3July
M'learned friend Mr. Pokery has been playing catch-up on some relatively old posts. To That's Overhang!, he comments,
There seem to be very clear advantages in knowing before the election how many members ought to be elected after it [...] and I am not yet convinced – though I think I could still be! – that the possible overrepresentation of one region is a price worth paying for this benefit.
The question might best be posed: is it better to have a parliament (assembly, council...) of absolutely fixed size, or one that does its best to accurately represent the voting patterns expressed by the electorate? Is the important thing to have a roughly equal number of representatives per member, or a roughly equal representation of their views? Expanding the Welsh Assembly from 60 to 64 seats would ensure a more representative democracy, but would give slightly more voice to those regions where Overhang seats arose. Fixing it at 60 ensures (roughly) the same number of electors per member, but reduces the representation for Overhang areas.
[Overhang seats] may be unlikely in any one region
The evidence suggests otherwise. Here's the complete list of Overhang seats in the Welsh Assembly elections:
1999
N Wales – Lab over LD
S Wales C – Lab over LD
S Wales W – Lab over PC, 2 seats
2003
N Wales – Lab over C
S Wales C – Lab over LD
S Wales W – Lab over C
2007
N Wales – Lab over C
S Wales C – Lab over LD
S Wales W – Lab over C, LD (1 seat each)
These Overhang seats arise because Labour does just enough to win the first-past-the-post seats – for instance, its '99 and '03 victories in Clwyd West were with 31% and 34.8% of the vote, both times less than 3% ahead of the next candidate. By not picking up a vast majority of list votes, the party gets no top-up seats, but there are very few top-ups in the first place. Slim majorities in many seats translate into a large majority in the final assembly.
The effect of these very naive Overhang rules is for Labour to consistently face three or four opposition members, and to consistently be over-represented in these three regions. In the first two elections, the German model would have denied Labour any reasonable chance of reaching an overall majority without a coalition, as happened this year.
There is a further question here, one more philosophical than psephological: is the democratic will expressed through the elections in forty individual constituencies, or through the elections in five regions, or through the election in the nation as a whole? The current model in Wales favours the constituency, as there are typically eight constituency members to four from each region. In Scotland, the model perhaps favours the region, with roughly nine constituencies adding seven regional members.
One reasonably simple way round this problem (and one that Snowy could explain at four in the morning ... not sure that Jez Vine would understand it at four in the afternoon) would be to replace some of the regional top-ups with national top-ups. Work out the constituency members as at present, and work out the regional top-up members, but for a slightly smaller number of top-ups – in Scotland, perhaps award six seats per region rather than seven; in Wales, three rather than four. Then add all the regional top-up votes into a national figure, and award the handful of national top-ups in the standard way. This proposal would eliminate Overhang seats from most elections, including all three Scottish Parliament elections. Where it does not – and all three elections to the Welsh Assembly leave one unresolved Overhang seat – it could be that the balance between constituencies and top-ups is incorrect in the first place.
