Auckland Maori seats: good idea? [updated]

The main stumbling block currently to Auckland's merger into a mega-city isn't worries about the loss of Tiebout competition - which seems to be only of concern to me, see here and here, though NotPC also agrees and I may have convinced Matt Nolan (see comments, oops) Rauparaha that this is worth worrying about.

Rather, the biggest concern seems to be whether city council will have a standard set of wards or whether it will follow the system used for national elections in having overlapping districts: one set for those voters who want to elect to be on the Maori roll, and another set of general electorates for those who don't.

Correct me if I'm wrong, but I don't think there's anywhere else in the world that uses this kind of system.

Does having these kinds of overlapping districts help Maori? I suspect not much under proportional representation except inasmuch as they help to facilitate overhang: many voters in the Maori electorates tick the Maori Party as their district Member of Parliament but give their Party vote to Labour. The Maori Party then wins more seats than those to which it is entitled by its list vote and Parliament is made bigger. In a PR system, this means that the Maori Party has a seat or two more than it otherwise would and so has a bit more clout as a minor party.

Auckland's council, though, as best I understand it, isn't planning to go PR. What are the effects of this kind of system when you have only ward-based representatives?

Imagine that there are two groups, A and B. Everyone that self-identifies as group A, 10% of the population, is 90% likely to vote for Party 1 and 10% likely to vote for Party 2; everyone in group B, 90% of the population, is 47% likely to vote for Party 1 and 53% likely to vote for Party 2. Suppose every district has 100 voters and groups A and B are evenly split across all districts. In that case, every district has 9 A-group and 42 B-group voters for Party 1, 1 A-group and 48 B-group voters for Party 2, and Party 1 wins all districts. Now, suppose that you instead create special A-group electorates that comprise 10% of the seats. Now, Party 1 wins all A-group electorates but Party 2 wins all other seats. You've made sure that Group A is represented, but at the cost of vastly reducing their influence.

Indeed, this has been found to be one of the effects in the US of judicial mandates for majority-minority districts. Heck, even Wikipedia knows it:

Gerrymandering may be advocated to improve representation within the legislature among otherwise underrepresented minority groups by packing them into a single district. This can be controversial, as it may lead to those groups' remaining marginalized in the government as they become confined to a single district. Candidates outside that district no longer need to represent them to win election.

As an example, much of the redistricting conducted in the United States in the early 1990s involved the intentional creation of additional "majority-minority" districts where racial minorities such as African Americans were packed into the majority. This "maximization policy" drew support by both the Republican Party (who had limited support among African Americans and could concentrate their power elsewhere) and by minority representatives elected as Democrats from these constituencies, who then had safe seats.

Enough of Wikipedia; turn to Google Scholar. Cameron, Epstein and O'Halloran, 1996, find that there is a very real tradeoff between advancing the number of minority representatives and advancing the issues that minority voters care about. Shotts, 2001, finds that a majority-minority mandate has no effect where Republicans control redistricting (because they already would have packed minority voters into a small number of districts, by my read) but reduces Democratic seats where Democrats control the redistricting and where the majority-minority mandate requires supermajorities; in this latter case, it can cause the Democrats to lose control of the state legislature.

I can see why the Maori Party would support this kind of move: it guarantees them a couple of seats (yes, yes, city council elections aren't party-based, you get the idea though). But it may well also guarantee that no other city councilor has to care at all about Maori issues because they won't have any Maori constituents. But that's too conspiratorially-minded. Most likely, nobody involved in the process has bothered to check the empirical literature, just like nobody involved in the process has bothered to worry about Tiebout.

Careful what you wish for, folks. The Law of Unintended Consequences is a stern one.