With the Scottish Parliament election and the AV referendum (yawn) on the same day this year, there’s been a bit of chat on Twitter regarding whether the constituency element of the Scottish Parliament election should also changeover to AV in the unlikely event that people vote Yes.  But I think there’s a better way to bring some more proportionality to the system as it currently stands, and that is to make the list component a Scotland-wide list, rather than dividing it into 8 regions.

This has several advantages – we’d be looking at consistent levels of requirement to be elected across Scotland.  In 2007, for example, 10,749 votes got Patrick Harvie elected 7th on Glasgow list, and 10,663 got the last SNP MSP (Dave Thompson) elected in H&I.  However, Lord George Foulkes needed 15,099 to be elected as the last MSP for Labour in Lothians and Stuart McMillan 15,191 to be the last MSP in West.  If we had national lists, in 2007,  you would have needed 14,700 votes to be the last MSP elected, and you this could compensate for big votes in some regions and smaller votes in others.  In other words – every vote would actually count.

Secondly, and probably more importantly, the result is much more proportional.  I’ve been playing with the numbers for a while (and I can email folk my working if anyone is remotely interested!), but based on the 2007 result (on the boundaries at the time) and assuming the constituency vote stayed the same, this is what we’d be looking at:

Now there are several points which are worth exploring here, and I’ll get to some of them now.

First, you’ll notice this would elect a BNP MSP.  That’d be a downside its true – but if people will vote for them, then they will win seats.  They were particularly close to getting an AM in the Welsh Assembly in North Wales in 2007 – so that’s something to watch out for.  But electing 1 extremist is not a reason not to consider this (and there are ways you could minimise this risk, should you want to, which I’ll come to in a moment).

Secondly, you’ll also probably notice much more fragmentation of the party system, certainly compared to what we have at the moment.  That’s simply because parties in most regions accumulate anything from 1,000 to 6,000 votes in any given region, but haven’t come close to the 10,000+ required to win a seat .  But add them all together in this kind of system, and suddenly it is enough.

There is a way around this fragmentation – and its a threshold.  In Germany, they use a similar system to here (though their split is 50-50 between constituency and regional members, and its a national list they elect from) and they put a threshold at 5% of the vote – don’t make that, and you get no members elected.  It is designed to stop extremists (especially since German political parties are state funded).  If we did that with this system, the threshold may have to be lower, since only the “Big 4” would make 5%.  I’m instinctively against thresholds – again, they reduce proportionality and, in my mind, are anti-democratic since they ignore some voter’s stated preferences – but I understand the arguments for them.  Not so much the ant-extremist angle, but the controlling fragmentation (and thus allowing efficient government) I get.

I mentioned before that this would be a more proportional system.  And it would be – here’s the numbers:

For all except the Greens (who didn’t – with one exception – stand constituency candidates) I’ve averaged the vote on both constituency and regional vote to give a reflection of party support as a whole.  And you can see how close the correlation is:  The SNP and Labour seat shares are half a percent higher than they “should” be, the Conservatives’ are half a percent lower, and the Greens (and the remaining “Others”) win the seats their vote share suggests they should. Only the Lib Dems are out, by 2% – a direct result of their winning more constituencies than their national share of the vote would dictate.  They get “punished” for it on the top-up element (which happens with the regional system too).

I’ve run the figures for 1999 and 2003 as well – they appear in the table below – and the results are consistent with what I found running the 2007 vote:

You can see that in both 1999 and 2003, if the lists had been national instead of regional, then there would have been slight differences in the actual outcome.  Starting with 1999 – Labour would have returned no list MSPs (meaning 3 fewer seats in total) while the SNP and Conservatives end up with 1 and 2 more list MSPs respectively.  In similar circumstance to 2007, the Lib Dems lose out a bit because they win an over-representation in FPTP seats while the Greens and the SSP would have gained more than they actually did.  In 2003, again Labour don’t return any list MSPs (and, again, like the Lib Dems, are over-represented because of their constituency wins) and the Greens/SSP add to their actual figures while the SNP are the same.

What is interesting to note is the comparison between vote share and seat share – and the difference that AMS makes when the national vote share is the deciding factor (rather than regions).  In both 1999 and 2003, Labour’s share of seats is still much higher than its share of the vote – this is because each of their seats was won through FPTP and not AMS.  In 2003, the SNP’s seat share is down on their vote share – but that’s because there were not enough regional seats to make up for their poor showing on the constituencies.  And the Lib Dems are constantly over-represented due to their winning more FPTP seats than their vote share would allow.  But look at the rest of the vote shares compared to share of seats (the 1999 and 2003 figures in the table directly above, 2007 figures in the one above that).  They are almost exactly correlated.

My point is simply this:  If we are considering a “small step” towards making the system more proportional, let’s forget about AV and simply make the AMS element of the Holyrood system a national – and not regional – list.  Sure this would make governing coalitions more difficult (no Lab-LD coalition makes 65 in 2003, for example) and fragments the party system further, but it IS more proportional.  If that is our priority, then surely it’s something we should be considering.