Background on Federal By-election Swings

Federal politics will soon kick into gear for 2024 with campaigning for the Dunkley by-election, likely to be held in late February or early March.

The by-election has been caused by the sad death of former Labor MP Peta Murphy, who succumbed to breast cancer at the end of 2023. It will be the third by-election since the election of the Albanese government in May 2022.

You can find more on the seat of Dunkley and the by-election in my seat profile on the ABC Elections site.

The by-election will be a test for Prime Minister Anthony Albanese and the Labor Party, keen to retain what is a marginal seat despite its on-paper electoral buffer of 6.3%.

It will also be a challenge for the Liberal Party and Opposition Leader Peter Dutton. After the Liberal Party’s historic loss at last April’s Aston by-election, the opposition needs a good result in Dunkley to confirm recent improvement in opinion polls.

But there are arguments for and against whether Dunkley will be a good test of the national electoral mood.

Being fought in Victoria, currently Labor’s strongest state, can Dunkley be viewed as representative of the national electorate? The 2022 federal election reduced the Liberal Party to just eight of 39 Victorian seats, since cut to seven seats by the Aston loss. Only two of those seats, Deakin and Menzies, are entirely suburban.

The Liberal Party has also performed badly in Victorian state politics, losing six of the last seven state elections. There was a swing to the Coalition at the November 2022 state election but the Liberal Party lost seats and has since been dealing with internal party recriminations.

Arguing for the by-election’s importance, Dunkley is the sort of outer-suburban seat the Liberal Party needs to start winning if it hopes to overcome the loss of once blue ribbon Liberal seats to Independents.

Dunkley includes some newer housing estates where interest rate rises have bitten. Across the electorate there are families who are feeling the effects of inflation.

Based on national opinion polls, there is not enough movement to predict the Liberal Party will win Dunkley.

But by-elections are an opportunity for voters to send a message when the government’s fate is not in play. Will the anti-government swing common at by-elections be large enough to deliver victory to the Liberal Party?

The swing needed is 6.3%, and Labor achieved a 6.4% swing the other way to win Aston. Covering the full period since Federation, the average anti-government swing is a little under 4%.

More recently there have been 52 by-elections since the election of the Hawke government in 1983. Of those, 28 were traditional two-party contests between Labor and the Coalition, the type of contest we will see in Dunkley.

Across the 28 two-party by-elections, the average anti-government two-party preferred swing was 3.5%. It was 4.7% against Labor governments in 17 contests, and 2.3% against Coalition governments in another 11.

Of the 28 by-elections, 15 were in government held seats and 13 in seats held by an opposition party. The average swing against government in government held seats was 5.4% compared to only 1.2% in Opposition held seats.

At the eight by-elections in Labor seats during Labor governments, the average swing was 8.2% compared to 2.3% in seven similar contests during Coalition governments.

The Labor Party had an astonishingly good result in Aston, in contrast to poor first term by-election results for Labor governments led by Bob Hawke and Kevin Rudd. Under both Hawke and Rudd, Labor was further ahead in polls than the Albanese government at the time of its Aston victory.

Not that average swings are a useful measure given the wide variety of swings at by-elections. Swings are more about the time specific circumstances of a by-election and are not always comparable with an average calculated over several decades.

Larger swings than required in Dunkley afflicted the Hawke government at third term by-elections in 1988. The Adelaide and Port Adelaide by-elections, fought on the now forgotten issue of timed local calls for home phones, cost the government Adelaide and produced a double digit swing in Port Adelaide. Another double digit swing struck at the Oxley by-election later the same year. Labor’s position in all three seats was restored at the 1990 election when the Hawke government was narrowly re-elected.

Not so with the 16.2% swing that delivered a rare Liberal win in the ACT at the 1995 Canberra by-election. Labor recovered Canberra at the 1996 election, but it was a pallid highlight amidst the wreckage of the Keating government’s defeat.

Going further back in time to June 1975, the famous Bass by-election produced a Liberal gain after a 14.3% swing, accurately predicting the Labor Party’s fate under Gough Whitlam later in the year.

The biggest anti-government swing under a Coalition government was in the Brisbane seat of Ryan in March 2001. Labor won the seat after a 9.7% swing. John Howard famously described the result as not a repeat of Bass and Canberra, and the Howard government recovered Ryan and was re-elected to office at the 2001 election.

So will the by-election produce an average anti-government swing and see Labor retain Dunkley, or will we see larger swing that delivers victory to the Liberal Party?

Either way, the Dunkley by-election will set the frame for politics in the first half of 2024.

For more on the Dunkley by-election, see my profile of the electorate and candidates at the ABC website.

And for more on by-elections and results, read on in this post.

By-Elections since 1901

The following summary draws on the publication “House of Representatives by-elections: 1901-2018” by Stepehen Barber and published by the Commonwealth Parliamentary Library as a 2018-19 Research paper. Where noted I have updated some of the figures.

  • Dunkley will be the 163rd by-election since federation, an average of 3.5 by-elections per parliament. (Updated)
  • the average number of nominations has grown over the years from 2.2 per by-election in the first two decades of the twentieth century to more than 11 since 2000. (See graph below)
  • There is an increasing tendency for governments to avoid contesting by-elections in their opponents’ safe seats.
  • In only ten cases has an opposition party failed to contest a by-election.
  • 87 of the by-elections followed a resignation and 69 by the death of a member. There have also been six voided elections, and one MP was expelled from the House (Updated)
  • Between 1901 and 1979, 61.5% of by-elections were caused by death compared to 33.7% for resignation. Since 1980, 88.1% of by-elections have been caused by resignation. (Updated)

Source: 1901-2019 data from Barber paper listed above, 2020-24 by author.

Looking at party contests and again drawing on the Barber paper with appropriate update -

  • Only 37 of 162 by-elections (22.8%) saw a seat change party status.
  • Of those 37, 28 (17.3%) were lost by a major party to another major party.
  • four (2.5%) were lost by a major party to a minor party (Corangamite 1918, Echuca 1919, East Sydney 1932 and Cunningham 2002)
  • three (1.9%) were lost by a major party to an independent (Wills 1992, Lyne 2008 and Wentworth 2018)
  • two (1.3%) previously held by independents were won by a major party (Franklin 1929 and Henty 1946)

Looking at the numbers on Government versus Opposition contests across 162 by-elections -

  • 25 by-elections (15.4%) saw a seat lost by the government of the day
  • four seats (2.5%) were lost by the opposition to a crossbench party or independent (Riverina 1904, Maranoa 1921, Cunningham 2002, Lyne 2008)
  • two seats (1.2%) were lost by the opposition to the government (Kalgoorlie 1920, Aston 2023)
  • four seats (2.5%) were lost by one Coalition party to another (Wide Bay 1928, Darling Downs 1936, Calare 1960 and Groom 1988)
  • two seats (1.2%) were won from Independents, one by the government of the day (Franklin 1929) and one by the opposition (Henty 1946)

By-elections since 1983

The calculations that follow are by me.

  • The Dunkley by-election will be the 53rd since the election of the Hawke government in 1983.
  • Of the 52 by-elections held, 28 finished as two-party contests between Labor and the Coalition, three saw one of the major party candidates excluded in the distribution of preferences, and 20 were not contested by the government of the day with North Sydney 2015 the only contest without an opposition party candidate.
  • Labor under Hawke and Keating contest 16 of the 22 by-elections between 1983 and 1996.
  • The Coalition under John Howard contested only three of nine by-elections between 1996 and 2007.
  • Labor under Rudd and Gillard contested only one of five by-elections between 2007 and 2013.
  • Coalition governments between 2013 and 2022 contested 8 of 14 by-elections.
  • Dunkley will be the third by-election in the current term contested by both Labor and the Coalition.

Looking at the 28 two-party by-elections between 1983 and 2023, the average swings were as follows:

  • The average swing against the government after preferences for all 28 by-elections was 3.5 percentage points.
  • The average swing was 4.7% against Labor governments and 1.6% against Coalition governments.
  • Of the 28 by-elections, 15 were in seats held by the government and the swing was an average 5.4% compared to 1.2% in seats held by the Opposition.
  • At the 8 by-elections in Labor seats during Labor governments, the average swing was 8.2% compared to 2.3% in Coalition seats under Coalition governments..
  • There were only seven by-elections that recorded a swing to the government.

Table of By-elections - 1983-2023

The table below includes only the 28 by-elections that finished as two party preferred contests. The 24 non-traditional by-elections are not included.

Explaining the table's columns , 'Govt' shows the party in office, 'Term' shows which term in the party's period in office, 'Held By' shows the party that won the seat at the preceding election, and 'Swing' shows two-party preferred swing towards (+) or away from (-) the government. So a (+) when Labor is in office is a swing to Labor, but to the Coalition under a Coalition in government.

Electorate names marked in bold changed party status at the by-election. See notes on all seats that changed party below table.

Federal By-elections - 1983-2023
Electorate Date PM Term Held By Margin Swing Govt
Wannon 7 May 1983 Hawke First LIB 9.7 -1.1 ALP
Bruce 28 May 1983 Hawke First LIB 0.7 -3.8 ALP
Moreton 5 Nov 1983 Hawke First LIB 1.6 -1.3 ALP
Corangamite 18 Feb 1984 Hawke First LIB 8.2 -1.2 ALP
Hughes 18 Feb 1984 Hawke First ALP 18.3 -5.3 ALP
Richmond 18 Feb 1984 Hawke First NAT 5.9 +0.1 ALP
Scullin 8 Feb 1986 Hawke Second ALP 27.6 -4.5 ALP
Adelaide 6 Feb 1988 Hawke Third ALP 6.5 -8.4 ALP
Port Adelaide 26 Mar 1988 Hawke Third ALP 16.3 -11.1 ALP
Oxley 8 Oct 1988 Hawke Third ALP 14.9 -11.8 ALP
Werriwa 28 Jan 1994 Keating Fifth ALP 15.8 -6.3 ALP
Fremantle 12 Mar 1994 Keating Fifth ALP 7.8 +1.0 ALP
Bonython 19 Mar 1994 Keating Fifth ALP 14.7 -7.8 ALP
Canberra 25 Mar 1995 Keating Fifth ALP 9.6 -16.2 ALP
Lindsay 19 Oct 1996 Howard First LIB 1.6 +5.0 L/NP
Ryan 17 Mar 2001 Howard Third LIB 9.5 -9.7 L/NP
Aston 14 Jul 2001 Howard Third LIB 4.2 -3.7 L/NP
Gippsland 28 Jun 2008 Rudd First NAT 5.9 -6.1 ALP
Griffith 8 Feb 2014 Abbott First ALP 3.0 +1.3 L/NP
Canning 19 Sep 2015 Turnbull First LIB 11.8 -6.5 L/NP
New England 2 Nov 2017 Turnbull Second NAT 16.4 +7.2 L/NP
Bennelong 16 Dec 2017 Turnbull Second LIB 9.7 -4.8 L/NP
Braddon 28 Jul 2018 Turnbull Second ALP 2.2 -0.1 L/NP
Longman 28 Jul 2018 Turnbull Second ALP 0.8 -3.7 L/NP
Eden-Monaro 4 Jul 2020 Morrison Third ALP 0.8 +0.5 L/NP
Groom 28 Nov 2020 Morrison Third LNP 20.5 -3.3 L/NP
Aston 1 Apr 2023 Albanese First LIB 2.8 +6.4 ALP
Fadden 15 Jul 2023 Albanese First LNP 10.6 -2.7 ALP

By-elections Since 1975 that produced a Change in Party Status

  • Bass (1975) - The Whitlam government appointed former Deputy PM Lance Barnard as Ambassador to Sweden triggering a by-election. The by-election held in the depths of winter and the Liberal Party's Kevin Newman swept to victory on a swing of 14.2%, accurately predicting the Labor Party's fate at the December 1975 election. (Not listed in above table)
  • Lowe (1982) - The resignation of former Liberal Prime Minister Sir William McMahon caused a March 1982 by-election that was won by Labor's Michael Maher. Maher was re-elected at the March 1983 election when Labor led by Bob Hawke defeated the Fraser government. (Not listed in above table)
  • Adelaide (1988) - the appointment of Labor's Chris Hurford as Australian Consul-General in New York triggered a by-election in February 1988 for his relatively safe seat of Adelaide. The Labor candidate was Don Farrell, these days a Labor Senator and Minister for Trade in the Albanese government. A swing of 8.4% delivered victory to Liberal Michael Pratt. Pratt was defeated and Adelaide reverted to the Labor Party at the 1990 election. A key issue of the campaign was a political debate on whether timed local calls should be introduced on home phones.
  • Groom (1988) - this by-election was held in April 1988 when Coalition relations still frosty after the Queensland National Party split the federal Coalition ahead of the 1987 election. 'Joh-for-Canberra' National MP Tom McVeigh resigned from Parliament and the by-election result saw the Liberal Party's Bill Taylor take the seat from the Nationals.(Not shown in above table)
  • Wills (1992) After being deposed as Prime Minister by Paul Keating, Bob Hawke resigned from Parliament causing an April 1992 by-election in his safe inner-Melbourne seat of Wills. With recession biting, and light industry in this part of Melbourne suffering from reductions in tariff protection, Labor's vote collapsed but not to the benefit of the Liberal Party. Independent Phill Cleary swept to Victory, winning a third of the first preference vote to lead the poll, easily elected on Liberal and other candidate preferences from a record field of 22 candidate. Labor's vote after preferences was 23.6%, but as an example of why swing is an unreliable measure when the final pairing of candidates changes, there was actually a 5.9% swing towards Labor as measured by the alternative two-party preferred count. Cleary was disqualified as candidate and MP by the High Court later in 1992 and narrowly elected at the 1993 federal election. (Not shown in above table)
  • Canberra (1995) - the resignation of former minister Ros Kelly led to a March 1995 in the southern ACT seat of Canberra. The unpopularity of the Keating government was demonstrated by a 16.1% swing, the Liberal Party's Brendan Smyth elected to serve a year in the House of Representatives. On new boundaries Labor recovered Canberra at the 1996 election but the government was heavily defeated across the country.
  • Ryan (2001) Long-serving MP and former Minister John Moore resigned from Parliament shortly after being dropped from Cabinet in a re-shuffle. It triggered a by-election resented by local voters. The by-election was held in March 2001, a month after the landslide defeat of the Court Coalition government in Western Australia and the landslide re-election of the Beattie Labor government in Queensland. At the time the Howard government was struggling with cost of living issues flowing from the introduction of the GST in mid-2000. Labor's Leonie Short eked out a narrow victory after a 9.7% swing. John Howard stated the result could not be compared with the past Bass and Canberra by-election defeats and went on to win a third federal election in November 2001.
  • Cunningham (2002) The resignation of former Labor MP Stephen Martin caused a by-election held in October 2002. Splits in local Labor ranks after a messy pre-election, combined with the Liberal Party choosing not to nominate a candidate, produced field of 13 candidates with most recommending preferences against Labor and towards the Greens. The seat was won by the Greens' Michael Organ, defeating Labor's Sharon Bird. The defeat led to the end of Simon Crean's Labor leadership. Bird was the Labor candidate at the 2004 Federal election when she defeated Organ.
  • Lyne (2008) The resignation of former Nationals Leader and Deputy Prime Minister Mark Vaile caused a September 2008 by-election. Popular MP for the local seat of Port Macquarie resigned from state Parliament and easily won the by-election. Oakeshott was re-elected at the 2010 election and held one of the crucial balance of power votes in the House of Representatives. (Not shown in above table)
  • Wentworth (2018) After being deposed as prime Minister, Malcolm Turnbull resigned as MP triggering an October 2018 by-election in this traditionally safe Liberal seat. New Liberal candidate Dave Sharma was defeated by Independent Dr Kerryn Phelps, costing the Morrison government its parliamentary majority. Sharma returned to defeat Phelps at the 2019 election before being defeated by a new Independent in Allegra Spender in 2022. (Not shown in above table)
  • Aston (2023) The resignation of former Liberal MP Alan Tudge triggered an April 2023 by-election in his eastern Melbourne seat of Aston. Liberal held since 1990, but on a reduced margin of 2.8% following the 2022 election, few had predicted that the seat would be won by Labor. A swing of 6.4% caused Aston to become the first by-election gain by government from the opposition since Kalgoorlie in 1920.

4 thoughts on “Background on Federal By-election Swings”

  1. I see that Peta Murphy’s primary vote in 2022 out-performed the ALP Senate vote in Dunkley by 8%. Impressive! Does that reflect her personal popularity, or some other Victoria-wide factor which depressed the ALP’s Senate vote.

    COMMENT: Those who know better credit it as personal vote for Murphy.

    1. Although Alan, always remember to be cautious when just using the normal ‘personal vote’ system of taking the FP House of Representatives figure and taking the Senate figure from it. Because in this instance there are some peculiarities changing this figure which narrows it down (somewhat) to a 2-3 % TPP personal vote.

      First there is the distribution of the 7.5% difference between Labor for Dunkley in the House vs the Senate. The most notable lack of gain from this is the Liberals, whose total Senate vote was almost on point to their House vote, so it is not like that ALP difference automatically assists the Libs who it definitely helps is Labor-preferencing minor parties who didn’t contest Dunkley in the House: HEMP, up 3.55%, Derryn Hinch’s Justice, (who did preference Labor overall by 2%), up 1.8%, GRN, 0.96%, Fiona Patten’s Reason, 0.76%. Only SFF broke this, because their Senate ticket got 0.95%. These Labor preferencing parties gaining so much probably means that a decent chunk of the ALP gap between their Senate and House figure in Dunkley were people who, without their preferred party on the ballot in the House, instead chose Peta Murphy (or, to a lesser extent, the Liberal candidate Sharn Coombes).

      Second, the lack of this 7.5% going to the Libs means that, in the all important TPP terms, the swing caused by losing Peta Murphy, as derived from the Senate would be a sizable, yet manageable amount. In fact, without taking account of the last paragraph, the direct swing, using the differences between the Senate and House, would be about 3.75% to the Libs, and when taking into account that of Labor’s losses moved to Labor preferencing parties. This would likely be downgraded into the 2-3 % range I gave at the top of my reply.

      Basically, the effect to the ALP’s all important TPP vote, as estimated by Senate figures, would likely be within a 2-3% range. Nevertheless, it is still nothing to be scoffed at, and could give the ALP a headache.

  2. Hi Antony,

    The federal byelections table is in a weird order, going Howard, Abbott, Morrison, Rudd, Turnbull rather than Howard, Rudd, Abbott, Turnbull, Morrison.

    The 2015 Canning by-election appears quite innocuously in your table, but may have contributed to Tony Abbott’s defeat by Malcolm Turnbull, as the leadership spill was just a few days before the by-election and may have been driven by the prospects of losing the seat (obviously just speculation, but I like to think WA can have some influence on things occasionally).

    I suspect there are far too many factors and far too few byelections to get any meaningful data, but I’ve always wondered whether the cause of a byelection has any impact on results, for example leaving due to scandal or for a corporate job vs death or serious health issues

    COMMENT: I had two separate tables initially and muddled the html lines while merging. Now fixed.

  3. Does the anti-government swing at by-elections tend to be smaller when it is caused by death (or resignation due to ill-health) rather than a member just no longer wanting to be an MP?

    COMMENT: By-elections caused by deaths are rare in recent decades. There have been three since 1983. The results were
    Isaacs (2000), Labor held, no Liberal candidate, swing irrelevant
    Aston (2001), swing against Howard government 3.7%
    Canning (2015), swing against Abbott/Turnbull government 6.5%

    The by-elections paper mentioned in the post does provide a calculation based on cause. The anti-government swing in by-elections caused by death averaged 2.6% compared to 3.8% for all by-elections since Federation. But as mentioned, by-elections caused by death have become rare in recent decades so the 2.6% figures is based on old data from a time when there were many more members of advanced age. Billy Hughes was 90 when he died and triggered the Bradfield by-election.

Leave a Reply