[Don't miss the great discussion on this issue that has been ongoing at the propagation bench!]
UK Labour Party chair, Hazel Blears, commenting on the possibility of proportional representation in a future elected House of Lords, says:
I’m always torn on this. When I was a candidate in Tatton people always used to say to me: ‘We’re really good Labour people, we’ll never win and our vote doesn’t count’ [indeed it does not--ed.].
But equally the bit that I really dislike about PR is if you lose your constituency link.
I know people have got some models where you can still have a constituency link [indeed we do!--ed.]. But if you look at a lot of PR systems, particularly in Europe, you end up again with a political elite who all live in London or Madrid or Paris…
This is a rather stubbornly held view. It is also wrong. Even leaving aside MMP and STV and turning our attention to strictly list PR, the claim that the “political elite” (which I am understanding in this context to mean legislators) hails from and permanently resides in the capital city is not supported by the evidence.
My reseach on this question shows that with lower-magnitude districts (by which I mean under about 10 seats) legislators tend to have high rates of district nativity and prior service on local councils–probably higher than in the US House! This is so even when lists are closed, and thus as a legislator, “you are dependent for your place on the party list,” as Blears goes on to say in the extension of the quotation above.
When lists are open, on the other hand, the rate of local origins among legislators is higher still–and increasingly so as the district gets larger–because the competition for preference votes gives parties even greater incentives to nominate candidates with a “constituency link.”
Even more remarkably, research by Michael Latner (who comments below) and Anthony McGann shows that even in Israel (single national district closed list PR) and the Netherlands (which has nearly irrelevant districting and preference voting), there is substantial regional representation. From their abstract:
Although central metropolitan areas are somewhat over-represented in the legislatures, so are the most peripheral regions. This is due to the fact that parties tend to choose representatives from the geographical regions where they expect to be electorally competitive.
In other words, even where it would be least expected due to “national” PR lists ranked by party leaders, parties cultivate constituency connections. Why? Presumably because voters want them, and the highly competitive environment of proportional representation means that parties ignore constituent demands at their peril.
Related to this discussion, see Espen’s comment about a proposal to reduce district magnitude in South Africa, while retaining overall national proportionality.