Saturday, May 7, 2022

Beautiful Losers: Australia's Leading Anti-Bellwether Seats

After the 2018 FIDE (chess) elections, a member of the English delegation told me that his nation had kept alive their proud tradition of always supporting the losing side.  Having voted for various opponents of the long-term FIDE President Kirsan Ilyumzhinov, England had switched to Ilyumzhinov's replacement Georgios Makropoulos just in time for the incumbent regime to be booted.  Perhaps this comment was distantly in my mind a couple of weeks ago when someone tweeted a comment referring to Bass as a bellwether seat.  Bass, I thought, is not a bellwether seat at all; it's more like a seat that hits the incumbent over the head with a bell no matter what is going on nationwide.  Then I thought, how about anti-bellwether seats?  Seats that have a history of nearly always voting for the party that loses?  I hope to have the time and energy for more central things in the coming week but for now I present some findings on this little amusement.

Bellwether or litmus seats (where there is a long history of whoever wins always or nearly always forming government) are much discussed, the idea being that since they have a history of being predictive they might contain some special insight into where a campaign is going.  Past champions have included Macarthur (voted for the winning party 23 times in a row 1949-2004) and Eden-Monaro (15 in a row 1972-2013).  The current leader is Robertson (14 in a row, 1983-present).   In theory, a bellwether is a perennial marginal seat that is a good microcosm of the nation, though in practice bellwether streaks involve a great amount of luck.   It might be possible to find a better predictive measure of 'as goes seat X, so goes the nation' than winning streak - some combination of average 2PP lean and correlation with the national swing, for example.

What makes a good anti-bellwether?  Going back far enough, historic accident has a lot to do with it (a seat redistributed favourably for one side just before that side started a run of losing is a very common scenario).  But a seat might also qualify if it is marginal enough to go backwards and forwards fairly readily, but also has some tendency to vote in a contrary fashion (for instance, since governments usually win, by often voting against the government of the day.)  It is interesting to note some seats on the list below that have been mentioned in dispatches as possible flips to the Coalition even if Labor wins, for instance McEwen, Greenway and Paterson (though Paterson speculation has gone off the boil lately).  

When looking at bellwethers, a common standard is the longest winning streaks.  But in terms of anti-bellwethers, these streaks don't go back very far, because it's much harder to keep swinging against the nation when there are landslide results and hence stay on the losing side.  Labor didn't make any gains from the Coalition in 2013, so there is no seat that has been won by the losing side at the last four elections in a row.  Starting from 2019 and working backwards, I look at the seats with the nearest things to a current perfect losing streak.  I should note that seats are not allowed to cheat by repeatedly electing independents or minor parties; the streak is defined by reference to the 2PP.  Scores are expressed by number of times the seat has gone with the government of the day, out of the number of elections.  The lower the better.  Also, by-elections are ignored.

0/3, 1/4 Many Seats

Every seat won by Labor in the last three elections (over 50, the exact number depending on how one counts various seat renamings with minimal redistribution) is on a streak of three consecutive losses.   The 1/4 group includes all these seats, provided they existed in 2010.  However it also includes some seats that voted Liberal in 2010 and 2013 but Labor in 2016 and 2019.  These are Cowan, Macarthur and Paterson (the last two of which were heavily redistributed prior to 2016).

1/5 McEwen, Greenway, Cowan, Macarthur, Paterson

Of the seats won by Labor in 2013, 2016 and 2019, two had not been won by Labor in 2007.  These were Greenway (which was then redistributed for 2010 prompting Louise Markus to move to Macquarie, and retained by Labor in 2013 because of this) and McEwen (which was retained by Liberal Fran Bailey in 2007 by a margin of 31 votes after a court battle). Those two therefore have 1/5 including the 2013-9 hat trick, and are joined by Cowan, Macarthur and Paterson, as discussed above.  

1/6, 1/7, 1/8 Cowan

Every seat that has existed that long has been won at least twice in the last six elections by the party forming government with the exception of Cowan, which has been won just once by the party forming government (2013) in the last eight.  Cowan may yet blot its copybook in 2022 but its form when it comes to flipping to the losing side in the last ten elections is very impressive:

1993: Cowan is one of five seats to flip to the Coalition Opposition in an election where Labor gained seats
1998: Cowan flips to the Labor Opposition in an election where Labor gains seats
2007: Sitting member Graham Edwards retires and Cowan is one of two to flip to the Liberals just as the Liberals lose office.
2016: Cowan again flips to the Labor Opposition after just one term of voting with the government

It might be thought that the 1998/2016 combination would be common, but it is rarer than may be expected.  Bass, Braddon and Paterson also flipped to Labor at both these elections.

2/9 Many seats

A painful point for the red team, this one: they've only won two of the last nine elections so any seat that has voted Labor all that time is the equal most common loser in that time.  Depending on how one handles renamings, there are about 37 seats batting 2 out of 9, and all but two of those have a history of voting Labor every time, hence going with the government in 2007 and 2010.  The two exceptions are our friend above Cowan (which voted with the government in 1996 and 2013 but was a Liberal seat in 2007-10) and Greenway (which voted with the government in 2004 and 2010, having been gained from Labor in 2004 and held for two terms.)  A few seats that are batting 2/8 missed the 2/9 club because the Liberals won the 2PP in 1996 but not 2013; these include Bendigo, Lilley, Griffith and (with a disendorsed candidate) Oxley.  

2/10 Cowan, Bruce, Isaacs

If a seat was Labor through 1996-2019 and even held by Labor during the 1996 and 2013 routs it's not that likely Labor would have lost it in 1993.  As a result, nearly all the 2/9 club become 3/10 at this point, with the exception of Cowan (which has the odd history discussed above) and also Isaacs and Bruce.  Isaacs was redistributed in the Liberals' favour prior to the 1990 election and they gained it with a large notional swing, but was then redistributed in Labor's favour prior to 1996 and gained by Labor against the tide.  Bruce, one of many divisions that has kept its name but moved a lot geographically, was Liberal all through the Hawke/Keating years but then also redistributed in Labor's favour prior to 1996, making it another gain against the tide (that said Labor did well to hold the swing low enough in Bruce in 1996 to win it).  

2/11 Isaacs, Bruce

See above.

2/12, 2/13, 2/14 Bruce

If a voter is 58 years old and has lived in Bruce continually, they have voted at just two elections at which their local MP was in government, 2007 and 2010.  Bruce voted for the loser nine times in a row between 1983 and 2004.  

Honourable Mention 3/14 Franklin

Franklin doesn't set any streak records but unlike Bruce and Isaacs (which had the benefit of redistribution and have therefore cheated) its anti-bellwether form in recent decades has been on merit.  Liberal Bruce Goodluck was elected in 1975 and retained the seat in 1983 off the back of the Franklin Dam boost for the Liberals in Tasmania at that election.  He then held it, largely on personal appeal, through to his retirement in 1993.  There was a disastrous Liberal preselection and Labor won the seat.  It has remained Labor ever since, and usually hasn't been close, meaning that since the start of the Hawke government it has only voted for the winner in 1993, 2007 and 2010.  Lyons (4/14) is another Tasmanian low-scorer, having voted for the winner in 2007, 2010 and 2013.

Hall of fame: Bendigo

Obviously Bruce's streak of nine Opposition wins is not an outright record, since any seat that voted Labor through the nine consecutive Coalition wins 1949-1969 ties it.  But the seat of Bendigo stands out here because it voted for the loser in 1943 and 1946, flipped to Labor with a redistribution as Labor lost in 1949, and flipped to Liberal as Labor won in 1972 - giving Bendigo 13 Opposition wins in a row 1943-1974.  I haven't verified that this is the longest anti-bellwether streak of all time but it seems hard to top.  After voting for the winner seven times in a row from 1975-1990 Bendigo has relapsed since, and is currently batting 3/11 (alongside Franklin and Cowan), with government wins in 1996, 2007, 2010.  

I will update this article sometime after the election with the new standings.  And if anyone feels inspired to go back further, please post any results in comments.  Corrections welcome.  

1 comment:

  1. An interesting follow up would be the seats that defy the trend at particular elections, based on a ratio of seats going in the other direction. Chisholm 2016 being the only L/NP gain from Labor in an election where Labor gained 16 seats from the L/NP is quite a stand out, as are Cowan/Swan in 2007.

    It would also be interesting to see state examples, which seem rarer (with states evidently being more uniform). Ripon in Victoria 2014 was a Liberal gain from Labor in an election where Labor returned to government, but that was after a redistribution that made the seat notionally Liberal. Liberals gained Miranda in NSW 2015, an election that was otherwise dominated by Labor gains from L/NP, but the seat only became Labor held in a byelection, and that ALP incumbent retired.