Wednesday, August 22, 2012

2008 Election Revisited

The Electoral Commission has recommended doing away with overhangs that increase the size of parliament - what I call 'external overhangs' - for sub-4%-threshold parties if the Commission's 'no one electorate seat exemption/waiver' recommendation is accepted.

Its reason for this linkage is:

For example, if the one electorate seat threshold had not applied at the 2008 General Election and the current provision for overhang seats had been retained there would have been eight overhang seats.
But this example and reasoning assumes the crudest possible way of doing away with the electorate seat exemption, one that makes all MPs for sub-threshold parties overhangers. Better to allow sub-threshold parties to keep any party share entitlements they have up to their number of electorate MPs, i.e., # of quotients for sub-threshold party i = min(# of i's electorate wins, # of quotients i's party vote share would entitle it to).That sort of proposal strips out tag-along list MPs of the ACT/United Future/Progressive sort while preserving genuine overhangers - the genuine inconsistencies with/violations of party vote share that so far only some Maori party MPs have been.

In the 2008 election, Progressive, United Future and ACT all earned enough party vote share to 'cover' their electorate successes, and the Maori party earned enough party vote share to 'cover' three out of its five electorate wins.
Here's what the 2008 election looks like under various regimes holding voting behaviors constant (click to enlarge):

The Commission's proposal is unjustified overkill, mostly leading to less proportional outcomes (LSQ(2008, Report policy) = 3.046 > 2.932 = LSQ(2008, external overhangs w/o tag-alongs), and it should be abandoned.

A simpler example than what the actual world provides us with may be valuable (click to enlarge):

In this toy case the disproportionalities are exaggerated and the separation between the Report's policy (III) and our proposal (IV) is correspondingly clearer (23.5 > 18.5). Indeed (III) gives the most disproportional outcome in this toy case, which isn't true in the real-world Election 2008 case. Finally, note that even if our proposal didn't generally have the edge in proportionality/non-distortingness, it might still be worth accepting for its basic intelligibility as a hack to avoid an allegedly nasty consequence (i.e., of too large overhangs given the maxially natural and intelligible policy).

Is trouble with its (internalize the) overhang prescription the Report's only infelicity? Decidedly not. More on that soon. But note that the allegedly fiendish 2008 Status Quo saw Act get 5 MPs for its 85K party votes while NZ First got 0 MPs for its 95K party vote. Quelle horreur! Under the Commission's policy, however, nearly exactly the reverse happens: Act gets 1 MP for its 85K party votes and NZ First gets 5 MPs for its 95K party votes. The more things change.... And in both cases the Maori party cheerily gets 5 MPs on 56K party votes. What progress!

If you were inclined to whine before, surely you must either be inclined to continue whining or to rethink your original inclination? Perhaps you were just being a baby about boundaries? More anon.

No comments: