There is something to be said for the design choice of federalism; pushing all possible decisions down to the smallest unit of population effected. However, our government has moved away from this since the Roosevelts time in power.
European Commission that basically runs European Union (500M people) has just around 25,000 employees [1].
This is tiny. Just to get an idea about the scale: New York City (8M people) government employs 250,000 people (though to be fair, I have no idea who is included in this total) [2].
EC seems to be doing something very right. They offer very good working conditions (pay, holidays, benefits, security), so they are able to attract good people. There is a tough competition for these civil servant jobs.
[1] http://en.wikipedia.org/wiki/European_Commission
[2] http://en.wikipedia.org/wiki/Government_of_New_York_City
Because they are massively, massively corrupt.
http://www.timesonline.co.uk/tol/comment/columnists/guest_co...
"For nine consecutive years the EU court of auditors has refused to sign off the budget. The numbers are huge. The annual EC budget is around ¤100 billion (£65 billion). The auditors cannot clear 95 per cent of that. We simply cannot tell what is happening to that money"
Or to put it another way, if this is true, it means that they are able to run real EC functions with just 5 billion Euro budget. That would quite efficient indeed.
My whole point was about low ranking civil servants that do the bulk of the nitty-gritty administration work and that all must pass through a highly selective competitive exams (1 in 29 applicants [1] cf. Harvard's 1 in 14 [2]).
And for these people, from my limited anecdotal evidence I got impression that they are indeed efficient, at least for bureaucracy standards.
For example, http://www.tfl.gov.uk/corporate/about-tfl/investorrelations/... shows that Transport for London (buses, trains and underground) alone employs about 18,000 people (and probably quite a bit higher now).
I've seen this commercially, too: large company buys small-medium company because it's good for both companies, even though it's demonstrably bad for employees and products of both. It's a tail-wags-dog situation.
What I was talking about is what economists call a true public service. This is a service that doesn't diminish in value when you add more users - a light tower is the classic example: The lighttower is as good a service to a ship whether it is the only one using it or there are a thousand others. Webpages are another good example.
Large countries have an advantage here: It costs the same to build a lighttower whether you are the US or Lichtenstein.
The military might have been a bad example though...
It seems ( I have no data to back this up) that the efficiency of bureaucracy decreases exponentially with it's size. This is obviously a bigger problem for large countries such as the US. There are advantages to size (setting the global agenda, advantages in areas where absolute amounts of money matter (for instance military - your military is twice as good if you spend twice the money no matter how large your population. Roughly...) . , etc.). Whether these advantages outweigh the disadvantage of a large bureaucracy I don't know.