

If you look at patents per head of population in the rich countries, it's the more equal ones that do better, partly because more unequal societies are wasting their talent.
So you have to compare rich countries only. "Rich countries" is also a relative term. I took the EU member states, plus USA, Canada, Australia, New Zealand, Switzerland, Iceland, Japan and South Korea. Most people would agree that these are relatively rich countries. The plot looks like this:Again, the data points are all over the place. There does not seem to be much of a (linear) relationship between the equality of income distribution and the number of patent applications per resident. Admittedly, this is not a sophisticated statistical analysis, but if there was a trend in the data, you should be able to see it in the scatter plots [the authors of "The Spirit Level" maintain that "There is a weak but statistically significant tendency for more equal societies to gain more patents per head than less equal ones"; this can't be disproved with the plots in this post. The key word here is "weak".]

What is definitely noteworthy are the extreme outliers of Japan and South Korea - they just have a lot more patent applications per capita than anyone else, including the USA.

Correction 2 Feb 2010: the WIPO data concerns patent applications, not granted patents. The post has been amended accordingly.


Luke Ueda-Sarson said...

Aren't those figures

a) for applications rather than actual patents, and

b) for applications per unit GDP, and not per unit population?

The spreadsheet is entitled "Resident patent filings per $billion Gross Domestic Product (1995-2007)" after all.

Cheers, Luke

Derek said...

Mark, I think your figures are patent applications rather than granted patents - according to the 2006 report on the trilateral site (, Japan had 360,000 first filings (or around 2800 filings/million) but only 123,000 grants (970/million) - which if you accept the same grant to application ratio for Japanese and non-Japanese filers (around 85% in 2008) means about 105,000 patents (820/million) granted to Japanese filers. Still, Japan and Korea will remain outliers, probably for the reason everyone assumes - many small applications rather than fewer big ones rather than any "economic equality".

Sam said...

Thanks for the scatter plots.

In "The Spirit Level: Why Equality is Better for Everyone", the authors plot equality v. outcome for the individual states of the USA. It may be interesting to try this for patents per head in the US (or perhaps Korea or Japan).

Mark Schweizer said...

@Luke: a) the link led to the wrong data; I did the plots with the per capita data, not the per unit GDP. The link is now corrected.

b) you (and Derek) are right, the WIPO data concerns applications, not granted patents. The post is amended to reflect this.

David said...

A nice bit of work, Mark. Thank you. I think the clue to the authors' predisposition to their conclusion of a tendency for more equal societies to gain more patents per head than less equal ones (which I guess they think is a good thing) is the title of the institute they work for. The real conclusion, however, is that there is no conclusion to be drawn, because the number of patents granted per head has very little to do with how equal a society is. The authors are simply making their conclusion fit their preconceptions.

Anonymous said...

Actually if you limit the analysis to English speaking countries - USA, Ireland, Australia, Canada, New Zealand, UK, the Log graph begins to resemble something approaching a straight line, tenuous I know, but interesting nonetheless.

Mind you, this doesn't seem to work for other languages, Austria and Germany are not in a straight line with the axis (oops poor choice of wording there, I mean the point where the two axes meet!). The same for Slovakia and the Czech republic. The line for the Scandinavian countries, (Norway, Sweden, Denmark) is pointing the other way (the less free the more innovative!).

What was it Disraeli said about statistics?

Anonymous said...

I'm sorry, but we're clearly confronted here with the old GIGO problem. Those WIPO stats are clearly bunkum, or at least they do not square up at all with related EPO stats. Case in point: according to WIPO, Spain is supposed to have a higher number of patent filings per unit population (72.8) than Belgium (42.73). However, not only does that go against anecdotal evidence, but is also directly contradicted by EPO stats, which show 29.8 European patent applications (including Euro-PCT) per unit population for Spain and 178.1 for Belgium.
Which raises the following questions:

a) Which "patent applications" is WIPO counting (national, regional, international)? Is it perhaps also including utility models or even designs?

b) Are WIPO stats perhaps just as reliable and well-founded as the age of its ex-president?

Michael Factor said...

As China and the USSR have become less egalitarian, the numbers of patent filings have gone up drastically, out of proportion with population fluctuations.

In the UK in the 19th century, it was the industrialists that created wealth spikes and patents.

Seems like another IP book not worth reading...

Subscribe to the IPKat's posts by email here

Just pop your email address into the box and click 'Subscribe':