big is not always best
In macroeconomic management, like so much in life, size matters. However big is not always best for, also like so much in life, it is possible for things to get too big, or that they simply be the wrong shape. From this you get the conglomerate discount, with the whole is worth less than the sum of it's parts, or the permanent crisis in the NHS because it is imply far too big to manage. Or even in completely different arenas such a software where both Apple with their Copland project and Microsoft with Longhorn found their projects becoming so complex that they had to either scrap the project completely, or wipe the slate clean and restart it.
The same is true of nation states. If you lump several smaller states of provinces together artificially then the result will not be as good as state that evolved naturally. Stumbling and Mumbling points to some economic research that shows how artificial states perform worse than natural ones with a view to the possible break up of Iraq.
Now there can be no more artificial a state than the state the EU was designed to become. The process that created the EU, of absorbing existing nations wholesale, makes trying to count partitioned groups a poor measure of artificiality since it's artificialness does not come from splitting groups, as was the case in the African countries studied in the paper that Stumbling and Mumbling is linking to, but in lumping them together. There is luckily another measure of artificiality than the ones that they used, Ethnolinguistic Fractionalization.
Fractionalisation can be calculated with the following formulae.
where there are N ethnic groups, group i has population pi and the national population is P. I have taken my data primarily from the CIA World Book. I have been forced to exclude France because of the French government's refusal to collect any statistics of ethnicity, so there are none of the populations of the ethnic groups of France included and the population of france has been removed from the EU total.
This leads to a fractionation value of between roughly 0.905 and 0.904 depending whether the populations consider themselves currently to be hypernated (like for example an Italian-American) or partitioned (like the Kurds who often consider themselves part of a single nation of Kurdistan which just happens to have been split between 4 different states) respectively. For comparison the fractionalisation value of the UK is 0.289, the UK being itself an artificial country which has several political parties that have been set up specifically to break it apart.
The Ethnolinguistic Fractionalization was used in William Easterly and Ross Levine's 1997 paper Africa's Growth Tragedy: Policies and Ethnic Divisions showed that per capita GDP growth is inversely related to Fractionalization. Luckily with good enough institutions these problems can be mitigated which despite the rampant fraud, inability to get their accounts signed off for 11 years, and massive democracy deficit the EU probably does have when compared to the Kleptocrats that have cursed Africa for so long. This is a good thing since the levels of fractionalization in the EU are also higher than in Africa, and as the Alberto Alesina, William Easterly and Janina Matuszeski paper that Stumbing and Mumbling linked to shows it is at the outside edges where these effects really take off.
But enough reasoned argument, back to ranting.